III / PROTEINS / Capillary Electrophoresis 4009
PROTEINS
Capillary Electrophoresis
S. P. Radko, National Institutes of Health, Bethesda,
MD, USA
Copyright ^ 2000 Academic Press
Electrophoresis, mostly in the slab gel format, and
high performance liquid chromatography (HPLC)
are the two techniques commonly employed for pro-
tein separation during the past decades. After emerg-
ing in the early 1980s, capillary electrophoresis
(CE) has been recognized as a new tool for protein
analysis and characterization: it combines a number
of aspects of both electrophoresis and HPLC. Based
on differences in charge-to-size ratio or isoelectric
point (pI) of protein macroions, the separations
in CE are inherently electrophoretic. However,
online detection producing quantitative information
in the form of peak area or height, single sample
analysis in a serial fashion, and the possibility
of performing separation in the presence of Sow
(generated by electroosmotic current) are features
of CE similar to those of HPLC. These features lend
CE to easy automation, in contrast to the labour-
intensive methods of conventional (gel) electrophor-
esis. Since heat dissipation by convection is effectively
suppressed in capillaries of less than 0.2 mm i.d., the
separation can be performed in free solutions without
a gel. The high heat dissipation properties of
thin fused silica capillaries also enable one to apply
Reld strengths of several hundred volts per
centimetre, thus greatly reducing the time of
separation.
To date, CE techniques analogous to a number
of conventional electrophoretic methods such as zone
electrophoresis, isoelectric focusing and sieving
(size-dependent) separations have been developed
and numerous applications of CE to separation and
characterization of proteins have been demonstrated.
In the particular mode of CE as capillary zone elec-
trophoresis, short separation times (often a few min-
utes) combined with relatively large diffusion
constants of proteins were expected to provide separ-
ation efRciences exceeding a million theoretical
plates. However, such potential efRciency has
never been practically achieved, and this mostly ap-
pears to be due to the interactions of proteins with the
capillary walls.
The approaches to reducing protein interactions
with the capillary walls and the modes of CE applied
to protein analysis are brieSy considered below.
Approaches to Reducing
Protein^Silica Surface Interactions
The high surface activity of fused silica at neutral pH,
combined with the high surface-to-volume ratio of
thin capillaries is a major problem in applying CE to
protein separation. In general, the protein}silica sur-
face interactions give rise to peak broadening and
asymmetry, compromising the resolution. In bare
fused silica capillaries, these interactions often result
in uncontrolled alterations of the electroosmotic Sow
(EOF) and irreproducible migration times, low mass
recovery of proteins or even their irreversible adsorp-
tion with loss of sample. Over the last 10 years, great
efforts have been made to develop conditions for
protein analysis under which the protein}capillary
wall interactions are minimized and the EOF is either
suppressed or stabilized.
In the pH range of 3}10, the charge density of the
capillary inner surface is known to increase progress-
ively due to the ionization of weakly acidic silanol
groups. The charge density on the wall is near zero at
pH(3 (silanol groups become fully protonated) and
saturated above pH 10 (silanol groups are fully disso-
ciated). Thus, protein species possessing a pI higher
than the pH of the electrophoretic buffer will experi-
ence an electrostatic attraction to the negatively
charged silanols. Beside silanol groups, fused silica
bears a variety of other active sites such as inert
siloxane bridges and hydrogen-bonding sites. These
active sites can join in the protein immobilization on
the inner surface of the capillary, by interacting with
the hydrogen-bonding and hydrophobic regions of
the protein.
The electrostatic attraction between protein mol-
ecules with a net positive charge at a given pH and
ionized silica is believed to play a key role in the
protein}capillary wall interactions. Thus, operating
at extremes of pH (at pH(3 where silanol ioniz-
ation is very low or at pH'11 where proteins carry
a net negative charge) appears to be the simplest
approach to their minimization. Though such an ap-
proach has been demonstrated to be successful in
a number of applications, operation at pH extremes
in general reduces charge diversity, diminishing the
separation selectivity. The pH extremes also tend to
4010 III / PROTEINS / Capillary Electrophoresis
denature proteins and induce formation of multiple
conformers. The electrostatic attraction between pro-
teins and the silica surface may be reduced by increas-
ing the ionic strength of electrolyte solutions
(100 mmol L\1 or greater). However, the high ionic
strength limits the applied voltage, consequently de-
creasing efRciency and increasing the analysis time.
Deactivation of silanol groups may also be achieved
by derivatizing them with organosilanes but the car-
bon moieties of organosilanes make the capillary wall
highly hydrophobic.
Two approaches appear to be the most successful
in rendering CE suitable for routine protein separ-
ations: Rrst, incorporation of appropriate additives
into the electrolyte solution to mask or compete for
either the silanol groups or the basic amino acid
residues of the protein which are exposed to the
solution; and second, use of capillaries with an inner
suface modiRed by an adsorbed or covalently at-
tached polymeric coating. A large variety of chem-
icals and modiRcation procedures currently exist
which can effectively reduce protein}wall interac-
tions and control the EOF so that a separation efR-
ciency of several hundred thousands of theoretical
plates has become practically achievable. Several
types of coated capillaries are commercially available,
and are described below.
The incorporation of buffer additives permitting
successful protein separations in bare fused silica ca-
pillaries has the advantage of simplicity. Ideally, the
buffer additive should not compromise the selectivity
of separation by interacting with the analyte, alter the
buffer pH or increase the operating current, and
should in general exhibit low UV absorbance. Or-
ganic compounds of different kinds have been exten-
sively examined as the buffer additives. Since
mostly they interact with the silica surface in a dy-
namic fashion, the method of modifying the capillary
walls by using buffer additives is known as dynamic
coating.
A large database of organic compunds and buffer
components suitable for improving CE performance
in protein separations has been established. This
database includes zwitterionic salts (methylglycine
and trimethylglycine, trimethylammoniumpropyl and
butyl sulfonates), an extensive number of mono-, di-
and polyamines, surfactants (ionic and zwitterionic
Suorosurfactants as well as nonionic surfactants of
the Brij and Tween series) and neutral polymers (cel-
lulose derivatives, dextran, polyvinyl alcohol, poly-
ethylene glycol). However, the effectiveness of
dynamic coatings is mostly evaluated with standard
mixtures containing a small number of proteins or
with variants of a single protein. Therefore, it is
not possible to predict how a particular additive
will act in conjunction with complex biological sam-
ples. Such samples may consist of a broad spectrum of
proteins ranging widely in their degree of hydropho-
bicity, pI values and molecular weight. Despite this
limitation, the present database of buffer additives
known to improve protein separations may be very
useful in developing methods for a targeted compon-
ent analysis like the purity control of recombinant
proteins, food analysis, or electrophoretic analysis of
haemoglobins, serum or urine proteins.
Though the main mechanism by which the buffer
additives improve protein separation appears to be
their interaction with the silica surface, they can play
an additional role } binding to the protein. In a num-
ber of cases, the additives have been shown to modu-
late selectivity by enhancing differences in
electrophoretic mobility (e.g. some surfactants upon
complexation to proteins, or alkyl diamines and their
derivatives upon binding to protein glycoforms).
The incorporation of diaminealkanes, polyamines
and Suorinated cationic and zwitterionic surfactants
in the electrophoretic buffer effectively controls both
the magnitude and direction of EOF.
Adsorbed coating differs from dynamic coating by
the degree of permanence, but the demarcation line
between them is arbitrary. In the case of an adsorbed
coating, the coating agent should not be present in the
electrophoretic buffer during a run. As a rule, poly-
meric species are used for adsorbed coating. Perma-
nence can result from the high binding afRnity of the
coating agent to the silica surface (polymeric amines,
polyethylene oxide) and may be enhanced by sub-
sequently cross-linking the adsorbed species into a
continuous, permanent Rlm (e.g. polyethyleneimines
treated with diepoxide after adsorption to capillary
walls). The permanence can also result from the abil-
ity of the coating agent to form, upon a particular
treatment, polymeric Rlms physically covering the
silica surface (cellulose acetate and polyvinyl alcohol
Rlms are examples). The polymers may be adsorbed
not directly to the silica surface but to a hydrophobic
layer formed by moieties of a surfactant covalently
attached to the capillary inner walls (hybrid coating).
Hybrid coating appears to be the most Sexible, since
the polymeric layer can easily be removed by rinsing
the capillary with an organic solvent and polymeric
species of other types may be adsorbed, depending on
the separation goal. An example of protein separation
in a capillary with the hybrid coating is presented in
Figure 1.
The covalently attached polymeric coating is usu-
ally carried out by grafting polymers to a silica
surface derivatized with organosilanes. In the sub-
sequent step, polymer chains may be cross-linked to
help stabilize the coating. Several neutral (cellulose

Figure 1 Electropherogram of some acidic and basic proteins
in a capillary with hybrid coating (after derivatizing with the or-
ganosilane, the capillary was coated with epoxybutane-modified
hydroxypropylcellulose). Other conditions: 0.05 mol L\1 NaH2PO4,
pH 3.0; detection at 210 nm, #21 kV, 30 mA. Capillary total length,
85 cm; effective length, 50 cm, 50
m i.d. Peak assignment: 1,
cytochrome c, pI 10.2; 2, lysozyme, pI 11.0; 3,
-lactoglobulin A,
pI 5.1; 4, conalbumin, pI 6.0; 5, haemoglobin, pI 5.6; ribonuclease
A, pI 9.3; 7, -chymotrypsinogen A, pI 9.2; 8, trypsin inhibitor, pI
4.2. Reproduced with permission from Yang C and El Rassi
Z (1998) Capillary zone electrophoresis of proteins with fused-
silica capillaries having polymers and surfactants adsorbed onto
surfactant moieties previously covalently bound to the capillary
column surface. Electrophoresis 19: 2278}84. Copyright: Wiley-
VCM.
derivatives, epoxy polymers, dextran) and cationic
(polyvinylimidazole, polyethyleneimine derivatives)
polymers have been employed for the covalently at-
tached coating. The most popular polymer used for
the polymeric coating is polyacrylamide. This coating
is mostly performed with polyacrylamide chains poly-
merized in situ.
It should be noted that, despite the variety of ma-
terials and procedures used to prepare the coated
capillaries, they do not appear to vary markedly in
separation properties. The diversity of chemistries
underlying the capillary coating is likely to reSect the
continuous search for a magic coating and inad-
equacy of any single approach to provide satisfactory
results for all applications. A particular problem
arises due to difRculties in optimizing the coating
process. The quality of coating appears to depend on
the quality of the fused silica surface, which may vary
between different sources of capillaries and even be-
tween different batches of silica; this requires a corre-
sponding adjustment in capillary pretreatment and
coupling chemistries.
Modes of Capillary Electrophoresis
Applied to Protein Analysis
Capillary zone electrophoresis (CZE), capillary
isoelectric focusing and sieving capillary electrophoresis
III / PROTEINS / Capillary Electrophoresis 4011
in polymeric media are the modes of CE most widely
used for separating and characterizing proteins.
A number of commercial kits for capillary isoelectric
focusing and sieving separations are now available.
Capillary isotachophoresis has in rare cases been em-
ployed for protein analysis. Micellar electrokinetic
chromatography has generally exhibited a poor selec-
tivity in separating proteins. This is probably due to
the inability of relatively large protein molecules to
partition into the detergent micelle.
Capillary Zone Electrophoresis
CZE is the simplest of the CE modes and straightfor-
ward to perform (Figure 1 depicts a typical example
of a CZE separation). When employing CZE for
protein separation, the choice of capillary (uncoated
or with the particular type of coating) and buffer
additives should be made carefully depending on
sample composition. The uncoated capillaries gener-
ally require a prior conditioning step. Detection based
on either UV adsorption or laser-induced Suorescence
(LIF) is most often employed in the CZE of proteins.
Depending on the detection mode, a sample pretreat-
ment may be necessary.
The sensitivity of the detection by UV absorbance
is limited since both the optical length ("capillary
internal diameter) and sample volumes (typically,
a few nanolitres) are very small in CZE. Though the
sensitivity can be greatly increased by detecting pro-
teins in the wavelength range of 200}220 nm, UV
detection still requires a relatively high concentration
of analyte in a sample. That is not always the case and
a preconcentration of the sample, often of a volume
of a few microlitres, is needed. Several online and
ofSine preconcentration techniques can be employed
in CZE. The Rrst and simplest approach to online
sample preconcentration is zone sharpening by stack-
ing. Proteins dissolved in a buffer with a conductivity
lower than that of the run buffer (commonly, the
diluted run buffer) become concentrated at the inter-
face between the sample and the run buffer due to
a high voltage drop in the sample zone. Preliminary
sample desalting is often necessary for this approach
and special methods have been developed for desalt-
ing (and concentrating) microlitre volumes of protein
samples, using small pore polyacrylamide gels.
Isotachophoresis is the other popular technique to
concentrate samples. The preconcentration may be
performed either online or in a coupled column, and
in the presence of salts. The gain in detection limit is
10- to 100-fold and can be increased up to 1000-fold
when a hydrodynamic counterSow is employed.
Another efRcient method of protein preconcentrat-
ion is selective accumulation of the proteins on
4012 III / PROTEINS / Capillary Electrophoresis
a solid-phase afRnity support. This method has been
used in both online and ofSine modes, with several
hundred-fold concentration.
After derivatization with a Suorophore, proteins
may be detected online by LIF. A number of Suor-
escent dyes capable of covalently binding to protein
molecules (e.g. Suorescein, naphthalenedicarboxal-
dehyde and Suorescamine) have been used, providing
mass detection limits in the attomole range (initial
sample concentrations of 10\8 to 10\10 mol L\1).
However, covalent binding of the dyes frequently
results in a broadening of protein peaks or even in the
formation of multiple peaks due to multiple derivatiz-
ation.
Capillary Isoelectric Focusing
Like conventional isoelectric focusing (IEF), capillary
isoelectric focusing (cIEF) is based on differences in
isoelectric points (pIs) of proteins. In cIEF, a stabiliz-
ing gel is not required and, due to the high Reld
strength, the focusing process usually takes only
5}15 min. The cIEF can provide resolution of up to
0.01 pH units, comparable with that of the most
successful applications of conventional IEF.
Sensitivity of detection based on UV absorbance (at
280 nm) is generally satisfactory for cIEF, due to the
concentration of proteins from a relatively large injec-
ted volume into a small volume of the focused zone.
Capillaries with a hydrolytically stable coating effec-
tively preventing protein adsorption and changes in
the EOF are required for successful cIEF separations.
As in conventional IEF, protein precipitation due to
the high protein concentration at the isoelectric point
is a potential problem in cIEF and is addressed in the
same way: using strong solubilizing agents, such as urea
and nonionic detergents, in the ampholyte mixture.
Size-dependent Separation of Proteins by
Capillary Electrophoresis
Although the use of narrow bore capillaries abolishes
the need for gel media to suppress convection, an-
other important feature of gels } their capability to
provide size-dependent separation of macromolecules
} is clearly beneRcial for protein analysis. Efforts to
adopt gels to the capillary format were made in the
early days of CE. However, technical difRculties, such
as bubble formation and the fast deterioration of
polyacrylamide gels during serial runs, limit the use of
gel-Rlled capillaries. These difRculties have been over-
come by using replaceable sieving media such as solu-
tions of entangled polymers. While gels are
polymerized in situ, polymer solutions are usually
prepared by dissolving commercially available
polymers in the run buffer and are pumped into the
capillary before each run. Solutions of dextran,
polyethylene oxide, polyvinyl alcohol and linear
polyacrylamide have been demonstrated to be suit-
able for protein analysis. Though the use of coated
capillaries is generally recommended, uncoated capil-
laries may also be employed if a polymer solution
produces sufRcient viscosity (typically '100 cP).
The main drawback of polymer solutions is that res-
olution is not as high as that obtained with gel-Rlled
capillaries.
Size-dependent separation by CE of protein}so-
dium dodecyl sulfate (SDS) complexes provides in-
formation similar to that obtained from conventional
SDS-polyacrylamide gel electrophoresis (SDS-PAGE),
as illustrated in Figure 2. The limits of UV detection
are comparable to those obtained in SDS-PAGE with
Coomassie blue staining, whereas total analysis time
for multiple samples is even shorter for CE than that
for, e.g. a 16-channel slab gel. Size-based analysis by
CE in sieving media under native conditions has been
demonstrated for proteins. Usually, such analysis em-
ploys constructing a Ferguson plot (the logarithm of
protein mobility vs. polymer (gel) concentration).
Such construction is extremely time-consuming in
traditional PAGE but becomes practical by using CE
in replaceable sieving media. The Ferguson plot-
based analysis may also be useful in estimating the
molecular weight of proteins whose binding with SDS
is anomalous (e.g. membrane proteins, glycoproteins,
highly basic proteins) and that of aggregates and
complexes of proteins.
Figure 2 Capillary electrophoresis sieving separation of a stan-
dard SDS}protein mixture. Sieving matrix, 3% solution of
polyethylene oxide. Inset shows the SDS-PAGE trace of the
same mixture. Numbers above the peaks correspond to protein
molecular weight. Buffer: 100 mmol L\1 Tris-CHES, pH 8.8, 0.1%
SDS. Condition: 300 V cm\1, 203C. Detection at 214 nm. Repro-
duced with permission from Guttman A (1996) Capillary sodium
dodecyl sulfate-gel electrophoresis of proteins. Electrophoresis
17: 1333}41. Copyright: Wiley-VCM.

Protein Characterization by
Capillary Electrophoresis
In addition to the targeted component analysis and
sample proRling, CE has been employed in a number
of speciRc electrophoresis-based approaches aiming
at the characterization of protein}ligand interactions,
protein functional activity and structure.
AfRnity capillary electrophoresis (ACE) has been
applied to the study of protein interactions with
drugs, carbohydrates, nucleic acids and other pro-
teins. In a typical ACE experiment, the receptor is
subjected to electrophoresis in a capillary containing
free ligand at different concentrations. The recep-
tor}ligand binding (Kon) and dissociation (Koff) con-
stants are estimated by Scatchard analysis of shifts in
the receptors mobility in response to ligand concen-
trations. The analysis of peak broadening has been
shown to be useful for estimating Kon and Koff con-
stants, if the characteristic times of receptor}ligand
interactions are comparable with migration time of
the analyte. The receptor}ligand complex equilib-
rium and separation process can be affected by capil-
lary wall effects and/or the use of buffer additives.
These limitations must be addressed when developing
an ACE method.
The potential of CE in analysing antibody}antigen
complexes has been extensively studied over the last
few years in order to develop CE-based immunoas-
says (IA). The CE-based IA offers advantages of high
speed of a single analysis, detection of antigen at trace
concentrations (10\10 mol L\1 if the LIF detection
and Suorescently derivatized antibodies are em-
ployed) and the potential for automation. Despite
some successful examples, CE has generally exhibited
the inability to perform direct IA due to the lack of
separation between bound and free antibodies.
Though this drawback has been overcome in the
competitive CE-based IA, the sensitivity of the com-
petitive IA does not meet the detection levels required
for many important clinical tests.
A CZE-based approach to microassaying enzyme
activity has recently been developed. In this in-tube
approach, the enzyme and substrate are elec-
trophoretically mixed inside the capillary under con-
ditions where their mobilities differ. Another new
application of CZE to protein characterization is
studying protein folding/unfolding transitions. Due
to recent advances in capillary coating and the in-
herent ability to perform electrophoresis in short time
intervals, CZE appears capable of reliably distin-
guishing different forms of protein conformations,
providing a new instrumental approach to the quant-
itative analysis of the conformational equilibrium of
proteins.
III / PROTEINS / Capillary Electrophoresis 4013
Coupling Capillary Electrophoresis to
Other Techniques for Protein Analysis
Several multidimensional separation systems for pro-
tein analysis, incorporating CE, have been proposed
over the last decade. Two-dimensional (2-D) tech-
niques such as CE-CE (using CE separations with two
different carrier systems), HPLC-CZE, size exclusion
chromatography (SEC)-CZE, and even a 3-D tech-
nique combining SEC-HPLC-CZE have been re-
ported. The multidimensional systems possess
tremendous resolving power and may be extremely
useful in analysing complex biological samples, but
there are drawbacks. Separation times are generally
long and can last 2}12 h. Beside the technical difRcul-
ties of interfacing different separation systems, the
compatibility of mobile-phase and run buffers as well
as maintaining the detection sensitivity adequate for
trace analysis in the sequential separations are the
most signiRcant problems.
By coupling CE with mass spectrometry (MS), the
molecular weight of separated proteins can easily be
determined. It is also possible, using MS, to detect
proteolysis and deamidation, Rnd glycosylation vari-
ants, and, with peptide mapping, conRrm the primary
structure of proteins. Though online interfacing of
CE with MS has progressed substantially in recent
years, the successful online combination of CE and
MS is still a challenging instrumental problem. In
addition to the interface design issue, practical con-
siderations concerning the compatibility between the
run buffer and the sheath liquid as well as differences
in their Sow rates at the interface must be addressed
when dealing with CE-MS coupling.
Conclusions
After an initial period of fast technical progress, over
the last decade CE has been increasingly focused on
developing practical methods for protein separation
and characterization. To date, due to advances in
capillary coating, CE may be viewed as a practical
tool for rapid, sensitive and quantitative analysis of
minute amounts of protein samples, with great utility
in targeted component analysis. For many applica-
tions, it can replace HPLC and conventional gel elec-
trophoresis but more often it should be used in
conjunction with other separation techniques, pro-
viding different selectivity or automated analysis.
Since the acceptance of CE in the clinical laboratory
for routine protein-based diagnostics depends con-
siderably on meeting adequate throughput, attempts
to develop high throughput CE systems are likely to
be intensiRed. With a decrease in the real cost and
improvement in sensitivity and resolving power of
4014 III / PROTEINS / Centrifugation
MS detection, the increasing use of a combined CE-
MS technique can be expected.
To be more widely accepted in the area of biomedi-
cal research, CE-based protein separations must dem-
onstrate a number of features that match the success
of conventional (gel) electrophoretic systems. Besides
proRling complex protein samples, these systems
allow for immunological and enzymatic assaying of
separated proteins as well as for simultaneous trans-
fer of sample components into another separation
dimension without altering the separation in the Rrst
one. All of these are achieved with minimal distur-
bance of zone integrity. Thus, the major efforts will
probably be made in developing both multidimen-
sional separation systems involving CE and CE-based
separation systems permitting post- or on-column
enzymatic and immunological analysis of the separ-
ated components of complex biological samples. In-
corporating immobilized enzymes or antibodies into
CE-MS systems will revolutionize the analysis of pro-
tein structure and, especially, glycoprotein analysis.
Further Reading
Camilleri P (ed.) (1998) Capillary Electrophoresis: Theory
and Practice, 2nd edn. Boca Raton: CRC Press.
Centrifugation
A. Yamazaki, Kresge Eye Institute, Wayne State
University, Detroit, MI, USA
Copyright ^ 2000 Academic Press
Introduction
Modern technological developments have made cen-
trifugation one of the most important and widely
applied techniques in experimental research. In bio-
logical studies centrifugation is used for the extrac-
tion and isolation of biological materials and for the
measurement of physical properties of macro-
molecules. Indeed, biological materials have been ex-
tracted and isolated for more than a thousand years
using centrifugal forces. In the 1920s, Svedberg and
other researchers developed motor-driven centrifuges
which had an optical system to observe sedimentation
of macromolecules during centrifugation, and used
these instruments for the measurement of physical
properties of macromolecules, especially proteins.
The molecular mass of most proteins was determined
using these analytical centrifuges until 1970, but they
El Rassi Z (ed.) (1997) Electrophoresis 18: No. 12/13.
Special Issue on Capillary Electrophoresis and Elec-
trochromatography.
Hjerten S (1996) Capillary electrophoretic separation in
open and coated tubes with special reference to proteins.
Methods in Enzymology 270: 296.
Karger BL, Chu YH and Foret F (1995) Capillary
electrophoresis of proteins and nucleic acids. Annual
Review of Biophysics and Biomolecular Structure 24:
579.
Khaledi MG (ed.) (1998) High-performance Capillary
Electrophoresis: Theory, Techniques, and Applications.
New York: Wiley.
Landers JP (ed.) (1997) Handbook of Capillary Elec-
trophoresis, 2nd edn. Boca Raton: CRC Press.
Lunte SM and Radzik DM (eds) (1996) Pharmaceutical and
Biomedical Applications of Capillary Electrophoresis.
Oxford: Pergamon.
Righetti PG (ed.) (1996) Capillary Electrophoresis in Ana-
lytical Biotechnology. Boca Raton: CRC Press.
Righetti PG and Deyl Z (eds) (1997) Journal of Chroma-
tography B 699, Special Volume: Proteins: Advanced
Separation Technologies.
Wehr T, Rodriguez-Diaz R and Zhu M (1999)
Capillary Electrophoresis of Proteins. New York:
Marcel Dekker.
have not been recently used for that purpose because
much easier methods for the measurement of molecu-
lar mass, such as size exclusion chromatography and
sodium dodecyl sulfate (SDS)-gel electrophoresis,
have been developed. More recently, centrifugation
has become an indispensable tool for the isolation of
proteins, nucleic acids and subcellular particles. The
use of centrifuges has also been revived for the
measurement of physical properties of proteins, espe-
cially for the characterization of protein associations
and protein}protein interactions. In this section, im-
portant points of theory and practice for the separ-
ation and isolation of proteins by centrifugation are
summarized.
Theoretical Basis of Centrifugation
Although a rigorous understanding of sedimentation
theory is not required for the separation and isolation
of proteins, a review of some basic principles will be
helpful for understanding the establishment of condi-
tions and the interpretation of experimental results
obtained. Because of their random thermal motion,
 III / PROTEINS / Centrifugation 4015

macromolecular particles in a solution do not show where No is Avogadros number. Thus, a particles
any perceptible sedimentation in a uniform gravi- volume, Vp, may be expressed in terms of its molar
tational Reld. However, these macromolecular par- mass:
ticles do sediment under a centrifugal force. If the
effect of diffusion is neglected, in a solution Vp"m
"
M/No [4]
(density ) the motion of a particle (mass m and
volume Vp) that is located a distance r from the axis where
is the particles partial speciRc volume.
revolving with angular velocity  can be expressed by Eqns [1], [3] and [4] may be combined to give:
the following equation:
M(1!
)2r
vf"m2r!2rVp [1] vf" [5]
No
where v is the velocity of the sedimenting particle, f its
frictional coefRcient; m2r the centrifugal force and When eqns [2] and [5] are combined, s may be ex-
2rVp the buoyant force. This equation may be pressed as:
rearranged to give:
v M(1!
)
v"dr/dt"s2r [2] s" 2 " [6]
r No f
where:
Since the particles partial volume,
, may be ex-
m!Vp pressed by the reciprocal of the buoyant density of the
s" particles, p, as
"1/p, s may also be expressed as:
f

This is the well-known sedimentation equation in v M(1!/p)

s" 2 " [7]
which s is the sedimenting coefRcient and has dimen- r No f
sions of time. For most biological macromolecules,
the magnitude of s is about 10\13 s. Therefore, the Since f"6rp, where  is the viscosity of the liquid
unit of sedimentation, the Svedberg (S), has been medium and rp is the radius of unsolvated spherical
deRned as being equal to 10\13 s. The standard sedi- particle, these equations indicate that the sedimenta-
mentation coefRcient (S20,w) is deRned as that equiva- tion velocity, v, is related to the sedimentation co-
lent to sedimentation in water at 203C. The efRcient s, which is mostly a function of particle size,
sedimentation coefRcients (S20,w) of some proteins are density of the particle p, density of the medium  and
shown in Table 1. the viscosity of the liquid medium, . In other words,
The sedimentation coefRcient s may be trans- for a given particle, sedimentation is directly related
formed to a more practical form. The mass of 1 mole to particle size, particle density and the centrifugal
of particles, M, is Reld, and inversely to the viscosity and density of the
liquid medium.
M"mNo [3]
Centrifugation for Protein Separation
Table 1 Sedimentation coefficients of some proteins
Centrifuges may be classiRed on the basis of the
Protein S20,w maximum speed, namely, low speed, high speed and
ultracentrifuges. Each of these can be used in the
Cytochrome C (bovine heart) 1.71
Egg-white lysozyme 1.9
various steps of protein separation and isolation from
Insulin 1.95 biological materials. Low speed centrifuges are used
Ribonuclease A (bovine pancreases) 2.00 routinely for the initial processing of biological sam-
Myoglobin (horse heart) 2.04 ples. This type of centrifuge can be mainly used to

-Chymotrypsin (bovine pancreases) 2.40 isolate cells and organelles that contain target pro-
Pepsin 2.8
g-Actin 3.7
teins by pelleting these materials. High speed centri-
Lactate dehydrogenase (pig heart) 6.93 fuges, with maximum speeds of 8000}25 000 rpm,
Catalase (horse liver) 11.2 are mainly used for the preparation of subcellular
Glutamate dehydrogenase (bovine liver) 26.6 fractions. These centrifuges can generate about
Fibrinogen (human) 7.63 60 000 g, which is enough to separate proteins from
Haemocyanine (octopus) 58.7
Haemocyanine (snail) 100
debris of cells and organelles. In order to isolate
proteins from other proteins, ultracentrifuges are
4016 III / PROTEINS / Centrifugation
required. Modern ultracentrifuges can generate ap-
proaching 1 000 000 g, which is sufRcient to pellet
even small proteins. Ultracentrifuges can be divided
into two types: analytical and preparative. Analytical
ultracentrifuges have a device by which the sedimenta-
tion rate of molecules can be optically measured dur-
ing centrifugation and can be used to obtain data on
the sedimentation properties of particles. The masses
of most proteins were determined by these ultracen-
trifuges before development of simpler molecular
mass determination methods. Eqn [4] indicates that
the particles mass m"M/No can be determined
from its sedimentation coefRcient s, if its frictional
coefRcient f, is known, as indicated in eqn [6].
Preparative ultracentrifuges are designed for
sample preparation. This kind of ultracentrifuge is
also commonly used for quantitative estimations of
sedimentation coefRcients of particles in a density
gradient, although the data obtained are not as accu-
rate as those obtained using analytical ultracen-
trifuges. Preparative ultracentrifugation can be
divided into two methods, namely differential ultra-
centrifugation and density gradient centrifugation.
Differential centrifugation is based on the differences
in the sedimentation rates of particles in samples. If
a suspension of particles is centrifuged in a tube
without a density gradient, each particle will move
toward the bottom of a tube. In this case, the rate of
sedimentation, v, is dependent upon s (eqn [2]). Since
s is mostly a function of particle size, the rate of
sedimentation is proportional to particle size. In the
course of the ultracentrifugation, two fractions can be
obtained from a solution of particles: a pellet contain-
ing sedimented particles and a supernatant solution
of the unsedimented fraction. A given particle in the
solution may sediment to the pellet or near the bot-
tom, as illustrated in Figure 1. As might be expected,
this centrifugation will Rrst sediment the largest par-
ticles in the sample solution to the bottom of the tube.
The only particle that is in puriRed form is the most
slowly sedimenting one, but the yield is very low.
The major problem with differential centrifugation is
that the centrifugal force necessary to pellet the
larger particles is also often sufRcient to pellet
the smaller particles initially near the bottom of the
tube (Figure 1). To separate one particle from another
effectively, a 10-fold difference in mass is usually
required. Thus, this centrifugation is recommended
for the separation of proteins from large particles
such as cells or organelles. However, it cannot be
used for the isolation of one protein from another
because the partial speciRc volume,
, of most pro-
teins (in eqn [6]) is not sufRciently different.
Eqn [6] assumes that centrifugation is performed in
a homogeneous medium. However, centrifugation
Figure 1 Fractionation of particles by differential centrifugation.
Reproduced with permission from Griffith (1979).
can be carried out in a solution of an inert substance
in which the concentration increases from the top to
the bottom of the centrifuge tube, i.e. density in-
creases from top to bottom. In such density gradient
centrifugation of a mixture of particles with different
sizes or buoyant densities, a particle will become
stationary when (1!
) in eqn [6] is zero. Thus,
various components will separate according to size or
buoyant densities, and form bands or zones of par-
ticles with similar densities. Thus, the use of such
density gradients greatly enhances the resolving
power.
There are two types of density gradient ultracen-
trifugation: isopycnic and rate-zonal ultracentrifuga-
tion. In isopycnic centrifugation, separation is based
on the centrifugation of particles in a density gradient
through which the particles move until their densities
are the same as those of the surrounding medium, i.e.
in eqn [7], p" (Figure 2). The sample is mixed
with a relatively concentrated solution of a low mo-
lecular mass substance, such as CsCl, and is centri-
fuged until the solution achieves equilibrium under
the high centrifugal Reld. The low molecular mass
substance forms a steep density gradient. It is not
obligatory to load the sample on top of the gradient.
In the centrifugation, particle size only affects the rate
at which particles reach their isopycnic position.
Since variations in amino acid composition give pro-
teins with only slightly different densities, isopycnic
centrifugation can be used only when proteins are
associated with nonprotein components such as lipids
or polysaccharides, and their density differences are
sufRcient for the separation. Various gradient media
can be used for the separation of these proteins be-
cause proteins form a band at low density in most
gradient media. For example, in addition to CsCl,
 III / PROTEINS / Centrifugation 4017

sedimentation rate is more affected by molecular size,

the rate-zonal ultracentrifugation separates similarly
shaped macromolecules largely on the basis of their
molecular masses. It should be noted that particles
separated by the rate-zonal centrifugation may not be
homogeneous because particles with similar mass,
even proteins, are sometimes heterogeneous.

Practical Aspects for Protein

Separation by Centrifugation
Since rate-zonal centrifugation is commonly used for
the separation of proteins, the following discussion
will focus on a practical approach for this technique.
Types of Rotor
Preparative centrifuge rotors can be classiRed into
four types: Rxed angle, swinging-bucket, vertical
and zonal. In Rxed-angle rotors, the tubes are posi-
tioned at a Rxed angle. These rotors are often used
for differential ultracentrifugation and are very efR-
cient for the separation of proteins from cells and
organelles. Typically, a sample is loaded atop a gradi-
ent which reorients as the rotor is spun (Figure 3).
During centrifugation, the larger particles are Rrst
sedimented across the tube, hit the wall of the
tube, and slide down to form a pellet at the bottom.

Figure 2 Types of density gradient centrifugation. (A) Rate-

zonal centrifugation. (B) Isopycnic centrifugation using
a preformed density gradient. (C) Isopycnic centrifugation using
a self-forming gradient. Reproduced with permission from Rick-
wood (1992) by permission of Oxford University Press.

RbCl, NaBr or KBr can also be used to form shal-

lower gradients for better resolution of these proteins.
Rate-zonal centrifugation is ideal for particles
of deRned size such as protein and RNA. In the
rate-zonal ultracentrifugation, a mixture containing
particles is layered on top of a density gradient.
Loading the concentrated samples to the top of the
gradient increases the eventual resolution of re- Figure 3 Operation of fixed-angle rotors. (A) The gradient is
covered particles. Sucrose is commonly used to form prepared, the sample is loaded and the centrifuge tubes are
placed into the rotor. (B) Both sample and gradient reorient during
a density gradient. During centrifugation, particles
acceleration. (C) Bands form as particles sediment. (D) Bands
move through the gradient at their characteristic and gradient are both reoriented when the rotor is at rest. Repro-
sedimentation rates, forming zones that can be re- duced with permission from Rickwood (1992) by permission of
covered at the end of the run (Figure 2). Because the Oxford University Press.
4018 III / PROTEINS / Centrifugation
EfRciency for the pelleting of particles is high due to
the short sedimentation path. However, Rxed-angle
rotors are not common for protein separation because
the pelleting process also disrupts sample zones
as particles sediment through the gradient. Thus,
Rxed-angle rotors are mainly used for the pelleting of
materials.
For the separation of proteins from other proteins,
especially for small scale separation, the swinging-
bucket rotor is widely used for rate-zonal centrifu-
gation. This type of rotor is also used for the estima-
tion of sedimentation coefRcients of proteins. As
shown in Figure 4, in the swinging-bucket rotor, the
sample tubes are loaded into individual buckets
which hang vertically while the rotor is at rest. When
the rotor begins to rotate, the buckets swing out
perpendicular to the axis of rotation. In these rotors,
resolution of particles is high because particles sedi-
ment with a relatively long path length. For the same
reason, run times are generally longer. Many types of
swinging-bucket rotors are commercially available.
The centrifuge tube should be as long as possible if
high resolution is the objective. For large volume
samples, swinging-bucket rotors with wider tubes
should be used because the sample can be loaded in
a narrow zone while still reducing particle interac-
tions during sedimentation.
Vertical rotors are suitable for isopycnic as well as
for rate-zonal separations. However, this type of
rotor is not practical for the separation of proteins.
As a result of diffusion and reorientation during
Figure 4 Operation of swinging-bucket rotors. (A) The gradient
is performed and the sample is loaded on the top of the gradient.
(B) Centrifuge bucket reorients as rotor accelerates to lie perpen-
dicular to the axis of rotation. (C) Bands form as the particle
sediment. (D) Rotor decelerates. Centrifuge bucket comes to rest
in its original vertical position. Reproduced with permission from
Rickwood (1992) by permission of Oxford University Press.
centrifugation, sample bands will be signiRcantly
broader than analogous bands in swinging-bucket
rotors. In addition, if the sample contains pellets or
Soats, these materials will distribute along the length
of the tube and can subsequently contaminate the
supernatant during reorientation at the end of run.
Choice of Density Gradient
A density gradient is essential for rate-zonal centrifu-
gation to support the zones of particles as they sedi-
ment. In addition, the sample can be loaded on to the
top of the gradient as a narrow zone and the increas-
ing density from the top to the bottom of the density
gradient suppresses mechanical disturbances. More-
over, the presence of a gradient of increasing viscosity
serves to sharpen the sample zones during centrifu-
gation. The density gradient material for protein sep-
aration requires the following properties.
1. The materials should be sufRciently soluble in
water to produce the range of densities needed.
2. Solutions of the gradient materials should be ad-
justable to a pH and ionic strength that are not
harmful to proteins in the sample.
3. The materials should not interfere with methods
of analysis of the target protein.
Sucrose has most often been used as a gradient
material. Sucrose is inexpensive and extremely sol-
uble in aqueous media and can be used to produce
density gradients ranging up to 1.35 g mL\1. Thus, it
is suitable for separation of almost all proteins in
cells. Although concentrated solutions of sucrose
have high osmotic potential that cause shrinkage of
certain cells and organelles, the high osmotic pressure
has relatively less effect on the biological properties of
proteins. Generally, sucrose is relatively inert to pro-
teins, although contaminants in many commercial
sources of sucrose may interact with proteins. Such
impurities can be removed by treatment with ac-
tivated charcoal. However, it is best to purchase
specially puriRed sucrose for density gradient work.
To sterilize sucrose solutions, autoclaving (1003C or
above) of the solution should be avoided and treat-
ment with 0.1% diethylpyrocarbonate is recommen-
ded. As described above, isopycnic centrifugation can
be used for the separation of different types of pro-
teins. However, it should be noted that the density of
sucrose, even of a saturated solution, is too low for
the separation. For this purpose, RbCl, NaBr or KBr
can be used to form shallower gradients for better
resolution.
Glycerol is used to stabilize some proteins, espe-
cially membrane-bound proteins, and provides gradi-
ent densities ranging up to 1.26 g mL\1. Thus,
glycerol gradients are widely used for the separation

of proteins by rate-zonal separations. However, it
should be noted that the high viscosity of glycerol
reduces the effective density range and glycerol ap-
pears to inhibit some enzyme activities.
Preparation of Gradients
Density gradients can be divided into two types: con-
tinuous and discontinuous. For protein separation,
continuous gradients are usually used in rate-zonal
centrifugation. The most common continuous gradi-
ent for protein separation is a linear gradient in
swinging-bucket tubes. A linear gradient is a gradient
in which the density increases linearly in a tube of
constant cross-sectional area with increasing distance
from the centre of rotation. Thus, in this conRgura-
tion the linear gradient can be deRned as one where
the density increases linearly with volume.
When designing a linear gradient in swinging-
bucket rotors, several points should be emphasized.
The density at the top of the gradient must be sufR-
cient to support the sample while the density of the
bottom of the gradient must not exceed the density of
proteins to be separated. In general, the greater the
slope of the gradient, the better the resolution ob-
tained because the viscous drag rises rapidly as the
sucrose concentration increases. Usually, as a Rrst
attempt, a 5}30% or 10}40% sucrose gradient
should be used. It should be emphasized that the
sample volume is related to the slope of the gradient
because a given slope of gradient can only tolerate
a limited amount of sample before gradient inversion
occurs. Poor resolution during rate-zonal centrifu-
gation almost always results from overloading.
Linear gradients are prepared using gradient
makers. Many conRgurations of gradient maker are
available. The simplest gradient makers consist of
two vessels of equal cross-sectional area joined by
a connecting channel with a stopcock. One chamber
is a reservoir and the other chamber has a mixing
device and an exit connected to the centrifuge tube.
There are two methods for preparing linear gradients:
1. The reservoir contains the less dense solution, the
mixing chamber contains the denser solution, and
the gradient is routed to the wall of the centrifuge
tube at the top. This method is readily applicable
to centrifuge tubes made of hydrophilic materials
such as cellulose nitrate and cellulose acetate
butyrate.
2. The reservoir contains the denser solution,
the mixing chamber contains the less dense solu-
tion, and the gradient is routed to the bottom of
the centrifuge tube. This method can be applied
to any type of centrifuge tube and it is much easier
to prepare the gradient without disturbance.
III / PROTEINS / Centrifugation 4019
The gradient should be prepared and maintained
at 43C.
Preparation of Sample
The sample should be ready for loading before the
gradient is prepared and should be kept cold for many
preparations. The sample is usually prepared in the
same buffer as the gradient. In addition, several
points are important in sample preparation:
1. The sample solution must have a density less than
that of the gradient.
2. Gradients should be centrifuged as soon as poss-
ible after the sample has been loaded to prevent
so-called droplet sedimentation.
3. For optimal resolution in rate-zonal centrifu-
gation, the sample must be loaded on to the top of
gradient and the sample volume should not exceed
2}3% of the gradient volume.
Loading of the sample on to the density gradient is
one of the most crucial steps in rate-zonal centrifu-
gation. The simplest method is to use a pipette to load
the sample directly to the meniscus at the tube wall.
Conditions During Centrifugation
Smooth acceleration and deceleration are important
for all gradient work. In addition, control of the
temperature of the sample and gradient are important
for reliable and reproducible sedimentation. Fortu-
nately, most modern ultracentrifuges are equipped
with programmed acceleration and deceleration
modes which minimize the disturbance of gradient
and temperature control. It should be emphasized
that, during the gradient reorientation phase of a run
using a swinging-bucket rotor, the rotor should be
accelerated as slowly as possible up to 1000 rpm, and
the brake switch should be off below 1000 rpm dur-
ing deceleration.
Recovery of Fractions from the Gradient
After centrifugation, gradients are fractionated to re-
cover protein bands. Great care must be taken at this
stage to avoid loss of resolution. Several points should
be emphasized.
1. All operations should be designed to minimize
disturbance of the gradient.
2. The volume of the tubing from the gradient to the
fraction collector should be minimized.
3. Care must be taken to avoid contamination of the
recovered fractions by pelleted materials.
4. The gradient should be fractionated at a slow Sow
rate, particularly if viscous materials are used for
the gradient.
4020 III / PROTEINS / Crystallization
In order to collect the entire gradient in a series of
fractions, several methods may be applied. The
simplest is to pierce the bottom of the tube with
a needle, and collect the gradient as it drops out.
Another method is to pump the gradient from the
bottom of the tube with a narrow capillary tube.
However, this method is not recommended because
of the potential to disturb the gradient and resulting
loss of resolution.
See also: III/Proteins: Capillary Electrophoresis; Crystal-
lization; Electrophoresis; Field Flow Fractionation; High-
Speed Countercurrent Chromatography; Ion Exchange.
Further Reading
GrifRth OM (1979) Ultracentrifuge Rotors: A Guide to
Their Selection. Palo Alto, Beckman Instruments.
Crystallization
M. Y. Gamarnik, Nanoscale Phases Research,
Bensalem, PA, USA
Copyright ^ 2000 Academic Press
Introduction
The Rrst protein crystals described in the literature
were obtained by Hunefeld in 1840. Hunefeld ob-
served hemoglobin crystals after slow drying of blood
pressed between two slides of glass. It is remarkable
that this Rrst result demonstrated the basic principle
used today, that protein crystals similar to inorganic
crystals may be produced by concentration of a pro-
tein in solution through slow dehydration. Through-
out the history of protein crystal growth, the
rationale for protein crystallization has been, Rrstly,
separation of proteins from complex extracts, and
then, starting in the 1930s, as puriRcation as deter-
mination of the three-dimensional structure of pro-
tein molecules.
Knowledge about the three-dimensional structure
is necessary to better understand the functions of
protein molecules in living systems and plants. Three-
dimensional structure can be determined by X-ray
diffraction. For X-ray diffraction, good quality pro-
tein crystals of appropriate sizes are required. Crystal
sizes in each direction should be at least 0.1 mm, if
using a strong beam of synchrotron radiation, or at
least 0.3 mm for conventional sources of X-rays.
Protein molecules in the crystalline state are more
stable than in solution. Therefore, crystallized pro-
Hsu HW (1981) Separation by Centrifugal Phenomena.
New York: John Wiley.
Laskin AI and Last JA (eds) (1974) Subcellular Particles,
Structures, and Organelles. New York: Marcel Dekker.
Neurath N and Hill RL (eds) (1975) The Proteins, 3rd edn.
New York: Academic Press.
Price CA (1982) Centrifugation in Density Gradient. New
York: Academic Press.
Rickwood D (ed.) (1983) Iodinated Density Gradient Me-
dia: A Practical Approach. Oxford: IRL Press.
Rickwood D (ed.) (1984) Centrifugation, 2nd edn, A Prac-
tical Approach. Oxford: IRL Press.
Rickwood D (ed.) (1992) Preparative Centrifugat-
ion, A Practical Approach. Oxford: Oxford University
Press.
Schachman HK (1959) Ultracentrifugation in Biochemis-
try. New York: Academic Press.
Sheeler P (1981) Centrifugation in Biology and Medical
Science. New York: John Wiley.
teins are more stable against denaturation and may be
preserved for a signiRcantly longer period of time
than in solution. That is the reason that protein crys-
tallization is often directed as much on preservation
as on separation and puriRcation.
This article comprises a brief description of general
principles of protein crystal growth and a description
of various techniques of protein crystallization with
the emphasis on methods using a small amount of
a crystallizing solution, from about 1 to 20 L. The
consumption of small amounts of protein is of value,
since screening and optimization tests of determina-
tion of crystallization conditions typically require
many portions of protein solution.
General Principles of Protein
Crystallization
Intermolecular Interaction
To crystallize a protein it should be Rrst of all dis-
solved to give a solution where the protein molecules
become close one to another to create a nucleus that
grows into a crystal.
An essential feature of protein solutions, associated
with the complex structure and large size of protein
molecules, is that the molecules may be charged by
electric charges, of the same polarity, resulting in
long-range electrostatic repulsion. This peculiarity is
mediated by the ability of macromolecules to acquire
protons from a solution or give up protons into the
solution depending on the pH. The charge value of

protein molecules increases with the difference be-
tween pH of the solution and pH in the isoelectric
point, pI, of the solution. At pH"pI, molecules
become neutral, and accordingly, the long-range re-
pulsion is absent. For most proteins pI is in the range
4.5}6.0.
At short distances van der Waals forces of attrac-
tion act between molecules. Competition between
repulsion and attraction determines the state of pro-
tein molecules in solution. If the repulsion dominates,
molecules remain apart in solution, protein is dis-
solved and no clusters or crystals are created. If at-
tractive forces dominate, the molecules gather into
clusters, that may form nuclei to yield crystals. For
creation of well-ordered nuclei the attractive forces
should be strong enough to provide slow clusteriz-
ation but not so strong as to impair the formation of
the crystal structure.
The Coulombic repulsive forces are affected usu-
ally by an addition to the protein solution of buffers,
salts, precipitants and other additives. Buffers inSu-
ence acidity; the pH of the solution alters the differ-
ence between pH and pI. This in turn changes charge
values of protein molecules and accordingly the long-
range repulsion. There are various buffers providing
different pH: 0.1 M sodium acetate, pH"4.6, 0.1 M
trisodium citrate dehydrate, pH"5.6, 0.1 M sodium
cacodylate, pH"6.5, 0.1 M HEPES, pH"7.5,
0.1 M tris hydrochloride, pH"8.5, etc.
Salts added to a protein solution may screen the
repulsive interaction between the molecules. For in-
stance, a 0.1}0.2 M aqueous solution of sodium
chloride essentially shields the repulsive electrostatic
forces. Precipitants such as polyethylene glycol (PEG)
of various molecular weights, isopropanol and
sodium formate, decrease the solubility of protein,
initiating creation of clusters or nuclei.
Nucleation
Nucleation is initiated by Suctuation of density in
a system of atoms or molecules. Crystallization in
protein solutions is initiated by protein concentration
Suctuation in a solvent. In regions of higher concen-
tration, molecules are associated in clusters because
of smaller intermolecular distances thus increasing
attractive forces. Generally, clusters may be stable or
metastable.
Reduction of repulsive interaction between protein
molecules is a necessary condition of nucleation pecu-
liar to proteins and other macromolecular substances.
But it is still not enough to create a stable nucleus which
becomes a seed for crystal growth. The mutual posi-
tion of protein molecules should correspond to a min-
imal energy to form thermodynamically stable nuclei.
III / PROTEINS / Crystallization 4021
Thermodynamic free energy of a cluster in solution
consists of two parts: volume energy,
Gv and surface
energy,
Gs. The volume energy is negative, so that it
stabilizes the cluster, but surface energy is positive
tending to make the cluster unstable. Competition
between these two parts of the overall free energy
determines the stability of clusters. The total free
energy of a cluster,
G"
Gv#
Gs depends on its
size, r, so that
G"
G(r). This dependence reveals
a maximum at a critical size, r"rc. At sizes larger
than rc clusters become a stable nucleus capable to
grow as a crystal. At sizes less than the critical size,
clusters tend to dissolve, since they are energetically
unstable.
The critical nucleus size, rc, decreases with an in-
crease of the supersaturation of protein solution
"c/csat, since
Gv&!ln  and accordingly rc&1/
ln . Here c is the concentration of protein in solution
and csat is the concentration of the protein in saturated
solution. In other words, csat is the solubility of the
protein. Consequently, nucleation may occur only at
'1, i.e. at protein concentration, c larger than the
concentration of saturated solution, csat. This condi-
tion indicates also that the chemical potential of
a cluster is lower than the chemical potential of the
solution, and accordingly the volume energy
Gv is
negative. An increase of the supersaturation,  also
reduces the maximum of the total free energy,
G(rc),
which should be overcome to create a stable nucleus,
because
G(rc)&1/(ln )2.
Nucleation is affected by many other parameters,
such as relative speciRc surface energy of protein
crystals in solutions, temperature, mobility and sub-
structure of protein molecules, impurities, rate of
supersaturation, etc. That is the reason that nuclea-
tion of proteins is an important step in protein crys-
tallization, which often requires many screening
experiments.
Crystal Growth
A zone of lower concentration is formed around a nu-
cleus after it starts to grow, consuming surrounding
molecules. The difference between the protein con-
centration in the bulk of the solution and in the
depletion zone becomes a driving force, transporting
the protein molecules from solution to the growing
crystal. This mass transport is realized generally by
diffusion and convection due to gradients of protein
concentration. Slow transport of the molecules is
required to yield good quality crystals. This is
achieved at low supersaturation and low rate of
dewatering of the protein solution and by crystalliza-
tion at lower temperatures, &4}63C.
Mechanisms of attachment of protein molecules to
the lattice of growing crystals have been investigated
4022 III / PROTEINS / Crystallization
intensively in recent years with the atomic force
microscope. It was demonstrated by Malkin et al.
with several proteins, lysozyme, thaumatin, cana-
valin, catalase and apoferritin, that macromolecules
grow by all surface integration mechanisms involved
in the crystallization of small molecules. The protein
crystals reveal growth on screw dislocations and by
two- and three-dimensional nucleation.
A screw dislocation generates steps that propagate
along the crystal surface. The steps contain kinks at
the crystal surface into which molecules are incorpor-
ated, building crystal layers in a spiral fashion around
the dislocation core. This mechanism is realized basi-
cally at lower supersaturations, about 1}1.5. At high-
er supersaturations, protein crystals grow typically by
two-dimensional nuclei formed on the surfaces. Ab-
sorption of three-dimensional nuclei have also been
detected. The three-dimensional clusters are de-
veloped into multilayer stacks or microcrystals.
Absorption of impurities may cause a cessation of
growth of macromolecular crystals. Dubrin and
Feher revealed parallel step trains on lysozyme crystal
surface accounted for by impurities, impeding the
subsequent crystal growth. Malkin et al. have detec-
ted surfaces of lysozyme crystals completely covered
by impurity absorption layers resulting in the cessa-
tion of growth.
Methods of Crystallization
Proteins may initially contain a large amount of
impurities, such as salts, other classes of macro-
molecules, denaturated molecules, solid particles and
other contaminants impeding crystallization. There-
fore, proteins should be puriRed for crystallization
tests. The same applies to buffers, salts and
Figure 1 Schematic presentation of hanging- and sitting-drop techniques. (A) Hanging drop; (B) sitting drop. (1) Drop of protein
solution, (2) reservoir solution, (3) glass vessel, (4) coverslip, (5) sealing rim, (6) inverted glass pot. (From Ducruix and Giedge (1992)
by permission of Oxford University Press.)
precipitants used in crystallizing solutions. Descrip-
tions of preparation and handling of proteins for
crystallization and methods of characterization of
macromolecules can be found in Further Reading.
There are many methods, techniques and appar-
atus for protein crystal growth depending on conRg-
uration of experiment } hanging drop, sitting drop or
crystallization in a capillary. The method used also
depends on the way the solution is supersaturated: by
vapour diffusion, liquid}liquid diffusion, dialysis,
mixing with a precipitant, change of temperature, etc.
Other factors include gravity conditions and mass
transport in solution, crystal growth in microgravity,
gel method, crystallization in drops and suspended in
heavy liquids.
Hanging and Sitting Drop
A hanging or sitting drop comprising a solute protein,
to be crystallized, with a crystallizing agent, is equi-
librated against a reservoir solution containing the
crystallizing agent at higher concentration than in the
drop (Figure 1). The reservoir, a glass vessel, is closed
by a glass cover slip and sealed by grease to prevent
the liquids evaporating. The drop volume is usually
from 1 to 10
L, the volume of reservoir solution is
&0.5}1 mL. An increase of concentration of crystal-
lizing agent in the drop leads to a supersaturation of
protein solution necessary for crystallization. Equili-
bration is achieved by vapour diffusion of water or
other volatile components and continued until va-
pour pressures in the drop and in the reservoir
become equal.
The crystallizing agent may comprise a buffer, such
as sodium acetate, tris hydrochloride, HEPES or so-
dium cacodylate, a precipitant, such as polyethylene
glycol, isopropanol or sodium formate, and a salt.

If there are no volatile components except water,
diffusion of water occurs from the drop to the reser-
voir solution. This decreases the drop volume and
accordingly increases the concentration of all the
components. In the presence of a volatile species, such
as isopropanol, in the crystallizing agent, diffusion
proceeds in two directions, from the reservoir to the
drop, and in the opposite way, from the drop to
reservoir, of water. Equilibrium is achieved after the
saturated partial vapour pressures of evaporating
components become equal in the drop and in the
reservoir. In this case, the drop may decrease or
increase in volume or remain the same.
The often used ratio between concentrations of the
crystallizing agent in the reservoir and the drop is 2.
This is obtained by mixing equal volumes of protein
solution and the reservoir solution. It may be, for
instance, a mixture of 2
L of the reservoir and 2
L
of the protein solution in the hanging or sitting drop.
If only water diffusion takes place in the absence of
other volatile species, at equilibrium the Rnal volume
of the drop will be half the original volume. Accord-
ingly, concentration of all components in the drop,
protein, buffer, salt and precipitant double.
Nucleation and crystal growth in a drop are affec-
ted essentially by rate of equilibration. The rate de-
pends on the difference between vapour pressures in
the drop and reservoir, which in turn depends on the
content of the crystallizing agent, temperature, and
the size and shape of the drop. SigniRcantly lower
equilibration rates with polyethylene glycol rather
than ammonium sulfate as a precipitant have been
demonstrated. A decrease of equlibrium rate with
increase in drop size and with decrease of temper-
ature has also been revealed. The time for &90%
equilibration varies from a day to about one month.
In the hanging-drop technique, crystals are basi-
cally created at the bottom of the drop, near the
surface. This may be caused by forming a layer of
supersaturated protein solution near the surface dur-
ing evaporation of water. The subsequent distribu-
tion of protein concentration over the whole drop is
diffusion limited.
The contribution of convection is small, since the
supersaturated layer is of higher density and located
at the bottom of the drop. Such inhomogeneity ap-
plies to nucleation as much as to crystal growth.
In the sitting-drop, distribution of protein concen-
tration is different. Evaporation forms initially a layer
of higher concentration near the surface at the top.
This excessive concentration is then distributed
quickly over the drop by convection. Accordingly,
nucleation and the start of crystal growth occur prac-
tically at the same protein supersaturation. This is
a disadvantage in comparison with the hanging-drop
III / PROTEINS / Crystallization 4023
technique, since crystals grow at higher supersatura-
tion. From another point of view, however, the sit-
ting-drop method is energetically more favourable
and heterogeneous nucleation on various substrates,
including minerals can take place.
For screening and optimization tests, multicham-
ber plates are utilized, containing many wells of the
type shown in Figure 1. Plastic Linbro boxes, normal-
ly used for tissue culture, are convenient for hanging-
drop tests. Each such box contains 24 wells. To
prepare tests, about 0.7}1.0 mL of crystallizing agent
is put in each reservoir. A drop comprising a mixture
of protein solution and reservoir solution is placed on
the glass coverslip. Then, the coverslip with the drop
is gently turned over and set on the vessel rim covered
with a thin layer of grease. In a similar way, the
sitting-drop tests are prepared, with the difference
that the drop is placed on an inverted glass pot
(Figure 1).
Method for Crystal Growth in a Capillary
A vapour diffusion method for growing crystals in-
side capillary tubes, invented by M. Y. Gamarnik and
U. R. Alvarado, is illustrated in Figure 2. The crystal-
lizing unit comprises a capillary tube containing a col-
umn made up of three layers. The Rrst layer is
a protein solution to be crystallized, which may con-
tain a buffer, salt, precipitant and other additives.
The second layer is an absorbent, which is typically
a liquid, such as glycerol or a highly concentrated salt
solution. In the preparation of the crystallizing unit,
the protein solution and absorbent are placed in the
capillary tube so that they are segmented by an air
section. The ends of the capillary tube are sealed by
end-caps. The liquids are placed sequentially in the
capillary tube by a syringe. The lengths of the protein
solution, absorbent and air layer segments are se-
lected by experimenter, but the usual lengths are from
2 to 20 mm.
The internal diameter of the capillary tube is se-
lected so that the capillary forces, acting on the
liquids, held within the sealed tube, are sufRcient to
prevent direct contact between the liquids during
handling or transportation. The internal diameter is
usually less than 3 mm, but for crystallization tests
from about 1}10
L of protein solution the diameter
should be about 0.6}1.5 mm.
The vapour of the solvent, which is normally
water, diffuses from the protein solution through the
air layer to the absorbent because of the difference in
the water vapour pressures in the protein solution and
absorbent. Evaporation of water from crystallizing
solution results in an increase of concentration of
protein and other components } buffer, salt and
4024 III / PROTEINS / Crystallization
Figure 2 Schematic presentation of vapour diffusion method for crystal growth inside a capillary tube. (A) Initial configuration,
(B) final configuration. (1) Capillary tube, (2) protein solution, (3) absorbent, (4) air layer, (5) end-caps, (6) protein solution with grown
crystal, (7) absorbent containing absorbed water.
precipitant } which were initially present. Nucleation
and growth of a protein crystal occur within the
crystallizing solution after it reaches the appropriate
supersaturation.
During the vapour transport, the length of the air
layer remains approximately constant but shifts to-
ward the crystallizing solution. The length of the
protein solution layer decreases. Accordingly, the
length of the absorbent, containing absorbed water,
increases (Figure 2B). At any time during crystal
growth, it is possible to calculate the concentration of
protein and other additives in the protein solution by
the relative change of its length.
A desired rate proRle of water evaporation can be
set by appropriate choice of concentration and vol-
ume of the protein solution and absorbent, the length
of the air layer and the temperature. Often it is desir-
able to set a relatively rapid rate of evaporation dur-
ing the initial phase of the growth process leading to
nucleation, followed by a lower rate of evaporation
during the crystal growth process. To accomplish this
proRle, the air layer length should be small, about
3}5 mm, and the absorbent layer length should be
also small, in comparison with the length of protein
solution layer. The short air layer establishes a rela-
tively fast dewatering followed by a rapid increase of
concentration of protein solution, while the small
volume of absorbent allows a relatively rapid de-
crease in the rate of evaporation after the initial phase
of evaporation, due to dilution of the absorbent by
the absorbed water. This proRle, initiating nucleation
with subsequent slow crystal growth, usually yields
a single crystal or a few crystals in each portion of the
protein solution. This is beneRcial since the dissolved
protein feeds only one or a few crystals during the
growth, resulting in their larger size.
Liquid+Liquid Diffusion Techniques
Interface diffusion In the interface-diffusion
method for protein crystal growth two liquids }
a protein solution and a solution of the crystallizing
agent } make contact and are separated at an inter-
face. Equilibration is achieved by diffusion of the
crystallizing agent and protein across the liquid}
liquid interface. Diffusion leads to a slow increase in
the concentration of the crystallizing agent in the
protein solution near the interface. Nucleation occurs
in the interface region after sufRcient supersaturation.
Subsequent slow crystal growth is provided by the
relatively small interface region of the higher protein
supersaturation.
A basic difRculty in this technique is bringing the
liquids into contact without convection, i.e. without
rapid mixing. Convection may be reduced if the less
dense solution is gently placed on the more dense
solution. Crystallization experiments conducted
under microgravity conditions in space avoided con-
vection during solution contact when the interface-
diffusion technique was used.
To use only a small amount of protein solution,
about 2}20
L, the liquids may be brought into con-
tact in a capillary tube of diameter in the range
&1}2 mm.
Dialysis In the dialysis technique, the protein solu-
tion and crystallizing-agent solution are separated by
means of a molecular membrane. Both solutions are
in contact with the membrane, which allows diffu-
sion of small molecules, but prevents macromolecules
passing. The maximum molecular weight of molecu-
les able to diffuse through the membrane pores is
determined by the molecular weight cutoff (MWCO).

The MWCO of contemporary dialysis membranes is
in the range from 100 to 300 000 Da.
During equilibration, molecules of buffer, salt,
precipitant or water may diffuse through the mem-
brane from the crystallizing agent to the protein solu-
tion and also in the opposite direction, resulting in
crystallization. For micro-quantities of protein solu-
tion, capillaries of small internal diameters, about
1}2 mm are used. The protein solution is placed in
a capillary, the end of which is capped by the mem-
brane. The membrane end of the capillary
is then put into an appropriate agent solution for
crystallization.
Batch technique Microbatch techniques include
crystallization under oil, with droplets of about
1}10
L of crystallizing solution. It was demon-
strated that application of parafRn and silicone oils
affect the rate of equilibrium, resulting in a better
quality of crystals. A containerless method has been
developed, where the crystallizing solution drop is
suspended between two oil layers of different density,
preventing the undesirable contact of the protein
solution with walls of a vessel.
A drawback of the batch method is that, typically,
many nuclei are formed resulting in the growth of
many crystals, and it takes much effort to Rnd the
conditions for growing just one or a few crystals in
a single batch.
Gel Method
In the gel method, crystallization in a gel results in the
protein solution being trapped by a loose network
which is stretched over the whole volume of the
solution. The gel comprises macropores, of about
100 nm in size, Rlled with the crystallizing solution.
The macropores are interconnected, by a dense sys-
tem of micropores, of about 10 nm in size.
Entrapping of the crystallizing solution by the gel
network prevents natural convection and sedimenta-
tion. Accordingly, equilibration between the protein
solution and the crystallizing agent is mediated by
diffusion only, through the gel pores. This results in
slow crystal growth, improving the quality of the
produced crystals if no gel structure is incorporated in
the crystal lattice, as is often the case. Various crystal
growth techniques can be used with a gel, such as
hanging-drop technique, liquid}liquid diffusion and
dialysis. Silica gel and agarose gel are most often used
for protein crystallization.
Crystallization in Microgravity
Protein crystal growth has been studied under micro-
gravity conditions conducted in space (satellites and
III / PROTEINS / Crystallization 4025
space shuttles). Space-grown crystals are frequently
of better quality than the same protein crystals grown
on the earth. Under microgravity conditions, convec-
tion and sedimentation are suppressed. Accordingly,
transport of molecules, supersaturating the protein
solution, is mediated by diffusion only, providing
slow crystal growth. Typically, crystals are suspended
and grow freely in different crystallographic direc-
tions, forming a well-ordered structure and equilib-
rium shape.
Various methods have been used for crystal growth
experiments in space, including interface-diffusion,
dialysis and vapour diffusion. Most methods have
a device or a means to connect or disconnect interac-
tion between the protein solution and crystallizing
agent so that mixing only takes place under zero
gravity. Typically, two wells, one of which is Rlled
with protein solution and a second one Rlled with
crystallizing agent, are brought into contact in orbit
through a connecting valve or by a relative movement
of the wells.
Concluding Remarks
Experiments, directed at crystallization of proteins
require an understanding of the main principles of
nucleation and growth and need much testing. Some
tests are associated with a decrease in critical super-
saturation, necessary for nucleation, through selec-
tion of appropriate buffers, salts and precipitants or
by exploration of substrates for heterogeneous nu-
cleation. Other methods follow the development and
use of crystallization cells, consuming small amounts
of protein solution, through decreasing the internal
diameter of capillaries (in the capillary method) or the
size of droplets in microbatch and hanging-drop ex-
periments.
Small volume crystallization cells will be adapted
for microgravity experiments to reduce space require-
ments on satellites and shuttles, accordingly reducing
their cost. This may be realized by modiRcation of the
capillary technique, shown in Figure 2. For instance,
two air layers may be used instead of one, separated
from each other by a water barrier layer. The water
barrier layer delays absorption of the vapour from the
protein solution until the barrier layer is eliminated
by absorption into the absorbent. The necessary delay
is typically one or two days, from preparation of the
tests until the spacecraft carrying the crystal growth
device has attained microgravity conditions.
Development of crystallization methods at lower
critical supersaturation seems to be supported by
a broadening of our knowledge of the main principles
governing nucleation and growth of macromolecular
crystals.
4026 III / PROTEINS / Electrophoresis
See also: II/Crystallization: Additives: Molecular
Design; Control of Crystallizers and Dynamic Behaviour;
Polymorphism. III/Supercritical Fluid Crystallization.
Further Reading
Chernov AA (1984) Modern Crystallography. III. Crystal
Growth. Berlin: Springer-Verlag.
Darby NJ (1993) Protein Structure. Oxford: IRL Press,
Oxford University Press.
Ducruix A and Giedge R (1992) Methods of crystallization.
In: Ducruix A and Giedge R (eds) Crystallization of
Nucleic Acids and Proteins. A Practical Approach.
Oxford: IRL Press, Oxford University Press.
Feher G and Kam Z (1985) Nucleation and growth of
protein crystals: general principles and assays. Methods
in Enzymology 114: 77d111.
Lorber B and Giedge R (1992) Preparation and handling of
biological macromolecules for crystallization. In:
Ducruix A and Giedge R (eds) Crystallization of Nucleic
Electrophoresis
M. J. Schmerr, National Animal Disease Center,
Ames, IA, USA
Copyright ^ 2000 Academic Press
Transmissible spongiform encephalopathies are neur-
odegenerative diseases found in both humans and
animals. The oldest known member of this family of
diseases is scrapie in sheep and goats and the most
infamous member is bovine spongiform encepha-
lopathy or mad cow disease. These diseases are fatal
for the individuals who become infected. As a result,
there is a considerable amount of interest in develop-
ing methods to detect early infection. This would
enable removal of animals from food chains and
by-products used for cosmetic and health care. For
humans, an early diagnosis may make it possible to
treat infected individuals with drugs to arrest the
course of the disease.
A feature of these diseases is the accumulation of
rod-shaped Rbrils in the brain that form from an
aggregated protein. This abnormal protein is a pro-
tease-resistant form of a normal host cell glycoprotein
(prion protein). When the aggregated protein is de-
natured in sodium dodecyl sulphate (SDS) and -mer-
captoethanol, a monomer form of Mr&27 kDa is
observed. This abnormal prion protein is used as
a marker for infection with a transmissible spongi-
form encephalopathy.
Acids and Proteins. A Practical Approach. Oxford: IRL
Press, Oxford University Press.
McPherson A (1982) Preparation and Analysis of Protein
Crystals. New York: John Wiley.
McPherson A (1991) A brief history of protein crystal
growth. Journal of Crystal Growth 110: 1d10.
McPherson A (1997) Recent advances in the microgravity
crystallization of biological macromolecules. Trends in
Biotechnology 15: 197d200.
Robert MC, Provost K and Lefaucheux F (1992) Crystalli-
zation in gels and related methods. In: Ducruix A and
Giedge R (eds) Crystallization of Nucleic Acids and
Proteins. A Practical Approach. Oxford: IRL Press,
Oxford University Press.
Rosenberger F, Vekilov PG, Muschol M and Thomas BR
(1996) Nucleation and crystallization of globular pro-
teins } what we know and what is missing. Journal of
Crystal Growth 168: 1d27.
Scopes RK (1987) Protein PuriTcation. Principle and Prac-
tice. New York: Springer-Verlag.
The abnormal prion protein is insoluble in most
biological buffers, whereas the normal prion protein
is soluble. In natural infections, the abnormal
prion protein is found in very low concentrations.
It is found in higher amounts in rodent-adapted
strains of the disease. These properties of insolubility
and low concentrations present quite a challenge
for the development of analytical methods to detect
this protein.
Most of the methods used to detect the prion pro-
tein are based on histological techniques and are used
postmortem. Immunoassays can be used to measure
the prion protein. Most of the antibodies that have
been produced to the abnormal prion protein have
been made to the denatured form. Removing these
denaturants is a major problem in the development of
immunoassays. Western blot can be used to detect the
prion protein but cannot be easily automated. Some
new approaches using chemiluminescence and time-
resolved Suorescence in plate assays have been de-
veloped. The amount of prion protein detected by
these assays is in the range of picomoles or
'500 fmol. To improve sensitivity, we have ap-
proached this problem using capillary electrophoresis
with laser-induced Suorescence. Fluorescent-labelled
peptides from immunogenic epitopes of the prion
protein can be detected in the attomole range using
this technique. A competition immunoassay using
Suorescein-labelled peptides was developed which is

able to detect the abnormal prion protein in the low
picogram range; this method can quantitate the
amount of prion protein. This assay was based on the
separation of the free peptide from the immunocom-
plexed peptide. Unlike most immunoassays which
measure only the antibody-bound ligand, both the
free and the bound peptide can be measured.
Method Development
Preparation of Tissues
Scrapie-infected tissues including brain, lymph node
and buffy coats were obtained from sheep conRrmed
positive for abnormal prion protein by Western blot.
Normal tissues were obtained from sheep from
a scrapie-free Sock and were conRrmed negative by
the above tests. BrieSy, the tissues were weighed, and
ground to a Rne powder in liquid nitrogen. Buffy
coats were prepared from blood and placed in 2 mL
of 20 mmol L\1 phosphate pH 7.0, 0.15 mol L\1
NaCl, and frozen at !703C until they were pro-
cessed as the tissues. After grinding, the tissues were
placed in 20 mmol L\1 Tris pH 7.4, 0.15 mol L\1
NaCl, 0.005 mol L\1 MgCl2 (10% w/v) and incu-
bated at 373C for 1 h in 50
g mL\1 DNase. After
incubation with DNase the tissue homogenates were
treated with proteinase K (50
g mL\1) for 1 h at
373C and held overnight at 43C. Sodium N-lauroyl-
sarcosine was added to the homogenate to make the
solution 10% in the detergent. The homogenate was
incubated for 1 h at 373C and then was centrifuged at
10 000 g for 20 min to remove particulates. The res-
ultant supernatant Suid was centrifuged at 230 000 g
for 1 h. The pellet was resuspended in 10 mmol L\1
Tris pH 7.4 (250
L g\1 of the initial brain sample).
The sample was solubilized in 0.01 mmol L\1 Tris
HCl, pH 8 containing 2 mmol L\1 ethylenediamine-
tetraacetic acid 5% SDS and 10% hexaSuoro-2-pro-
panol at 1003C for 10 min.
Chromatography
To remove the SDS, the sample was applied to a poly-
HYDROXYETHYL A (PolyLC, Inc., Columbia,
MD, USA) high performance liquid chromatography
column (200;4.6 mm) in 95% acetonitrile, 5%
water containing 0.1% triSuoroacetic acid and
50 mmol L\1 hexaSuoro-2-propanol (buffer A). The
Sow rate was 0.5 mL min\1. The conditions for elut-
ing abnormal prion protein were buffer A for 8 min
and then a linear gradient to 100% water containing
0.1% triSuoracetic acid and 50 mmol L\1 hexa-
Suoro-2-propanol (buffer B) in 15 min, 100% buffer
B for 10 min. Peak fractions were collected and dried
in a vacuum centrifuge. Fractions were resuspended
III / PROTEINS / Electrophoresis 4027
in 10
L of dH2O and tested for abnormal prion
protein in the capillary electrophoresis assay.
Peptide Synthesis and Antibody Preparation
Four peptides from the prion protein were synthesized.
The peptide sequences were CGQGGGTHNQWNKPSL
(spanning amino acid positions 89}103), CNDWED-
RYYRENMYR (142}154), (CRYPNQVYYRPVDRYSNQNNFVHD
(155}177) and RESQAYYQRGASVIL (218}232) (Multiple
Peptide Systems, San Diego, CA, USA). The peptides
were labelled with Suorescein through a -butyric
acid linkage on the N-terminus during synthesis. The
peptide 218}232 is used here as a representative
sample.
Rabbits were immunized with each peptide and
speciRc antibodies were produced for each peptide.
These antiserums reacted with scrapie-infected brain
but not with normal brain on Western blot analysis.
Rabbit IgG was prepared by passing each antiserum
over an afRnity column. BrieSy, 10 mg of a peptide
was coupled to agarose resin modiRed with an N-
hydroxyl succinimide ester in 1.0 mL dimethyl for-
mamide at room temperature for 20 min. After coup-
ling, the resin was washed with 5 mL of 0.1 mol L\1
3-(N-morpholino)propanesulfonic acid (MOPS) pH
7.5 (column wash buffer). Unreacted ester groups
were deactivated with 0.1 mol L\1 N-(2-
hydroxyethyl)piperazine-N-(4-butanesulfonic acid)
(HEPBS) pH 8.0 and 0.1 mol L\1 NH4Cl for 15 min.
Before antibodies were applied to the peptide column
they were puriRed using protein G chromatography.
After diluting 1 : 2 in column wash buffer, the anti-
bodies were then applied to the afRnity column and
recycled over the column for several cycles. The col-
umn was washed with column wash buffer and the
antibodies eluted with 0.1 mol L\1 NaH2PO4, pH 2.5
into tubes containing 50
L of 1 mol L\1 HEPBS, pH
8.5. The absorbance was measured at 280 nm. Frac-
tions with absorption at 280 nm were pooled,
aliquoted and frozen at !203C.
Immunocomplex Formation
Fifteen microlitres containing &0.2}20 pmol of the
Suorescent-labelled peptide were mixed with varying
amounts (0.5}5
g) of puriRed rabbit IgG to deter-
mine the antibody concentration that forms &50%
of the total immune complex formation. The Rnal
volume of the sample was adjusted to 20
L with
capillary running buffer. After mixing the compo-
nents, the samples were incubated at 253C for
&10 min and at 43C overnight. The height of the
immune complex peak was measured and replicate
samples of the peaks varied less than 1%. Slight
changes in the antibody reactivity, temperature and
4028 III / PROTEINS / Electrophoresis

Figure 1 A chromatogram showing hydrophilic interaction chromatography on poly-HYDROXETHYL A. The peak that was positive
for abnormal prion protein (PrP) is indicated. The gradient conditions for running the column are shown by the dashed line.

preparation of the running buffer caused small vari-
ations in the height of the immunocomplex peak from
day to day. Known concentrations of unlabelled pep-
tides corresponding to the Suorescent-labelled pep-
tides were used to generate standard curves.
Free Zone Capillary Electrophoresis
Free zone capillary electrophoresis was performed on
a Beckman P/ACE 5500 controlled by P/ACE Station
software. Laser-induced Suorescence detection was
done using an air-cooled argon laser with excitation
at 488 nm and emission at 520 nm. An unmodiRed
capillary 20 cm (length to the detector) ;20
m i.d.,
total length 27 cm capillary was used with
a 200 mmol L\1 Tricine buffer that was adjusted to
pH 8.0 by 6 mol L\1 NaOH. This buffer was selected
after studying the effect of higher pHs and other
concentrations of the buffer on the separation, im-
munocomplex formation and Suorescence. To pre-
vent the abnormal prion protein from adhering to the

Figure 2 An electropherogram showing the immunocomplex peak for the fluorescein-labelled peptide 218}232 and the free peptide
peak.
 III / PROTEINS / Electrophoresis 4029

Figure 3 A plot of the peak height of the immunocomplex peak versus the amount of antibody added to the assay.

capillary walls, 0.1% n-octylglucoside (Boehringer Results

Mannheim, Indianapolis, IN, USA) and 0.1% bovine
serum albumin (Sigma Chemical Co., St. Louis, MO,
USA) were added to the buffer. In preparation for
separation, the capillary was rinsed for 1 min with
0.25 mol L\1 NaOH, rinsed for 2 min with water
and then rinsed for 2 min with buffer. The separating
conditions were 30 kV for 3 min at 203C with a cur-
rent of &20
A. The sample was injected for 15 s
followed by a 5 s injection of running buffer. The
sample volume was &0.95 nL. Rinses were carried
out under high pressure and sample injection carried
out under low pressure.
Hydrophilic Interaction Chromatography
Figure 1 shows a chromatogram of the results of
hydrophilic interaction chromatography. This chro-
matography removes the SDS and other interfering
compound so that the reproducibility of the assay is
improved.
Capillary Electrophoresis Immunoassay
By the addition of Suorescein at the amino terminal
during synthesis, the sensitivity of this assay is en-
hanced 100-fold relative to the chemical addition of

Figure 4 Plot of the ratio of the height of the immunocomplex peak/height of the free peptide peak versus the amount of unlabelled
peptide added to the assay.
4030 III / PROTEINS / Electrophoresis

Figure 5 Representative electropherograms of antibody control (continuous line), normal brain sample (dotted line) and scrapie-
infected brain sample (dashed line).

Suorescein after synthesis of the peptide. An elec-
tropherogram showing the peptide and the im-
munocomplex peak when antibody is added to the
assay is shown in Figure 2. A titration curve of antibody
amount versus the height of the immunocomplex peak
is shown in Figure 3. The amount of antibody that
binds &50% of the Suorescein-labelled peptide was
chosen as the amount to be used in the competition
assays of the abnormal prion protein with the labelled
peptide for binding sites on the speciRc antibody.
Competition is determined by measuring the ratio of
the height of the immunocomplex peak and of the
free peptide peak. A standard curve was determined
by adding known amounts of unlabelled peptide into
the assay. This curve is shown in Figure 4. Elec-
tropherograms representing the immunocomplex peak,
a preparation from a normal sheep and a preparation
from a scrapie-infected sheep are shown in Figure 5
(peptide 218}232). An electropherogram represent-
ing a sample from a lymph node of an infected sheep
is shown in Figure 6. Figure 7 depicts three
electropherograms showing the antibody control,

Figure 6 Representative electropherograms of antibody control (dashed line) and a sample from an infected lymph node
(continuous line).

Figure 7 Representative electropherograms of (A) antibody control; (B) buffy coat from a scrapie-negative sheep; (C) buffy coat from
a scrapie-positive sheep.
samples extracted from buffy coats of a normal sheep
and from a buffy coat of a scrapie-infected sheep.
Concluding Remarks
The capillary electrophoresis assay described in this
study is reproducible, more sensitive and faster than
other analytical tests. The samples used in the capil-
lary electrophoresis assay were obtained from brain
and the lymphoid system of the animals. The sensitiv-
ity of this assay made it possible to test samples from
other tissues that contain much less abnormal prion
protein than brain samples. This assay has the poten-
tial to use tissues and Suids from live animals and
diagnose animals prior to the onset of clinical signs of
disease. Automation of this test could lead to more
economical and efRcient methods for testing for ab-
normal prion protein.
Further Reading
Altria KD (ed.) (1996) Capillary Electrophoresis Guide-
book: Principles, Operation and Applications. Totowa,
New Jersey: Humana Press.
Field Flow Fractionation
R. Hecker and H. Coi lfen, Max-Planck-Institut fuR r
Kolloid und GrenzflaR chenforschung (Kolloidchemie),
Am MuR hlenberg, Golm, Germany
Copyright ^ 2000 Academic Press
III / PROTEINS / Field Flow Fractionation 4031
Landers JP (ed.) (1997) Handbook of Capillary
Electrophoresis, 2nd ed. London: CRC Press.
Prusiner SB (ed.) (1996) Prions, Prions, Prions.
Current Top. Microbiol. Immunol. vol. 207. Berlin:
Springer.
Prusiner SB (1996) Prion biology and diseases}laughing
cannibals, mad cows, and scientiRc heresy. Medical Re-
search Review 16: 487}505.
Prusiner SB (1997) Prion diseases and the BSE crisis.
Science 278: 245}251.
Schmerr MJ and Jenny AL (1998) A diagnostic test for
scrapie-infected sheep using a capillary electrophoresis
immunoassay with Suorescent-labelled peptides, Elec-
trophoresis 19: 409}414.
Schmerr MJ et al. (1999) Use of capillary eletrophoresis
and Suorescent labeled peptides to detect the abnormal
prion protein in the blood of animals that are
infected with a transmissible spongiform en-
cephalopathy. Journal of Chromatography A 853:
207}214.
Weissmann C (1996) The ninth Datta lecture. Molecular
biology of transmissible spongiform encephalopathies.
FEBS Letters 289: 3}11.
Introduction
This review focuses on the use of Reld-Sow fractiona-
tion (FFF) for the characterization of proteins and
protein assemblies such as protein aggregates, DNA
4032 III / PROTEINS / Field Flow Fractionation

and viruses. FFF is based on the differential transport
rates of solutes in a ribbon-like channel when interac-
ting with an applied Reld. The type of Reld may be
chosen from a wide range, for example an electrical
potential, sedimentation, a hydrodynamic cross-Sow,
a thermal gradient and so forth. A schematic of this is
shown in Figure 1. The solute will therefore occupy
a region above the sample wall, with a mean position
determined by the balance between the solutes diffu-
sion and the sample}applied Reld interaction. Al-
though there exist further complications for solutes
greater than &0.5
m diameter, they are not relevant
given the small hydrodynamic diameter of proteins.
Positioned at the outlet of the channel is a sample
detector of some sort, typically a traditional
high-performance liquid chromatography (HPLC)
spectrophotometric detector, although a signiRcant
development has been with the application of a num-
ber of detectors providing complementary informa-
tion about the sample. Such detectors include
spectophotometric and refractive index types, and
more recently light scattering for molecular mass,
electrospray}mass spectrometry, and inductively
coupled plasma, although the last two have not yet
been applied to protein studies.
There are a number of advantages offered by the
FFF methods over other contemporary protein analy-
sis methods. FFF is often more rapid than analytical
ultracentrifugation, and the range of Relds available
provide FFF with greater versatility. In comparison
with gel-permeation chromatography, FFF is not im-
peded by a size exclusion limit, the low exposed
surface area limits sample loss through adsorption on
to the exposed surface, and the availability of
Reld programming allows a wide range of materials
to be analysed in a single channel. The open channel
geometry usually allows FFF to characterize samples
without need for pretreatment, such as Rltration, and
provides a very high upper limit to the protein size
range. Similarly, the open channel allows the theoret-
ical basis of FFF to provide direct access to funda-
mental physical constants of proteins, often without
the need for calibration. Finally, both FFF and gel
electrophoresis may separate a protein mixture, but
sample collection is simpler in FFF.
Flow FFF
The universal nature of (cross)-Sow FFF (Fl-FFF)
has led to its wide use for protein characterization.
The free choice of carrier liquid, whether a buffer or
a simulated native environment, avoids denaturing
the protein. Flow FFF has two conRgurations, the
original symmetric form and the newer asymmetric
method, differing only how the Reld is generated in
the fractionation cell and sample-loading protocols.

Figure 1 (A) Schematic of the mechanism of FFF separation of proteins. The smaller protein, with greater diffusivity, competes more
successfully with the applied field and occupies a mean position further from the accumulation wall. Samples occupying the higher
mean position is subject to more rapid flow laminae, and elutes earlier. The particle sizes represented, the channel thickness and the
extent of back-diffusion are not to scale. (B) For the subtechnique flow FFF, the accumulation wall is a liquid-permeable porous
material, typically ceramic. A membrane exists over the accumulation wall to prevent the samples from leaving the cell through this wall.
The upper wall may or may not be porous as well, depending whether the symmetrical or asymmetrical variant is used.
 III / PROTEINS / Field Flow Fractionation 4033

Table 1 Compilation of flow FFF physicochemical data relevant for selected common proteins and with comparison to commercial
polystyrene latexes

Sample Molecular mass (Da) Diffusion coefficient

(;1011 m2 s\1, ;107 cm2 s\1)

Cytochrome c (bovine heart) 13 400 11.4

Ovalbumin (chicken egg) 45 000 8.71
Bovine serum albumin 64 000 6.89
Catalase (horse liver) 221 000 4.30
Apoferritin 450 000 3.84
Urease 483 000 3.46
Ferritin 622 000 2.91
Tobacco mosaic virus +40 000 000 0.46
Polystyrene latex, H 0.090
m 0.45
Polystyrene latex, H 0.311
m 0.22

In both cases, the separation is a direct function of the than 15 min. One further advantage of the focusing
diffusion coefRcient, where the most highly diffusive method is the immobilization of the sample prior to
components are the least retained. A compilation of elution. For a very dilute sample, multiple injections
common biological samples and their diffusion coefR- subject to these opposing Sows produce an on-chan-
cients are provided in Table 1. nel concentrating effect, where the protein is retained
Fl-FFF is capable of separating proteins with only on the membrane at the focus point.
a 15% size difference within 3}10 min. Reported During any form of chromatography, sample dilu-
results for animal proteins and biopolymers include tion is inevitable. For small quantities of proteins this
albumins (human and bovine serum, egg), globulins may challenge the limits of the detectors used. FFF
(
-globulin, haemoglobin, thyroglobulin), ferritin, offers an advantage over other methods through the
apoferritin, lysozyme, casein, blood products (human ability to skim off the atmosphere of carrier liquid
and rat blood plasmas, lipoproteins) and nucleic and greatly reduce sample dilution before detection.
acids. Proteins from an industrial perspective are rep- Sample enhancement was Rrst mentioned in the liter-
resented by a growing body of work emerging on the ature in the early 1990s and now enjoys routine use in
characterization of proteins from Sours used for
bread-making purposes.
In all of the above cases, no sample treatment is
needed prior to injection, such as exhaustive dialysis
or Rltration. This is to be expected, as the permeable
membrane acts as a dialysis cell, and the open channel
will not become clogged and require a Rlter. Since the
sample is not manipulated beforehand, the presence
of aggregate structures remains unaltered. Figure 2
shows baseline resolution of a biological mixture.
Protein dimers elute as satellite peaks at &1.4 reten-
tion times of the monomer, followed similarly by the
higher aggregates eluting later. Most signiRcantly, the
entire separation takes place in only four minutes.
The asymmetric Sow FFF variant does not inject
the sample directly into the inlet line. Rather,
a sample pump introduces the sample into the cell
and opposing Sows from both ends of the cell hy-
drodynamically focus the sample into a narrow band
across the channel before elution. This allows for
remarkably well-resolved and efRcient protein separ-
ations. Figure 3 illustrates the sensitivity of the tech- Figure 2 Separation of a monoclonal antibody from its higher
clusters showing separable peaks up to pentameter aggregation.
nique. Two plasmid fragments were injected at low (Reproduced with permission from Giddings (1993) Science
concentration (0.1
g
L\1) and volume (1
L) while 260: 1456, Copyright the American Association for the Advance-
exhibiting both baseline resolution and elution in less ment of Science.)
4034 III / PROTEINS / Field Flow Fractionation
Figure 3 Separation of (1) 2390 bp and (2) 4320 bp plasmids
by asymmetrical FFF. (Reproduced with permission from LitzeH n
A and Wahlund KG (1989) Journal of Chromatography 476: 413
Copyright Elsevier Science BV.)
contemporary practice. Both symmetrical and asym-
metrical variants have been successfully applied to
proteins. Of special interest is the frit inlet}frit outlet
modiRcation. These methods in tandem enhance de-
tectability and aid fractionation stability. The combi-
nation of frit inlet and outlet has been reported as
recently as 1999, for the automation of wheat protein
fractionation.
One rarely discussed drawback to the Fl-FFF
method is the requirement of a membrane for sample
retention. For adhesive protein samples, this demands
compatibility between the sample, membrane and the
carrier solution. Biopolymers can strongly adsorb on
to particular membranes and at modest ionic
strengths may be completely adsorbed. The simplest
method to test this is to inject samples over a range of
concentrations and/or volumes and ensure there is
Figure 4 Superposition of two elution profiles for cytochrome c (0.82 mg mL\1, 25
L) in 0.05 mol L\1 2-[N-morpholino]propanesul-
fonic acid (mops) buffer at pH 6.2. The membranes are both regenerated cellulose, with cited 10 000 molecular weight nominal pore
sizes from different suppliers. From Hecker, unpublished results.
proportionality between detected signal size and the
amount of sample. A partial, reversible adsorption
leads to an increased retention and this would indi-
cate that the sample is erroneously large, or induce
a number of fractionation proRle artefacts. Clearly
the chemistry of the system, between the sample,
membrane and carrier, must be known before any
statements may be made.
The membranes physical characteristics may also
be signiRcant. Firstly, membrane compressibility and
protrusion into the channel reduce the channel thick-
ness and elution is more rapid, although this is easily
detected by measuring the channel void volume with
an unretained probe. More subtle effects include sur-
face roughness and membrane pore size, as demon-
strated by Figure 4. Although the experimental
arrangement, carrier sample chemistry and Sow rates
are the same for both experiments, the effect of the
membrane is clear. The molecular mass of cyto-
chrome c is only just greater than the membrane size
cut-off (12 500 versus 10 000), and the delayed reten-
tion from the poor membrane may be attributed to
a partial physical entrapment in the pores. For the
poor membrane the pore size distribution may be
particularly wide, leading to a greater proportion of
the sample suffering excessive retention. The mean
size and size distribution of the pores of the mem-
brane are clearly an issue of importance. A simple
solution is to choose a much Rner membrane, for
example a cut-off at 3000 or 5000 is appropriate for
cytochrome c, but the pressure drop across the chan-
nel may be incompatible with high Reld Sow rates
needed for sufRcient retention of small species. The

presence of the membrane therefore determines the
smallest-sized species capable of being retained in
a Fl-FFF channel.
Such membrane effects have been used to advant-
age, however. Proteins have been characterized with
a separation based on both standard FFF principles
and enhanced retention for some species by
sample}membrane interactions. This offers a remark-
ably wide scope for characterizing systems with
subtle differences in physical sizes but dissimilar
chemistries, but assigning peaks in the fractionation
proRle calls for a number of pure standards and
calibration processes.
Of particular interest to protein science is the ob-
servation and quantiRcation of protein}ligand or pro-
tein}protein interactions. Such an example is
provided in Figure 5 for the interaction between im-
munoglobulin IgC and an interacting ligand, poly-
glutamic acid, with the conjugate peak showing
a small amount of free ligand. Quantifying such an
interaction to measure the binding constant is a more
difRcult task. It is necessary to be able to produce
fractionation proRles of the components as a function
of concentration, implying that sample loss on to the
membrane must be prevented. Furthermore, at least
two from the protein, ligand or complex peaks must
be well separated for quantiRcation if the
stochiometry is known prior to the experiment,
otherwise all three must be resolved. This precludes
Figure 5 Elution profiles of the components of a protein}poly-
mer ligand mixture, immunoglobulin IgG and polyglutamic acid,
and their covalent conjugate. The fractionation of the conjugate
suggests that a quantity of the polyglutamic acid remains un-
bound, and offers a method of determining the binding constants
of such mixtures. (Reproduced with permission from Giddings JC
et al. (1992) Journal of Liquid Chromatography 15: 1729 Copy-
right Marcel Dekker.)
III / PROTEINS / Field Flow Fractionation 4035
many simple systems, for example bovine serum al-
bumin (BSA)/anti-BSA, or ovalbumin/concavalin A,
where the hydrodynamic sizes of these species are too
similar for reliable quantiRcation.
The application for protein interaction studies is
limited to processes in which the interaction time is
insigniRcant compared to the transport time, effec-
tively making protein studies with a kinetic barrier to
interaction difRcult. Further, the use of FFF to investi-
gate sample}sample interactions has been criticized,
in that during transport dilution will occur so equilib-
rium in the FFF channel will be different to that of the
mixing conditions. These limitations are clearly not
relevant for rapid, near-irreversible interactions.
The opportunity for the study of protein shape
by Fl-FFF is possible. Like other hydrodynamic
methods, the information available from these
methods renders them primarily as complementary
methods to high resolution crystallography or mag-
netic resonance. Nevertheless, both theory and prac-
tice, discussed by CoK lfen and Pauck, demonstrate that
retention is a function of molecular shape (Figure 6),
with the retention decreasing with the degree of
asymmetry.
All these examples show that Fl-FFF is a powerful
technique for protein characterization, as it is both
very rapid and requires only microgram or smaller
amounts of sample. Future potential can be seen in
the quantiRcation of interactions between proteins.
However, potential factors affecting the results and
possibly producing artefacts, such as membrane}
sample interaction or sample shape, must be
considered when interpreting the results.
Sedimentation FFF
The technique of sedimentation FFF balances the
back-diffusion of the sample against sedimenting for-
ces, a function of the samples hydrodynamic dia-
meter, density difference and the rotation rate
applied. Sedimentation FFF offers signiRcantly
greater size-based sensitivity over Fl-FFF, with a cor-
responding greater resolution. The method is also free
of the complications arising from the membrane re-
quired by Fl-FFF, although the possibility of electros-
tatic effects between the sample and cell cannot be
ruled out. Unfortunately for protein applications,
where the hydrodynamic diameters are of the order of
a few nanometers and sample density is close to that
of the buffering liquid, the rotation rate of the chan-
nel, and thereby the applied force Reld, must be high.
None the less, successful application of the sedi-
mentation FFF method to the characterization of
biopolymers has been reported. The samples of inter-
est tend to be among the larger biopolymers, and
4036 III / PROTEINS / Field Flow Fractionation

Figure 6 Temperature-corrected diffusion coefficients for a variety of proteins, using both analytical ultracentrifugation (A) and
asymmetric FFF (B). The molecular weight}diffusion coefficient relationship is linear for the globular proteins, represented as open
circles. Less spherical samples (filled circles) show a deviation from the linearity, with increasing deviation with eccentricity.
(Reproduced with permission from Pauck T and CoK lfen H (1998) Analytical Chemistry 70: 3886 Copyright the American Chemical
Society.)

reported examples include DNA, proteoglycans, and amylopectin, in dimethylsulfoxide as carrier

Rbrinogen and myohemerythrin.
Thermal FFF
Thermal FFF, employing the Soret effect, is also suit-
able for the separation of biomolecules. Unfortunate-
ly, the thermodiffusion effect is extremely poor in
water. The use of organic solvents restricts statements
about the native state in aqueous-based buffer, and
furthermore extensive conformational changes and
even denaturation may occur which signiRcantly re-
strict the range of applicable samples. Reported uses
of thermal FFF for biological samples have been lim-
ited to the polysaccharides, dextrans, Rcolls, pul-
lulans and cellulose, and the starch polymers amylose
liquid. Partially aqueous carriers have been investi-
gated but it is the fraction in organic solvent that
explicitly determines retention.
Electrical and Magnetic FFF
Electrical FFF is a subtechnique devoted to the frac-
tionation of proteins, as reSected in the number of
examples with protein applications. The narrow
channel leads to high electrophoretic gradients across
the cell, so samples with similar electrophoretic
mobilities and differences in diffusion may be separ-
ated. As such, electrical FFF exists as a complement to
electrophoresis. As early as 1972, a paper by Cald-
well et al. Rrst demonstrated the possibilities of

electrical FFF for the separation of albumin, ly-
sozyme, haemoglobin and
-globulin in buffer solu-
tions at different pH.
Later, the performance of an electrical FFF channel
with Sexible membranes, a channel with rigid mem-
branes and a circular channel for the separation of
proteins was described. In these studies, human and
bovine serum albumin, bovine
-globulin, cyto-
chrome c, egg white lysozyme and soluble ribonucleic
acid (t-RNA) as well as denatured proteins were suc-
cessfully separated. Unfortunately, the electrical Reld
induces charge polarization of carrier liquid species,
such that they migrated adjacent to the electrodes and
then screen the electrical Reld. These early experi-
Figure 7 The coating of streptavidin on to a standard 165 nm
diameter polystyrene (PS) latex bead affects the elution of the
latex substrate by electrical FFF. Under pH 7.2 fractionation
conditions the latex has a negative surface charge while the
protein is isoelectric. The lower net surface potential is reflected in
the poorer retention of the coated bead (A). The magnitude of this
peak shift quantifies the degree of surface coating, as shown by
the correlation in retention with the protein adsorption isotherm
(B). (Reproduced with permission from Schimpf and Caldwell
(1995) American Laboratory 27: 64I68.)
III / PROTEINS / Field Flow Fractionation 4037
mental conRgurations of electrical FFF utilized ion-
permeable membranes separating the channel volume
from the electrode compartments. These conditions
led to difRculties in forming a homogeneous electric
Reld, and from the late 1970s the technique entered
a period of quiescence. Results published in the early
to mid 1990s using conductive, rigid walls of either
graphite or gold-plated glass, have allowed reproduc-
ible separations, while the addition of a redox couple
in the carrier liquid, such as quinone-hydroquinone,
reduced the polarization effects. Due to these delays
in experimental development, electrical FFF is less
mature than other FFF techniques.
Electrical FFF is also well suited to measuring pro-
tein adsorption on to surfaces. The thin layer provides
only subtle differences to the hydrodynamic size and
net density, making Sow or sedimentation FFF analy-
sis difRcult. However, the adsorption dramatically
inSuences the surface charge and thereby inSuences
both sample}Reld interaction and retention, as shown
in Figure 7.
Although not formally FFF, dielectrophoresis in
combination with Suid Sow through an open chamber
with interdigitated sinusoidally corrugated electrodes
has been used for the separation of proteins and DNA.
A minor method, magnetic FFF, has been applied
to study the retention behaviour of BSA in the pres-
ence and absence of nickel nitrate. In the presence of
nickel ions, the retention time of the BSA sample was
6% higher with the magnetic Reld than it was without
the Reld. Retention times reported for BSA samples
both with and without a magnetic Reld did not differ
in the absence of Ni (II). However, the application
range of magnetic FFF for protein separations is very
limited, and the method can only be applied in excep-
tional conditions.
Micropreparative FFF Applications
A variant of the FFF apparatus, the split-Sow thin cell
(SPLITT) permits continuous separation of milli- or
even gram quantities of material. The apparatus is
similar to a FFF cell equipped with both frit inlets and
outlets. Initial conRgurations fed a mixture of large
particles into the upper wall and carrier liquid into
the base, while at the other end of the cell the liquid
Sowed out of two opposing exits. The ratio between
the Sows produces a hydrodynamic splitting plane
in the cell. During passage the larger particles could
sediment sufRciently to exit at the other end of the
SPLITT cell through the base, while smaller, less
dense particles did not pass the splitting plane and
eluted through the top. For protein applications,
an electrical potential applied across such a cell
in a range of buffers allows proteins with greater
4038 III / PROTEINS / Field Flow Fractionation
electrophoretic mobility to pass the splitting
plane. The separation of a mixture of model proteins
by such a method has been reported. The relatively
high throughput reported (15 mg h\1) makes
this an interesting development for routine puriRca-
tion, but it requires a difference in protein pI
of about two units as a necessary precondition for
separation.
Miniaturization of FFF
There is a drive to produce the equivalent of
hand-held devices for sample analysis based on the
FFF principles, the chip laboratory. Advantages
of such methods include the ability to analyse
freshly sampled, or to undertake a number of simulta-
neous parallel analyses. For such miniaturized
devices the injection volume is a signiRcant propor-
tion of the channel volume, with commensurate
band-broadening problems, while theory predicts
that some quantities, such as retention ratio and plate
height, degrade with decreasing size. None the less,
the reported developments for microfabricated
electrical and dielectrophoretic FFF show healthy
progression.
Concluding Remarks
The early development of FFF was hindered by the
experimental complexity of the method and a focus
on theory over practice. Over the last ten years,
a number of simplifying experimental features such as
the frit inlet}outlet system, and a fuller understand-
ing of the theoretical background have led to a
dramatic worldwide rise in the number of applica-
tions. It seems unlikely that more novel Relds will
be introduced into this family of techniques, but
the subtlety of application is increasing. Methods and
procedures are developing, from the analysis of
simple proteins and mixtures, to protein aggregates,
Glycoproteins: Liquid Chromatography
See III / GLYCOPROTEINS: LIQUID CHROMATOGRAPHY
proteins in complex matrices and increasingly
fragile samples such as liposomes, where the
open channel has few, if any, real analytical
competitors.
The other exciting branch of development is
increased commercial application, where the FFF
method becomes a black box technique. Leading
the way is the Fl-FFF method, but with the recent
innovations in electrical FFF, the dominance of
gel electrophoresis for protein analysis may be
passing.
See also: III/Proteins: Centrifugation.
Further Reading
CoK lfen H and Antionetti M (2000) Field-Sow-fractionation
techniques for polymer and colloid analysis. Advances in
Polymer Science 150: 67}187.
Giddings JC (1991) UniTed Separation Science. New York:
John Wiley.
Giddings JC (1993) Field-Sow fractionation: analysis of
macromolecular, colloidal, and particulate materials.
Science 260: 1456}1465.
Janca J (1988) Field-Flow Fractionation. Chromatographic
Science Series vol. 39. New York: Marcel Dekker.
Liu MK, Li P and Giddings JC (1993) Rapid protein separ-
ation and diffusion coefRcient measurement by frit inlet
Sow Reld-Sow fractionation. Protein Science 2:
1520}1531.
Martin M (1998) Theory of Reld-Sow fractionation.
Advances in Chromatography 39: 1}138.
Myers MN (1997) Overview of Reld-Sow fractionation.
Journal of Microcolumn Separations 9(3): 151}162.
Schimpf ME and Caldwell KD (1995) Electrical Reld-Sow
fractionation for colloid and particle analysis. American
Laboratory 27(6): 64}68.
Wahlund K-G and LitzeH n A (1989) Application of an
asymmetric Sow Reld-Sow fractionation channel to
the separation and characterisation of proteins,
plasmids, plasmid fragments, polysaccharides, and
unicellular algae. Journal of Chromatography 461:
73}87.
 III / PROTEINS / High-Speed Countercurrent Chromatography
High-Speed Countercurrent Chromatography
Y. Shibuswa, Tokyo University of Pharmacy and Life
Science, Tokyo, Japan
Y. Ito, National Institutes of Health, Bethseda, MD, USA
Copyright ^ 2000 Academic Press
Introduction
Countercurrent chromatography (CCC) is essentially
a form of liquid}liquid partition chromatography. Its
unique feature among other chromatographic sys-
tems is derived from the fact that the method uses no
solid support and the stationary phase is retained in
the column with the aid of gravity or centrifugal
force. The method has been termed after two classic
partition techniques } countercurrent distribution
and liquid chromatography.
A great advance in the CCC technology was made
with the discovery of a new hydrodynamic phenom-
enon in a rotating coiled tube, which provided the
basis for developing a highly efRcient CCC system
called high-speed CCC (HSCCC). In the last decade,
types XL, XLL, XLLL and L cross-axis coil planet
centrifuges (CPCs) have been developed to perform
CCC with highly viscous polar solvent systems, such
as polyethylene glycol (PEG) potassium phosphate,
PEG dextran aqueous}aqueous two-phase systems.
The absence of a solid support eliminates various
complications that might arise from this in conven-
tional chromatographic systems and the CCC has the
ability to preserve the functional and enzymatic activ-
ity of proteins.
Apparatus
The cross-axis coil CPCs, which include types X and
L and their hybrids (see Figure 3 in the article
on Countercurrent chromatography), are used for
protein separation. These modiRed versions of the
Figure 1 Orientation of the column holder on the axis of rotation in five different types of the cross-axis coil planet centrifuges. ;, axis
of revolution; dashed line, axis of rotation; L, lateral shift; R, revolution radius.
4039
HSCCC centrifuge have a unique feature among the
CPC systems in that the system provides reliable
retention of the stationary phase for viscous polymer-
phase systems. Figure 1 shows Rve different types of
cross-axis CPCs. A series of studies has shown that
the stationary-phase retention is enhanced by
laterally shifting the position of the coil holder along
the holder shaft, apparently due to the asymmetry of
the laterally acting force Reld between the upper and
the lower halves of the rotating coil. The degree of the
lateral shift of the coil holder is conventionally ex-
pressed by L/R, where L is the distance from the
middle point of the rotary shaft to the coil holder and
R is the distance from the centrifuge axis to the holder
axis. Types XL, XLL, XLLL and L have been effec-
tively used for protein separations with various poly-
mer-phase systems. For example, the polymer-phase
system composed of PEG and potassium phosphate
has a relatively large difference in density between the
two phases, so it can be retained in XL-XLL column
positions which provide efRcient mixing of the two
phases. On the other hand, the viscous polymer-phase
system composed of PEG and dextran has an ex-
tremely low interfacial tension and small density dif-
ferences between the two phases so that they tend to
emulsify under vigorous mixing. Therefore, the type
XLLL or L column position, that provides less efR-
cient mixing under a strong centrifugal force Reld, is
required to achieve satisfactory retention of the sta-
tionary phase for this polymer-phase system.
Figure 2 shows the XLLL cross-axis CPC
(L/R"3.5) equipped with a pair of multilayer coil
separation columns. Each column consists of 2.6 mm
i.d. polytetraSuoroethylene (PTFE) tubing wound on
to a coil holder hub, forming multiple layers of left-
handed coils. Table 1 lists various CPC models that
have been used for the preparative separation of pro-
teins, together with various parameters, including the
4040 III / PROTEINS / High-Speed Countercurrent Chromatography

Figure 2 Type XLLL cross-axis CPC equipped with a pair of multilayer coils connected in series.

dimensions of columns and column holders, and

values of multilayer coils.
Polymer-phase Systems for
Preparative Separation of Proteins
CCC utilizes a pair of immiscible solvent phases pre-
equilibrated in a separatory funnel where one phase is
used as the stationary phase and the other as the
mobile phase. There are two typical polymer-phase
systems available for protein separation: PEG dextran
and PEG potassium phosphate systems.
PEG Dextran Systems
The polymer-phase system composed of PEG and
dextran has a characteristic feature: small molecules
are partitioned fairly evenly between the two phases,
whereas macromolecules such as DNA and polynuc-
leic acids are distributed unilaterally in one phase or
the other, depending on the pH of the solvent system.
Consequently, the system can be used effectively for
separation of these macromolecules using pH gradi-
ent elution. The PEG dextran system forms two layers
without addition of high salt concentration, which
tends to be precipitated in PEG phosphate systems
(see below) at high salt concentrations. On the other
hand, the PEG dextran system has a serious drawback
in its CCC application. At high dextran concentra-
tions the viscosity of the lower phase increases, and
the similar polarity of the two polymers reduces inter-
facial tension between the two phases, resulting in
a high probability of emulsiRcation. A typical PEG
dextran polymer system contains 4.4% (w/w) PEG
8000, 7% (w/w) dextran T500 and 10 mmol L\1

Table 1 Type of apparatus and dimensions of columns used for protein separation

x-axis CPC (L /R)a Coil holder Columns

Diameter Width i.d. Length Layers Capacity
Values b

(cm) (cm) (mm) (m) (mL)

Type XL (1.25) 10.0 5.0 2.6 31 3 165 0.50}0.60

15.2 5.0 2.6 11 1 60 0.76
15.2 5.0 2.6 64 4 340 0.76}0.90
Type XLL (2.0) 3.8 5.0 2.6 47 8 250 0.25}0.60
7.6 5.0 2.6 53 6 280 0.50}1.00
Type XLLL (3.5) 3.8 5.0 2.6 57 9 300 0.50}1.30
3.4 5.0 2.6 83 12 440 0.45}1.50
Type L (infinity) 3.6 5.0 2.6 25 5 130 0.16}0.27c
a
L"Distance from the centre of the holder shaft to the coil holder; R"distance from the centrifuge axis to the holder shaft.
b

"r /R.
c

"r /L, where r is the distance from the holder axis to the coil.
 III / PROTEINS / High-Speed Countercurrent Chromatography 4041

Table 2 Preparation of polymer two-phase solvent systems most free from contamination by low-molecular-
weight impurities that either elute immediately after
Concentration (%, w /w) pH the mobile phase front or remain almost permanently
PEG 1000 K2HPO4 KH2PO4 in the column.
Table 2 shows the composition of seven different
1 12.5 12.5 0 9.0 PEG 1000 potassium phosphate systems. The ratio of
2 12.5 9.375 3.125 7.7 the monobasic and dibasic potassium phosphate de-
3 12.5 8.33 4.17 7.3
termines the pH of the solvent system and the parti-
4 16.0 12.5 0 9.2
5 16.0 10.4 2.1 8.0 tion coefRcient of the protein samples. In all these
6 16.0 8.33 4.17 7.3 solvent systems, the upper layer is rich in PEG and the
7 16.0 6.25 6.25 6.8 lower layer is rich in phosphate.
Pro\lin+actin Complex Puri\cation from
Crude Acanthamoeba Extract
potassium phosphate buffer at proper pH. This two-
phase system consists of a PEG-rich upper phase and Using the type L cross-axis CPC equipped with a pair
a dextran-rich lower phase. of multilayer coils (130 mL capacity), proRlin}actin
complex has been puriRed directly from an Acan-
PEG Potassium Phosphate Systems
thamoeba extract with a polymer-phase system com-
The PEG potassium phosphate system is com- posed of 4.4% (w/w) PEG 8000, 7% (w/w) dextran
plementary to the PEG dextran system in that it tends T500 at pH 6.8. The lower dextran-rich phase was
to distribute low-molecular-weight compounds uni- used as the stationary phase. The sample solution was
laterally in either the upper or lower phase while prepared by adding the correct amounts of PEG 8000
macromolecules such as proteins are more evenly and dextran T500 to 12.5 g of the Acanthamoeba
distributed between phases. Consequently, once a crude extract to adjust the two-phase composition to
suitable partition coefRcient for the target protein is that of the solvent phases used for separation. The
obtained, the system yields high-purity fractions al- separation was carried out by pumping the PEG-rich

Figure 3 Purification of profilin}actin complex from Acanthamoeba soluble extract. Experimental conditions: column is a 2.6 mm i.d.
PTFE multilayer coil;2,
"0.16}0.27; 130 mL capacity; sample consists of 12.5 g Acanthamoeba soluble extract; solvent system is
4.4% (w/w) PEG 8000, 7% (w/w) dextran T500 in a 10 mmol L\1 potassium phosphate buffer at pH 6.8; mobile phase is the upper
phase; flow rate: 0.5 mL min\1; revolution: 900 rpm; SF, solvent front.
4042 III / PROTEINS / High-Speed Countercurrent Chromatography

upper phase into the head of the column at
0.5 mL min\1 under a high-revolution speed of
1000 rpm. The results are shown in Figure 3A. The
solvent front emerged at the 14th fraction (3 mL per
fraction) whereas the proRlin}actin complex was
eluted in fractions 20 to 28, well separated from other
components. The impurities were mostly eluted later
with a retention volume close to the total column
capacity (around fractions 33}60), while some were
also found near the solvent front (fraction 15). Identi-
Rcation of the proRlin}actin complex was made
by 12% sodium dodecyl sulfate}polyacrylamide gel
electrophoresis (SDS-PAGE), as illustrated in
Figure 3B. The retention of the lower stationary
phase was 69% of the total column capacity.
Countercurrent Chromatographic Fractionation
of Lipoproteins from Human Serum
The performance of the XLL cross-axis CPC has been
evaluated by the direct separation of high- and low-
density lipoproteins (HDLs and LDLs) from human
serum. The effects of the molecular weight of the PEG
was studied with a polymer-phase system composed
of 16% (w/w) PEG, 12.5% (w/w) potassium phos-
phate. Figure 4 shows the chromatograms of
human serum (4 mL) obtained from four solvent
systems containing different molecular weight PEGs
(600, 1000, 2000 and 4000).
In each experiment, the CCC column was Rrst
entirely Rlled with the PEG-rich upper stationary
phase, and the sample solution (a mixture of 4 mL
human serum and 2 mL each of upper and lower
phases, to which the required amounts of PEG and
potassium phosphate were added to adjust the two-
phase composition) was injected through the sample
port. Then, the potassium phosphate-rich lower mo-
bile phase was eluted through the column at a Sow
rate of 2 mL min\1 while the apparatus was rotated
at 500 rpm. The lipoprotein fractions obtained in the
CCC were characterized using polyacrylamide gel disc
electrophoresis (disc PAGE). Serum proteins in the
CCC fractions were also characterized by SDS-PAGE.
In the PEG 600 system (Figure 4A), all proteins
including HDLs, LDLs and serum proteins were
strongly retained in the PEG-rich stationary phase
and eluted together when the column was eluted in

Figure 4 Countercurrent chromatographic fractionation of HDL-LDL and VLDL-serum protein fractions from human serum with four
different aqueous polymer-phase systems containing (A) PEG 600; (B) PEG 1000; (C) PEG 2000; (D) PEG 4000. Experimental
conditions: column is a 2.6 mm i.d. PTFE multilayer coil;2,
"0.76}0.90, 340 mL capacity; sample is a mixture of 4 mL volume of
human serum, 2 mL of the upper and the lower phases, to which the required amounts of PEG and potassium phosphate were added to
adjust the two-phase composition; solvent system consists of 16% (w/w) PEG 1000, 12.5% (w/w) K2HPO4 (pH 9.2); mobile phase is
the lower phase; flow rate: 2.0 mL min\1 revolution: 500 rpm; SF, solvent front; UP, starting point of the reversed elution mode with the
upper phase mobile.
 III / PROTEINS / High-Speed Countercurrent Chromatography 4043

a reversed elution mode with the PEG-rich upper

phase. Similarly, when PEGs with molecular weights
higher than 2000 were used in the solvent system, all
proteins including HDLs, LDLs and serum proteins
were mostly distributed to the potassium phosphate-
rich lower phase and eluted together at the solvent
front (Figure 4C and D). Successful separation of the
combined HDL and LDL fraction was achieved with
the 16% (w/w) PEG 1000, 12.5% (w/w) potassium
phosphate solvent system at pH 9.2, where both
HDLs and LDLs were eluted together near the solvent
front, while other proteins, including very-low-den-
sity lipoproteins (VLDLs) and serum proteins were
retained in the column for much longer. The separ-
ation time of these two lipoproteins was 3 h. The
VLDLs were eluted by the PEG-rich upper phase in
the second peak or its shoulder (Figure 4B).
These results show that both HDL-LDL and
VLDL-serum protein fractions were fractionated
within 3 h by CCC with a polymer-phase system
composed of 16% (w/w) PEG 1000 and 12.5%
(w/w) dibasic potassium phosphate at a relatively
high Sow rate of 2 mL min\1.
Puri\cation of HDLs, LDLs and VLDLs from
Human Serum by Combined Use of CCC and
Hydroxyapatite Chromatography
Figure 5 Countercurrent chromatographic separation of HDL-
In the previous section, two lipoprotein fractions LDL and VLDL-serum protein fractions from human serum by
small-capacity columns. Experimental conditions: column is
(HDL-LDL and VLDL-serum proteins) were ob-
a 2.6 mm i.d. PTFE multilayer coil;2,
"0.76, 60 mL capacity;
tained from human serum using a polymer-phase sample is a 4 mL volume of human serum containing 0.9 g PEG
system by the type XL cross-axis CPC equipped with 1000 and 0.7 g dibasic potassium phosphate; solvent system
a large-capacity column (340 mL). A small-capacity consists of 16% (w/w) PEG 1000, 12.5% (w/w) K2HPO4 (pH 9.2);
column (60 mL) mounted on the same apparatus can mobile phase is the lower phase; flow rate: 0.5 mL min\1; revol-
ution: 500 rpm; SF, solvent front; UP, starting point of the
be employed to shorten the separation time. Figure 5
reversed elution mode with the upper phase mobile.
shows a chromatogram of human serum (4 mL) ob-
tained with the cross-axis CPC using 16% (w/w) PEG
1000, 12.5% (w/w) dibasic potassium phosphate (pH buffer used for hydroxyapatite chromatography.
9.2). The separation was performed at 500 rpm and The concentrates of both fractions were chromato-
at a Sow rate of 0.5 mL min\1 using the lower phase graphed separately on the hydroxyapatite column.
as the mobile phase. Both HDLs and LDLs were Figure 6 shows the elution proRle on hydroxyapatite
eluted together near the solvent front, while other obtained from CCC-fr. 1. A 1.4 mL volume of the
proteins were retained in the column much longer. concentrate was loaded on the Bio-Gel HTP DNA-
After collecting the HDL-LDL fraction (CCC-fr. 1), grade column (5.0;2.5 cm i.d.) and eluted at
VLDLs were eluted together with serum proteins 1.0 mL min\1 with 75 and 290 mmol L\1 potassium
(CCC-fr. 2) by pumping the upper phase in the re- phosphate buffer at pH 7.4. Two lipoprotein peaks
verse direction. The separation was completed within were eluted: the Rrst peak (HA-fr. 1) contained HDLs
4.5 h. The lipoproteins in each peak were conRrmed and the second peak (HA-fr. 2) contained LDLs.
by agarose gel electrophoresis with Oil Red 7B stain- The concentrate (1.5 mL) of CCC-fr. 2 was sim-
ing, and the serum proteins were also detected by ilarly chromatographed (Figure 7). The separation
10% SDS-PAGE with Coomassie Brilliant Blue pro- was performed with two-step elution with 290 and
tein staining. 650 mmol L\1 potassium phosphate buffers at pH
The CCC fractions 1 (HDL-LDL) and 2 (VLDL- 7.4. Most of the serum proteins, including albumin
serum proteins) were each separately dialysed against and globulins, were eluted with 290 mmol L\1 potas-
distilled water until the concentration of the potassi- sium phosphate buffer (HA-fr. 3) at pH 7.4.
um phosphate was reduced to that in the starting The VLDLs, on the other hand, were retained in the
4044 III / PROTEINS / High-Speed Countercurrent Chromatography

Figure 6 Stepwise elution profile of HDLs and LDLs of the CCC fractions by hydroxyapatite chromatography. Experimental
conditions: column is Bio-Gel HTP DNA-grade hydroxyapatite (5.0;2.5 cm i.d.); eluents are 75 and 290 mmol L\1 potassium
phosphate buffers at pH 7.4; flow rate: 1.0 mL min\1 sample is the 1.4 mL concentrate of HDL-LDL CCC fraction containing 13.9 mg
total proteins (CCC-fr. 1).

column for much longer and were eluted with fractions were conRrmed by agarose gel electrophor-
650 mmol L\1 potassium phosphate buffer (HA-fr. 4). esis. The results of agarose gel electrophoresis
Lipoproteins in the hydroxyapatite chromatographic indicated that HDLs, LDLs and VLDLs were present

Figure 7 Stepwise elution profile of VLDL-serum proteins fraction of the CCC fractions by hydroxyapatite chromatography.
Experimental conditions: column is Bio-Gel HTP DNA-grade hydroxyapatite (5.0;2.5 cm i.d.); eluents are 290 and 650 mmol L\1
potassium phosphate buffers at pH 7.4; flow rate: 1.0 mL min\1; sample is the 1.5 mL concentrate of serum protein-VLDL CCC fraction
containing 41.8 mg of total proteins (CCC-fr. 2).
 III / PROTEINS / High-Speed Countercurrent Chromatography
Figure 8 Countercurrent chromatographic purification of purine
nucleoside phosphorylase (PNP) from crude extract of Escherichia
coli (A) and SDS gel electrophoresis profile of CCC fractions (B).
Experimental conditions: column is a 2.6 mm i.d. PTFE multilayer
coil;2,
"0.25}0.60, 250 mL capacity; sample consists of crude
PNP in 10 mL solvent; solvent system consists of 16% (w/w) PEG
1000, 6.25% (w/w) K2HPO4 6.25% (w/w) KH2PO4 (pH 6.8); mobile
phase is the upper phase; flow rate: 0.5 mL min\1; revolution:
750 rpm; SF, solvent front. Lane 1, molecular weights markers;
Lane 2, crude extract; Lane 3, HSCCC fraction 46 (solvent front);
Lane 4, HSCCC fraction 71 (213 mL).
in HA-fr. 1, HA-fr. 2 and HA-fr. 4, respectively.
From the results of SDS-PAGE of the hydroxyapatite
fractions, HA-frs. 1, 2 and 4 are free from serum
proteins and HA-fr. 3 contained only serum proteins.
Puri\cation of Recombinant Enzymes from
Crude Escherichia coli Lysate
The capability of the XLL cross-axis CPC was further
examined in the puriRcation of some recombinant
enzymes from a crude extract of Escherichia coli
lysate. The polymer-phase system used was 16%
4045
(w/w) PEG 1000, 6.25% (w/w) monobasic and
6.25% (w/w) dibasic potassium phosphate (pH 6.8).
The phosphate-rich lower phase was used as the sta-
tionary phase. About 1.0 mL of crude lysate contain-
ing purine nucleoside phosphorylase (PNP) in 10 mL
of the above solvent system was loaded into the
multilayer coil and eluted with the PEG-rich upper
phase at a Sow rate of 0.5 mL min\1. Figure 8A
shows the chromatogram of crude PNP lysate ob-
tained. A 3 mL volume was collected in each fraction.
The solvent front emerged at the 46th fraction
(138 mL retention volume) and puriRed PNP was
obtained from fractions 65}80 (195}240 mL).
Figure 8B shows the 12% SDS gel electrophoresis
patterns of the CCC fractions obtained from the
crude PNP lysate. Gel electrophoresis clearly demon-
strates that PNP in the crude E. coli lysate was highly
puriRed by CCC via a single pass through the column.
PuriRcation of recombinant uridine phosphorylase
(UrdPase) from E. coli lysate has been performed
similarly, as shown in Figure 9. The polymer phase
system was the same as that used for the puriRcation
of recombinant PNP described above. About 2.0 mL
of the crude lysate in 4 mL of the solvent, 1 mL of
upper phase and 3 mL of lower phase containing
16% PEG 1000 and 12.5% potassium phosphate,
was loaded on the column and eluted with the PEG-
rich upper phase at 0.5 mL min\1. In Figure 9, pro-
tein concentration in the eluted fractions (solid line) is
plotted against the retention volume. The chromato-
gram shows four protein peaks. Most of the protein
mass was eluted immediately after the solvent front in
fractions 35}55 (105}165 mL), whereas the enzyme
activity of the UrdPase coincides with the fourth
protein peak corresponding to fractions 75}95
(230}285 mL). These results indicate that recom-
binant UrdPase can be highly puriRed from the crude
E. coli lysate in a one-step operation within 10 h by
the XLL cross-axis CPC.
Puri\cation of Lactic Acid Dehydrogenase
from Bovine Heart Crude Extract
CCC has been applied to the puriRcation of lactic
acid dehydrogenase (LDH) from a crude bovine heart
Rltrate using the XL cross-axis CPC. The separation
was performed with a polymer-phase system, com-
posed of 16% (w/w) PEG 1000, 12.5% (w/w) potas-
sium phosphate at pH 7.3.
Figure 10A shows chromatograms of the bovine
heart crude extract obtained, where the PEG-rich
upper phase was used as the stationary phase. The
separation was performed at 500 rpm at a Sow rate
of 1.0 mL min\1 using the phosphate-rich lower
phase as the mobile phase. The enzymatic activity of
LDH was detected between the second and third
4046 III / PROTEINS / High-Speed Countercurrent Chromatography

Figure 9 Countercurrent chromatographic purification of uridine phosphorylase (UrdPase) from crude Escherichia coli lysate.
Experimental conditions: column is a 2.6 mm i.d. PTFE multilayer coil;2,
"0.25}0.60, 250 mL capacity; sample consists of 2 mL
crude UrdPase in 4 mL solvent; solvent system consists of 16% (w/w) PEG 1000, 6.25% (w/w) K2HPO4 and 6.25% (w/w) KH2PO4
(pH 6.8); mobile phase is the lower phase; flow rate: 0.5 mL min\1; revolution: 750 rpm; SF, solvent front.

peaks. These fractions were analysed by 12% (w/v)
SDS-PAGE with Coomassie Brilliant Blue staining
(Figure 10B), indicating that the LDH is actually con-
tained in 30 mL of eluent (fractions 140}170 mL)
without detectable contamination from other pro-
teins. The traditional techniques used for puriRcation
of LDH require several steps, including precipitation
with ammonium sulfate, centrifugation and dialysis;
hence they are very tedious and time-consuming. By
combined use of the XL cross-axis CPC and the
aqueous polymer-phase system described above,
LDH is puriRed within 3 h.
These results show that, with relatively simple
manipulation of several parameters (buffer, polymer
molecular mass, rotation speed), CCC is well suited to
the rapid puriRcation of enzymes from very crude

Figure 10 (See Colour Plate 116) (A) Countercurrent chromatography of bovine heart homogenate and (B) SDS-PAGE profile of the
fractions. Experimental conditions: column is a 2.6 mm i.d. PTFE multilayer coil;2,
"0.50}0.60, 165 mL capacity; sample is
a mixture of 3 mL bovine heart crude extract, 3 mL solvent system (1.5 mL each phase); solvent system consists of 16% (w/w) PEG
1000 12.5% (w/w) potassium phosphate (pH 7.3); mobile phase is the lower phase; flow rate: 2.0 mL min\1; revolution: 500 rpm; SF,
solvent front. LMW, low molecular weight protein markers.

tissue extracts. Because of the protective effect
of a high concentration of polymers and potassium
phosphate, the native structure of the proteins is
preserved at room temperature during separation,
and the support-free partitioning eliminates sample
loss and deactivation of enzymes which is often
caused by using the solid support in conventional
chromatography. We expect that these merits of
the method will apply in the puriRcation of other
enzymes.
Conclusion
The capability of the cross-axis CPCs for performing
CCC has been demonstrated in the separation and
puriRcation of proteins. The unique feature of
the apparatus is that it provides sufRcient retention
of the stationary phase for viscous, low interfacial
tension polar solvent systems, such as aqueous-
aqueous polymer phase systems. Consequently, the
method can be utilized for the fractionation of a wide
variety of proteins without adsorptive sample
loss and denaturation of proteins caused by the
solid support. The CCC method may be further
extended to the puriRcation and fractionation of
other biopolymers.
See Colour Plate 116.
See also: II/Chromatography: Countercurrent Chromato-
graphy and High-Speed Countercurrent Chromatography:
Instrumentation. Chromatography: Liquid: Countercur-
rent Liquid Chromatography. Appendix 1: Essential
Ion Exchange
P. R. Levison, Whatman International Ltd,
Maidstone, Kent, UK
Copyright ^ 2000 Academic Press
Introduction
Proteins are polymers of amino acids, the so-called
building blocks of nature and are found in all living
matter be it of animal, microbial or vegetable origin.
By their very structure proteins have an electrical
charge and can therefore be fractionated by ion ex-
change processes. This paper brieSy reviews the prin-
ciples underlying protein puriRcation by ion exchange
and addresses some of the process issues associated
with their puriRcation.
III / PROTEINS / Ion Exchange 4047
Guides for Isolation/Purification of Enzymes and
Proteins; Essential Guides for Isolation/Purification
of Immunoglobulins.
Further Reading
Albertsson P-A> (1986) Partition of Cell Particles and Mac-
romolecules, 3rd edn, New York: Wiley Interscience.
Conway WD (1990) Countercurrent Chromatography,
Apparatus, Theory and Applications. New York:
VCH.
Shibusawa Y (1996) Separation of proteins by high-
speed countercurrent chromatography. In: Ito Y
and Conway WD (eds) High-Speed Countercurrent
Chromatography, ch. 16, pp. 385}414. New York:
Wiley Interscience.
Shibusawa Y, Chiba T, Matsumoto U and Ito Y (1995)
Countercurrent chromatographic isolation of high-
and low-density lipoprotein fractions from human
serum. In: Conway WD and Petroski RJ (eds) Modern
Countercurrent Chromatography (ACS Monographs),
ch. 11, pp. 119}128. Washington, DC: American Chem-
ical Society.
Shibusawa Y, Mugiyama M, Matsumoto U and Ito Y
(1995) Complementary use of countercurrent
chromatography and hydroxyapatite chromatography
for the separation of three main classes of lipoproteins
from human serum. Journal of Chromatography,
Biomedical Applications 664: 295}301.
Shibusawa Y, Eriguchi Y and Ito Y (1997) Lactic
acid dehydrogenase puriRcation from bovine heart
crude extract by counter-current chromatography.
Journal of Chromatography, Biomedical Applications
696: 25}31.
Proteins
As living cells reproduce, genetic material is passed
from parent cells to daughter cells in the form of
DNA. DNA is a template coding for the various
proteins required for the developing organism. As
the organism grows, cells differentiate to form the
various organs of the mature organism. Each cell has
the capability to express every single protein of the
organism, but in life only a small fraction of proteins
are expressed. For example muscle cells produce actin
and myosin to facilitate movement, the pancreas pro-
duces chymotrypsinogen and trypsinogen to facilitate
digestion and lymphocytes are responsible for the
expression of immunoglobulins which provide im-
munity from infection and disease.
4048 III / PROTEINS / Ion Exchange
Because of their functional and structural roles in
nature, proteins have signiRcant commercial poten-
tial in many areas including food and beverage,
biological detergents, diagnostic enzymes, veterinary,
agricultural and pharmaceutical applications. How-
ever, because of their diversity, the challenges of their
puriRcation are immense and their isolation from
a particular tissue or organ, regardless of host, may be
regarded as searching for a needle in a haystack.
Protein Structure
Proteins are polymers of amino acids bonded together
through amide linkages. There exist 20 common
amino acids in nature ranging in molecular mass from
75 to just over 200 Da. Proteins range in molecular
mass from around 10 000 up to '1 000 000 Da, and
consequently their amino acid sequence or primary
sequence may be hundreds of residues in length. Of
the 20 amino acids, several have positively or nega-
tively charged side chains, while others have neutral
side chains, which may have hydrophilic or hydro-
phobic properties. The primary sequence of a protein
results in a zwitterion with the positively charged
N terminus balancing the negatively charged C termi-
nus. However, the charges of the side chains of the
charged amino acids and the pKa values of their
functional groups give, at least in principle, an overall
positive or negative charge at a given pH. However,
proteins are not simple structures and certain
sequences of amino acids fold to give secondary
structures such as helices and pleated sheets. This
secondary structure scrambles up to give a three-
dimensional tertiary structure. Some proteins exist as
an assembly of subunits giving a quaternary struc-
ture. Many proteins are glycosylated to aid with mo-
lecular recognition in vivo and this inSuences their
shape and surface properties.
The net effect of the three-dimensional structure of
proteins is that their theoretical charge or hydropho-
bicity based on a primary sequence bears little rela-
tion to the actual properties of the molecule in its
native state. If, for example, all the charged groups
are buried inside a pocket in the molecule, then its
response to an ion exchanger may be quite weak. The
three-dimensional structure of a protein is associated
with function and for the puriRed protein to have
intrinsic value, its three-dimensional structure should
be retained. This presents practical difRculties in
terms of puriRcation, because a denatured protein
may not readily, if at all, refold back to its native
state. For mammalian systems, typical physiological
conditions are pH 7.4 and 0.15 M NaCl and most
proteins would be stable and active under these con-
ditions. However, deviations in operating pH and, to
a lesser extent, ionic strength, may irreversibly denature
the protein of interest, which can severely restrict the
mode of puriRcation available to the chromatographer.
Methods of Protein Puri\cation
Prior to carrying out any practical studies, the protein
chemist is provided with a range of chromatographic
techniques, the use of which should enable effective
puriRcation to be achieved. Those techniques suitable
for low pressure operation include those listed in
Table 1. While all of these techniques are suitable for
laboratory-scale use, those typically scaled-up include
ion exchange, hydrophobic interaction, afRnity and
size exclusion. These techniques each exploit differ-
ing physicochemical properties of the protein molecu-
les as manifest by their three-dimensional structure.
Ion exchange chromatography and hydrophobic in-
teraction chromatography rely on electrostatic inter-
actions between a charged stationary phase and
charged surfaces of the protein or hydrophobic inter-
actions between a hydrophobic stationary phase and
hydrophobic surfaces of the protein respectively. Af-
Rnity separations rely on a biospeciRc interaction, for
example the interaction of an enzyme with its immo-
bilized substrate or an immunoglobulin with its im-
mobilized antigen. Size exclusion chromatography is
a molecular sieving effected by the three-dimensional
size and shape of the protein. One or more of these
techniques should be suitable for protein puriRcation
with their choice inSuenced by the selectivity require-
ments of the process in terms of both the target and
contaminants.
Ionic Properties of Proteins
For the reasons described above, all proteins will have
ionic properties, and their three-dimensional struc-
ture imparts a subtle uniqueness to their ionic charge,
which is available for exploitation by ion exchange
chromatography during their puriRcation. In a sim-
ilar manner to small molecules, such as organic acids,
which vary their charge with pH, as prescribed by
their pKa, proteins have an isoelectric point or pI. The
Table 1 Techniques available for low pressure chromato-
graphy of proteins
Salt precipitation
Ion exchange
Size exclusion
Hydrophobic interaction
Thiophilic interaction
Affinity
Chiral
Copyright ^ 1998 Whatman International Ltd., reproduced with
permission.

pI of a protein is the pH where it has a net charge
of zero.
When the pH is greater than the pI, the protein will
have a net negative charge and may bind to an anion
exchanger, provided that the pH is less than the pKa
of the functional group. If the pH is less than the pI
the protein will have a net positive charge and may
bind to a cation exchanger, provided the pH is greater
than the pKa of the functional group. As a rough
guide most proteins have a pI of less than 7, ranging
typically from 4.5 to 6.5. The major exceptions are
the immunoglobulins which typically have a pI of
greater than 7.
Ion Exchange Chromatography of
Proteins
Principles
On the basis of the physicochemical issues discussed
above, it might appear that by following a few simple
rules, centred around pH, an ion exchange separation
can be designed. The Rrst barrier to overcome is to
Rnd out the pI of the protein. Unless the protein is
well characterized, this may not be documented, and
although it can be readily determined by techniques
such as isoelectric focusing, this presupposes that it
can be obtained in a relatively pure state. A second
barrier is the pH and ionic strength stability ranges of
the protein. This can be determined readily by a para-
metric study, centred around a robust assay for the
protein, typically linked to its biological function.
A third, and often underestimated barrier, is the inSu-
ence of the other contaminants within the feedstream
and how they may impact on the efRciency of the ion
exchange separation. For example, competitive ad-
sorption of an unwanted contaminant to the adsor-
bent can signiRcantly hamper the selectivity of the
separation and the process economics. Other key con-
siderations include the mobile phase composition as
deRned by the preceding chromatographic step and
its impact on an ion exchange separation and the
eluent composition and its impact on the subsequent
downstream step.
Method Scouting
In the light of the issues highlighted above, the
chromatographer can start to develop the ion ex-
change process. A broad strategy for ion exchange is
represented in Figure 1. Intuitively, it would seem
reasonable to expect ion exchange steps to be of the
positive form, whereby the target is retained by the
exchanger, and assuming elution volume is less than
feedstock volume, has the potential to effect product
concentration. Such an approach may be preferable if
III / PROTEINS / Ion Exchange 4049
Figure 1 Approaches to development of an ion exchange
chromatographic process. Copyright ^ 1998 Whatman Interna-
tional Ltd., reproduced with permission.
the target binds more strongly to the exchanger than
the contaminants, so they are displaced by the target
during loading. The isolation of ovalbumin from hen
egg-white by anion exchange chromatography is one
such example (Figure 2), where we have shown oval-
bumin to displace the less anionic conalbumin com-
ponent during adsorption. However, the desorbed
protein product is typically in a mobile phase contain-
ing up to 1 M NaCl and this may be unsuitable for
adsorption in a subsequent step, for example an afRn-
ity interaction. If this is the case then another unit
process, such as diaRltration, needs to be introduced
which may be costly, time-consuming and could re-
sult in additional yield/activity loss.
An approach which may often be dismissed, but
in fact can be highly efRcient is the negative step. In
this case the contaminants bind to the adsorbent and
the target passes unretained during loading. Since
there is no volume reduction, product concentration
remains constant, but purity increases. If the target is
present in excess, then a modest adsorbent volume
may sufRce, which has an impact on cost, but perhaps
as important, the mobile phase composition remains
unchanged, which may facilitate the subsequent
chromatographic step. Negative steps are routinely
employed during immunoglobulin isolation from
serum or plasma, where the anionic albumin con-
taminants adsorb to an anion exchanger at neutral
pH, while the cationic immunoglobulin fraction
passes unretained through the exchanger, as repre-
sented in Figure 3.
4050 III / PROTEINS / Ion Exchange
Figure 2 Chromatography of hen egg-white proteins on Whatman DE52 using 0.025 M Tris/HCl buffer, pH 7.5 in a column (45 cm
i.d.;16 cm) at a flow rate of 1 L min\1. Copyright ^ 1998 Whatman International Ltd., reproduced with permission.
These are some of the fundamental considerations
during method scouting and are based on two ques-
tions. Firstly, at a given pH will the target bind to an
anion exchanger or a cation exchanger? Secondly, is
this what is wanted?
Figure 3 Chromatography of a goat serum preparation
on Whatman QA52 using 0.02 M Tris/HCl buffer, pH 7.5 in
a column (32.5 cm i.d.;12 cm). A, denotes a wash in load-
ing buffer; B, denotes a wash using loading buffer containing
0.5 M NaCl. Copyright ^ 1998 Whatman International Ltd., repro-
duced with permission.
Given that the mobile phase of the feedstock is
determined by the upstream process and any signiR-
cant adjustments add cost and complexity, the
chromatographer can now start to address these two
questions. Due to the nature of proteins and the
inSuences of protein : protein interactions which may
occur in the feedstream, it is difRcult, if not im-
possible, to predict the optimal chromatographic
conditions without practical study. It is therefore
common practice to carry out a parametric study,
investigating the inSuence of pH and ionic strength
with different ion exchangers. This is potentially
a time-consuming process, but one that will aid in
process optimization. This traditionally was per-
formed manually, but more recently automated
workstations have become commercially available
which can be programmed to carry out multivariable
parametric studies automatically.
Media Selection
A key consideration during method scouting is which
ion exchanger to use. The protein chromatographer is
offered a range of strong or weak anion or cation
exchangers, from several suppliers. The functional
groups are broadly similar across the range, but the
base matrices range from polysaccharides including
agarose, cellulose and dextran to polymeric supports
and advanced composites. Given that each manufac-
turer has proprietary chemical processes, the offer-
ings available are quite diverse. In order to assess the
impact of media selection on method scouting and
development, we screened 70 different commercially

available anion and cation exchangers, which may be
considered for process-scale protein separations.
Each medium was screened under identical condi-
tions. Perhaps not surprisingly, our data demon-
strated that 70 different media had 70 different
properties. Our data, descriptive rather than prescrip-
tive, suggest that media effectiveness is process de-
pendent rather than supplier dependent and the
thought process of it worked last time is not an
appropriate rationale for developing a second process
using the same adsorbent.
Method Development
When an appropriate ion exchanger and mobile
phase system have been identiRed, it must then be
decided whether to conduct the separation in either
a column contactor or a batch stirred tank system.
The former technique, being contained, lends itself to
automation and control, but the latter technique,
albeit classical, should not be dismissed. If, for
example, the feedstream/adsorbent volume ratio is
high, perhaps '20 : 1, then the time to pump the
feedstream through a packed bed of adsorbent would
be several hours. A batch stirred tank adsorption
including medium collection by centrifugation should
take less than 1 h. Similarly in a large scale process
where several hundreds of kilograms of ion ex-
changer are used, columns are costly and often cum-
bersome to use, so a large batch system may be
preferred due to its simplicity.
For highly selective separations where desorption
and elution conditions are critical then a column-
based technique would be appropriate, typically us-
ing gradient elution.
As stated earlier, proteins are large molecules with
a size up to hundreds of angstroms and consequently
diffusion into and out of the pores of an ion ex-
changer is the rate limiting step both in terms of
capture efRciency during adsorption and selectivity
during desorption and elution. In order to enhance
each of these parameters, one needs to maximize
residence time of the adsorbate with the absorbent to
increase the capture efRciency of the target protein
and in order to maximize selectivity, one needs to
provide adequate time for the desorbed protein to
diffuse into the bulk liquid phase, so it elutes as
a sharp peak. Flow rate is clearly the critical factor to
regulate these adsorption/desorption rates and this
equates with process-time. Unlike ion exchange of
small ions, where pore diffusion rates while limiting
have minimal criticality, they are crucial for effective
ion exchange chromatography of proteins and must
be considered during methods development. Typical
linear Sow rates for polysaccharide-based ion ex-
changers would be 30}300 cm h\1 and for advanced
III / PROTEINS / Ion Exchange 4051
polymeric-based media, a further 10-fold increase in
Sow may be possible. However, it should be noted
that the maximum Sow rate speciRcation of the ion
exchanger and an operational Sow rate for effective
protein binding and elution may be widely different,
and will likely to be determined by the nature of the
protein separation itself.
Scale-down Studies
Having deRned the ion exchanger, the mobile phases
and the mode of operation, a series of small scale
studies will be carried out at the laboratory bench to
assess process economics and perhaps to carry out
sensitivity analysis and validation support studies.
Scaled-down studies are very valuable and enable
a substantial amount of data to be generated and
collated in a cost-effective manner, although the time-
scale may be similar to that required for large scale
work. The key feature of a small scale study is that the
contactor is a scaled-down version of the process
system. For a batch process the aspect ratio of the
tank and tip speed of the agitator, etc., would be the
same for both scales of operation. The ratio of feed-
stock volume to mass of ion exchanger would be
constant as would the contact times for adsorption,
washing and elution. For a column separation, col-
umn bed height would be identical at both scales of
operation and linear Sow rate would be maintained
throughout the process. Provided that all mobile
phase volumes used were in proportion to the amount
of ion exchanger used, and that feedstock and buffer
compositions remain unaltered, a small scale study
should be a good model of the large scale process
separation. In the anion exchange chromatography of
hen egg-white proteins, for example, we have re-
ported the 1000-fold scale-up of a column process
from a 25 mL column to a 25 L column.
Process validation is a critical area in the isolat-
ion of therapeutic proteins. In these applications
it is crucial that for multiple uses of the adsorbent,
the eluting fraction containing the target protein
has a consistent composition from run to run and
that it meets a speciRcation in terms of microbial
bioburden, endotoxin levels, pyrogenicity and
viral contamination. These aspects of process valida-
tion have been adequately reviewed by Sofer and
Nystrom.
A widely used mobile phase for regeneration of ion
exchangers following protein chromatography is so-
dium hydroxide. It is well established that exposure
of a column of ion exchanger to 0.5}2.0 M NaOH for
up to 12 h is an effective means of regenerating the
medium. Importantly, these conditions are also ac-
knowledged to be effective at sanitization of the ion
exchange system, and we have conRrmed this to be
4052 III / PROTEOMICS: ELECTROPHORESIS
the case following gross microbial contamination of
columns of both anion and cation exchangers.
A key element of process validation is the chemical
stability of the ion exchanger to the cleaning/sanitiz-
ing regent systems. We have developed protocols for
monitoring hydrolysis of functional groups, referred
to as ligand leakage, and also matrix dissolution, in
order to address these concerns.
These process validation studies are typically con-
ducted in scale-down mode, and conRrmatory checks
made subsequently at process scale.
Process Scale Ion Exchange
Chromatography of Proteins
Having deRned the feedstock and mobile phases, se-
lected the ion exchanger and selected a batch or column
mode of operation, the chromatographer should Rnd
himself or herself in a position to scale-up the process.
It is difRcult to predict cost information on process
scale ion exchange separations since much depends
on the upstream and downstream process require-
ments and reusability of the ion exchanger, etc.
We have reported a cost estimate for single usage
of 25 kg of DE52 in batch and column operation
(see Ganetsos and Barker), but this is exemplary only.
Unfortunately information of this type, while in exist-
ence, is proprietary and therefore withheld.
Industry has carried out large scale ion exchange
chromatography of proteins for the last few decades
Metalloprotiens: Chromatography
See III / METALLOPROTEINS: CHROMATOGRAPHY
Thin-Layer (Planar) Chromatography
See III / PEPTIDES AND PROTEINS: Thin-Layer (Planar) Chromatography
PROTEOMICS: ELECTROPHORESIS
M. J. Dunn, National Heart and Lung Institute,
Imperial College of Science, Technology
and Medicine, Harefield Hospital,
Middlesex, UK
Copyright ^ 2000 Academic Press
and with the established principles described above,
there is no reason to assume that things will change
signiRcantly over the short to medium term. In the
longer term, developments in Suidized/expanded
beds and membrane adsorbers may offer new oppor-
tunities in this area of chromatography.
See also: II/Affinity Separation: Rational Design, Syn-
thesis and Evaluation: Affinity Ligands. III/Proteins: Cen-
trifugation; Electrophoresis; High-Speed Countercurrent
Chromatography.
Further Reading
Ganetsos G and Barker PE (eds) (1993) Preparative and
Production Scale Chromatography. New York: Marcel
Dekker.
Levison PR, Mumford C, Streater M, Brandt-Nielsen A,
Pathirana ND and Badger SE (1997) Performance com-
parison of low-pressure ion-exchange chromatography
media for protein separation. Journal of Chromatogra-
phy A 760: 151.
Shillenn JK (ed.) (1996) Validation Practices for Biotech-
nology Products. West Conshohocken: ASTM.
Sofer GK and Nystrom L-E (1991) Process Chromatogra-
phy. A Guide to Validation. London: Academic Press.
Stryer L (1981) Biochemistry, 2nd edn. San Francisco:
Freeman.
Subramanian G (ed.) (1991) Process-scale Liquid
Chromatography. Weinheim: VCH.
Verrall MS (ed.) (1996) Downstream Processing of Natural
Products. Chichester: John Wiley.
Introduction
The Rrst complete genome sequence, that of Haemo-
philus inUuenzae, was published in 1995. Intense effort
over the last three years has resulted in the completion
of the genomes for a further 12 micro-organisms
 III / PROTEOMICS: ELECTROPHORESIS 4053

ranging in complexity from Mycoplasma genitalium, functions. A further limitation of the genomic ap-
with a genome size of only 0.58 Mb encoding less proach is that it does not provide any insights into the
than 500 proteins, to Escherichia coli, with a genome ways an organism may modify its pattern of gene
size of 4.6 Mb encoding more than 4000 proteins expression in response to different conditions.
(Table 1). The complexity of the eukaryotic genomes
has resulted in slower progress, with only one organ-
ism, the yeast Saccharomyces cerevisiae, having been
Analysis of Gene Expression
completed (Table 1). However, signiRcant progress is These problems can only be solved by direct invest-
being made for a variety of other species, with the igation of gene expression, which can be achieved at
estimated date for the completion of the human either the level of messenger RNA (mRNA) or pro-
genome currently being 2001 (Table 1). tein. A variety of techniques such as cDNA micro-
The vast amount of information being generated by arrays and serial analysis of gene expression (SAGE)
the various genome sequencing programmes has the make it possible to undertake mass screening of
potential to contribute signiRcantly to our under- mRNA expression and establish which particular
standing of how an organism functions and its evolu- mRNAs are expressed in an organism under any
tionary relationships with other life forms. However, given condition. However, recent studies have high-
it has already become clear that genomics alone will lighted an important limitation of this approach in
not provide all of the answers. For those organisms that there is not always a direct correlation between
whose genomes have been completed, typically RNA abundance and the amount of the correspond-
around 30% of the genes can be assigned deRnite ing functional protein within the cell.
functions with up to a further 30% being attributed A further major limitation of studies at the level of
functions on the basis of homology with other genes mRNA is that they are unable to provide any in-
of known function. The remaining 40% of the struc- formation of processes of co- and post-translational
tural genes can often not even be attributed putative modiRcation of proteins. The modiRcation of pro-
teins by processes such as phosphorylation, glyco-
Table 1 Some organisms whose genomes have been com- sylation, sulfation, hydroxylation, N-methylation,
pletely sequenced and others which are the subject of active acylation, prenylation and N-myristoylation, can re-
genome sequencing programmes sult in signiRcant modulation of their functional
properties. Knowledge of these processes is therefore
Organism Size ORFs Year
(Mb) completed important for a complete understanding of gene ex-
pression.
Microorganisms
Mycoplasma genitalium 0.58 470 1995
Ureaplasma urealyticum 0.75 640 Proteome Analysis
Mycoplasma pneumoniae 0.81 679 1996
Treponema pallidum 1.14 1000 The realization that these problems can only be ad-
Borrelia burgdorferi 1.44 843 1997 dressed through studies at the level of protein expres-
Aquifex aeolicus 1.50 1512 1998 sion has resulted in increasing interest in the area
Helicobacter pylori 1.66 1590 1997 which has become known as proteome analysis. The
Methanococcus jannaschii 1.66 1738 1996
term proteome was Rrst coined by a collaborative
M. thermoautotrophicum 1.75 1855 1997
Haemophilus influenzae 1.83 1743 1995 team of scientists at Macquarie and Sydney Universi-
Streptococcus pyogenes 1.98 1900 ties in 1995 and is deRned as the protein complement
Archaeoglobus fulgidis 2.18 2436 1997 of the genome of an organism. Increasing genomic
Nisseria gonorrhoreae 2.2 2100 complexity together with the potential for co- and
Pyrobaculum aerophilum 2.22 1900
post-translational modiRcations make proteome
Synechocystis PCC6803 3.57 3168 1996
Bacillus subtilis 4.20 4100 1997 analysis a difRcult task for higher organisms. As a
Mycobacterium tuberculosis 4.41 3924 1998 consequence, active proteome programmes are cur-
Escherichia coli 4.60 4288 1997 rently restricted to some of the simpler organisms
Eukaryotes such as mollicutes (M. genitalium, Sprioplasma melli-
Saccharomyces cerevisiae 13.0 5885 1996 ferum), prokaryotes (E. coli, Chlamydia trachomatis)
Dictyostelium doscoideum 70 12500 and yeast (S. cerevisiae).
Arabidopsis thalania 70 14000 The complexity of eukaryotic proteomes has result-
Caenorhabditis elegans 80 17800
ed in the term proteomics or proteome analysis
Drosophila melanogaster 170 30000
Homo sapiens 2900 50000 being used in a narrower context in which it is used to
characterize patterns of protein (and thereby gene)
ORFs"open reading frames. expression in particular cell type and tissues. This
4054 III / PROTEOMICS: ELECTROPHORESIS
approach can provide new insights into a variety of
biological processes such as development, apoptosis
and the cell cycle and add to our knowledge of the
mechanisms that control gene expression. There is
also considerable interest in applying proteomics to
the study of diseases, where it promises further under-
standing of these processes at the molecular level and
may lead to the discovery of new diagnostic markers
and novel therapeutic targets. The pharmaceutical
industry is also expressing considerable interest in the
potential of proteomics in the process of drug dis-
covery, as well as for analysis of the pharmaceutical
and toxicological effects of existing therapeutics.
Need for Protein Separation
The Rve main steps of proteome analysis are shown in
Figure 1. The primary requirement is that we must be
able to separate the very complex protein mixtures
present in lysates of cells, tissues and organisms. It is
generally accepted that the best method currently
available is two-dimensional polyacrylamide gel elec-
trophoresis (2D electrophoresis). While there are sev-
eral possibilities for combination of electrophoretic
procedures, the most effective approach is a combina-
tion of a Rrst-dimension separation by isoelectric fo-
cusing (IEF) under denaturing conditions with
a second-dimension separation by sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-
PAGE). This results in the sample proteins being
separated according to their charge properties (i.e.
isoelectric point, pI) in the Rrst dimension followed
Figure 1 The five main steps in proteome analysis.
by a size-based (molecular weight) separation in the
second dimension. As the charge and size properties
of proteins are essentially independent parameters,
this results in the sample proteins being distributed
across the whole area of the 2D separation (Figure 2).
OFarrell Method of 2D
Electrophoresis
The method of 2D electrophoresis (2DE) described by
OFarrell in 1975 has formed the basis of almost all
subsequent developments in 2-DE. This method used
a combination of IEF in cylindrical gels (cast in capil-
lary tubes) containing 8 M urea and 2% of the
nonionic detergent, Nonidet P-40 (NP-40), with the
SDS-PAGE system of Laemmli. However, for effective
use in proteome analysis, 2DE must be capable of
consistently reproducible high resolution protein sep-
arations. This proved to be problematic largely due to
the nature of the synthetic carrier ampholytes (SCA)
used to generate the pH gradients for IEF. SCA are
small molecules which are freely mobile within the IEF
gel, and the electroendosmotic Sow of water which
occurs during IEF results in their migration towards
the cathode. This process is known as cathodic drift
and results in pH gradient instability, with loss of the
more basic proteins from the Rnal 2D gel pattern.
2DE using IPG IEF
This problem has been largely overcome with the de-
velopment of the Immobiline reagents (Amersham
Pharmacia Biotech) for the generation of immobilized
pH gradients (IPG) for IEF. The Immobiline reagents
are acrylamide derivatives which give a series of buf-
fers with different pK a values distributed across the
range pH 3}10. The appropriate Immobiline reagents,
calculated from the extensive series of published
recipes, are added to the mixture used for gel polym-
erization, resulting in the buffering groups which will
form the pH gradient being covalently attached via
vinyl bonds to the polyacrylamide backbone. This
immobilization of the pH gradient renders it immune
to the effects of electroendosmosis, resulting in highly
stable IEF separations.
Despite early problems which were encountered in
the application of IPG IEF to 2DE separations,
this method has now become the method of choice for
the Rrst dimension of 2DE. The procedure which
is now used is largely based on the work of GoK rg
and her colleagues (see Further Reading). BrieSy,
IPG IEF is performed in individual gel strips, 3}5 mm
wide, cast on plastic supports. After IEF, the gel
strips are equilibrated to allow the separated proteins
to interact with SDS, and then applied either to the

Figure 2 A 2DE separation of 100
g of human heart using a nonlinear pH 3.5}10 IPG IEF gel was used in the first dimension and
12% SDS-PAGE in the second dimension. The separated proteins were visualized by silver staining.
surface of a horizontal SDS-PAGE gel to the top of
a vertical. Interlaboratory studies of heart, barley
and yeast proteins have unequivocally demonstrated
the excellent reproducibility of both protein spot posi-
tion and quantity that can be achieved with this
method.
Separation Capacity of 2DE
The ability of 2DE to resolve complex mixtures of
proteins is dependent on the gel area used for the
separation. Thus, the standard combination of 18 cm
IPG strips with 20 cm SDS-PAGE gels is capable of
routinely separating 2000 proteins from lysates of
whole cells and tissues. It has been shown that very
large gel formats (up to 30 cm in each dimension) are
capable of separating up to 10 000 proteins from such
samples, but this is achieved at the expense of the ease
of gel handling and processing. In contrast, only a few
hundred proteins can be separated using mini-format
2DE, but this approach has the advantage of rapid
separations for screening purposes.
IPG IEF also provides great Sexibility in the choice
of pH gradient used for the separation, providing an
additional approach to maximize the efRciency of sep-
aration of the particular protein mixture under invest-
igation. Thus, IPG IEF gels spanning the range pH
3.5}10 are ideal for examining the diversity of protein
III / PROTEOMICS: ELECTROPHORESIS 4055
in a sample (Figure 2). Optimal resolution of proteins
in a particular pH range can be achieved using nar-
rower pH gradients (Figure 3).
A further advantage of IPG IEF is that it has a very
high capacity for micropreparative 2DE protein separ-
ations, particularly using a recently described method
in which IPG strips are reswollen directly in a solution
containing up to several mg of the protein sample to be
analysed.
Protein Detection
The next requirement for effective proteome analysis is
detection of the separated proteins at high sensitivity.
The Coomassie brilliant blue dyes have been the most
commonly used reagents for detecting protein zones
separated by gel electrophoresis, but their limited sensi-
tivity (around 0.5
g per protein spot) necessitates the
use of relatively high sample loadings ('500
g).
Much higher sensitivity of detection (0.1 ng per protein
spot) can be achieved by silver staining (Figures 2 and
3), but there can be problems in using this method as
a quantitative procedure as it is known to be non-
stoichiometric. Silver staining intensity is linear
over the range from 0.04 to 2 ng protein/mm2,
but above this limit the stain becomes nonlinear,
resulting in saturation and even negative staining
effects.
4056 III / PROTEOMICS: ELECTROPHORESIS
Figure 3 A 2DE separation of 100
g of human heart using a linear pH 4}7 IPG IEF gel was used in the first dimension and 12%
SDS-PAGE in the second dimension. The separated proteins were visualized by silver staining.
Many of these problems can be overcome using
detection methods based on the use of Suorescent
compounds. Such methods are highly sensitive and
generally exhibit excellent linearity and a high dy-
namic range, making it possible to achieve excellent
quantitative analysis if a suitable imaging device
is used. A variety of Suorescent compounds are avail-
able for labelling proteins prior to electrophoresis.
However, such pre-electrophoretic staining often re-
sults in charge modiRcation, resulting in alterations
in spot positions during 2DE. Recently, compounds
which react with cysteine or lysine residues have
been used successfully for 2DE. The cysteine-reactive
reagent, monobromobimane was found to have a sen-
sitivity of 1 pg protein per spot when imaged
using a cooled CCD camera. The amine-reactive cy-
anine dyes, propyl Cy3 and methyl Cy5, have been
used to label E. coli proteins. These dyes are claimed
to have an inherent positive charge, thereby preserv-
ing the overall charge of the proteins after coupling.
Due to their spectral properties, the two dyes can
be distinguished when present together, allowing
two different samples each labelled with one of the
dyes to be mixed together and separated on the same
2DE gel. This method, which has been termed differ-
ence gel electrophoresis (DIGE), has great potential
for improving the efRciency of detection of differ-
ences in 2DE protein patterns between different
samples.
One approach to overcoming the problems asso-
ciated with charge modiRcation during the IEF di-
mension is to label the proteins while present in the
Rrst dimension gel after IEF, prior to the second
dimension separation by SDS-PAGE. The Suorescent
compound 2-methoxy-2,4-diphenyl-3(2H)-furanone
(MDPF) and a Suorescent maleimide derivative
have been used in this way. The alternative approach
is to label the proteins after the 2DE separation has
been completed. Recently, two post-electrophoretic
Suorescent staining reagents, SYPRO orange and red
have been described. These stains have a very high
sensitivity (2 ng protein per spot) and excellent lin-
earity with a high dynamic range.
Metabolic labelling of proteins with a radiolabelled
amino acid prior to their separation by 2DE provides
a very sensitive method for protein detection. This
approach is most commonly used with in vitro cell
culture systems, but it is also possible to radiolabel
synthetically the proteins in small pieces of fresh
tissue. While proteins can be readily radiolabelled
postsynthetically by methods such as radioiodination
with [125I] or reductive methylation with [3H]-sodium
borohydride, these result in signiRcant charge modiR-
cations precluding their use in proteome analysis.
 III / PROTEOMICS: ELECTROPHORESIS 4057

Qualitative and Quantitative Analysis prot.html), but the uncertainty of molecular weight
estimation by SDS-PAGE (typically around $10%)
The Rrst step in the analysis of 2DE protein proRles is
makes this process unreliable. Recently, mass spectro-
to produce a digitized image. Stained gels can be
metry (MS) has been used to measure directly the
digitized using a Sat-bed scanning laser densitometer
mass of proteins separated by 2DE. In this approach
or a suitably modiRed document scanner. Autoradio-
the proteins are transferred by Western blotting onto
graphic Rlm images of 2DE separations of radiolabel-
the surface of a nitrocellulose or PVDF membrane
led proteins can also be imaged in this way, but more
which is then treated with a matrix required for MS.
accurate quantitation can be achieved using a phos-
The protein spot of interest is excised, mounted dir-
phorimaging scanner. Fluorescently labelled protein
ectly into a matrix-associated laser desorption ioniz-
separation patterns can be visualized using either
ation mass spectrometer (MALDI-MS) and the mass
a dedicated scanning laser densitometer (Suorimager)
of the intact protein measured (Figure 4). We have
or a camera system Rtted with cooled CCD array.
found that this method is very accurate, usually with-
Several commercial software systems for the analysis
in 1% of the predicted mass, but requires a MALDI-
of 2DE gels are now available running on desktop
MS Rtted with an infrared laser. While such mass
workstations (Unix, PC, Mac). These systems make it
data can be invaluable in characterizing post-transla-
possible to extract quantitative and qualitative in-
tional modiRcations of proteins, it is unlikely on
formation from individual 2DE gels, to match protein
its own to result in unequivocal protein identiRcation.
patterns from large collections of 2DE gels, and there-
Fortunately, over the last few years, several
by establish comprehensive databases which can be
methods have been developed which make it possible
used to investigate quantitative protein expression in
to identify and characterize proteins separated by
cells, tissues and organisms.
2DE (Figure 5).

Protein Identi\cation and Western Blotting

Characterization
It is clear from the foregoing that 2DE provides in-
formation on the pI, molecular weight and relative
abundance of the separated proteins. However, it
provides no direct clues to their identities or func-
tions. The pI and molecular weight information can
be used to search sequence databases for proteins
with similar properties, for example using the
TagIdent tool (http://www.expasy.ch/www/guess-
The Rrst major advance in the characterization of
proteins separated by gel electrophoresis was the de-
velopment of Western blotting. In this technique, the
separated proteins are transferred (blotted) from the
restrictive gel matrix, usually by application of an
electric Reld perpendicular to the plane of the gel
(electroblotting), onto the surface of a suitable mem-
brane support such as nitrocellulose or PVDF. The
proteins can then be probed with a variety of ligands,

Figure 4 IR-MALDI mass spectrum of a protein spot from a 2D gel electroblotted onto a PVDF membrane. Mass peaks indicated are
multiply charged or dimers of the molecular ion. The protein spot is known to be cardiac
-actin. The measured mass of the molecular
ion (41842.1) is very close to the theoretical value determined from its amino acid sequence (41784.6).
4058 III / PROTEOMICS: ELECTROPHORESIS
Figure 5 Methods currently used for the identification and characterization of proteins separated by 2DE.
particularly poly- and monoclonal antibodies. This
approach has been used quite extensively for the
identiRcation of known proteins separated by 2DE,
but is a very time-consuming process that is depen-
dent on the availability of a suitable panel of speciRc
antibodies reactive with the denatured proteins in
2DE gels.
Amino Acid Sequence Determination
by Edman Degradation
Amino acid sequence, even if this is only a few resi-
dues in length, is the most speciRc method of protein
identiRcation. The chemical sequencing of proteins
has been possible for half a century since the develop-
ment of the method known as Edman degradation in
1949. This remained a laborious manual procedure
until the development of the Rrst automated protein
sequenators, with the Rrst commercial spinning cup
instrument becoming available in 1971. This instru-
ment was relatively insensitive, requiring at least
10 nmol of sample (equivalent to 500
g for a 50 kDa
protein). However, progress in sequenator tech-
nology has resulted in the current generation of
gas}liquid sequenators which are capable of generat-
ing N-terminal sequence information from low
picomole quantities of protein (1 pmol is equivalent
to 0.05
g for a 50 kDa protein). This level of
sensitivity is compatible with the amount represented
by many of the spots present on micropreparative
2DE gels and this method remains the method of
choice if extended runs of N-terminal protein se-
quence are required. This is a particularly important
consideration for the analysis of apparently novel
proteins, i.e. sequences not present in protein and
nucleotide databases. Although chemical protein se-
quencing is a sensitive and informative method of
protein identiRcation, throughput is very low,
typically one or two samples per day. Thus, there is
a need for alternative approaches which allow rapid
and sensitive screening of gel proteins separated by
2DE, so that only those which cannot be identiRed
unequivocally or appear to be novel require further
detailed characterization.
Problem of N-Terminal Blockage
A major problem with protein sequencing by Edman
degradation is that many proteins lack a free
-amino
group, due to co- or post-translational modiRcation.

Such N-terminal blockage occurs typically in up to
50% of eukaryotic proteins and results in no se-
quence being obtained. This problem can be over-
come by subjecting the separated protein, either in
situ within the gel matrix or after Western blot trans-
fer onto a nitrocellulose or PVDF membrane, to
chemical (e.g. CNBr) or enzymatic (e.g. trypsin)
cleavage to generate shorter peptides which can
be isolated and sequenced. The cleavage products
are then usually separated by RP-HPLC, and selected
peptide fractions directly applied to the protein
sequenator. This procedure is highly efRcient, but
the determination of multiple stretches of sequence
usually requires two to three times more protein than
does N-terminal protein sequence analysis.
Amino Acid Compositional Analysis
Amino acid compositional analysis (AAA) is the best
method for the absolute measurement of protein
abundance. Current methods for the analysis of Suor-
escently derivatized amino acids have sub-pmole sen-
sitivity and so can be applied directly to proteins
separated by 2DE. An individual proteins have more
or less unique amino acid compositions, AAA can be
an excellent method for the rapid identiRcation of
proteins separated by 2DE, in which the experimental
amino acid composition is compared with amino acid
sequences generated in silico from protein and nu-
cleotide sequence databases. The major drawback of
this approach is that the output is a ranked list of
possible protein identities (Figure 6) and the correct
protein does not necessarily occur as the Rrst ranked
entry. This method is better used in conjunction
with another rapid method of protein identiRcation
such as peptide mass proRling (see later) and this
orthogonal approach has been found to be useful for
identifying proteins even across the species barrier.
Another approach to improving protein identiRcation
by AAA is to generate a short N-terminal protein
sequence tag by Edman degradation and to use
this in combination with the AAA data for protein
identiRcation.
Peptide Mass Pro\ling
It has long been realized that the peptides generated
by chemical (e.g. CNBr) or enzymic (e.g. trypsin)
digestion of a protein are characteristic of that pro-
tein and such peptide Rngerprints or maps analysed
by chromatography or electrophoresis have been used
for investigating the relatedness of proteins. The ad-
vent of MS methods for the analysis of peptides has
made this into a much more powerful approach for
protein identiRcation. In this method the peptide
III / PROTEOMICS: ELECTROPHORESIS 4059
masses obtained by MS of a protein digest are used to
interrogate databases of peptide masses generated
in silico from protein and nucleotide sequence
databases. As in the case of AAA, this technique of
peptide mass proRling or Rngerprinting produces
a list of possible protein identities ranked in order of
probability (Figure 7). The reliability of this method
can be increased by combining peptide mass proRling
data from two or more separate digests (e.g. trypsin,
Lys-C) or by adopting an orthogonal approach in
combination with AAA (see above).
The enzymatic cleavage of the 2DE protein spot
can be carried out either in situ within the gel matrix
or after electroblotting to a suitable membrane
(nitrocellulose or PVDF). After recovery, the unfrac-
tionated peptide can be readily analysed by MALDI-
MS (Figure 7). Alternatively, the peptide mixture can
be fractionated by high performance liquid
chromatography (HPLC), with the fractions being
analysed either ofSine by MALDI-MS or online by
electrospray ionization (ESI)-MS using a quadrupole
or ion-trap instrument.
Amino Acid Sequence Determination
by Mass Spectrometry
Recently alternative techniques for the determinat-
ion of the primary sequence of peptides and proteins
have been developed based on the use of mass spec-
trometry (MS). This can be achieved by peptide
fragmentation within the spectrometer or by
a method termed ladder sequencing. In the
latter approach, Edman chemistry or enzymic degra-
dation with aminopeptidase or carboxypeptidase
is used under limiting conditions to produce an
overlapping series of truncated peptides. These
differ in size according to the number of amino
acid residues which have been removed from their
N- or C-terminus, allowing the sequence to be
deduced by measurement of the peptide masses
by MALDI-MS. A high mass accuracy is required
and it is not possible to distinguish between leucine
and isoleucine as these residues have an identical
mass.
The alternative approach is to take advantage of
the ability of two-stage mass spectrometers, either
MALDI-MS with post-source decay (PSD) or ESI-
MS/MS triple-quadropole or ion-trap instruments, to
induce fragmentation of peptide bonds. It is possible
to use this approach to generate extended de novo
amino acid sequence information, but it requires con-
siderable expertise to interpret the complex spectra
that are obtained. However, partial sequence data is
an extremely powerful adjunct to the identiRcation of
proteins by peptide mass proRling.
4060 III / PROTEOMICS: ELECTROPHORESIS

Figure 6 The identification of a protein spot from a 2DE separation by amino acid compositional analysis. (A) HPLC analysis of
pre-column derivatized amino acids resulting from hydrolysis of the protein spot. (B) Amino acid composition determined from the HPLC
analysis. (C) Result of the amino acid composition database search indicating that the protein is cardiac fatty acid binding protein.

The method known as peptide sequences tagging
is based on interpretation of a portion of the ESI-
MS/MS or PSD-MALDI-MS fragmentation data to
generate a short partial sequence or tag, which is
used in combination with the mass of the intact par-
ent peptide ion, and provides a signiRcant amount of
additional information for the homology search.
A powerful extension of this approach has been the
development of a nano-electrospray ion source that
allows spraying times of more than 30 min from
about 1
L of sample. The sensitivity of this method
is in the low femtomole range and silver stained 2DE
protein spots containing as little as 5 ng protein have
been successfully sequenced. Moreover, using this
 III / PROTEOMICS: ELECTROPHORESIS 4061

Figure 7 The identification of a protein spot from a 2DE separation by peptide mass fingerprinting. (A) MALDI-MS spectrum of the
tryptic digest of the protein spot. (B) Result of the peptide mass database search indicating that the protein is the M-chain of creatine
kinase.

method it is possible to sequence multiple peptides
from a digest without the need for their prior separ-
ation by RP-HPLC.
An alternative approach is based on the automated
interpretation of ESI-MS/MS fragmentation data
which is used to directly search sequence databases.
In the Rrst step in this process all those peptides that
can be generated from proteins in the sequence
databases and whose masses match those of the mea-
sured peptide ion are identiRed. The fragment ions
expected for each of the candidate peptides are then
calculated and the experimentally determined
MS/MS spectrum is then compared with the pre-
dicted spectra using cluster analysis algorithms. This
4062 III / PROTEOMICS: ELECTROPHORESIS
method has been fully automated and sensitivity is at
the low femtomole level.
Database Construction
The Rnal requirement for proteome analysis is the
construction of databases to store the data generated
and to make this readily available within the laborat-
ory and, where possible, accessible to other scientists
worldwide. The best approach to this is currently
offered by the World Wide Web (WWW) on the
Internet. There is currently no international standard
for the construction of such databases, but in order
that there should be good connectivity between them
it has been suggested that they are constructed ac-
cording to a set of fundamental rules. Databases
conforming to these rules are said to be Federated 2D
Databases and a list of these can be viewed at
WORLD-2DPAGE (http://www.expasy.ch/ch2d/2d-
index.html).
Conclusions
From the foregoing discussion it can be seen that
proteomics provides an interface with genomics
which can provide information on protein expression
in biological systems. This information will aid
our understanding of complex cellular processes
and the way in which they react to varying condi-
tions. Proteomics will also provide insights into pro-
cesses of disease at the molecular level and should
result in the development of novel diagnostics and
therapeutics.
Further Reading
Dunn MJ (1987) Two-dimensional polyacrylamide gel elec-
trophoresis. In: Chrambach A, Dunn MJ and Radola BJ
(eds) Advances in Electrophoresis, vol. I, pp. 1}109.
Weinheim: VCH.
GoK rg A, Boguth G, Obermaier C, Posch A and Weiss
W (1995) Two-dimensional polyacrylamide gel elec-
trophoresis with immobilized pH gradients in the Rrst
dimension (IPG-Dalt): The state of the art and the
PURGE-AND-TRAP: GAS CHROMATOGRAPHY
See III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
PYROLYSIS: GAS CHROMATOGRAPHY
See II / CHROMATOGRAPHY: GAS / Pyrolysis Gas Chromatography
controversy of vertical versus horizontal systems.
Electrophoresis 16: 1079}1086.
Humphery-Smith I, Cordwell SJ and Blackstock WP (1997)
Proteome research: Complementarity and limitations
with respect to the RNA and DNA worlds. Electrophor-
esis 18: 1217}1242.
Lamond AI and Mann M (1997) Cell biology and the
genome projects: A concerted strategy for characterizing
multiprotein complexes by using mass spectrometry.
Trends in Cell Biology 7: 139}142.
Patterson SD (1994) From electrophoretically separated
protein to identiRcation: Strategies for sequence and
mass analysis. Analytical Biochemistry 221: 1}15.
Patterson SD and Aebersold R (1995) Mass spectrometric
approaches for the identiRcation of gel-separated pro-
teins. Electrophoresis 16: 1791}1814.
Pennington SR, Wilkins MR, Hochstrasser DF and Dunn MJ
(1997) Proteome analysis: From protein characterization
to biological function. Trends in Cell Biology 7: 168}173.
Sutton CW, Wheeler CH, U S, Corbett JM, Cottrell JS and
Dunn MJ (1997) The analysis of myocardial proteins by
infrared and ultraviolet laser desorption mass spectro-
metry. Electrophoresis 18: 424}431.
Wan JS, Sharp SJ, Poirer GM et al. (1996) Cloning differen-
tially expressed mRNAs. Nature Biotechnology 14:
1685}1691.
Wasinger VC, Cordwell SJ, Cerpa-Poljak A et al. (1995)
Progress with gene-product mapping of the Mollicutes:
Mycoplasma genitalium. Electrophoresis 16: 1090}1094.
Wheeler CH, Berry SL, Wilkins MR et al. (1996) Charac-
terisation of proteins from two-dimensional elec-
trophoresis gels by matrix-assisted laser desorption mass
spectrometry and amino acid compositional analysis.
Electrophoresis 17: 580}587.
Wilkins MR, Pasquali C, Appel RD et al. (1996) From
proteins to proteomes: Large scale protein identiRcation
by two-dimensional electrophoresis and amino acid
analysis. Biotechnology 14: 61}65.
Wilkins MR, Williams KL, Appel RD and Hochstrasser DF
(1997) Proteome Research: New Frontiers in Functional
Genomics. Berlin: Springer.
Wilm M, Shevchenko A, Houthaeve T et al. (1996) Fem-
tomole sequencing of proteins from polyacrylamide gels
by nanoelectrospray mass spectrometry. Nature 379:
466}469.
Yates JR (1998) Database searching using mass spectro-
metry data. Electrophoresis 19: 893}900.
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
QUANTITATIVE STRUCTURE^RETENTION
RELATIONSHIPS (QSRR) IN CHROMATOGRAPHY
R. Kaliszan, Medical University of Gdan& sk,
Gdan& sk, Poland
Copyright ^ 2000 Academic Press
Introduction
To relate structure and chromatographic retention an
approach is needed that lacks the rigour of thermo-
dynamics but which provides otherwise inaccessible
information. Such an approach is a combination
of detailed models with certain thermodynamic
concepts.
Linear free-energy relationships (LFER) may
be regarded as linear relationships between the
logarithms of the rate or equilibria constants for
one reaction series and those for a second reaction
series subjected to the same variation in reactant
structure or reaction conditions. Retention
parameters can be assumed to reSect the free-energy
changes associated with the chromatographic distri-
bution process. Accordingly, a chromatographic
column can be treated as a free-energy transducer,
translating differences in chemical potentials
of analytes, arising from differences in their
structure, into quantitative differences in retention
parameters.
Assuming LFER it is possible to determine relative
inputs of individual structural groups, fragments or
features, to a property measured for a series of com-
pounds in various chemical, physical, physicochemi-
cal and biological experiments. Such structural
parameters (descriptors) can then be related to reten-
tion parameters.
The existence of LFER is normally proved statis-
tically. The basic methodology of employing LFER to
predict differences in pharmacological activity within
a series of related agents was proposed in 1964 by
Hansch and Fujita (QSAR, quantitative structure}ac-
tivity relationships). Multiple regression analysis was
applied in 1977 to chromatographic data (QSRR,
quantitative structure}retention relationships). Other
chemometric methods of data analysis have since
been introduced to QSRR. QSRR are now one of the
most extensively studied manifestations of LFER and,
at the same time, the most common application of
chemometrics.
4063
Methodology and Goals of QSRR
Analysis
To undertake QSRR studies two kinds of input data
are needed (Figure 1). One is a set of quantitatively
comparable retention data (dependent variable) for
a sufRciently large (for statistical reason) set of
analytes. The other is a set of quantities (independent
variables) assumed to account for structural differ-
ences among the analytes being studied. Through the
use of chemometric computational techniques, reten-
tion parameters are characterized in terms of various
descriptors of analytes (and/or their combinations) or
in terms of systematic knowledge extracted (learnt)
from these descriptors.
To obtain statistically signiRcant and physically
meaningful QSRR, reliable input data are required
and stringent mathematical analysis must be carried
out. If this is not done, formally valid correlations
may be developed for chemically invalid principles.
Once good QSRR have been obtained, one can
exploit them for:
1. prediction of retention of a new analyte;
2. identiRcation of the most informative structural
descriptors possessing the highest retention predic-
tion potency;
3. insight into the molecular mechanism of separ-
ation operating in individual chromatographic
systems;
4. evaluation of physicochemical properties of
analytes, e.g. their hydrophobicity (lipophilicity);
5. prediction of relative biological (pharmacological)
activities within a set of drugs and other xenobiotics.
Retention Parameters for QSRR
The great advantage of the QSRR analysis over other
quantitative structure}property relationship studies is
that chromatography can readily produce a large
amount of relatively precise and reproducible data.
In a chromatographic process all conditions may be
kept constant and hence the structure of an analyte
becomes the single independent variable in the
system.
The most commonly used retention parameter in
gas chromatography is the KovaH ts retention index.
When the adjusted retention times are used to calcu-
late retention indices, parameters are obtained that
4064 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY

Figure 1 Methodology and goals of studies of quantitative structure}retention relationships (QSRR). (Adapted with permission from
Kaliszan R (1992) Quantitative structure}retention relationships. Analytical Chemistry 64: 619A}631A. Copyright 1992 American
Chemical Society.)

depend only on the column temperature and the sta-
tionary phase used. KovaH ts retention indices are high-
ly reproducible; with a well-designed experimental
technique, an interlaboratory reproducibility of one
unit is possible. Sometimes in QSRR studies the log-
arithms of retention volumes of solutes are used in-
stead of KovaH ts indices.
Classical thin-layer chromatographic (TLC) reten-
tion parameters are of rather limited reproducibility.
The use of well-deRned small-diameter stationary
phase particles and a better knowledge of the para-
meters that determine the efRciency of chromato-
graphic systems have led to the development of high
performance TLC (HPTLC). An advantage of TLC
over column chromatography, from the point of view
of QSRR studies, is that tens of analytes can be
simultaneously chromatographed on the same plate.
The retention parameter from TLC (and also from
paper chromatography) that is normally used in
QSRR is the RM value. The RM value is deRned as log
((1/RF )!1), where RF is the distance migrated by the
sample from the origin compared with the distance
migrated by the solvent front from the origin.
The LFER-based retention parameter in high per-
formance liquid chromatography (HPLC) is the log-
arithm of the retention factor k. The retention factor
is deRned as in eqn [1].
k"(tR!tM)/tM"(VR!VM)/VM [1]
where tR and VR are the retention time and the reten-
tion volume of the analyte. The quantities tM and
VM denote the elution time and the elution volume of
an unretained compound.
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
HPLC retention data for QSRR analysis are usually
obtained by measuring log k at several Rxed eluent
compositions (isocratic conditions) and then by ex-
trapolating the dependence of log k on a binary eluent
composition to a common mobile phase composition
based on the SoczewinH ski}Snyder model:
log k"log kw!S
[2]
In eqn [2] S is a constant for a given analyte and
a given HPLC system and
is the volume fraction of
one of the mobile phase components. In the case of
reversed-phase HPLC, kw is a hypothetical reten-
tion factor expected for pure water (buffer) mobile
phase (
"0).
The curvature often observed in plots of log k ver-
sus
leads to a quadratic relationship:
ln k"A
2#B
#C [3]
where A, B and C are constants for a given analyte
and a given chromatographic system. The ln k value
calculated from eqn [3] by assuming
"0 is only
occasionally used in QSRR analysis.
In spite of considerable effort, the relationships
between retention and mobile phase composition are
approximate. Often the values of log kw extrapolated
from a number of isocratic measurements in
water/organic modiRer eluents of varying compo-
sitions to a pure water eluent (the intercepts in
eqn [2]) are different from those determined experi-
mentally (when this is possible). Reversed-phase
HPLC log kw data are also usually different when
derived from aqueous systems modiRed with different
organic solvents. Still, the determination of log kw
appears to be the most reliable means of normalizing
the retention parameters for QSRR analysis.
It should be noted here that some workers advocate
using as the dependent variable the parameter S from
eqn [2] or its ratio to log kw.
The electrophoretic mobility,
el, of spherical par-
ticles is described by a simple equation:

el"(z)/(6aN) [4]
where z is the effective charge,  is the charge per
mole of protons,  is the viscosity of the medium, a is
the radius of the charged species and N is the
Avogadro number.
A parameter normally measured in capillary elec-
trophoresis (CE) is migration time, t. In a given CE
system this parameter is inversely proportional to the
electrophoretic mobility,
. A normalized parameter,

(cm3 V\1) allows comparison of data obtained in
different CE systems. If a series of analytes is analysed
4065
under the same conditions, then 1/t and
are equiva-
lent.
Chemometric Procedures in QSRR
Assuming LFER, a given chromatographic retention
parameter may be described (statistically) by a set of
analyte structural descriptors:
Retention parameter"f(a1x1,2, anxn) [5]
The coefRcients a1}an for individual n descriptors
are calculated by multiple regression. There are
computer programs available commercially that
are able to derive regression coefRcients and to
evaluate a statistical value of the regression model
assumed.
Whether or not any of the possible models are
statistically signiRcant is judged on the basis of sev-
eral statistical signiRcance parameters. Among them
are: the correlation coefRcient (R); the standard error
of estimate (SE); the value of the F-test of the
overall signiRcance (F); the values of t-test of signiR-
cance of individual regressors (t); and the cross-
correlation coefRcients between the independent
variables in the regression QSRR equation. Even
if the values of these statistical parameters are
within the acceptable range, one cannot exclude
a chance correlation. This may result when too many
variables are surveyed to correlate too few retention
data.
Multivariate methods of data analysis, like dis-
criminant analysis, factor analysis and principle com-
ponent analysis, are often employed in chemometrics
if multiple regression methods fail. The most popular
chemometric method in QSRR is principle compon-
ent analysis (PCA). By PCA one reduces the number
of variables in a data set by Rnding linear combina-
tions of those variables that explain most of the
variability.
Commercially available software packages tabu-
late the component weights and the values of indi-
vidual principal components. Plots of component
weights for each variable (structural descriptor) are
useful in QSRR analysis. Analogously, scatterplots
for the Rrst two principal components illustrate the
distribution of objects (analytes) according to their
inputs to the principal components.
There is an approach is QSRR in which principal
components extracted from analysis of a large table
of structural descriptors of analytes are regressed
against the retention data in a multiple regression, i.e.
principal component regression (PCR). The partial
least squares (PLS) approach with cross-validation
also Rnds application in QSRR.
4066 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY

Neural networks (NN) is a method of data analysis Table 1 Structural descriptors in QSRR
that emulates the brains way of working. NNs are
considered powerful tools and techniques for carry- Classification Descriptors
ing out signal processing, modelling, forecasting and Molecular bulkiness Carbon number
pattern recognition. A NN has its input neurons that descriptors Molecular mass
load the system with descriptor values. Next, there Refractivity
are the hidden layers that weight and mix the incom- Polarizability
Van der Waals volume and area
ing signals, and an output layer with neurons predict-
Solvent-accessible volume and area
ing the calibrated response values. The advantage of Total energy
NNs lies in nonlinear transformations of signals oc-
curring at every neuron. The NNs are trained to Molecular geometry Length-to-breadth ratio
respond properly using a representative set of struc- descriptors Moments of inertia
Shadow area parameters
tural data and the corresponding retention para-
meters. The well-trained (but not an overtrained) NN Physicochemical Hammett constants
predicts retention based on input information of an empirical Hansch constants
analyte. and semiempirical Taft steric constants
parameters Hydrophobic fragmental parameters
Solubility parameters
Selection of Structural Descriptors Linear solvation energy relationship
(LSER) parameters
The translation of molecular structures into numer- Partition coefficients
ical descriptors is important not only in QSRR but Boiling temperatures
also to many subdisciplines of chemistry and pharma- pKa values
cology.
Molecular polarity Dipole moments
A popular strategy for identifying structural descriptors Atomic and fragmental electron
parameters in QSRR analysis is to start from the excess charges
accepted theories of chromatographic separation. Orbital energies of HOMO and LUMO
Such structural parameters should quantify the Partially charged areas
Local dipoles
abilities of analytes to take part in the postulated
Submolecular polarity parameters
intermolecular interactions that determine chromato-
graphic separations. Empirical or semiempirical Molecular topological Molecular connectivity indices
structural parameters of analytes based on the sol- descriptors Kappa indices
vatochromic comparison method and on linear solva- Information content indices
Topological electronic index
tion energy relationships (LSER) belong to that
category of structural descriptors. Also, reliable pre- Indicator variables Zero-one indices
dictions of retention have been demonstrated using
the LFER-based experimental substituent or fragmen- Ad hoc designed
tal constants. descriptors
The structural descriptors that are commonly used
in QSRR analysis are classiRed in Table 1.
The structural descriptors related to molecular size the QSRR equation apparently loses statistical signiR-
may be related to the ability of an analyte to take part cance.
in nonspeciRc intermolecular interactions (dispersive What is more or less intuitively understood as mo-
interactions or London interactions) with the compo- lecular polarity of an analyte is difRcult to quantify
nents of a chromatographic system. They are the unequivocally. The descriptors of polarity are ex-
factors most often found signiRcant in QSRR analy- pected to account for differences among analytes re-
sis. The bulkiness parameters are decisive in the de- garding their dipole}dipole, dipole-induced dipole,
scription of separations of closely congeneric ana- hydrogen bonding and electron pair donor}electron
lytes. For example, carbon number normally sufRces pair acceptor (EPD}EPA) interactions. To Rnd a good
to differentiate the members of a homologous series. descriptor of these chemically speciRc interactions is
On the other hand, when dealing with the set of difRcult; the more so since changes in the polarity of
analytes of the same size (e.g. isomers), they may an analyte also change its ability to take part in
appear not to be signiRcant in QSRR analysis. This size-related interactions and affect analyte geometry.
does not mean that dispersive interactions are mean- Obviously, geometry-related or molecular shape
ingless for separation of congeners but just that they parameters are difRcult to quantify one-dimen-
are closely similar, and hence the respective term in sionally. Single numbers reSecting molecular shape
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
differences are adequate only in the case of rigid and
planar solutes. They become signiRcant in QSRR
equations if the range of analytes considered com-
prises compounds of similar size and polarity.
Physical meaning of the molecular graph-derived
descriptors is never clear a priori. It is rather that
good QSRR allow for assigning physical meaning to
individual topological indices.
The empirical physicochemical parameters have
good informative value for determining the mecha-
nism of retention operating in a given chromato-
graphic system. The problem is, however, the lack of
such descriptors for the analytes of interest in actual
QSRR studies.
Indicator values (dummy variables) 0}1 are as-
signed depending on the presence or absence of
a given structural feature in an analyte molecule.
They serve to improve statistics but help occasionally
to identify a structural descriptor of real physical
signiRcance.
The established structural descriptors listed in
Table 1 seldom sufRce to derive QSRR for the actual
chromatographic data and often ad hoc descriptors
have to be designed and included. QSRR analysis
helps to test the predictive potency of the proposed
structural descriptors, which may also appear suit-
able for deriving other kinds of structure}property
relationships.
Prediction of Retention
Prediction of retention within homologous series is
feasible owing to the linear relationships that are
normally observed between retention parameters,
log k, and carbon numbers of analytes, n (Figure 2).
The slopes of lines, B, for various homologous series
chromatographed under the same conditions are very
similar, whereas the intercepts, A, may vary:
log k"A#Bn [6]
Occasionally linear correlations are observed be-
tween retention parameters and molecular bulkiness
descriptors of analytes that are not homologues.
A good prediction of retention within a series of
related nonpolar analytes, such as polyaromatic hy-
drocarbons (PAH) or alkylbenzenes, can be obtained
using van der Waals volume as the structural descrip-
tor.
The bulkiness descriptors can account for separ-
ation of analytes when dispersive interactions
(London interactions) are the only interactions
effective in a given chromatographic system, or when
differences in polar interactions among analytes are
not signiRcant.
4067
Figure 2 Plots of log k versus carbon number, n, of analyte for
HPLC on a polyfunctional C18-bonded silica with pure methanol
eluent at 253C: n-alkanes (), methyl-n-alkanoates () and n-
alkanols (*). (Reprinted from Tchapla A, Herson S, Lessellier
E and Colin H (1993) General view of molecular interaction
mechanisms in reversed-phase liquid chromatography. Journal of
Chromatography A 656: 81}112, with permission of Elsevier
Science.)
The ability of an analyte to take part in polar
interactions is difRcult to characterize by means of
a single descriptor. Hence simple QSRR involving
analyte polarity descriptors, e.g. dipole moments, are
rare.
Normally in chromatography (excepting afRnity
chromatography) molecular shape effects on reten-
tion are of minor importance in comparison with the
effects of molecular size and molecular polarity. In
the case of planar/nonpolar PAH isomers, retention is
linearly related to a shape descriptor (the degree of
elongation of the analyte molecule).
There are numerous reports on good performance
of the molecular connectivity index (RandicH index)
and its modiRcations in predicting retention of con-
generic analytes, including isomers. The correlations
are good when retention is on nonpolar stationary
phases, but not when polar phases are involved.
Whereas on the nonpolar phases the nonspeciRc dis-
persive interactions determine differences in retention
among the analytes, the more speciRc polar interac-
tions become discriminative in the case of
polar phases (and polar analytes).
Using substituent electronic constants to derive
simple QSRR with a real retention prediction ability
has seldom succeeded. A wider application in that
respect is found in Hansch substituent hydrophobic
constants, , and Rekker or Hansch}Leo fragmental
hydrophobic constants, f. The sums of these constants
(plus corrections due to intramolecular interactions)
account well for retention in reversed-phase liquid
chromatographic systems.
Regarding the latter systems, even better predic-
tions are provided by an empirical parameter } the
4068 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
logarithm of the n-octanol}water partition coefRc-
ient, log P. Another useful empirical retention pre-
dictor appears occasionally to be boiling point, Tb,
for example the boiling point of isomeric hydrocar-
bons in the gas chromatography.
Prediction of retention of variously substituted de-
rivatives of parent compounds in a given separation
system can be based on the Martin rule:
n
log kS"log kP# i [7]
i"1
In eqn [7] kP is the retention parameter of a parent
compound, kS is the corresponding value for the de-
rivative carrying n substituents and the summation
represents the retention increments due to individual
substituents i. Having appropriate values for func-
tional groups of interest, one needs only to determine
retention of the parent structure to be able to calcu-
late retention of a derivative. To get reliable predic-
tions, correction factors are introduced in eqn [7] to
account for mutual interactions between substituents
(electronic, steric, hydrogen bonding). However, in
the case of polyfunctional analytes, interactions be-
tween substituents make retention predictions of
rather limited value.
A semiempirical description of reversed-phase
HPLC systems, allowing for the prediction of the
relative retention and selectivity within a series of
analytes, has been developed by Jandera. The
approach consists of determining the interaction
indices and the structural lipophilic and polar
indices. A suitable set of standard reference analytes
is necessary to calibrate the retention (or selectivity)
scale.
The multiparameter QSRR based on linear solva-
tion energy relationships (LSER) possess a high
predictive power regarding reversed-phase HPLC
retention. The model developed by Abraham and
co-workers to predict the n-octanol}water partition
coefRcient, log P, appears to be useful also in the case
of log k from reversed-phase liquid chromatography:
log k"c0#c1Vx#c2H
2 #c3 2 #c4 2 #c5R2
H H
[8]
In eqn [8] Vx is the so-called McGowans character-
istic volume, which can be calculated simply from
molecular structure; H2 is the dipolarity/polarizability
of the analyte, which can be determined through
gas-chromatographic and other measurements; H 2 sed-phase HPLC retention of simple aromatic solutes
is the effective or summation hydrogen bond acidity;
H 2 is the effective or summation hydrogen bond
basicity; and R2 is an excess molar refraction, which
can be obtained from refractive index measurements
and is an additive quantity. The LSER-based struc-
tural descriptors are available for a large number of
compounds.
Experimentally determined ionic radius, Ir, and en-
ergy of ionization, Ei, accompanied by atomic mass,
Am, produce a three-parameter regression equation
predicting capillary electrophoretic mobility of metal
cations. The QSRR equation indicates that atomic
mass approximates to the retardation factors (nega-
tive input to mobility) whereas the ionic radius is an
approximate measure of the effective charge on the
analyte. Energy of ionization can play the role of
a secondary, but signiRcant, correction factor to the
effective charge. Unfortunately, there are no good
QSRR to predict the CE retention of organic analytes.
A typical multiparameter approach to predicting
retention of an unknown compound based on struc-
tural features and chromatographic properties of
other representative compounds consists of generat-
ing a multitude of analyte descriptors that are next
regressed against retention data. The structural
descriptors are usually derived by computational
chemistry methods for the energy-optimized confor-
mations. Software systems have been developed that
produce and process hundreds of quantum chemical,
molecular modelling, topological and semiempirical
additive}constitutive descriptors after sketching the
molecule on the computer. Observing all the rules
and recommendations for meaningful statistics, the
minimum number of descriptors (uncorrelated) is se-
lected that are needed to produce a QSRR equation
with a good predictive ability. The descriptors that
eventually serve to predict retention of new analytes
are sometimes of obscure physical meaning. For
example, it is difRcult to ascribe deRnite physical
sense to such descriptors reported in predictive QSRR
as the surface area of the positively charged portion
of the molecule divided by the total surface area or
total entropy of the molecule at 300 K divided by the
number of atoms. Nevertheless, for several groups of
compounds, prediction of retention by means of
QSRR is reliable enough for identiRcation purposes,
especially when there is no better alternative. Exemp-
lary predictive QSRR are for polychlorinated diben-
zofurans and biphenyls, anabolic steroids, stimulants
and narcotics used as doping agents, barbituric acid
derivatives, polyaromatic and nitrated polyaromatic
hydrocarbons, etc.
There are QSRR of useful predictive potency that
comprise only physically interpretable terms. Rever-
on typical octadecyl silica columns has been related to
a molecular bulkiness descriptor (total energy), a po-
larity descriptor (local dipole) and the energy of the
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY 4069

highest occupied molecular orbital of analytes. Good
prediction of liquid chromatographic retention of
about 50 aromatic acids was realized using as re-
gressors the calculated theoretical logarithm of the
n-octanol}water partition coefRcient (log P), the
dipole moment, the principal ellipsoid axes, the sum
of the charges on the oxygen and nitrogen atoms, the
energy of the highest occupied molecular orbital
(HOMO) and the electrophilic superdelocalizability
for the aromatic carbon atom.
In Figure 3 is illustrated the predictive performance
of QSRR for 216 HPLC retention data points. The
points are for 36 analytes chromatographed in six
eluents on a diol stationary phase. The eluents were
heptane containing 0.5% of tetrahydrofuran, dioxane,
ethanol, propanol, octanol and dimethylformamide.
In Figure 3 the log k data experimentally measured
are plotted against the values predicted by eqn [9]:
log k"0.100 polarizability (analyte)
!0.400 log P (analyte)
!0.330EHOMO (analyte)
#1.106EHOMO (eluent)
#0.401ELUMO (eluent)
with the values n"216, R"0.97, s"0.097 and
F"655. In this equation n is the number of data
points used to derive regression equation, R is the
multiple correlation coefRcient, s is the standard error
of estimate and F is the value of F-test of signiRcance;
ELUMO denotes energy of the lowest unoccupied mo-
lecular orbital and EHOMO is energy of the highest
occupied molecular orbital.
Figure 3 reSects realistically the actual predictive
power of QSRR. The predictive QSRR equations
normally hold within the family of analytes for which
they were derived and may be used for tentative
identiRcation of chromatographic peaks.
In recent years a three-dimensional quantitative
structure}biological activity relationship method
known as comparative molecular Reld analysis
(CoMFA) has been applied to construct a 3D-QSRR
model for prediction of retention data. The CoMFA
3D-QSRR model is obtained by systematically samp-
ling the steric and electrostatic Relds surrounding
a set of analyte molecules and then correlates the
differences in these Relds to the corresponding differ-
ences in retention.
Several reports have recently appeared on predic-
tions of retention data from structural descriptors by
means of neural networks (NN). By now the predic-
tions provided by NN are of similar reliability to
those obtained from regression models.
QSRR and Molecular Mechanisms
of Retention
[9]
The QSRR equations that comprise physically inter-
pretable structural descriptors can be discussed in
terms of the molecular mechanisms involved in the
chromatographic process. There is evidence that dif-
ferent structural parameters of analytes account for
retention differences in GC on polar versus nonpolar
stationary phases. Also, the structural descriptors in

Figure 3 Plot of log k predicted by eqn [9] against experimental data determined on a diol column for 36 chalcone derivatives with
heptane eluent containing 0.5% tetrahydrofuran, dioxane, ethanol, propanol, octanol or dimethylformamide. (Reprinted with permission
from Azzaoui K and Morin-Allory L (1996) Comparison and quantification of chromatographic retention mechanisms on three stationary
phases using structure}retention relationships. Chromatographia 42: 389}395. Copyright Friedrich Vieweg & Sohn.)
4070 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
QSRR equations that are valid for normal and for
reversed-phase HPLC systems are different. In
the case of apparently similar chromatographic sys-
tems, the differences in retentive properties of station-
ary phases may be reSected by the magnitude of the
regression coefRcients for analogous descriptors.
Comparative QSRR studies are especially valuable
when new chromatographic phases are introduced.
A general rule is that QSRR equations are charac-
terized by two kinds of structural descriptors: one
that accounts for the bulkiness or size of an analyte
and one that encodes its polar properties. Size de-
scriptors are always signiRcant in GC on nonpolar
phases and in reversed-phase HPLC, whereas the sig-
niRcance of polar descriptors increases as polarity of
both the stationary phases and the analytes increases.
Publications give evidence that in GC on polar
phases and in normal-phase (adsorption) liquid
chromatography (HPLC and TLC) the chemically
speciRc, molecular size-independent intermolecular
interactions are assumed to play the main retention-
determining role. For example, the HPLC retention
parameters determined for substituted benzenes on
porous graphitic carbon are described by QSRR
equations comprising polarity descriptors but no bulk
descriptors. Because, in general, it is difRcult to quan-
tify polarity properties precisely, the QSRR for GC
on polar phases and for normal-phase HPLC are
usually of lower quality then for GC on nonpolar
phases and in reversed-phase liquid chromatography.
QSRR differentiate in a quantitative (statistical)
manner stationary phase materials of different chem-
ical nature. However, when the stationary phases that
belong to the same chemical class are compared, such
as hydrocarbon-bound silicas for reversed-phase
HPLC, the results obtained are ambiguous.
The proper QSRR strategy aimed at objective char-
acterization of differences in retentive potency of in-
dividual chromatographic systems should employ
a well-designed set of test analytes. The analytes
should be selected in such a way that, within the test
set, the intercorrelations are minimized among the
individual analyte structural descriptors. At the same
time, the selection of test analytes should provide
a wide range and even distribution of individual
structural descriptor values. In addition, the series of
analytes should be large enough to assure statistical
signiRcance of the QSRR equations but not too large
so as to remain experimentally manageable.
Often the retention parameters of test analytes are
Rrst linearly regressed against the reference log P
values from the n-octanol}water partition system.
Good correlations obtained are usually interpreted as
evidence of a partition mechanism operating in the
chromatographic system under study.
Several QSRR studies aimed at comparison of
retention mechanisms on individual alkyl silica
reversed-phase materials for HPLC have employed
LSER-based analyte parameters. It was observed gen-
erally that the most important analyte parameters
that inSuence retention are bulkiness-related (molar
volume, molar refraction) and hydrogen bonding
basicity, but not hydrogen bonding acidity. The
analyte dipolarity/polarizability appeared to be a
minor but often signiRcant factor. However, on
poly(st yrene}divinylbenzene) stationary phases the
dipolarity/polarizability term provides an important
positive input to QSRR.
The results of QSRR studies in which eqn [8] was
applied to the retention parameters, log kw, from
measurement on alkyl silica phases with methanol}
water and acetonitrile}water eluents are instructive.
The most signiRcant parameters appeared to be hy-
drogen bond basicity (H 2 ) and McGowan volume
(VX) of analytes. The third signiRcant parameter in
QSRR equations is either dipolarity/polarizability
(H
2 ) in the case of methanolic eluents or hydrogen
bond acidity (H 2 ) in the case of acetonitrile-modiRed
mobile phases.
The rationalization of these results might be as
follows. The dispersive interactions of analytes (char-
acterized by VX) and hydrogen bonding interactions
in which an analyte molecule is a hydrogen-bond
acceptor (characterized by H 2 ) signiRcantly affect
the retention of analytes in both water}meth-
anol/stationary phase and water}acetonitrile/station-
ary phase equilibrium systems. However, in
methanolic systems the third signiRcant factor deter-
mining equilibrium is the ability of an analyte mol-
ecule to be preferentially attracted by polar molecules
of methanol owing to the dipole}dipole and dipole-
induced dipole interactions (characterized by H 2 ). In
the systems containing acetonitrile, the H 2 descriptor
becomes insigniRcant in QSRR equations. Instead,
the ability of the analyte to be preferentially attracted
by the eluent owing to hydrogen bonding in which the
analyte serves as a hydrogen bond donor (character-
ized by H 2 ) becomes more signiRcant. The well-
known hydrogen bond acceptor properties of
acetonitrile manifest themselves in eqn [8] as a reten-
tion-decreasing term k4 H 2 with a negative value of
the k4 regression coefRcient.
Most readily interpretable would appear to be the
molecular mechanism of retention in terms of QSRR
equations comprising the parameters of analytes ob-
tained from molecular modelling. One can easily as-
sign physical meaning to van der Waals surface area
or solvent-accessible molecular surface area (SAS) as
differentiating the strength of dispersive interactions
between the analyte and the molecules forming
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
chromatographic systems. Dipole moment (
) should
also account for differences among analytes regarding
their dipole}dipole or dipole-induced dipole interac-
tions. Energies of the lowest unoccupied molecular
orbital (ELUMO) and the highest occupied molecular
orbital (EHOMO) should explain the differences in the
tendency of analytes to take part in the charge trans-
fer interactions. Yet reliable QSRR employing these
structural descriptors are rare and hold only for se-
lected sets of analytes.
In QSRR concerning reversed-phase HPLC reten-
tion parameters, the net positive effects on retention
are due to the analyte bulkiness descriptors. The
dispersive attractions of an analyte are stronger with
the bulky hydrocarbon ligand of the stationary phase
than with the small molecules of aqueous eluent. The
net effect on retention provided by dipole moment
(or its square) is negative. This is because the
dipole}dipole and dipole-induced dipole attractions
are stronger between the polar (polarized) analyte
and polar molecules of eluent than between the same
analyte and the nonpolar hydrocarbon ligand of the
stationary phase. Unfortunately, these types of QSRR
are not precise enough to differentiate individual al-
kyl silica stationary phase materials in a quantitative,
statistically signiRcant manner. They are signiRcant
enough, however, to reSect the differences in reten-
tion mechanism operating in the reversed-phase and
in the normal-phase HPLC systems or in GC on
nonpolar and polar phases.
Factorial methods of data analysis (principal com-
ponent analysis, correspondence factor analysis, spec-
tral mapping analysis) provide classiRcation of
stationary phases based on retention data determined
for short series of test analytes. Among the commer-
cially available materials for HPLC those can be
selected that possess closely similar retention proper-
ties. Also, a stationary phase with clearly distinctive
properties can be identiRed, which can be useful for
speciRc method development.
Chromatographic Methods
of Determination of Hydrophobicity
Hydrophobicity or lipophilicity is understood to be
a measure of the relative tendency of an analyte
to prefer a nonaqueous over an aqueous environ-
ment. The partition coefRcients of the substances
may differ if determined in different organic}
water solvent systems but their logarithms are often
linearly related. Octanol}water is a reference system
that provides the most commonly recognized hydro-
phobicity measure: the logarithm of the partition
coefRcient, log P. The standard shake-Sask method
for determining partition coefRcients in liquid}liquid
4071
systems has several disadvantages. Having appropri-
ate QSRR, the chromatographic data can
be used to predict log P. Many good correlations of
reversed-phase liquid chromatographic (HPLC or
TLC) parameters with log P have been reported for
individual chemical families of analytes and
chromatographic methods for assessing the hydro-
phobicity of drugs and environmentally important
substances have ofRcially been acknowledged
and included in the OECD Guidelines for Testing
Chemicals.
On the other hand, the partition chromatographic
systems are not identical with the n-octanol}water
partition system. Each chromatographic system pro-
duces an individual scale of hydrophobicity. Hence
attempts to reproduce log P by means of liquid
chromatography are only partially successful. Centri-
fugal countercurrent chromatography (CCCC) pro-
vides a better chance of mimicking log P but the
inconvenience of this method and the need for special
equipment hinder its wider application.
The versatility of chromatographic methods of hy-
drophobicity assessment can be attributed to the use
of organic modiRers in aqueous eluents. Normally,
the retention parameters determined at various or-
ganic modiRer}water (buffer) compositions are ex-
trapolated to zero organic modiRer content.
The extrapolated parameters (log kw from HPLC
and R0M from TLC) depend on the organic modiRer
used.
Alkyl silica stationary phases and methanol}water
eluent are most often used in hydrophobicity studies.
The problem with these phases is that the hydropho-
bicity of nonionized forms of organic bases cannot
be determined because of the chemical instability
of silica-based materials at higher pHs (above about
pH 8). Also, speciRc interactions of analytes with the
free silanols of alkyl silicas disturb partition pro-
cesses.
The limitations of standard reversed-phase mate-
rials have been partially overcome by introducing
modern specially deactivated hydrocarbon-bonded
phases, immobilized on alumina or zirconia supports
and on polymeric materials. Using the latter two
types of stationary phase materials one can determine
HPLC retention factors under acidic, neutral and
alkaline conditions. That way a universal, continuous
chromatographic hydrophobicity scale can be con-
structed, as is the standard log P scale.
Hydrophobic properties of xenobiotics are as-
sumed to affect their passive diffusion though biolo-
gical membranes and binding to pharmacological
receptors. If the hydrophobicity measuring system
is to model a given biological phenomenon, then
similarity of the component entities is a prerequisite.
4072 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY

Figure 4 Chemical structures of ligands of three types of immobilized artificial membrane (IAM) columns of Pidgeon (Liu H, Ong S,
Glunz L and Pidgeon C (1995) Predicting drug}membrane interactions by HPLC: structural requirements of chromatographic surfaces.
Analytical Chemistry 67: 3550}3557. Copyright 1995 American Chemical Society.) and a schematic model of a biological membrane.

Hence the partition system to model the transport
through biological membranes should be composed
of an aqueous phase and an organized phospholipid
layer. The immobilized artiRcial membranes (IAM)
introduced by Pidgeon as a packing material for
HPLC (Figure 4) appear to be reliable and convenient
models of natural membranes.
Correlations between log k data determined on
IAM-type columns and log P values are generally not
high nor are the correlations between log k from IAM
columns and log kw determined by liquid chromato-
graphy employing standard stationary phase mater-
ials. This means that retention data determined on
IAM columns contain information on the properties
of analytes that is distinct from that provided by the
n-octanol}water system and by the hydrocarbon}
silica reversed-phase columns. There is evidence that
the hydrophobicity characteristics provided by IAM
columns are better suited for modelling the phar-
macokinetics of drug processes.
Retention Parameters in Predicting
Bioactivity of Analytes
Biological processes of drug absorption, distribution,
excretion and drug}receptor interaction are dynamic
in nature as are the analytes distribution processes in
chromatography. The same fundamental inter-
molecular interactions determine the behaviour of
chemical compounds in both biological and
chromatographic environments. Modern techniques
and procedures of HPLC and CE allow for inclusion
 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
of biomolecules as active components of separation
systems and QSRR processing of appropriate sets of
chromatographic data can reveal systematic informa-
tion regarding the xenobiotics studied. This informa-
tion can be used to elucidate molecular mechanisms
of pharmacological action and to facilitate rational
drug design.
It is often sufRcient to identify and employ a speci-
Rc chromatographic system yielding hydrophobicity
values of analytes best conforming to log P data.
Normally, chromatographic systems that produce re-
tention parameters less correlated to log P are dis-
carded but information extracted from diversiRed
retention data may be more appropriate for predic-
tion of pharmacological properties of analytes than
information based on an individual hydrophobicity
scale. To extract meaningful information from diver-
siRed (yet often highly mutually intercorrelated) sets
of data, multivariate chemometric methods of data
analysis are employed. Large matrices of retention
data determined for test series of analytes in many
chromatographic systems differing in type of station-
ary and/or mobile phases, are processed by
factorial methods, usually by principal component
analysis (PCA). If two to three extracted abstract
factors (principal components) account for most of
the variability in a large set of retention data then the
distribution of test analytes can be presented graphi-
cally. Clustering of analytes owing to similarity of
their chromatographic behaviour in diverse separ-
ation systems is usually observed. If that clustering
agrees with the pharmacological classiRcation of the
test agents, then recalculations are done after includ-
ing the retention data for drug candidates. Indications
on potential pharmacological activity of new analytes
can be obtained even before biological experiments.
This approach can facilitate preselection of drug can-
didates, especially among a multitude of compounds
produced by combinatorial chemistry. The challenge
is to design and select the chromatographic systems
yielding retention data of sufRcient classiRcation
potential.
Figure 5 shows the distribution of drugs belonging
to several pharmacological classes on the plane deter-
mined by the Rrst two principal components, which
together account for 81.5% of the variance in the
retention data measured in eight HPLC systems. The
HPLC systems employed different stationary phases
(standard and specially deactivated hydrocarbon
bonded silicas, polybutadiene-coated alumina, immo-
bilized artiRcial membranes and immobilized 1-acid
glycoprotein). Methanol}buffer eluents of varying
composition and pH were used. The clustering of
analytes is consistent with their established pharma-
cological classiRcation. Also, the partial overlap of
4073
Figure 5 Pharmacologically consistent distribution scatterplot
of drug classes on the plane determined by two first principal
components extracted from a 8;83 (drugs;HPLC systems)
matrix of diversified retention data. Roman numbers denote: I,
psychotropic drugs; Ia, inactive phenothiazines; II, -adrenolytics;
III, histamine H1 receptor antagonists; IV, histamine H2 receptor
antagonists; V, drugs binding to -adrenoceptors. (Reprinted from
Nasal A, BucinH ski A, Bober L and Kaliszan R (1997) Prediction of
pharmacological classification by means of chromatographic
parameters processed by principal component analysis. Interna-
tional Journal of Pharmaceutics 159: 43}55, with permission of
Elsevier Science.)
individual clusters is interpretable in terms of par-
tially overlapping pharmacological properties of indi-
vidual drugs.
There are individual processes of drug action that
are satisfactorily modelled by HPLC on immobilized
artiRcial membrane (IAM) columns. QSRR equations
have been reported predicting several pharmaco-
kinetic parameter of -adrenolytic drugs from their
log k parameters determined on IAM columns. Good
predictions by means of log kIAM have also been
reported regarding antihaemolytic activity of
phenothiazine neuroleptics. The human skin per-
meation of steroids also correlates better with
log kIAM than with log P.
The log kIAM alone will not sufRce to predict bind-
ing of basic drugs to the serum protein, 1-acid glyco-
protein (AGP). However, combining that parameter
with atomic excess charge on aliphatic nitrogen, Nch,
and a size parameters, ST, in a multiple regression
equation results in a good prediction of AGP binding.
The ST parameter is the area of a triangle having one
vertex on the aliphatic nitrogen and the two remain-
ing vertices on the extremely positioned atoms in the
drug molecule (Figure 6). The QSRR equation has
4074 III / QUANTITATIVE STRUCTURE-RETENTION RELATIONSHIPS IN CHROMATOGRAPHY
the form:
log kAGP"0.6577 ($0.0402) log kIAM
#3.342($0.841)Nch
!0.0081($0.0030)ST
#1.688($0.245) [10]
with the values n"49, R"0.929, s"0.163,
F"92 and p(10\5.
Equation [10] may be useful as a Rrst approxima-
tion to the relative binding of a drug to AGP without
the need to perform biochemical experiments. It can
help to identify structural features of the binding site
of basic drugs on AGP (Figure 6). The site can be
modelled as a conical pocket. Its internal surface
contains hydrophobic regions at the base of the cone
and an anionic region close to the apex of the cone.
Protonated aliphatic nitrogen guides drug molecules
towards the anionic region. Hydrophobic hydrocar-
bon fragments of the interacting drugs provide an-
choring in the hydrophobic regions of the binding
site. There is a steric restriction for the molecule to
plunge into the binding site.
QSRR analysis of HPLC data determined on an
immobilized human serum albumin (HSA) column
helps to suggest the topography of two binding sites
of different afRnity to benzodiazepine enantiomers.
Also, the mechanism of interaction of phenothiazine
neuroleptics with melanin can be rationalized by
means of QSRR analysis of HPLC retention data.
Another QSRR study concerns interactions of drugs
with immobilized keratin and collagen.
In general, QSRR analysis of retention parameters
determined on immobilized biomacromolecules can
yield reliable predictions of activity and identiRcation
of the required binding structural properties of
Figure 6 Mode of binding of the organic base drugs derived
from QSRR analysis of HPLC data determined on an immobilized
1-acid glycoprotein column. (Adapted with permission from Kalis-
zan R, Nasal A and Turowski M (1995) Binding site for basic
drugs on 1-acid glycoprotein as revealed by chemometric analy-
sis of biochromatographic data. Biomedical Chromatography 9:
211}215. Copyright John Wiley & Sons Limited.)
drugs and drug candidates. The approach appears
especially promising now that biotechnologically
produced pharmacological receptors are becoming
available.
Concluding Remarks
In 1991 Giddings wrote Because pure theory is im-
practical, progress in understanding and describing
molecular equilibrium between phases requires
a combination of careful experimental measurements
and correlations by means of empirical equations and
approximate theories. This has been realized in a sys-
tematic manner over a period of 20 years through
QSRR analysis. During that time a consistent re-
search startegy has been developed and established
within the area. Easy access to computers equipped
with advanced statistics and molecular modelling
software has ensured rapid progress and engendered
a wide interest in QSRR among chromatographers
and other specialists.
QSRR are employed by analytical chemists to help
identify unknown members of individual classes of
analytes of pharmacological, toxicological, environ-
mental or chemical interest. At the same time, QSRR
of good retention prediction capability help to ident-
ify structural descriptors for analytes that provide
acceptable estimates of properties other than
chromatographic ones. In this way, chromatographic
systems allowing for fast and convenient evaluation
of analyte hydrophobicity can be identiRed. Also,
QSRR conRrm the suitability of the LSER-based de-
scriptors for property predictions.
Well-designed QSRR studies are helpful in identi-
fying the structural features within a family of
analytes that affect retention in a given separation
system. In this way molecular mechanisms of reten-
tion may be explained. With a designed test series of
analytes the QSRR derived for retention data deter-
mined in individual separation systems provide objec-
tive, numerical characteristics for these separation
systems. This is especially useful for quantitatively
comparing retention properties of various stationary
phase materials.
Chromatographic retention data can be employed
to predict pharmacological properties of analytes. By
employing chromatographic systems comprising bio-
macromolecules, large amounts of data can be ob-
tained that reSect differences among analytes with
regard to their interactions with given biomac-
romolecules. These data can be used to derive QSRR
explaining the mechanism of drug}biomacro-
molecule interactions. In effect, the topography
of binding sites for drugs on individual biomacro-
molecules can be characterized. By employing

biotechnologicallly acquired pharmacological recep-
tor proteins to generate drug}receptor interaction
data and by applying QSRR analysis, the preselection
of drug candidates can be facilitated and experiments
on animals reduced.
See also: II/Chromatography: Liquid: Mechanisms:
Reversed Phases.
Further Reading
Carr PW, Martire DE and Snyder LR (eds) (1993) The
retention process in reversed-phase liquid chromatogra-
phy. Special Volume of Journal of Chromatography
A 656: 1}618.
ForgaH cs E and CserhaH ti T (1997) Molecular Bases of
Chromatographic Separations. Boca Raton, FL: CRC
Press.
REACTIVE DISTILLATION
S. M. Mahajani, Monash University, Clayton,
Victoria, Australia
S. P. Chopade, Michigan State University, East Lans-
ing, MI, USA
Copyright ^ 2000 Academic Press
Introduction
Reactive distillation is a combination of separation
and reaction in a single process. Commercial reactive
distillation processes for the manufacture of methyl
t-butyl ether (MTBE) and methyl acetate were suc-
cessfully commissioned in 1981 and 1983, respective-
ly. These processes have a distinct edge over their
conventional predecessors. The reactive distillation
process is particularly advantageous in the case of
reversible reactions where the conversion is limited by
thermodynamic equilibrium. Some of the important
beneRts of reactive distillation are: reduced capital
cost; employment of low mole ratios of reactants;
energy saving owing to utilization of the heat of
reaction; and automatic temperature control and
elimination of hot spots. The commercial process of
MTBE manufacture has shown that heterogeneous
catalysts such as ion exchange resins can be advantage-
ously used in reactive distillation columns. Innovative
techniques of conRning the small size resin particles
(0.3}2 mm) in the column, allowing efRcient
solid}liquid contact and high void fraction, have been
III / REACTIVE DISTILLATION 4075
Giddings JC (1991) UniTed Separation Science. New York:
Wiley.
Jinno K (ed.) (1997) Chromatographic Separations Based
on Molecular Recognition. New York: Wiley-VCH.
Jurs PC (1996) Computer Software Applications in Chem-
istry, 2nd edn. New York: Wiley.
Kaliszan R (1987) Quantitative Structure}Chro-
matographic Retention Relationships. New York: Wiley.
Kaliszan R (1997) Structure and Retention in Chro-
matography. A Chemometric Approach. Amsterdam:
Harwood Academic Publishers.
Kier LB and Hall LH (1986) Molecular Connectivity in
Structure}Activity Analysis. Letchworth: Research
Study Press.
Plis\ ka V, Testa B and van de Waterbend H (eds) (1996)
Lipophilicity in Drug Action and Toxicology. Wein-
heim: VCH.
Smith RM (1995) Retention and Selectivity in Liquid
Chromatography. Amsterdam: Elsevier.
developed by CDTech, Sulzer, Koch Engineering and
BASF. An alternative approach is to prepare a cata-
lyst in the form of conventional column packing and
pack it directly into the reactive distillation column.
Recognizing the potential of reactive distillation
for a particular process is a difRcult task, as not all the
reactions can be conducted effectively in this way.
Once its potential has been identiRed, the next step is
to design the reactive distillation column for the re-
quired task. The simultaneous existence of multiple
processes such as mixing, mass transfer and reaction
are involved, and the design method requires thor-
ough knowledge of both chemical and physical equi-
libria as well as the reaction kinetics. Graphical
representations of liquid phase compositions, called
residue curve maps or distillation maps, are com-
monly used to analyse the reactive distillation pro-
cess. Though some efforts have been made to study
the underlying theory of the design method, the work
is still at its preliminary stage. Another approach to
understanding the behaviour of this process is to
perform computer simulations and predict the perfor-
mance of a column of known conRguration.
In this article the important aspects of commercial
reactive distillation processes of MTBE and methyl
acetate manufacture are described in detail. Recent
trends in the experimental and theoretical investiga-
tions in this area are also outlined. The potential
importance of reactive distillation in some industrial
4076 III / REACTIVE DISTILLATION
processes such as hydrolysis of methyl acetate and
recovery of chemicals from aqueous streams is dis-
cussed.
MTBE Production
The area of particular interest where reactive distilla-
tion can be used is the production of fuel ethers such
as MTBE. Gasoline reformulation using these ethers
as environmentally benign octane boosters has been
driven by various Clean Air Acts, which have boosted
MTBE production to a new level. By 2001 the pro-
duction of MTBE is expected to be 25;106 tonnes
per annum worldwide. t-Amyl methyl ether (TAME)
and ethyl t-butyl ether (ETBE) are also emerging as
promising fuel additives. In addition to its property as
an antiknock agent to enhance the octane number of
the fuel, MTBE improves the water tolerance limit
of the fuel and has a higher caloriRc value than that of
other additives such as methanol.
Another important aspect of carrying out the
etheriRcation to near complete conversion is its efR-
cient use in separating the iso-oleRns from the reRnery
stream containing both normal and secondary bu-
tenes (C4 or C5), which are otherwise very difRcult to
isolate. A reactive distillation column can handle the
mixed oleRns quite effectively and exploits the pres-
ence of inert butenes to improve performance. This
separation is necessary because n-butenes are re-
quired in the pure form for homopolymerization and
as a feed for the oxidative production of butadiene.
Reaction Details
MTBE is a product of the liquid-phase reaction of
isobutylene and methanol, catalysed by a strong
acidic macroreticular ion exchange resin. The
reaction is highly selective, so that methanol reacts
only with isobutylene in the presence of other
C4 oleRns [I].
CH3OH #(CH3)2C"CH2 0 (CH3)3COCH3 [I]
methanol isobutylene MTBE

H0298"!37.7 kJ mol\1
The favourable temperature and pressure ranges for
the reaction to occur are 323}373 K and 5}15 atm,
respectively. The useful side reactions are the dimeriz-
ation and oligomerization of isobutylene and bu-
tadiene as well as the formation of codimers. Since,
until recently, only butadiene-extracted C4 reRnery
streams were used for MTBE production, the only
important by-product is diisobutylene, which consists
of the isomers 2,4,4-trimethyl-1-pentene and 2,4,4-
trimethyl-2-pentene. The other side reactions, which
are of less importance, are formation of t-butanol by
reaction of isobutylene with water present as a feed
impurity, the formation of traces of dimethyl ether by
methanol condensation, and the double bond isomer-
ization of 1-butene. Amberlyst 15, a macroporous
cation exchange resin, is widely used as a catalyst for
this reaction. Numerous investigations on the kinetics
of this reaction system have been reported in the
literature. A model based on systematic studies of
reaction kinetics (eqn [1]) and equilibrium of this
system incorporating the activities of the compounds
has been developed and is used by many investigators
for column simulation studies.


aIB 1 aMTBE
r"mcat qacidkf ! [1]
aMeOH Keq a2MeOH
In eqn [1], mcat is the catalyst loading, qacid is the ion
exchange capacity and ai is the activity coefRcient of
the corresponding component (IB, isobutylene,
MeOH, methanol). The forward reaction rate con-
stant kf and the equilibrium constant Keq have been
Rtted experimentally. As a result of the high polarity
of methanol, the reaction mixture is highly nonideal
and involves formation of two binary azeotropes and
one ternary reactive azeotrope. The activities of the
components can sometimes be up to 20 times their
mole fractions.
Commercial Process
Conventional processes for the manufacture of
MTBE (see Figure 1) use a catalytic reactor with
a slight excess of methanol (methanol/isobuty-
lene"1.05}2). The products correspond to the
near-equilibrium conversion of about 90}95%. The
reaction mixture is separated using distillation, but
suffers from complications resulting from the forma-
tion of binary azeotropes methanol}MTBE and iso-
butylene}methanol. The unreacted isobutylene is also
difRcult to separate from other volatile C4 products.
With the reactive distillation process, almost com-
plete conversion of isobutylene is obtained, thereby
eliminating the separation and recycle problems.
Figure 2 provides a schematic representation of this
process. A Rxed-bed pre-reactor is used to achieve
near-equilibrium conversion. The product stream
equivalent to the equilibrium conversion is fed to the
reactive distillation column, wherein, the residual
amount of isobutylene is reacted with methanol. The
reactive distillation column is composed of three sec-
tions, the middle of which is a reactive zone packed
with a solid catalyst. The top nonreactive rectifying
section performs the separation of inert gases and

Figure 1 Conventional process for MTBE manufacture.
excess methanol, while the bottom section separates
out MTBE in pure form. The boiling points of MTBE
and methanol are 328 K and 337.5 K, respectively.
This may seem surprising, as MTBE is the bottom
product while unreacted high boiling methanol is
collected through the distillate. The behaviour is
caused by the formation of an MTBE}methanol low
boiling azeotrope, which lifts methanol from the
stripping section of the column.
The pioneering work to commercialize this techno-
logy was performed by Smith from Chemical Re-
search and Licensing Company, who has been
awarded several patents for different catalyst struc-
tures, column internals design and Sow schemes.
Some patents have also been assigned to researchers
from ELF who claim to have used alternating cata-
lytic and noncatalytic zones successfully to carry out
the etheriRcation. The efforts in these studies were
directed towards minimizing the pressure drop in the
catalyst bed and providing maximum residence time
Figure 2 Reactive distillation process for MTBE manufacture.
III / REACTIVE DISTILLATION 4077
for the liquid in the catalytic zone. This was achieved
by providing separate free passage to the up Sowing
vapour stream either by packing the catalyst in the
downcomers or by providing annular space in the
catalyst bed, thereby isolating reaction and distilla-
tion zones in a single column. UOP, Koch Engineer-
ing and HuK ls AG have jointly developed the
Ethermax process for producing ethers by reactive
distillation. The process uses Koch Engineerings
Katamax packing, where a solid acid catalyst is con-
Rned in screen envelopes.
Simulation Studies
Following the successful commercialization of the
MTBE process, numerous studies simulating a cata-
lytic distillation column have appeared in the litera-
ture. The basic idea behind simulation studies is to
predict the overall conversion of either isobutylene or
methanol, and examine the product purity at a steady
state for a known column conRguration and feed
composition. Various software packages such as
ASPEN PLUS, SPEEDUP, etc, have been used
effectively for this purpose.
The interesting discovery of multiple steady states
for a column operated under identical conditions has
attracted the attention of many researchers in the
recent past. Several studies examining the reasons for
the existence of these steady states have been re-
ported. Experimental Rndings conRrmed this fact and
showed that the same column conRguration operated
under similar conditions can give rise to two dif-
ferent conversions. Simulation studies using an
ASPEN PLUS Sowsheet simulator for a column
with a total of 17 reactive and nonreactive stages,
operated at 11 atm with two different feed streams of
methanol and butenes, result in either 36% or 99%
isobutylene conversion when methanol is fed to the
10th stage. The methanol feed plate was varied by
following either top-to-bottom or bottom-to-top se-
quence and it was found that only at certain feed
plates (9}12) were multiple conversions realized (see
Figure 3). The steady-state conversion in this multi-
plicity region depends upon which sequence is fol-
lowed to simulate the column. In the upgoing
sequence low conversions are obtained, while the
downgoing sequence is associated with high conver-
sions. Subsequent efforts on column simulation have
conRrmed this Rnding. Installation of the methanol
feed at more than one location has been suggested to
avoid the unwanted steady state caused by column
misoperation.
A mechanistic explanation has been provided as to
why MTBE production by reactive distillation may
yield multiple solutions. It was found that the initial
estimates for temperature and composition proRles
4078 III / REACTIVE DISTILLATION
Figure 3 Steady-state multiplicity behaviour of the MTBE
process.
decide whether a steady-state simulation would con-
verge to a high conversion or a low conversion solu-
tion. In order to obtain a high conversion solution,
the lower section of the column must contain sufR-
cient MTBE to lift the entire amount of methanol to
the reactive zone. Second, in the reactive zone, the
reaction mixture must be diluted to avoid a substan-
tial amount of MTBE decomposition. Initial esti-
mates of the composition at the lowest stage in the
reactive zone are crucial in deciding the nature of the
steady state. This is expected to be due to the inherent
coupling between lift and dilution effects that takes components even if one of the reactants is used in
place on this stage.
Recent simulation studies on reactive distillation of
MTBE and TAME indicate two types of multiple
steady states. The Rrst, discussed above, arises out of
the interaction between reaction and vapour}liquid
equilibrium. The second multiple steady state is re-
lated exclusively to the chemical reaction and arises
because of the highly nonlinear concentration de-
pendence of methanol activity at low operating pres-
sures. The only experimental evidence of multiple
steady states reported so far comes from work on
etheriRcation for TAME synthesis in a pilot plant of
Nestle Oy. It is therefore necessary to perform dy-
namic simulations during the Rrst steps of the design
process in order to avoid dynamic surprises.
Another interesting Rnding of MTBE simulation
studies is the oscillatory behaviour of the reactive
distillation column. Sustained oscillations of boiling
temperature and reSux have been reported in experi-
mental studies on reactive distillation. It has been
proved that the nonreactive and nonideal interactions
k(aHOAcaMeOH(!aMeOAcaH2O/Keq))
r"
(1#KHOAcaHOAc#KMeOHaMeOH#KMeOAcaMeOAc#KH2OaH2O)
between methanol and isobutylene are responsible for
these effects.
Methyl Acetate Production
Methyl acetate is another high volume commodity
chemical that is manufactured commercially using
reactive distillation. It Rnds applications as an inter-
mediate in the manufacture of a variety of polyesters
such as photographic Rlm base, cellulose acetate,
Tenite cellulosic plastics and Estron acetate.
Reaction Details
The reaction of methanol and acetic acid to
give methyl acetate (reaction [II]) has equilibrium
limitations.
CH3OH #CH3COOH 0 CH3COOCH3 #H2O
methanol acetic acid methyl acetate water
[II]

H0298"!8.0 kJ mol\1
aMeOAcaH2O
Keq" "5.2 [2]
aAcOHaMeOH
Equation [2] gives the equilibrium constant Keq as
a function of the activity coefRcients aMeOAc (methyl
acetate), aH2O (water), and aAcOH (acetic acid).
Thus the reaction product will contain all four
excess. The reaction can be conducted in the temper-
ature range 310}393 K and at a pressure of 1 atm.
The only important side reaction is the formation of
dimethyl ether by the condensation of methanol. This
reaction is predominant at high temperatures.
Though the reaction has been commercialized in
a reactive distillation column, it is surprising that
a systematic study on the kinetics of this reaction in
the presence of sulfuric acid as catalyst is not evident
in the open literature. As in the MTBE system, the
rate expression in the form of activities is strongly
preferred because the high polarity of water and
methanol compared to that of methyl acetate leads to
strongly nonideal solution behaviour. Because of the
commercial success of reactive distillation and
the proven potential of the ion exchange resins, some
efforts have been made to propose a rate expression
for an ion exchange resin-catalysed reaction. The
expression for the rate, r, based on kinetic data gener-
ated over a range of molar feed ratios more typical of
reactive distillation conditions, is given by:
[3]
where k is the rate constant, Keq is the equilibrium
constant and the Kis are the adsorption coefRcients
 III / REACTIVE DISTILLATION 4079

involved in the Langmuir}Hinshelwood/Hougen}

Watson model (HOAc, methyl acetate, MeOH,
methanol; MeOAc, methyl acetate, H2O, water). The
expression has been successfully used to verify the
experimentally observed residue curve maps of this
system. The residue curve maps shows no distillation
boundaries and hence, ultrahigh purity methyl acet-
ate and water can be obtained through a proper
design of reactive distillation column.
Commercial Process
Conventional processes before the 1980s used mul-
tiple reactors with a large excess of one of the react-
ants to achieve high conversion of the other.
The product is difRcult to purify because of the
formation of methyl acetate}methanol and methyl
acetate}water azeotropes. Different means to break
the methyl acetate}methanol azeotrope were em-
ployed, such as use of several atmospheric and vac-
uum distillation columns or extractive distillation.
A typical process contained two reactors and eight
distillation columns, making it complex and capital
intensive.
Eastman Kodak has developed a reactive distilla- Figure 4 Reactive distillation process for methyl acetate
manufacture.
tion process for the manufacture of high purity and
ultrahigh purity methyl acetate. The remarkable fac-
tor is that, in spite of the reaction having unfavour- purity methyl acetate as the product. The whole pro-
able equilibrium limitation, high purity product is cess is integrated in a single column, eliminating the
obtained using a near-stoichiometric mole ratio of need for a complex distillation column system and
methanol and acetic acid. The reactive distillation recycle of the methanol}methyl acetate azeotrope.
column used in the process is shown in Figure 4. In A single reactive distillation column at Eastman
order to explain the process, the column can be Kodaks Tennessee plant produces 180 000 metric
divided in four stages starting from the top as: tons per year of high purity methyl acetate. The
(1) methyl acetate enrichment; (2) water extraction; composition proRle of the column shown in Figure 5
(3) reaction; and (4) methanol stripping. The reaction demonstrates that the methyl acetate can be manufac-
occurs in the middle section (section 3) in a series of tured in a single column without need for additional
countercurrent Sashing stages with sulfuric acid as puriRcation steps.
the catalyst. In section 2, acetic acid acts as an ex-
tracting agent and extracts water (breaking the
methyl acetate}water azeotrope) and some methanol.
Hydrolysis of Methyl Acetate
Acetic acid and methyl acetate are separated above Methyl acetate}water mixture is produced in large
the acetic acid feed, in the methyl acetate-enriching quantities from puriRed terephthalic acid (PTA)
section (section 1), allowing pure methyl acetate to be plants. The manufacture of poly(vinyl alcohol) (PVA)
recovered as the overhead product. Methanol is strip- also produces large quantities of methyl acetate
ped from water in the bottom section (section 4) and (1.68 kg per kg PVA). Since methyl acetate is a com-
water is the bottom product. Some intermediate paratively low value solvent, it has to be sold at
boiling compounds are formed because of the impu- a lower price; hence it would be a better idea to
rities present in feed. Hence, a small stream is with- hydrolyse it economically and recover methanol and
drawn just above the catalyst feed point and treated acetic acid for reuse in the process.
separately in an impurity-removal system. The impu- Conventional processes for the hydrolysis
rities are stripped and concentrated, and the meth- of methyl acetate use a Rxed-bed reactor followed
anol#methyl acetate stream is recycled to the by a complex arrangement of several distillation/
reaction zone. The reactive distillation column has extraction columns. The conversion is limited by
been successfully operated at a near-stoichiometric unfavourable equilibrium (equilibrium constant
mole ratio of acetic acid and methanol, yielding high 0.14}0.2) and a large amount of unconverted methyl
4080 III / REACTIVE DISTILLATION
Figure 5 Composition profile in methyl acetate reactive distillation column.
acetate has to be separated and recycled. A schematic
diagram of a typical conventional process is given in
Figure 6. The reaction is carried out in a Rxed-bed
reactor and the product stream contains all four com-
ponents. Four additional columns are required to
separate methanol and acetic acid streams and recycle
unconverted methyl acetate, along with methanol, to
the reactor.
The above has shown how reactive distillation sim-
pliRes the process in the case of the manufacture of
methyl acetate. A similar concept can be applied to
the hydrolysis reaction. A reactive distillation process
has been developed on a laboratory scale for the
hydrolysis of methyl acetate using an ion exchange
resin catalyst in a special form. Converting the pro-
cess from conventional to reactive distillation offers
the possibility of eliminating many complicated steps.
The use of solid acid catalysts obviates the need for
recovery of the spent acid and the use of exotic
Figure 6 Conventional process for hydrolysis of methyl acetate.
construction materials. Resin was moulded into
7 mm;7 mm pellets using polyethylene powder. The
distillation column was directly packed with these pel-
lets, which played the role of both catalyst and packing.
A schematic diagram of the proposed reactive dis-
tillation process is shown in Figure 7. Water is fed at
the top of the reactive section and methyl acetate is
introduced at the bottom of the reactive section. The
column is operated under total reSux of methyl acet-
ate}methanol azeotrope. The stripping section strips
all the methyl acetate and the bottom product is
essentially free of methyl acetate. The bottom product,
which now contains only methanol, water and acetic
acid, can be easily separated using two distillation
columns in series giving methanol and acetic acid as
products. Thus, this process eliminates two main
pieces of equipment from the conventional process: (1)
a water wash column for the separation of methanol
from methyl acetate, and (2) a methanol-enriching

Figure 7 Reactive distillation process for hydrolysis of methyl
acetate.
column for recovery of water-diluted methanol. Con-
versions to the tune of 99% are achieved in this
process. The estimated heat savings are 50% that of
the conventional process.
Recovery and Puri\cation
of Chemicals
The esteriRcation reaction has also been successfully
employed for the recovery of acetic acid from aque-
ous streams. Dilute acetic acid is produced in large
quantities in many processes, such as the manufacture
of cellulose esters, terephthalic acid and dimethyl
terephthalate; and also in reactions such as acetyla-
tion and nitration. The recovery of acetic acid from
these streams is a daunting problem. The conven-
tional methods for recovery are azeotropic distilla-
tion, simple distillation and liquid}liquid extraction.
With the advent of reactive distillation processes,
esteriRcation of acetic acid with methanol seems an
attractive alternative. Laboratory experiments have
been carried out to recover acetic acid in a reactive
distillation column. The column contained commer-
cially available ion exchange resin along with Rashig
rings. The use of a solid acid catalyst offers noncor-
rosive conditions so that a less expensive construction
material can be used. Up to 84% recovery of acetic
acid as methyl acetate was achieved. Hoechst Cel-
anese Corporation has recently described a reactive
distillation process for the recovery of acetic acid
from aqueous solutions as methyl acetate. With the
use of acidic ion exchange resin as catalyst, more than
90% recovery from 5}30% aqueous acetic acid is
claimed. They also suggest the use of Koch Engineer-
ings Katamax packing as catalyst.
Reactive distillation can be applied for the recovery
of many other chemicals from dilute streams. The
polymer industry is often faced with the challenge of
treatment of aqueous formaldehyde solutions, as it is
a nuisance to the environment and it is difRcult to
III / REACTIVE DISTILLATION 4081
remove trace quantities of formaldehyde. Reactive
distillation with methanol, ethanol or ethylene glycol
not only brings down the formaldehyde concentra-
tion to the ppm level, but also yields useful acetal
products. Similarly, nonboiling chemicals such as
glyoxal and glyoxylic acid can be recovered from
their aqueous solutions through the formation of
their corresponding acetals or esters, which can be
separated by distillation. CDTech has recently de-
veloped a reactive distillation process for hydrodesul-
furization, called the CDHDS process, which is aimed
at producing low sulfur fuels to meet stringent future
environmental regulations at the lowest cost.
Reactive distillation has reportedly been employed
for the puriRcation of bisphenol A of polycarbonate
grade, where impurities in the form of carbonyl com-
pounds such as acetone, mesityl oxide, hydrotropal-
dehyde, etc., have to be reduced from about
3000 ppm to (10 ppm. A continuous reactive distil-
lation column has been claimed to be a versatile
method to achieve this objective.
Concluding Remarks
Reactive distillation offers several beneRts
over conventional processes for MTBE and methyl
acetate manufacture. The commercial success of
MTBE manufacture by reactive distillation has led to
numerous investigations in the recent past on almost
every aspect of this process. The generation of kinetic
and equilibrium data at boiling temperatures, simula-
tion and design studies, control strategies and identi-
Rcation of new reactions as candidates for reactive
distillation, are some of the areas being investigated.
Simulation studies of catalytic distillation for etheriR-
cation have highlighted the important aspects of
steady state multiplicity. This concept carries
a special signiRcance and plays an important role in
design methods. Future work on simulation will see
other reactions displaying this unusual phenomenon.
Eastman Kodak has demonstrated the feasibility and
advantages of reactive distillation at the commercial
scale for methyl acetate manufacture. The process has
scope for improvement in the sense that solid acid
catalyst can be employed instead of sulfuric acid.
Different techniques of conRning the small beads of
ion exchange resin in Rbre glass cloth, wire mesh or
structured packing have been developed. These cata-
lysts offer very good vapour}liquid contact and activ-
ity but replacing the deactivated catalyst would be
labour-intensive and time-consuming. The future focus
should be on development of a catalyst in the form of
a conventional column packing, such as Rachig rings,
which would have good mechanical strength, activity
and stability under the reaction conditions. Reactive
4082 III / RESINS AS BIOSORBENTS: ION EXCHANGE
distillation may Rnd a place in many other processes
such as hydrolysis of methyl acetate, recovery of
carboxylic acid from their aqueous solutions, hy-
drodesulfurization and puriRcation of phenols.
See also: II/Distillation: Energy Management; Historical
Development; Instrumentation and Control Systems;
Theory of Distillation.
Further Reading
Agreda VH, Partin LR and Heise WH (1990) High purity
methyl acetate via reactive distillation. Chemical Engin-
eering Progress 86(2): 40}46.
Ancillotti F, Pescarollo E, Szatmari E and Lazar L (1987)
MTBE from butadiene-rich C4s. Hydrocarbon Process-
ing 66: 50}53.
Bravo JL, Pyhalahti A and Jarvelin H (1993) Investigations
in a catalytic distillation pilot plant: vapour/liquid
equilibrium, kinetics, and mass transfer issues.
Industrial and Engineering Chemistry Research 32:
2220}2225.
Chopade SP and Sharma MM (1997) Reaction of ethanol
and formaldehyde: use of versatile cation exchange
resins as catalysts in batch reactors and reactive distilla-
tion columns. Reactive and Functional Polymers 32(1):
53}65.
Chopade SP and Sharma MM (1997) Acetalization of ethy-
lene glycol with formaldehyde using cation exchange
resins as catalysts: batch versus reactive distillation. Re-
active and Functional Polymers 34(1): 37}45.
DeGarmo JL, Parulekar VN and Pinjala V (1992) Consider
reactive distillation. Chemical Engineering Progress
88(3): 43}50.
Fuchigami Y (1990) Hydrolysis of methyl acetate in distilla-
tion column packed with reactive packing of ion ex-
change resin. Journal of Chemical Engineering of Japan
23: 354}359.
Hauan S, Hertzberg T and Lein KM (1995) Why methyl
tert-butyl ether production by reactive distillation may
yield multiple soultions. Industrial and Engineering
Chemistry Research 34: 987}991.
Jacobs R and Krishna R (1993) Multiple solutions in react-
ive distillation for methyl tert-butyl ether synthesis. In-
dustrial and Engineering Chemistry Research 32:
1706}1709.
RESINS AS BIOSORBENTS:
ION EXCHANGE
S. Belfer, The Institutes for Applied Research,
Ben-Gurion University of the Negev, Beersheva,
Israel
Copyright ^ 2000 Academic Press
Kolah AK, Mahajani SM and Sharma MM (1996) Acetaliz-
ation of formaldehyde with methanol in batch and
continuous reactive distillation columns. Industrial
and Engineering Chemistry Research 35(10):
3707}3720.
Mohl KD, Kienle A, Gilles ED, Rapmund P, Sundmacher
K and Hoffman U (1997) Nonlinear dynamics of react-
ive distillation processes for the production of fuel
ethers. Computers and Chemical Engineering 21:
S989}S994.
Neumann R and Sasson Y (1984) Recovery of acetic acid by
esteriRcation in a packed chemorectiRcation column.
Industrial and Engineering Chemistry Process Design
and Development 23: 654}659.
Nijhuis SA, Kerkhof FPJM and Mak ANS (1993) Multiple
steady states during reactive distillation of methyl tert-
butyl ether. Industrial and Engineering Chemistry Re-
search 32: 2767}2774.
Nocca JL, Leonard J, Gaillard JF and Amigues P (1989)
Process for manufacturing a tertiary alkyl ether by react-
ive distillation. US Patent 4 847 431.
RehRnger A and Hoffman U (1990) Kinetics of methyl
tertiary butyl ether liquid phase synthesis catalysed by
ion exchange resin } I. Intrinsic rate expression in liquid
phase activities. Chemical Engineering Science 45(6):
1605}1617.
Scates MO, Parker SE, Lacy JB and Gibbs RK (1997)
Recovery of acetic acid from dilute aqueous streams
formed during a carbonylation process. US Patent 5 599
976.
Sharma MM (1995) Some novel aspects of cationic ex-
change resins as catalysts. Reactive and Functional Poly-
mers 26: 3}23.
Smith LA (1980) Catalyst system for separating isobutene
fron C4 streams. US Patent 4 215 011.
Smith LA (1981) Catalytic distillation process. US Patent
4 307 254.
Song W, Venimadhavan G, Manning JM, Malone MF and
Doherty MF (1998) Measurement of residue curve maps
and heterogeneous kinetics in methyl acetate system.
Industrial and Engineering Chemistry Research 37:
1917}1928.
Sundmacher K and Hoffmann U (1995) Oscillatory
vapour}liquid transport phenomena in a packed reactive
distillation column for fuel ether production. Chemical
Engineering Journal and the Biochemical Engineering
Journal 57: 219}228.
Introduction
The term biosorbent is usually applied to solid poly-
meric media employed in the puriRcation, separation
or isolation of biotechnological products. To assure
 III / RESINS AS BIOSORBENTS: ION EXCHANGE
efRcient sorption, these materials must meet certain
requirements: they must have a high sorption
capacity combined with ease of regeneration, good
kinetic properties and mechanical stability over many
sorption}regeneration cycles. Both ion exchange
resins and their precursors, the inert polymer ma-
trices, are extensively used to isolate fermentation
products, including low molecular weight com-
pounds, such as acetic acid, and high molecular
weight compounds, such as enzymes and proteins.
Ion exchange resins have been traditionally used in
water treatment technologies, for example for desali-
nation and softening and for wastewater treatment.
Their Rrst application to pharmaceuticals may be
dated to the 1950s and 1960s, although the greatest
surge in interest in terms of papers and patents pub-
lished occurred in the period 1960}75.
Pharmaceuticals
Of the various pharmaceutical products processed by
ion exchange technologies, antibiotics are probably
the most important. Because antibiotics mostly con-
sist of charged molecules, they lend themselves read-
ily to isolation with ion exchange resins, and with
cation exchangers in particular. In January 1945, Van
Dolah, Christenson and Shelton Rled a US patent
application claiming the use of organic cation ex-
changers for the puriRcation of streptomycin and
streptothricin. First to be used for this purpose were
the phenol-sulfonic acid-type cation exchangers (Am-
berlite IR-100, Ionac C-200, Dowex 30). These were
followed by high capacity carboxylic acid exchangers
for commercial applications (Amberlite IRC-50).
Both groups are characterized by a gel structure. They
have no open pores in the dry state, but when placed
in contact with aqueous solutions they undergo swell-
ing and acquire the ability to uptake large ions.
Commercialization of the macroporous sorbents of
the Amberlite XAD series by Rohm and Haas in the
1960s was a revolutionary step in ion exchange tech-
nology and opened up new possiblities for the isola-
tion of antibiotics. Macroporous sorbents had the
necessary mechanical strength, provided large surface
areas for sorption, and had appropriate pore sizes for
rapid transport. Macroporous resin sorbents such as
the polyaromatics Amberlite XAD-4,-16 and -1180,
Diaion HP20, media consisting of aliphatic esters
(Amberlite XAD-7) and nitrated aromatics (nitrated
Amberlite XAD-16) were recommended for large
scale application for antibiotics.
Vitamins constitute another class of pharmaceut-
icals that are puriRed by ion exchange resins. Vitamin
B12, for example, is produced by microbial fermenta-
tion and can be separated from the broth using a car-
boxylic acid exchanger.
4083
Proteins are based on copolymers of amino acids
and may thus be regarded as polyionic materials. At
a given pH they bear either a positive or a negative
charge depending on their isoelectric point. Proteins are
therefore eminently suitable for isolation by ion
exchange technology. Exchangers based on matrices
consisting of cross-linked polyacrylic and phenol-for-
maldehyde polymers have been used for large scale
protein puriRcation. However, the traditional ion ex-
changers are generally unsuitable for the adsorption of
proteins due to their hydrophobicity, high charge den-
sity and high degree of cross-linking, which result in low
protein capacities and a tendency towards denaturation
of sorbed molecules. After the introduction in 1956 of
the Rrst ion exchanger speciRcally designed for proteins,
a number of highly hydrophilic polysaccharide matrices
have been proposed, all of them less rigid and more
hydrophillic than the polystyrene type of biosorbents.
Synthesis of Resins
Todays ion exchange technology is based on organic
polymer matrices. The typical spherical ion exchange
beads are made by suspension polymerization of styrene
with divinylbenzene to form an insoluble polymer gel.
The mixture of monomers to which an initiator of
radical polymerization has been added is stirred into
an aqueous suspension under conditions designed to
give the desired droplet size. This mixture is heated
for several hours to yield solid spherical beads, which
are then treated with concentrated sulfuric acid at
about 803C to obtain cation exchange resin. The Rnal
product is a sulfonated cross-linked polystyrene } the
strong-acid cation exchanger most widely used com-
mercially. It has a capacity of 5.25 mmol g\1 calcu-
lated for oven-dried resin. The structural formula of
the resin is given below (Structure 1), together with the
formula of a weak-acid cation exchanger based on
acrylic acid copolymerized with DVB (Structure 2).
4084 III / RESINS AS BIOSORBENTS: ION EXCHANGE

Anion exchange resins are produced by a two-step illustration, a list of synthetic resins manufactured by
process. First, chloromethylation is applied to intro- Mitsubishi and designed for protein separation is
duce chloromethyl groups. The second step is amina- given in Table 1, together with the relevant recom-
tion. When a tertiary amine such as trimethylamine is mendations.
used, the product is a strong-base quaternary am- As an alternative to the highly hydrophobic organic
monium compound (Structure 3). This resin is the polymeric matrices, ion exchange materials for biolo-
anionic equivalent of the sulfonic cation exchange gical compounds have also been developed from
materials. The capacity of a typical strong-base resin cross-linked dextran, agarose and beaded crystalline
is 3.9}4.2 mmol g\1 of dry resin. The use of a second- cellulose polymers. The functional groups typically
ary amine, such as dimethylamine or other multifunc- added to such matrices are shown below.
tional amine, gives various weakly basic resins, for
example the one shown in Structure 4. Anionic functional groups
Aminoethyl (AE) }OCH2CH2NH#3
Diethylaminoethyl (DEAE) }CH2CH2N(CH2CH3)2
Quaternary aminoethyl }OCH2CH2N#(C2H5)2-
(QAE) CH2CH(OH)CH3

Cationic functional groups

Carboxymethyl (CM) }OCH2COO\
Phospho }PO4H\
2
Sulfopropyl (SP) }CH2CH2CH2SO\
3

DEAE-cellulose, an anion exchanger contain-

The resins mentioned above are among those most
commonly used as ion exchangers. However, a wide
range of resins tailored for speciRc needs is available;
further information may be found in commercial
catalogues as well as in relevant monographs. For
ing diethylaminoethyl groups attached to the cellu-
lose, is applied extensively. An exchanger of this
type having a content of basic groups of only
1 mmol g\1 adsorbs three-quarters of its own
weight of bovine plasma albumin from 0.2% solution
in 0.01 mol L\1 sodium phosphate at pH 7.0. CM-
cellulose, a cation exchanger, which contains car-
boxymethyl groups, adsorbs its own weight of horse
carbon monoxide haemoglobin from 0.2% solution
in 0.01 mol L\1 sodium phosphate at pH 6.0. Cellu-
lose ion exchangers with improved characteristics are
now available, and numerous studies on their use in

Table 1 Sepabeads FP series product lista

Grade Functional group Pore sizeb Chromatography mode

Small Medium Large GFC CEC AEC HIC AFC

FP-HG }OH FP-HG20 FP-HG13 FP-HG05

FP-CM }CH2COOH FO-CM13
FP-QA }N#(CH3)2C2H4OH FP-QA13

FP-DA }N(C2H5)2 FP-DA20 FP-DA12 FP-DA05

FP-DA13 O
FP-HA }NH(CH2)6NH2 FP-HA20 FP-HA13
FP-BU }O(CH2)3CH3 FP-BU13 FP-BU05
FP-OT }O(CH2)7CH3 FP-OT13
FP-PH }OC6H5 FP-PH13 O O
FP-CL }N(CH2COOH)2 FP-CL13
FP-BL Cibacron blue 3G-A FP-BL13

a
Average particle size approximately 120
m.
b
The second digit in the product name refers to the pore size. GFC, gel filtration chromatography; CEC, cation exchange chromato-
graphy; AEC, anion exchange chromatography; HIC, hydrophobic interaction chromatography; AFC, affinity chromatography.
From Paion, Manual of Ion-Exchange Resins and Synthetic Absorbents.
 III / RESINS AS BIOSORBENTS: ION EXCHANGE
the separation of biologicals have been reported in
the last 5 years.
Characteristics of Resins
Selection of the exchange resin for a given application is
a process of compromise based on examination of many
factors, such as the polar nature of the sorbate, the size
of the sorbate, resin capacity, equilibrium relationships,
elution properties and Sow characteristics.
Adsorption Isotherm
In order to design a puriRcation process based on an
ion exchange technique, it is essential to know some-
thing about the capacity of the exchanger. Equilib-
rium sorption capacity is commonly determined with
the help of the sorption isotherm, which gives the
sorption uptake (q) and the Rnal equilibrium concen-
tration of the residual solute in solution (c). Sorption
isotherms are measured by placing solutions with
different concentrations of solute in contact with
a known weight of the resin at a constant temperature
until equilibrium is attained. Calculation of the dif-
ference between the concentration of product before
and after equilibrium, cH, gives the sorbed protein
mass qH. Plotting qH versus cH yields the equilibrium
sorption isotherm. Assuming that single-site interac-
tion occurs between bioproduct and sorbent, and also
that nonspeciRc interactions are absent, the apparent
constant KA and the maximum product-binding capa-
city qm may be evaluated by Rtting the experimental
data to the well-known Langmuir model:
qm ) KA ) cH
qH"
1#KA ) cH
Non-Langmuirian behaviour may point to multiple
interaction sites. In such cases, appropriate models
may be worked out to Rt the experimental data and
used to determine whether this behaviour may be due
to additional nonspeciRc interaction sites from the
sorbents surface, or to product}product interactions
with the Rrst adsorption layer.
Kinetics of Adsorption
Another important factor of sorption performance is
the kinetics of the adsorption/desorption reactions.
The rates of these reactions dictate the length of time
that has to be allowed to attain equilibrium. For
example, the adsorption of protein on to packed beds
involves three processes.
First, the protein is transported from the bulk Suid
to the outer surface of the adsorbent particles by Rlm
mass transport. Second, intraparticle transport occurs
4085
by diffusion. Finally, the protein binds to ligand at-
tached to the inner surface of the particle. It is impor-
tant to determine which of these processes is the
rate-limiting step.
Process Design
Isolation of bioproducts by ion exchange processes
can be carried out either batchwise or by traditional
packed-bed techniques. In the former, the exchanger
is added to the product solution in a vessel which is
mixed until sorption has occurred.
Packed-Bed Column
In a packed-bed column the movement of liquid
through the bed approximates to plug Sow, resulting
in a maximum number of theoretical equilibrium
stages within the column and hence good adsorption
and chromatographic performance. The overall Sow
performance is strongly related to the length and
shape of the ion exchange zone evolving during sorp-
tion and regeneration. This zone appears between the
section of column saturated with product and the
section that still contains fresh sorbent. As loading or
regeneration progresses, the zone moves along the
column in the direction of the liquid Sow. Break-
through occurs when the zone approaches the end of
the column and the concentration in the outlet stream
increases sharply. Breakthrough proRles provide
a measure of the performance of different ion ex-
changers in packed-bed operations. A sharp break-
through proRle is desirable in order to achieve
efRcient use of sorbent. Figure 1 shows breakthrough
proRles for two hypothetical adsorbents with identi-
cal equilibrium capacities. It can be seen that a greater
proportion of bed capacity is used in the case of sharp
breakthrough.
Figure 1 Hypothetical breakthrough curves for two sorbents.
The unfavourable breakthrough curve (triangles) is flat and trail-
ing, while the favourable breakthrough curve (circles) is sharp and
steep.
4086 III / RESINS AS BIOSORBENTS: ION EXCHANGE
Fluidized Bed Column
In a Suidized bed, liquid upSow through the column
causes the resin particles to become separated from
each other. This technology has attracted attention
for biochemical separation processes because it en-
ables direct treatment of crude feedstocks from fer-
mentation reactors. There are two important criteria
that must be met before Suidized bed sorption can be
considered a viable method for separating products
from unRltered fermentation broths. First, broth
solids must have a lower terminal settling velocity
than the resin, and the terminal velocity of the resin
must be sufRcient to achieve reasonable time cycles.
Terminal velocity is deRned as the upSow velocity at
which particles will not remain in the column. Sec-
ond, the dynamic adsorptive capacity of the resin for
the product must be of such a magnitude that optimal
yield, purity and cycle time can be achieved.
Determining optimum resin terminal velocity and
dynamic sorptive capacity for a speciRc product is
a complex process. The breakthrough curves are usu-
ally obtained for a variety of design and operating
conditions (column size, distributor, bed type, bed
height, Sow rate and number of stages). It is also
essential to Rnd an appropriate mathematical model
for simulation and optimization of the processes. An
extensive literature exists describing the mode of op-
eration of Suidized beds with reference to bioproduct
Figure 2 Schematic representation of fluidized bed separation.
separation. A schematic representation of Suidized
bed separation is given in Figure 2.
New Developments
Although ion exchangers remain the most frequently
used media for separation of biological mixtures,
some novel approaches have emerged. Perfusion
chromatography is one of them. This method exploits
the fact that particle resins have very large pores
(600}800 nm) that permit convective Sow. A high
surface area for sorption is provided by the presence
of numerous small diffusive pores. Thus, convection
rather than diffusion dominates the mass transport of
the sample molecules. This makes the process 10
times faster than the usual separation process without
much loss in capacity or resolution.
Another approach which has emerged as a power-
ful separation tool is immobilized metal afRnity
chromatography (IMAC). In this method, immobi-
lized ligands, like iminodiacetic acid, produce
chelates with transition metal ions (such as Ca2#,
Zn2# and Fe3#) which, when exposed to a protein,
form a ternary complex on the protein surface. Fur-
ther isolation is then accomplished with ease.
In conclusion, the latest developments in sorption
media and separation technology provide a broad
and varied basis for identiRcation of appropriate
sorbents and selection of contact mode between feed-
stock and sorbent.
See also: II/Chromatography: Protein Separation. Ion
Exchange: Organic Ion Exchangers.
Further Reading
Calmon C and Kressman TRE (1957) Ion-exchangers in
Organic and Biochemistry. New York: Interscience.
Chase HA (1994) PuriRcation of proteins by adsorpt-
ion chromatography in expanded beds. TIBTECH 12:
296.
Cowan GH, Gosling IS and Sweetenham WP (1987) Mod-
elling for scale-up and optimization of packed-bed col-
umns in adsorption and chromatography. In: Kerral MS
and Hudson MJ (eds) Separations for Biotechnology, p.
152. Chichester: Ellis Horwood.
Draeger N and Chase HH (1990) Modelling of protein
adsorption in liquid Suidized bed. In: Pyle DC (ed.)
Separations for Biotechnology, p. 325. London/ New
York: Elsevier.
Gailliot FP, Cleason C, Wilson JJ and Zwarick J (1990)
Fluidized bed adsorption for whole broth extraction.
Biotechnology Progress 6: 370}375.
Garcia AA (1991) Strategies for the recovery of chemicals
from fermentation: A review of the use of polymeric
adsorbents. Biotechnology Progress 7: 33}42.
 III / RESTRICTED-ACCESS MEDIA: SOLID-PHASE EXTRACTION 4087

Gelfferich F (1962) Ion-exchange. New York: McGraw-Hill.
Graf H, Rabaud JN and Egly UM (1994) Ion-exchange
resins for the puriRcation of monoclonal antibodies
from animal cell culture. Bioseparation 4: 7}20.
Greig JA (ed.) (1996) Ion-exchange Developments and Ap-
plications. Proceedings of IEX 96. Cambridge: Royal
Society of Chemistry.
Levison PR (1993) Process scale liquid chromatography. In:
Kennedy JF, Philips GO and Williams PA (eds) Cellu-
losics: Materials for Selective Separation and Other
Technologies. Chichester: Ellis-Horwood.
Pirotta M (1991) Ion-exchangers in pharmacy, medicine
and biochemistry. In: Dorfner K (ed.) Ion Exchangers.
New York.
Rossomando EF (1990) Ion-exchange chromatography. In:
Deutscher MP (ed.) Methods in Enzymology, vol. 182,
Guide to Protein PuriTcation, pp. 309, 409. New York:
Academic Press.
Streat M and Cloete FLD (1987) Ion exchange.
In: Rousseau RW (ed.) Handbook of Separation Process
Technology. New York: Wiley.

RESTRICTED-ACCESS MEDIA:
SOLID-PHASE EXTRACTION
J. Haginaka, Mukogawa Womens University, propyl (i.e. diol) phases by attachment of the tripep-
Nishinomiya, Japan tide GFF, bonded via the amino groups to the glycer-
Copyright ^ 2000 Academic Press ylpropyl groups. The phenylalanine moieties were
then cleaved from the external surface of the silica
with carboxypeptidase A, which is too large to enter
the small pores. After this enzymatic treatment, the
Introduction glycerylpropyl moieties and glycine residues remain
For the determination of drugs and their metabolites on the external surface. Because the ISRP concept was
in serum or plasma by high performance liquid innovative for drug determinations in serum, many
chromatography (HPLC), tedious and time-consum- RAM materials were subsequently developed.
ing pretreatment procedures such as liquid}liquid ex- Another RAM material based on silica gels is shielded
traction, solid-phase extraction (SPE) or membrane- hydrophobic phase (SHP), which consists of a
based extraction are often required. Among those
pretreatment procedures, SPE is the most widely used
for extraction of target compounds in biological
Suids. However, direct injection of serum or plasma
samples onto HPLC or SPE materials causes protein
denaturation with accumulation of materials on the
sorbent, resulting in undesired loss in the capacity
and selectivity of the sorbent. Thus, it is essential to
remove serum or plasma proteins before loading the
samples onto the HPLC or SPE sorbents. Recently,
restricted access media (RAM) materials were intro-
duced for direct injection of proteinaceous samples
onto the HPLC or SPE materials. With RAM mate-
rials large molecules such as proteins are eluted in the
void volume without destructive accumulation be-
cause of restricted access to some surfaces, while
allowing small molecules such as drugs and their
metabolities to reach the hydrophobic, ion-exchange
or afRnity sites and be separated. One approach uses Figure 1 Schematic representation of an internal-surface
an internal-surface reversed-phase (ISRP) material, reversed-phase (ISRP) material. Proteins do not adsorb on the
produced from porous silica gels, which has hydro- hydrophilic exterior surfaces and do not penetrate into the hydro-
phobic interior surfaces, while analytes can reach the interior
phobic interior and hydrophilic exterior surfaces, as
surfaces and be separated. (Reproduced with permission from
shown in Figure 1. The ISRP}GFF material Perry JA (1991) The internal surface reversed phase. Concept
(GFF"glycine- L-phenylalanine- L -phenylalanine) and applications. Journal of Liquid Chromatography 13:
was prepared from covalently modiRed glyceryl- 1047}1074.)
4088 III / RESTRICTED-ACCESS MEDIA: SOLID-PHASE EXTRACTION
Figure 2 Schematic representation of a shielded hydrophobic
phase (SHP) material. S"silica gel matrix; R"hydrophobic
pocket; P"hydrophilic network; G"large unretained protein;
A"small retained analyte. (Reproduced with permission from
Gisch DJ, Hunter BT and Feibush B (1988) Shielded hydrophobic
phase: a new concept for direct injection analysis of biological
fluids by high-performance liquid chromatography. Journal of
Chromatography 433: 264}268.)
hydrophilic polyoxyethylene network embedded with
phenyl groups, bonded to both the external and inter-
nal surfaces of the particles (Figure 2). Other RAM
materials based on silica gels include semipermeable
surface, dual zone and mixed function phase materials.
RAM materials based on polymer beads have also
been developed. For example, one polymer-based RAM
material was prepared from porous uniformly sized
poly(glycidyl methacrylate-co-ethylene dimethac-
rylate) beads. Hydrolysis of the epoxide groups to
diols can be carried out exclusively within the large
pores of the medium through the use of a polymeric
catalyst, polystyrenesulfonic acid (average molecular
weight, 141 000 Da). The epoxide groups remaining
in the small pores after hydrophilization of the large
pores were then reacted with either hydrophobic
Figure 3 Schematic representation of a polymer-based RAM
material modified in pore-size selective fashion using a polymer
catalyst. (Reproduced with permission from Smigol V, Svec F and
FreH chet JMJ (1994) Novel uniformly sized polymeric stationary
phase with hydrophilized large pores for direct injection HPLC
determination of drugs in biological fluids. Journal of Liquid
Chromatography 17: 891}911.)
C18 or phenyl groups, or more polar diethylamino
groups. The pore-size selective modiRcation of por-
ous materials provided the RAM materials as shown
in Figure 3.
RAM materials could be used for direct serum
injection assays of drugs as HPLC or SPE materials.
The former materials have been designed and used
preferentially as packings for large-size (150;
4.6 mm i.d.), i.e. analytical columns. In this case the
extraction and separation of analytes take place
simultaneously. For SPE RAM, sorbents were packed
into a small (typically 5}30 mm;3}4.6 mm i.d.)
precolumns connected to an analytical column via
a six-port valve, i.e. switching valve. In the coupled-
column mode this is a sequential process. The two
approaches for the extraction and analysis of the
target compounds in biological Suids by HPLC are
compared in Table 1.
Single-column Mode
When using the RAM materials, the ionic strength
and pH of an eluent, and the content of organic
modiRer are limited in order to prevent precipitation
of serum proteins. For the ISRP}GFF materials, the
recommended eluent pH range was 6.0}7.5. The re-
covery of serum proteins was low at acidic pH. This is
due to the electrostatic attractions of the serum pro-
teins having a net positive charge (isoelectric point, pI
of serum albumin, 4.7) and the external glycine resi-
dues having a negative charge. Taking into account
the pKa values of the bound glycine (between 2.3 and
3.0), the recovery of serum proteins might be higher
with an eluent pH below 2. However, chemically
bonded columns cannot be used for long periods at
this pH because of the hydrolysis of the bonded
phase. However, for RAM materials such as SHP,
whose external surface has no charges, there is no
eluent pH limitation. These materials can be used at
any pH suitable for siloxane-bonded silicas (pH 2}8).
These results demonstrated that external uncharged
surfaces should be suitable for the external layers of
RAM materials. With regard to the eluent, an ionic
strength of 0.05}0.2 was used. The preferred organic
modiRers are acetonitrile, 2-propanol, tetrahyd-
rofuran and methanol because they can afford a wide
selectivity in controlling solute retention on the ac-
cessible hydrophobic surfaces. The content of the
organic modiRer should be (20%.
Direct serum injection assays of drugs were carried
out on the ISRP}GFF materials. The chromatograms of
plasma spiked with probenecid or lidocaine at clinical
levels (50
g mL\1 for probenecid, 5.94
g mL\1 for
lidocaine) are shown in Figures 4 and 5, respectively,
together with those for methanolic solutions of the
 III / RESTRICTED-ACCESS MEDIA: SOLID-PHASE EXTRACTION 4089

Table 1 On line sample extraction and analysis: comparison of single-column and coupled-column modes. (Reproduced with
permission from Boos K-S and Grim C-H (1999) High-performance liquid chromatography integrated solid-phase extraction in
bioanalysis using restricted access precolumn packings. Trends in Analytical Chemistry 18: 175}180)

Parameter Single-column mode Coupled-column mode

Matrix elimination and analyte separation Simultaneous Sequential

Peak capacity Low High
Selectivity Low High
Incidence of interferences High Low
Sample volume (100
L 100
L
Analyte enrichment No Yes
Limit of quantification Increased Decreased
Mobile-phase composition Restricted (pH, additives) Variable
Detection UV'240 nm No limitation
Fluorescence detection } yes
Electrochemical detection } no
Column lifetime Short Long
Cost/analysis High Low

same concentration. The recovery was calculated the difference in the detection wavelengths. Nat-
from the peak}area ratio of a given concentration of urally, the shorter wavelength (220 nm for lidocaine)
the drug dissolved in plasma and methanol. Despite reveals more matrix peaks at a higher intensity than
the differences in the bound fractions (83}94% for the longer wavelength (254 nm for probenecid). In
probenecid and 65}77% for lidocaine), both drugs
were almost completely recovered from plasma sam-
ples. The large difference in the intensities of the
background peaks in these chromatograms is due to

Figure 4 Separation of probenecid from human plasma. Mobile
phase, 0.1 M potassium phosphate bufferItetrahydrofuran
(95 : 5), pH 7.0; flow rate, 1.0 mL min\1; stationary phase,
ISRPIGFF column, 150 mm;4.6 mm i.d.; detection, UV
(254 nm); injection volume, 10
L. (Reproduced with permission
from Nakagawa T, Shibukawa A, Shimono N et al. (1987) Reten-
tion properties of internal-surface reversed-phase silica packing
and recovery of drugs from human plasma. Journal of
Chromatography 420: 297}311.)
Figure 5 Separation of lidocaine from human plasma. Mobile
phase, 0.1 M potassium phosphate bufferItetrahydrofuran (9 : 1),
pH 7.2; flow rate, 0.8 mL min\1; stationary phase, ISRPIGFF
column, 150 mm;4.6 mm i.d.; detection, UV (220 nm); injection
volume, 10
L. (Reproduced with permission from Nakagawa T,
Shibukawa A, Shimono N et al. (1987) Retention properties of
internal-surface reversed-phase silica packing and recovery of
drugs from human plasma. Journal of Chromatography 420:
297}311.)
4090 III / RESTRICTED-ACCESS MEDIA: SOLID-PHASE EXTRACTION

1.97
g mL\1, which agreed with the free CBZ con-

centration determined by means of ultraRltration
(2.08
g mL\1). Furthermore, it is interesting that the
CBZ concentration calculated from the area of this
plateau was 8.06
g mL\1, in agreement with the
total CBZ concentration of this plasma sample. This
implies that both free and total drug concentrations
can be determined simultaneously by a single analysis
based on the height and area of the drug plateau,
respectively. However, it is required to inject a large
sample volume (in this case, 1.8 mL plasma sample)
in order to observe the plateau peak. Further, the
plateau peak cannot always be separated from blank
peak, dependent on the drug separated.

Coupled-column Mode
As shown in Table 1, the advantages of a coupled-
column mode include separation selectivity (ability to
Figure 6 Chromatograms of fetal bovine serum (A) and couple precolumns and analytical columns of differ-
phenytoin (Ph)-spiked fetal bovine serum (B) at pH 2.5. Chromato- ent selectivity), detection sensitivity (analyte enrich-
graphic conditions: column, SHP column (150 mm;4.6 mm i.d.); ment due to larger sample volumes and reduced
mobile phase, acetonitrile I 50 mM KH2PO4 (pH 2.5) (15 : 85); flow
number of interfering peaks), and higher variability
rate, 2.0 mL min\1; temperature, ambient; detection, UV at 254 nm,
0.008 aufs; injection volume, 25
L. (Reproduced with permission of mobile phases and detection modes. In recent times
from Gisch DJ, Feibush B, Hunter BT et al. (1989) A new HPLC RAM materials have mainly been used in the
concept for direct analysis of drugs in biological matrices: shiel- coupled-column mode.
ded hydrophobic phase. BioChromatography 4: 206}215.) Figure 8 shows a representative chromatogram
obtained after the direct injection of an untreated
both cases the eluent pH was about 7. However, human serum sample onto a RAM precolumn. The
phenytoin was not eluted under the neutral condi- injection was followed by fully automated online
tions on the SHP materials, but when the eluent pH extraction and subsequent separation of antiepileptic
was adjusted to 2.5, it was eluted and resolved from
serum matrix components (Figure 6). Because of the
presence of secondary amines on phenytoin, the re-
tention factor of phenytoin was decreased by reduc-
ing the eluent pH. Since the SHP materials had no
charged groups on the external surface as described
above, they could be used at eluent pH 2.5.
In the above applications, less than 100
L of
serum sample was injected. At higher sample vol-
umes, analyte peaks were broadened and a plateau
peak was observed. The plateau peak is dependent on
the unbound fraction of analyte. Whether the
broadened or plateau peak appears; that is, when the
unbound drug fraction is higher, we can inject a lar-
ger sample volume for direct serum injection assays of
Figure 7 Determination of unbound and bound concentrations
the drug without peak-broadening.
of carbamazepine (CBZ) in human plasma. Total concentration of
On the other hand, both free and total drug con- CBZ is 8
g mL\1. Stationary phase: ISRPIGFF column
centrations could be simultaneously determined by (150 mm;4.6 mm i.d.). Mobile phase: potassium phosphate buf-
injecting such a larger sample volume that the plateau fer (pH 7.4, I"0.17). Flow rate: 1.2 mL min\1. Detection: UV
peak of a drug appears. Figure 7 shows the chromato- 300 nm. Column temperature: 373C. Injection volume: 1.8 mL.
(Reproduced with permission from Shibukawa A, Nakagawa T,
gram of 8.00
g mL\1 carbamazepin (CBZ) in hu-
Nishimura N et al. (1989) Determination of free drug in protein
man plasma. CBZ was well separated from the blank binding equilibrium by high-performance frontal analysis using
peak and gave a clear and wide plateau. The CBZ internal-surface reversed-phase silica support. Chemical & Phar-
concentration calculated from this plateau height was maceutical Bulletin 37: 702}706.)
 III / REVERSED-FLOW GAS CHROMATOGRAPHY
Figure 8 Coupled-column analysis of antiepileptic drugs in
serum. Precolumn: 25 mm;4 mm i.d., LiChrospher RP-18 ADS
(particle size, 25
m); analytical column: 250 mm;4 mm i.d.,
LiChrospher 60 RP-Select B (particle size, 5
m); loading mobile
phase: 0.5 M monobasic sodium phosphate (pH 4.0) for 0 min
at 0.5 mL min\1; transfer mobile phase: 5 : 95 (v/v) aceto-
nitrileIwater for 5 min at 0.5 mL min\1; separation mobile phase:
acetonitrileIwater with a linear acetonitrile gradient from 5% to
34% in 34 min at 0.5 mL min\1; detection: UV absorbance at
205 nm; sample: 100
L of analyte-spiked serum. Peaks:
1"ethosuximide (10.3 nmol), 2"primidone (1.8 nmol),
3"phenobarbital (6.6 nmol), 4"carbamazepine (1.3 nmol),
5"phenytoin (2.5 nmol). (Reproduced with permission from
Boos K-S and Rudolphi A (1997) The use of restricted-access
media in HPLC. Part I. Classification and review. LCIGC 15:
602}611.)
REVERSED-FLOW GAS
CHROMATOGRAPHY
N. A. Katsanos, University of Patras, Patras, Greece
F. Roubani-Kalantzopoulou, National Technical
University of Athens, Zografou, Greece
Copyright ^ 2000 Academic Press
4091
drugs on a conventional analytical reversed-phase
HPLC column. The precolumn (25 mm;4 mm i.d.)
packed with one of ISRP materials (particle size,
25
m) can tolerate 2000 injections or more of 50
L
of human plasma.
Future Trends
Since the invention of the ISRP}GFF materials, vari-
ous RAM materials have been developed for direct
serum injection assays of drugs with single- and
coupled-column modes. However, they are lacking in
selectivity because hydrophobic or ion-exchange
ligands are used as the analyte binding ligands. In
the future, more selective ligands such as immuno-
afRnity or chemoafRnity ligands or molecularly im-
printed polymers have the potential to further im-
prove selectivity and sensitivity in bioanalysis.
Further Reading
Boos K-S and Grim C-H (1999) High-performance liquid
chromatography integrated solid-phase extraction in
bioanalysis using restricted access precolumn packings.
Trends in Analytical Chemistry 18: 175}180.
Boos K-S and Rudolphi A (1997) The use of restricted-
access media in HPLC. Part I. ClassiRcation and review.
LC}GC 15: 602}611.
Haginaka J (1991) Drug determination in serum by liquid
chromatography with restricted access stationary
phases. Trends in Analytical Chemistry 10: 17}22.
Rudolphi A and Boos K-S (1997) The use of restricted-
access media in HPLC. Part II. Applications. LC}GC 15:
814}823.
Shibukawa A, Kuroda Y and Nakagawa T (1999) High-
performance frontal analysis for drug}protein binding
study. Journal of Pharmaceutical and Biomedical Analy-
sis 18: 1047}1055.
Thurman EM and Mills MS (1998) Solid-phase Extraction:
Principles and Practice. New York: Wiley-Interscience.
Introduction
Gas chromatography (GC) is a technique that is used
not only to separate substances from each other, but
also to separate physicochemical quantities by
4092 III / REVERSED-FLOW GAS CHROMATOGRAPHY
measuring the value of one in the presence of another,
e.g. the rate of a chemical reaction in the presence
of diffusion phenomena. Several books, reviews and
original papers have been published dealing with
physicochemical measurements by GC. Some of the
properties measured pertain to the moving gas
phase, e.g. diffusion coefRcients of solutes into the
carrier gas, and the emphasis is on determining the
properties of the solutes. However, the majority
of physicochemical properties studied by GC
relate to the stationary phase and its interaction
with well-known probe solutes, e.g. the catalytic
properties of the solid stationary phase for reactions
between gases. This is termed inverse gas chromatog-
raphy and has the stationary phase of the system as
the main object of investigation. The procedures em-
ployed are the same as those used in direct GC, but
the results are used to derive properties of the station-
ary phase. The main source of information obtained
experimentally is the broadening of the chromato-
graphic elution peaks, mainly due to nonfulRlment of
the assumptions under which the central chromato-
graphic equation is derived, namely: (1) negligible
axial diffusion of the solute gas in the chromato-
graphic column; (2) linearity of the distribution iso-
therm; and (3) instantaneous equilibration of the
solute between the mobile and the stationary phase.
Classical chromatographic systems are not usually in
true equilibrium during the retention period, so that
extrapolation to inRnite dilution and zero carrier gas
Sow rate is required to approximate true equilibrium
parameters.
Another approach to extracting information about
the physicochemical properties of the stationary
phase from the elution peaks is based on the analysis
of the statistical moments of the peaks. Both of these
approaches, i.e. measurement of the broadening of
the peaks and analysis of their statistical moments,
Figure 1 Schematic representation of columns and gas connections showing the principle of RF-GC. Reprinted from Journal of
Chromatography A 775; 1997, 211}224, with permission from Elsevier Science.
are dynamic measurements because of the convective
movement of the carrier gas inside the chromato-
graphic column. In many cases achieving an accept-
able precision for the quantities determined is
a difRcult, if not impossible, task. The reversed-Sow
gas chromatography (RF-GC) technique offers an ad-
ditional route from experimental measurements to
the properties of the stationary phase, as described
below.
Physical Description and
Experimental Arrangement of the
System
Although there would be no GC without a mobile gas
phase, i.e. a carrier gas, in most studies its Sow rate
remains constant throughout a single experiment.
The magnitude of the Sow rate is usually treated as an
adjustable parameter. There are, however, two Sow
rate perturbation methods, the stopped-Uow and the
reversed-Uow techniques. Both are very simple to
apply and consist of either stopping the carrier gas
Sow for short time intervals, or reversing the direc-
tion of its Sow from time to time. Experimentally,
this is easily done by using shut-off valves in the Rrst
technique and a four-port valve in the second, as
shown in Figure 1. By switching the valve from the
position shown by the solid lines to that indicated by
the broken lines, the carrier gas, Sowing initially from
end D1 to D2 of the sampling column, now Sows in
the opposite direction, i.e. from D2 to D1. The samp-
ling column is a usual 1/4 in (6.35 mm) stainless-steel
or glass chromatographic tube of total length l
#l
ranging from 40#40 cm to 100#100 cm, either
straight or bent and empty of any solid or liquid
material.
The system also contains two other columns: a sep-
aration column and a diffusion column.
 III / REVERSED-FLOW GAS CHROMATOGRAPHY
A separation column of any suitable diameter and
length is placed before the detector. The carrier gas
Sows normally in one direction through this column
without any reversal, as shown by the gas connec-
tions of the valve. The separation column is Rlled
with ordinary stationary-phase material and is used
to effect the separation of two or more substances
reaching the valve from the sampling column. A refer-
ence injector at the inlet end of the separation column is
used for identiRcation purposes. If only one substance
reaches the detector, a simple restrictor sufRces.
A diffusion column of the same diameter as the
sampling column and length 30}100 cm is connected
perpendicularly at about the middle of the sampling
column; it is closed at the other end, and therefore no
carrier gas Sows through it.
The whole system of the three columns described
above is placed within the oven of an ordinary gas
chromatograph, with the columns and the connection
tubes properly bent. The question naturally arising is:
what happens when the direction of the carrier gas
Sowing through the sampling column is reversed, if it
is not reversed in the separation column and there is
no Sow at all through the diffusion column? If the
diffusion column does not contain any stationary
phase, but only stagnant carrier gas, nothing would
happen. If a probe gas is introduced through the
injector into the empty diffusion column, the repeated
Sow reversals of duration 10}100 s would create nar-
row and symmetrical peaks like those shown in Fig-
ure 2 (sample peaks). From the height H of these
peaks in arbitrary units, measured as a function of
time, t, that elapses between the injection of probe
substance and the respective Sow reversal, the diffu-
sion coefRcient of the substance into carrier gas can
be calculated. Finally, if a certain length L2 (3}12 cm)
of the diffusion column is Rlled with a stationary
phase, the diffusion band obtained by plotting H or
ln H against t is different from before, owing to the
interactions of this phase with the gaseous probe.
This is a way of separating the pure chromatographic
process from the rate or equilibrium processes involv-
ing the stationary phase.
The RF-GC method was Rrst introduced in 1982,
after preliminary studies on heterogeneous catalysis.
It was later developed for measuring various
physicochemical quantities pertaining to both,
substances contained in the moving gas phase and
stationary-phase behaviour. The technique was
reviewed in 1988 and again more recently. It does not
have any of the disadvantages connected with the
carrier gas Sow and the instrumental spreading of the
chromatographic bands, because the phenomena
being studied take place inside the diffusion column
L1#L2 (cf. Figure 1), with no carrier gas Sowing
4093
Figure 2 Sample peaks of propene in nitrogen carrier gas, with
section L2 (9.6 cm) containing 0.67 g TiO2, at 323.2 K. FID, flame
ionization detector.
through it. The Sow reversals are used merely as
a means for repeated sampling of the concentrations
at the point x"l
, i.e. at the exit of the section L1.
This sample is transported to the separation column
(containing another stationary phase for separation
purposes only), and then to the chromatographic de-
tector, without measuring the elution velocity of the
sample peaks or the carrier gas Sow rate, provided it
is steady. An obvious difference between conven-
tional elution gas chromatography and RF-GC is that
in the former longitudinal gaseous diffusion currents
are parallel to the chromatographic current, while in
the second method diffusion is, from the outset, phys-
ically separated from the chromatography by placing
the diffusion process perpendicular to the chromato-
graphic process.
Mathematical Models
Inverse gas chromatography has been used for study-
ing many properties of solids, such as the glass
transition of polymers, their crystallinity and melting
point, and the thermodynamics of their solutions.
The RF-GC technique is a useful tool for inverse
chromatographic studies, particularly when gas}solid
interfaces are involved. Several such studies have been
published. Here a short outline of its possibilities is
given using some characteristic examples. Take for
instance the determination of diffusion parameters,
adsorption}desorption rate constants and isotherms,
and catalytic rate constants of some gaseous probes
on typical solid catalysts, supported or not. All these
physicochemical properties can be determined simul-
taneously in a single gas chromatographic experiment
lasting a few hours. The experiment resembles those
conducted with a plug Sow reactor, but without a gas
Sowing through it, as Figure 1 shows. It is only
gaseous diffusion of the reactant(s) and product(s) that
4094 III / REVERSED-FLOW GAS CHROMATOGRAPHY
causes the movement of these substances through the
solid bed along the length coordinate y, and then along
length coordinate z to the junction x"l
of the samp-
ling column. Two diffusion coefRcients of the probe
molecules are involved in these movements: D1 in
section L1 and another D2 in section L2 inSuenced by
the bed obstruction factor . These are taken care of
implicitly in the mathematical analysis, without using
the actual values of D1 and D2 from the literature.
The mass balance equation of the probe reactant
A in the region y (cf. Figure 1) Rlled with the station-
ary phase under study is:
cy 2cy as
"D2 2 !k 1 (cHs !cs) [1]
t y \ ay
where cy is the gaseous concentration of the probe
A in region y; t is the time measured from the moment
of injection of A as an instantaneous pulse (delta
function, ) onto the solid bed, at y"L2; k 1 is the
\
rate constant for desorption of A from the bulk solid;
as is the amount of stationary phase per unit length of
solid bed; ay is the cross-sectional area of the void
space in region y; and cs, cHs are the concentrations of
A adsorbed on the solid at time t and at equilibrium,
respectively.
An analogous mass balance equation is valid for the
gaseous concentration cz of A in diffusion column z:
cz 2cz
"D1 2 [2]
t z
It is noteworthy that the usual convective terms
!v(cy/y) and !v(cz/z) of gas chromatographic
mass balances (v being the average linear velocity of
the carrier gas) are missing from both the above
equations, since no carrier gas Sows through column
sections y and z. This makes the solution of differ-
ential eqns [1] and [2] much easier, and all
disadvantages connected with the carrier gas Sow
mentioned before disappear. No competition be-
tween v and rate processes on the gas}solid interface
takes place. No corrections of v are required, not even
the measurement of its approximate value is needed,
except only to calculate a calibration factor g which is
needed in the isotherm determinations. Both eqns [1]
and [2] have the form of Ficks second law for diffu-
sion in one dimension, the Rrst equation being modi-
Red only by inclusion of the far right-hand side term
describing the overall rate of adsorption of the probe
gas A onto the surface of the stationary phase.
The system of partial differential equations,
eqns [1] and [2], is complemented by:
1. the rate of change of the adsorbed concentration cs:
cs
"k 1(cH !cs)!k2cs [3]
t \ s
now including a rate constant k2 of a possible Rrst-
or pseudoRrst-order surface reaction of the adsor-
bed substance A; and
2. a local experimental adsorption isotherm:

t
ns ay
cHs " (y!L2)# k1 cy() d [4]
as as 0
where ns is the initially adsorbed amount of A, k1 is
a dynamic adsorption rate constant, and  is
a dummy variable for the time t. The adsorption
equilibrium is local and instantaneous, the adsorp-
tion parameter k1 transforming the area under the
cy versus t curve into cH
s .
Equation [4] describes the actual experimental iso-
therm, not necessarily a linear one, without the help
of an a priori isotherm equation (Langmuir, BET,
etc.). The graphical experimental isotherm can be
constructed in detail if desired, but this is not neces-
sary. Only the basic deRnition by eqn [4] sufRces to
incorporate the exact isotherm into the mathematical
calculations. The nonlinearity in general is automati-
cally taken into account.
The use of a dummy variable  for the time t simply
facilitates the notation in the integral of eqn [4] with-
out any special meaning attached to .
The system of eqns [1]}[4] is solved under the
initial conditions:
n
cz(0, z)"0, cy(0, y)"  (y!L2)
ay
and:
cs(0, y)"0 [5]
where n is the total amount of A injected. The

solution is effected by using double Laplace trans-
forms of all terms with respect to time and length
coordinates, under the given initial conditions and the
isotherm eqn [4], and subject to the appropriate
boundary conditions at z"0, z"L1 and y"L2. By
means of various approximations this leads to the
expression:
H1/M"g cz(0, t)"A1 exp (B1t)#A2 exp (B2t)
#A3 exp (B3t)#A4 exp (B4t) [6]
where H is the height (in arbitrary units, say cm) of
sample peaks measured from the ending baseline of
the chromatogram, like that of Figure 2, M is the
response factor of the detector (M"1 for the FID),
and g is a calibration constant for the detector, trans-
forming cm of H to mol cm\3 of the concentration
cz(0, t) of probe substance A at z"0 and time t.
 III / REVERSED-FLOW GAS CHROMATOGRAPHY 4095

A steady-state approximation for cs in eqn [3], 2V1k1k 1k2"B1B2B3B4"W [11]

\
dcs/dt"0, leads to:
2(1#V1)"!(B5#B6#B7)"X1 [12]
H1/M"gcz(0, t)"A5 exp (B5t)#A6 exp (B6t)
#A7 exp (B7t) [7] k1k 1k2
\ "B5B6#B5B7#B6B7"Y1
12#
k 1#k2
instead of eqn [6]. \
The detailed assumptions on which the derivation [13]
of eqns [6] and [7] was based have been published
(see Further Reading). 2V1k1k 1k2
\ "!(B5B6B7)"Z1 [14]
It is seen from eqns [6] and [7] that cz(0, t), i.e. the k 1#k2
gaseous concentration of the probe A at the point \
z"0 of the diffusion column (cf. Figure 1), as mea-
1"2D1/L21, 2"2D2/L22,
sured by H, should be Rtted to the sum of four or
three exponential functions of time. This can be V1"2V
G(empty)/VG#L22/L21 [15]
achieved by a nonlinear least-squares PC program in
the Journal of Chromotography A (1998) papers where VG and V
G are the gaseous volumes of the
listed in the Further Reading, by entering the pairs void spaces in regions z and y, respectively and 
H (peak height), t (time of reversal) in the DATA the external porosity of the solid bed. The
lines. The exponential coefRcients of time B1, B2, B3 X, Y, Z, W, X1, Y1 and Z1 are just auxiliary para-
and B4 of eqn [6] and B5, B6 and B7 of eqn [7], are meters to facilitate the calculation of the primary
calculated, together with the respective pre-exponen- kinetic parameters k1, k 1 and k2 from Bi. This is
tial factors A1, A2, A3 and A4 of eqn [6], and A5, A6 \
done as follows: the sum k 1#k2 is obtained from
and A7 of eqn [7]. This PC program is based on the \
the difference X!X1, from the ratio W/Z1, or as
so-called exponential stripping method of Sedman a mean value of these two. Subtraction of k 1#k2
and Wagner. The calculation of the Bi and Ai values \
from X gives the value of 2(1#V1). From this, 2V1
(i"1, 2,2, 7) is guided by the overall goodness of Rt is easily computed since V1 is given by the relation [15].
expressed by the square of the correlation coefRcient The calculation of k1k 1 follows from the relation:
r2. This universally accepted criterion, calculated and \
printed, is in the range 0.990}0.999 in most cases,


showing a remarkable goodness of Rt for a nonlinear Z
k1k 1" Y!2(1#V1)(k 1#k2)!
regression analysis. A t test of signiRcance for the \ \ k 1#k2
\
smallest r2 usually found shows that it is highly signif-

icant, with a probability smaller than 0.05% of being
exceeded. The program also prints, together with the
B values, their standard errors, which are usually
reasonable for physicochemical measurements.
The question naturally arises as to the meaning and
physical content of the parameters Ai and Bi so deter-
mined. These are given directly and explicitly by the
solution of the system of eqns [1]}[5] that led to
eqns [6] and [7]. They are written below:
2(1#V1)#k 1#k2"!(B1#B2#B3#B4)"X
\ m and ms, but the analytical form of this dependence
[8]
2(1#V1)(k 1#k2)#12#k1k 1
\ \
"B1B2#B1B3#B1B4#B2B3#B2B4#B3B4"Y
[9]
12(k 1#k2)#2V1k1k 1#k1k 1k2
\ \ \ adsorption isotherm, instead of the real experimental
"!(B1B2B3#B1B2B4#B1B3B4#B2B3B4)"Z
[10]


W 2V1
# 1! [16]
2V1(k 1#k2) k 1#k2
\ \
Then, dividing W by 2V1 and by k1k 1 gives the
\
value of k2. The value of k 1 follows from the differ-
\
ence (k 1#k2)!k2, and that of k1 from the ratio
\
k1k 1/k 1.
\ \
The pre-exponential factors A1, A2, A3 and A4 of
eqn [6] and A5, A6 and A7 of eqn [7] are explicit
functions of gm12/VQ , where VQ is the volumetric Sow
rate of the carrier gas, and of B1, B2, B3, B4, k 1, k2,
\
is not required in the calculations.
It is well known that the values of the three rate
constants k1, k 1 and k2, especially determined at
\
various temperatures, are of primary importance in
characterizing solid adsorbents and catalysts. The
experimental data, i.e. the H, t pairs from the
chromatogram, can be analysed by using a linear
isotherm (eqn [4]) without bothering whether it
is linear or nonlinear. Both methods lead to the
4096 III / REVERSED-FLOW GAS CHROMATOGRAPHY

simultaneous determination of three primary kinetic being related only to the local adsorption isotherm
parameters, namely, k1, k 1 and k2. From these (k1), the desorption rate constant (k 1) and the sur-
\ \
primary rate constants, the method based on the face reaction rate constant (k2).
linear isotherm calculates the overall mass transfer The only physicochemical assumptions made con-
coefRcients KG and KS for the gaseous and the solid cerning the gas}solid interactions are that all para-
phases, respectively, while the real isotherm method meters measured directly or calculated indirectly
computes also the deposition velocity Vd and the refer to elementary steps at equilibrium. Thus the
overall reaction probability  of the solute with the ratio k1/k 1 represents the equilibrium distribution
\
stationary phase, by means of the relations: constant K, according to the principle of microscopic
reversibility. Note that it is not easy to measure simul-
k1V
G(empty) k2 taneously rate constants of adsorption (k1) and de-
Vd" ; [17] sorption (k 1) at a dynamic equilibrium state like that
As k 1#k2 \
\ justiRed experimentally in the experiments described
here.


1/2
1 RgT 1 1 Finally, the explicit calculation of the isotherms has
" ; # [18]
 2MB Vd 2 been described in the literature. However, eqn [6]
provides an easy route to the calculation of the distri-
where Rg is the ideal gas constant, T the absolute bution of surface adsorption energies, , the local
temperature, As the total surface area of the solid, monolayer capacities, i.e. the maximum adsorbed
and MB the molar mass of the probe gas. The original concentrations, cHmax, of the probe substance on each
deRnition of Vd was the Sux of gas to the solid kind i of adsorption sites, and the local adsorption
surface divided by the concentration difference isotherms i(p, T, ) on heterogeneous surfaces of
between the bulk gas-phase and the surface. The solids. Many papers and books have recently been
values of k1, k 1, k2, Vd and  are calculated and published on this subject. From cH max, in mol per g of
\
printed directly by running the PC program adsorbent, multiplying by the Avogadro number
mentioned before. NA and the cross-sectional area of the probe molecu-
It is clear from the deRnitions of Vd and  that both les in m2, one can Rnd the speciRc surface area of the
parameters are independent of molecular diffusion, stationary phase in m2 g\1.

Table 1 Adsorption, desorption and surface reaction parameters for various hydrocarbons on four metal oxides, at 323.2 K, based on
a linear and a nonlinear isotherm model

CxHy k1 (10\4 s\1 ) k 1 (10\4 s\1 ) k2 (10\4 s\1 ) KG (10\9 cm s\1 )Vd (10\9 cm s\1 )  (10\13 )
\
Linear Nonlinear Linear Nonlinear Linear Nonlinear Linear Nonlinear Nonlinear

Metal oxide: Fe2O3

C2H2 1.56 16.0 3.51 47.8 6.17 21.6 16.4 16.3 12.7
C2H4 131 0.0182 1.26 397 3.41 444 1380 0.561 0.454
C2H6 92.0 13.4 4.99 11.3 3.25 20.3 969 28.3 23.7
C3H6 1.47 0.0094 3.16 521 4.13 560 15.4 0.448 0.444
1-C4H8 2.64 527 2.94 0.0821 2.67 0.338 27.8 2280 2610

Metal oxide: Cr2O3

C2H2 6.76 15.5 2.69 99.9 0.340 0.383 0.196 0.132 0.151
C2H4 7.55 14.2 2.38 167 0.270 0.385 0.219 0.0733 0.0839
C2H6 6.06 9.4 3.04 81.5 0.211 0.429 0.176 0.110 0.126
C3H6 3.94 9.02 2.20 62.5 0.016 0.133 0.114 0.0428 0.0490
1-C4H8 4.91 2.63 1.08 33.4 0.218 0.805 0.142 0.138 0.158

Metal oxide: ZnO

C2H2 11.8 6.90 3.52 54.3 0.211 0.936 0.440 0.374 0.291
C2H6 7.76 7.80 2.85 124 0.370 0.695 0.295 0.137 0.114

Metal oxide: PbO

C2H2 0.180 5739 0.280 0.0322 4.52 28.9 21.2 244 800 190 900
C2H6 45.3 10.0 11.9 64.5 3.18 15.9 5250 84.5 70.8
C3H6 0.510 6.37 7.61 35.2 3.33 14.9 58.5 81.0 80.3
1-C4H8 1.51 5.11 3.59 5.75 2.13 23.6 138 289 331
 III / REVERSED-FLOW GAS CHROMATOGRAPHY
Some Representative Results
Applying the previous theoretical analysis for the
characterization of some metal oxides used as solid
catalysts gave the values of the physicochemical para-
meters k1, k 1, k2, KG, Vd and  for Rve probe hydro-
\
carbons. These are given in Table 1, calculated for
both the linear and nonlinear isotherm models. It is
seen that the values of some physicochemical
parameters are signiRcantly different for the two
models; it is clear that for an inorganic solid of
the kind used as a catalyst the linear isotherm model
is inadequate, due to the nonuniformity of the sur-
face. The two physicochemical parameters k1 and
k 1 characterize any newly prepared catalyst.
\
These constants can be measured easily and accurate-
ly and their values supply a reliable catalyst
characterization.
An a priori acceptance of an adsorption isotherm
equation is not needed, which is preferred to the
Langmuir or the BET isotherms for this purpose. Any
new catalyst sample can be tested by measuring
adsorption and desorption rate constants before
utilization. From Table 1, it can be seen that many
values obtained from the nonlinear isotherm model
are 1}4 orders of magnitude different from those
corresponding to the linear model. All physicochemi-
cal quantities based on the nonlinear model are
intended to characterize newly prepared catalysts,
on the basis of accurately deRned physicochemical
concepts.
In some cases in Table 1 the ratios of the same
parameters determined with linear and nonlinear iso-
therms are inverted in going from one gas}solid sys-
tem to another. This is probably due to the fact that
solid surfaces are very heterogeneous as regards ad-
sorption energy distribution and this heterogeneity
may differ greatly from one solid to another for the
same hydrocarbon, and from one hydrocarbon to
another for the same solid. The local adsorption iso-
therm of eqn [4] is very sensitive to the energy distri-
bution function on the surface, whereas simple
overall linear isotherms are not.
Of course, it is not possible from the measurements
at one temperature to conclude whether the gas}
solid interactions represent physical adsorption
or chemisorption. Experiments at various temper-
atures, however, will give heats of adsorption
and activation energies, from which it may be pos-
sible to draw conclusions about the nature of the
adsorption processes.
4097
Conclusion
The RF-GC technique is a useful tool in inverse GC
studies. It has also some advantages for adsorption
isotherm determinations in that: (1) the gaseous
diffusion and resistance to mass transfer are not
neglected, (2) the sorption effect in dynamic
chromatographic systems is nonexistent, (3) the pres-
sure gradient is negligible along the solid bed, and (4)
the experimental isotherm can be determined in the
presence of a surface reaction of the adsorbate. This is
very important when dealing with catalysts, when
many determinations based on chemisorption must
follow every catalyst preparation.
The method brieSy outlined here has been used for
several published physicochemical measurements.
Results have also been obtained for deposition para-
meters of air pollutants on solid surfaces with a simul-
taneous determination of the apparent rate constant
of gaseous reactions above the solid, and also for
the measurement of various rate coefRcients and
equilibrium constant by utilizing the solid phase as an
internal wall coating of denuder tubes in place of the
diffusion column.
See also: II/Chromatography: Gas: Theory of Gas
Chromatography.
Further Reading
Abatzoglou Ch, Iliopoulou E, Katsanos NA, Roubani-Ka-
lantzopoulou F and Kalantzopoulos A (1997) Journal of
Chromatography A 775: 211}224.
Gilbert SG (1984) Advances in Chromatography 23:
199}228.
Katsanos NA (1982) Journal of the Chemical Society, Fara-
day Transactions 178: 1051}1063.
Katsanos NA (1988) Flow Perturbation Gas Chromatogra-
phy. New York: Marcel Dekker.
Katsanos NA and Roubani-Kalantzopoulou F (1995) Jour-
nal of Chromatography A 710: 191}228.
Katsanos NA, Thede R and Roubani-Kalantzopoulou
F (1998) Journal of Chromatography A 795: 133}184.
Rudzinski W and Everett DH (1992) Adsorption of Gases
on Heterogeneous Surfaces. New York: Academic Press.
Sedman AJ and Wagner JG (1976) Journal of Pharmaceut-
ical Sciences 65: 1006}1010.
Sotiropoulou V, Vassilev GP, Katsanos NA, Metaxa H and
Roubani-Kalantzopoulou F (1995) Journal of the Chem-
ical Society, Faraday Transactions 91: 485}492.
Vassilakos Ch, Katsanos NA and Niotis A (1992) Atmo-
spheric and Environment 26A: 219}223.
4098 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS
RIBONUCLEIC ACIDS: CAPILLARY
ELECTROPHORESIS
J. Skeidsvoll, University of Bergen, Bergen, Norway
Copyright ^ 2000 Academic Press
Introduction
With the introduction of capillary electrophoresis
(CE), a new generation of electrophoretic tech-
niques has seen the light of day. The scientiRc
literature today describes a large number of applica-
tions of this powerful analytical technique in
the analysis of nucleic acids. For nucleic acids, as
for most other analytes, CE offers signiRcant
advantages over many of the conventional elec-
trophoretic techniques. In general, CE is character-
ized by short analysis time, high resolution, accuracy
and reproducibility, quantitative online detection and
automation. The small sample volumes required
and the extreme sensitivity CE offers, represent a
large analytical potential for samples of biological
origin. The fundamental analytical and operational
parameters for the separation of nucleic acids by CE
were identiRed around 1990. A decade later, CE is
considered a fully developed technology for the anal-
ysis of DNA. The rapid development of this applica-
tion of CE seems to have been driven by the many
practical applications of electrophoretic separation
and detection of DNA in both basic and applied
science.
The Rrst reported analysis of RNA by CE was
published in 1993 by Reyes-Engel et al. and describes
the separation and quantiRcation of a speciRc
messenger RNA by capillary zone electrophoresis.
To date, only a limited number of articles have
been published which focus on the application of CE
in the analysis of RNA. The reason for this is not
obvious, considering the widespread use of conven-
tional gel electrophoresis of RNA throughout the
biomedical scientiRc Reld. The fact that RNA,
in many respects, displays similar characteristics
as DNA, should constitute the basis for signiRcant
efforts in the development of RNA analyses
based on CE. However, the scientiRc literature holds
promise for a substantial increase in the use of CE in
RNA analyses. The following sections intend to give
a basic introduction to CE of RNA, with emphasis on
important analytical and operational parameters in
the analyses. Finally, examples from a diverse group
of applications are presented.
Capillary Electrophoresis of RNA
In general, electrophoretic separation of RNA is
based on the differences in electrophoretic mo-
bilities of the analytes. As in conventional elec-
trophoresis, the rate of migration of a RNA molecule
in CE depends on the mass and the dimensions of the
molecule, the charge carried, the applied current and
the resistance of the medium. In an electric Reld, at
moderate pH, negatively charged RNA migrates to-
ward the anode. A number of parameters affect
the separation of RNA in CE (see below). CE of
RNA can be divided into two separate categories
based on the principle by which the molecules are
separated: capillary zone electrophoresis (CZE) and
capillary gel electrophoresis (CGE). In CZE, the RNA
molecules are mainly separated by their charge to
mass ratio. From the fact that nucleic acids larger
than a few nucleotide units have approximately iden-
tical charge to mass ratio, CZE provides little or no
separation power. Consequently, only single RNA
species can be identiRed by this technique, unless
multiple labelling is being used. In CGE, the RNA
molecules are separated mainly by their molecular
dimensions, i.e., the ability of the different
analytes to migrate through a gel matrix. CGE is
by far the most common technique for RNA ana-
lyses. A description of CGE of RNA is given in the
following section.
Capillary Gel Electrophoresis of RNA
Analytical parameters of signiRcance for the
separation of DNA by CGE, including gel polymer
concentration, electrical Reld strength and
temperature, have been investigated and optimized
for the analysis of RNA molecules ranging from
oligomers (10 to 40 bases) to several kilobases
(Figure 1).
RNA Migration in Capillary Gel Electrophoresis
In conventional gel electrophoresis, the migration of
a RNA molecule is inversely related to the log10 mo-
lecular mass. However, base composition (primary
structure) and secondary structure can affect this
relationship. In CGE, separation is achieved because
large molecules move more slowly through the gel
than small molecules. Separation within a given RNA
molecular range is obtained by selecting a gel of
 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS
Figure 1 Electropherogram of RNA molecular-mass marker.
The sample was denatured, injected at 300 V cm\1 for 10 s and
subjected to CGE at 200 V cm\1 in 1;TBE/8 mol L\1 urea con-
taining 0.3% HPMC. AU, arbitrary units. (Reprinted from Skeid-
svoll J and Ueland PM (1996) Analysis of RNA by capillary
electrophoresis. Electrophoresis 17: 1512}1517. Copyright 1996,
with permission from Wiley-VCH Verlag GmbH.)
appropriate pore size. Experiments have dem-
onstrated that CGE of higher molecular mass RNA
(in the range from 100 bases to more than
6 kb) to a large extent resembles CGE of single-
stranded DNA. An interesting Rnding is that RNA
Figure 2 Comparison of migration of RNA and single-stranded
DNA. A molecular-mass marker containing RNA and DNA com-
ponents was denaturated by pre-incubation at 953C for 3 min in
the presence of 80% formamide and subjected to electrophoresis
in a separation buffer containing 8 mol L\1 urea and 0.3% HPMC.
Migration time is plotted versus molecular mass for RNA (*) and
DNA (). (Reprinted from Skeidsvoll J and Ueland PM (1996)
Analysis of RNA by capillary electrophoresis. Electrophoresis 17:
1512}1517. Copyright 1996, with permission form Wiley-VCH
Verlag GmbH.)
4099
and single-stranded DNA of identical length display
different migration when co-analysed under com-
pletely denaturing conditions, DNA having a slightly
higher migration rate than RNA (Figure 2). The shift
in migration for DNA vs. RNA is found constant for
molecules ranging from 100 to approximately 1000
bases. The phenomenon is explained by the higher
charge to mass ratio of single-stranded DNA.
An inherent property of the (single-stranded) RNA
molecule is the potential to form secondary structures
or intramolecular and intermolecular hydrogen
bonds. To what extent the reaction takes place is
primarily a function of the RNA sequence. The pre-
dictable determination of RNA molecular mass is
essential in most RNA techniques based on elec-
trophoretic separation. Consequently, in order to pre-
vent unpredictable migration of RNA due to the
formation of secondary structures, CE should be car-
ried out under completely denaturing conditions.
Such conditions can be accomplished through optim-
ization of physical and/or chemical parameters. For
example, heating the sample in the presence of a
denaturant prior to electrophoresis and addition of
a denaturant in the electrophoresis and separation
buffers combined with high temperature during
electrophoresis should have a strong denaturing
effect. Denaturants are chemical compounds
that disrupt hydrogen bonds. The most commonly
used denaturant, urea, is often added to the separ-
ation buffer in very high concentrations (up to
8 mol L\1). Despite an extensive use of buffer
additives, data from both conventional RNA gel elec-
trophoresis and CE of RNA indicate that even the
presence of 8 mol L\1 urea in the separation buf-
fer is not sufRcient to completely disrupt intra-
molecular or intermolecular hydrogen bonds.
Addition of the stronger denaturant formamide in
concentrations up to 30% (in addition to 3.5 mol L\1
urea) and performing CE at 603C has been necessary
to disrupt extensive secondary structures in a ham-
merhead ribozyme (37 nucleotides) and to provide
appropriate separation from its substrate (17 nucleo-
tides). In addition, a decrease in ionic (cationic)
strength and an increase in pH are known to have
a denaturing effect on RNA. Common problems
related to inefRcient separation, detection and
identiRcation of RNA in CE, probably result from
incomplete denaturation of RNA.
Important Analytical and Operational
Parameters
From the comprehensive scientiRc literature describing
CE of nucleic acids, it is obvious that operational para-
meters like capillary dimensions (m) electrical Reld
4100 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS
strength (E, V cm\1) and temperature (t, 3C) have to be
chosen carefully to optimize the separation of RNA.
In most applications of CE in RNA analyses, the
electroosmotic Sow is eliminated through the use of
surface-modiRed (coated) capillaries. This consider-
ably simpliRes the experimental design and leaves the
scientist with a limited number of variable analytical
and operational parameters.
Buffer Composition
In general, all buffer systems that are used for
CZE can also be used for CGE. The most common
buffers are the Tris-borate buffers (i.e., TBE)
with a pH range of 7.5}9.0. Buffer additives like
methanol and acetonitrile are used in separation buf-
fers optimized for low-molecular-mass RNA. Urea
and formamide are mainly added as denaturants.
Moreover, the addition of urea to the separation
buffer has been found to increase the resolution
of RNA (except for oligoribonucleotides less than
5 bases).
Gel-forming Polymers
A number of different gel-forming polymers
have successfully been used in both DNA and RNA
separations by CE. The separation matrices comprise
both cross-linked gel polymers like polyacrylamide
and noncross-linked gel polymers like linear polyac-
rylamide and cellulose derivatives. Through the op-
timization of composition and concentration,
noncross-linked polymers have now taken over as the
predominant separation matrices for most RNA ana-
lyses. These materials have demonstrated signiRcant
advantages over cross-linked polymers, including
ease of preparation and use, physical stability and
uncomplicated washing and reRlling procedures be-
tween analyses. The resolving power of these gels
mainly depends on the concentration of the dissolved
polymer } dilute gels are used for high-molecular-
mass RNA molecules and more concentrated gels for
low-molecular-mass RNA.
A systematic study of the electrophoretic separ-
ation of RNA at different concentrations of
a noncross-linked polymer gel demonstrated that
high concentrations ('0.3%) hydroxypropylmethyl-
cellulose (HPMC) were optimal for the separation of
RNA less than 1000 bases and low concentrations
were optimal for the separation of higher molecular-
mass RNA. The results are consistent with data from
the separation of DNA by CE.
A number of separation matrices, optimized for
different ranges of RNA molecular mass, are
commercially available. Additionally, matrices are
available which contain denaturants.
Electrical Field Strength
Electrical Reld strength is recognized as an important
operational parameter in CE of RNA. An in-
crease in electrical Reld strength is found to result
in a logarithmic decline in migration times. EfR-
ciency, N (number of theoretical plates) and resolu-
tion, Rs, are found to have a more complex relation
to the electrical Reld strength, although a clear
tendency towards a decline in both parameters
with increased electrical Reld strength has been
demonstrated. In general, low electrical Reld
strengths are preferable for the optimal separation
of RNA molecules larger than 100 bases. With the
increase in electrical Reld strength, an increased
current will result in the production of heat (Joule
heating), which, if excessive, adversely affects
the separation by causing broadening of the
migrating zones.
Temperature
Temperature, an important analytical and opera-
tional parameter, inSuences both total analysis
time and system efRciency. The effect is
mainly mediated by a decrease in the separation buf-
fer viscosity. A linear decrease in migration time for
RNA molecules ranging from 200 to 2000 bases has
been observed for temperatures ranging from 20 to
503C. The separation efRciency and resolution
were found essentially constant over the temperature
range. In addition, temperature is a parameter of
signiRcant importance in the CE of RNA due to
its denaturing effect on intra- and intermolecular
hydrogen bonds.
Quantitative Aspects
For a general description of the quantitative aspects
of injection in CE, see DNA: Capillary Electrophor-
esis. Electrokinetic injection is the most common
injection mode for RNA in CE. In order to obtain
quantitative data, an external reference should be
added to or co-injected with the RNA sample. Ideally,
the external reference should resemble the sample of
interest, but be readily identiRable. Hydrodynamic
injection is often used in experiments for the deter-
mination of reaction kinetics or in studies of enzy-
matic activity. Hydrodynamic injection provides
representative samples for analysis.
UV absorbance is the most common detection prin-
ciple for RNA in CE. Despite its general usefulness,
the technique suffers from low sensitivity as com-
pared to other detection principles (e.g., laser-
nduced Suorescence) and represents a limiting factor
in some RNA analyses. Detection of RNA based on
(laser-induced) Suorescence confers the selectivity
 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS
and sensitivity required for a number of analyses
where the concentration of analytes is low. However,
this detection principle normally requires the co-
valent attachment of Suorophores to target mole-
cule(s) or Suorogenic buffer additives.
Applications
The application of CE to RNA includes a diverse
group of analyses, which often includes one or a com-
bination of the following elements:
E Characterization of RNA molecular dimensions
(mass or spatial structure).
E Characterization of RNA sequence.
E Characterization of RNA reaction kinetics.
E Characterization of RNA-binding constants.
An example of a group of CE-based RNA analyses
that combines more than one of these elements is
the hybridization techniques, which both rely on
molecular mass determination and sequence-speciRc
detection of the RNA of interest. In the applica-
tions described, RNA samples originate either from
chemical synthesis (oligoribonucleotides) or are ex-
tracted from biological material. The last group
comprise RNA of eukaryotic, prokaryotic and viral
origin.
Characterization of RNA Molecular Dimensions
(Mass or Spatial Structure)
Capillary electrophoresis analysis of synthetic short-
chain oligoribonucletides (Figure 3) Thirty synthetic
oligoribonucleotides, ranging from 3 to 18 nucleotides
were analysed by CE in a nondenaturing noncross-
linked gel polymer. An equation was developed, based
on the experimental data which, under Rxed condi-
tions, accounts for the inSuence of charge to mass ratio
(i.e., net charge and base composition) on migration
time. High resolution (1 nucleotide unit) was
Figure 3 CGE analysis of a mixture of 12 oligoribonucleotides from 4 to 18 units under nondenaturing conditions. (Reprinted from
Kolesar JM, Allen PG and Doran CM (1997) Direct quantification of HIV-1 RNA by capillary electrophoresis with laser-induced
fluorescence. Electrophoresis 697: 189}194. Copyright 1997, with permission from Elsevier Science.)
4101
obtained for homologous series of oligoribonucleo-
tides, and, to some extent, for groups of oligoribonuc-
leotides of identical length, but different sequence.
CGE is often used to determine the quality of
chemically synthesized oligoribonucleotides and can
be used in conjunction with HPLC to develop an
effective method for the puriRcation of crude
oligonucleotide solutions.
Low-molecular-mass RNA Vngerprinting of bacteria
by capillary electrophoresis RNA proRling provides
a direct genotypic Rngerprint technique for the identi-
Rcation and classiRcation of bacteria by generating an
electropherogram including three groups of mole-
cules of taxonomic signiRcance, small tRNAs, large
tRNAs, and 5S rRNA (ranging from 70 to 135 nu-
cleotides). The technique is of particular importance
for molecular ecology and taxonomic studies, and
can also be applied directly to analyses of environ-
mental samples. CGE using both noncross-linked
polymer gels (HPMC) and cross-linked polyacrylam-
ide gels have been investigated and optimized for
their applicability in the separation of this class of
RNA molecules. Good resolution was obtained only
for small tRNAs up to approximately 80 nucleotides
using cross-linked gels, larger tRNAs and 5S rRNA
could not be resolved with this experimental set-up.
The use of noncross-linked polymer gels resolved
tRNAs and 5S rRNA under nondenaturing condi-
tions, even when they possessed only different
secondary structure or small differences in length
(1}5 nucleotides). CE using HPMC in the separation
buffer resulted in both good peak resolution and
reproducibility and was suitable for routine Rnger-
printing of bacterial low-molecular-mass RNA.
Investigation of intracellular metabolism of a stabil-
ized ribozyme by CGE after uptake by transfection or
as free ribozyme CGE has been used to investigate
4102 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS
cellular uptake and degradation of a Suorescein label-
led chemically stabilized ribozyme (37-mer). After
internalization by transfection or uptake of free
ribozyme, electrophoretic peaks of intact ribozyme
and different degradation products were easily
resolved and the amount of intracellular intact
ribozyme quantiRed. Using laser-induced Suores-
cence for detection, the method offered extreme
sensitivity with estimated limit of detection: 10 and
200 pmol L\1 ribozyme from cell extracts and cell
medium, respectively.
A third example include the direct quantiRcation
(by UV absorbance measurements) of HIV-1 RNA in
human plasma by CZE.
Characterization of RNA Sequence
An important and diverse group of analytical RNA
techniques is based on sequence-speciRc hybridiza-
tion between two single-stranded nucleic acids
and the electrophoretic separation, detection and
quantiRcation of the intermolecular reaction product
(hybrid). Consequently, the analyses involves charac-
terization of RNA in two dimensions, size and se-
quence. The Northern (RNA) blotting technique, the
nuclease- (S1 or RNase) protection assays and other
RNA-hybridization techniques play an important
role in the qualitative and quantitative analysis of all
classes of RNA in biological systems. The techniques
often involve use of radioisotope labels in detection.
CE-based hybridization analyses of RNA has been
successfully demonstrated for a number of applica-
tions. In general, the hybridization reactions are car-
ried out pre-column (in solution) and the separation
and detection of the hybrids on-column. It is demand-
ing to transfer the conventional hybridization tech-
niques to the capillary format and important
challenges are related to the development of selective
and compatible conditions for both the pre-column
and on-column elements of the analyses. Addition-
ally, the low sample volumes injected in CE represent
signiRcant analytical and instrumental challenges.
Direct quantiVcation of a speciVc messenger RNA by
capillary zone electrophoresis Total RNA was
isolated from whole blood and hybridized with a
biotinylated oligonucleotide speciRc for a receptor
mRNA (angiotensin II). The hybrid was Rrst captured
on streptavidin-conjugated magnetic beads, then
eluted and Rnally quantiRed by UV absorbance in
CZE. Using this procedure, quantiRcation of the ex-
pression of low expressed genes is easy and fast and
subject to two limiting factors: the speciRcity of the
capturing oligonucleotide or probe selected and the
amount of total RNA. The procedure represents an
nonradioactive alternative to conventional RNA ana-
lyses like Northern blotting, RT-PCR or the nuclease-
(S1 or RNase) protection assays.
Direct quantiVcation of HIV-1 RNA by capillary elec-
trophoresis with laser-induced Wuorescence (LIF) de-
tection (Figure 4) A hybridization method using a
HIV-speciRc probe with analysis by CE}LIF was de-
veloped. Plasma samples from HIV-seropositive pa-
tients were lysed to obtain RNA, hybridized with
a Suorescein-labelled HIV-speciRc DNA probe, diges-
ted by a speciRc RNase to remove nonhybridized
RNA and analysed by CE-LIF in presence of the
Suorescent intercalator thiazole orange (TO). 19 fg
(corresponding to 1710 copies per mL of starting
plasma) of HIV RNA was quantitatively detected.
The technique, analogous to the conventional RNase
protection assay, takes advantage of signal ampliRca-
tion by using the RNA-binding Suorescent intercala-
tor TO. Calibration is done through the analysis of
a Suorescein-labelled RNA marker. The actual ap-
proach appears to be an efRcient, sensitive and
reliable method to speciRcally and quantitatively ana-
lyse RNA from a variety of sources.
Detection of oligonucleotide N3}P5 phosphor-
amidate/RNA duplexes with capillary gel electro-
phoresis The DNA analogues N3}P5
phosphoramidates (3-phosphoramidates) has dem-
onstrated favourable properties as hybridization
probes, including high melting temperature of du-
plexes with RNA and high reaction rate at low ionic
strengths. The RNA hybridization technique takes
advantage of the 3-phosphoramidate oligomer prop-
erties as hybridization probes through duplex forma-
tion with short complementary strands of RNA of
identical length (9 nucleotides). Hybrids were found
to have unique relative mobilities in CGE, compared
to the reactants. The ability of CGE to detect the
presence of, and to discriminate between, perfect du-
plexes and duplexes that contained a base mismatch
were demonstrated under routine electrophoretic
running conditions. In conclusion, the study indicates
that 3-phosphoramidate oligonucleotides may have
application in nucleic acid based diagnostics.
Characterization of RNA Reaction Kinetics
Current commercial CE instrumentation offers
the scientist operational functions like thermostated
sample compartments, automatic sampling and
thermostated analyses. These functions increase
the potential of CE technique developments, com-
pared with most conventional gel electrophoresis sys-
tems, and are especially useful in studies of reaction
kinetics, for the determination of association and
dissociation constants and in enzymatic assays.
 III / RIBONUCLEIC ACIDS: CAPILLARY ELECTROPHORESIS 4103

sampling operation, respectively. CE has developed

into an effective technique, for example, deter-
mination of apparent equilibrium constants for mo-
lecular association in solution. Examples of CE being
used in the characterization of RNA reaction kinetics
are described below.

Determination of the catalytic activity of a hammer-

head ribozyme (Figure 5) Ribozymes are sequences
of catalytic RNA. The catalytic activity of a synthetic
hammerhead ribozyme (37-mer), i.e., the hydrolysis
of its RNA substrate (17-mer), has been determined
by regular injection from the reaction vial. Kinetic
parameters like km and kcat were calculated from the
integrated area of the decreasing substrate peak. Ex-
perimental conditions compatible with both ribozyme
activity and CE separation of ribozyme and substrate
were determined by careful optimization of the reac-
tion mixture (sample) and the separation buffer.
A combination of thermal and chemical denaturation
was necessary to separate the oligoribonucleotides.
Kinetic analyses were performed in the range where

Figure 4 Electropherogram from a hybridization experiment.

RNA samples obtained from a HIV-seropositive patient and
a seronegative volunteer were hybridized with a HIV-specific
probe and analysed by CGE: (A) HIV RNA/probe complex (HIV-
positive patient); (B) seronegative volunteer; (C) negative control
containing all reaction components except RNA. (Reprinted from
Kolesar JM, Allen PG and Doran CM (1997) Direct quantification
of HIV-1 RNA by capillary electrophoresis with laser-induced Figure 5 Typical electropherograms demonstrating different
fluorescence, Journal of Chromatography B 697: 189}194. Copy- stages in a ribozyme-mediated catalytic breakdown of a RNA
right 1997, with permission form Elsevier Science.) oligonucleotides substrate. (Reprinted from Saevels J, Schepdael
AV and Hoogmartens J (1999) Capillary electrophoresis of RNA
A thermostated and closed sample vial and a oligonucleotides: catalytic activity of a hammerhead ribozyme.
computer-controlled injection system is equivalent Analytical Biochemistry 266: 93}101. Copyright 1999, with per-
to a chemical reaction chamber and an automatic mission from Academic Press.)
4104 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
the substrate exhibited linear detector response. RNA
detection by UV absorbance was found to be a limit-
ing factor in the Michaelis}Menten analysis.
Characterization of pre-tRNA maturation by RNase
using capillary gel electrophoresis A CGE-based
technique has been developed in order to characterize
the reaction kinetics and mechanism for maturation
of a set of pre-tRNAfMet mutants. At all steps of the
study of RNase P, including the preparation of the
pre-tRNA (quality), the kinetic analysis and the con-
trol and yield of the puriRcation steps, CGE was
found appropriate and reliable.
Analysis of a ribonuclease H digestion of N3}P5 phos-
phoramidate}RNA duplexes by capillary gel elec-
trophoresis The activity of a ribonuclease H (RNase
H)-mediated RNA hydrolysis of duplexes formed by
oligodeoxyribonucleotides or their analogue, N3}P5 Further Reading
phosphoramidates and complementary RNA strands,
have been investigated. The enzymatic assay conditions
were carefully optimized enabling sampling directly
from the reaction mixture. CGE electropherograms
revealed that RNA}N3}P5 phosphoramidates du-
plexes remained intact and therefore did not appear
to be a substrate for RNase H.
Conclusion
Today, CE of nucleic acids has become an important
analytical technique for biochemists and molecular
RNA
See III / DEOXYRIBONUCLEIC ACID PROFILING: Capillary Electrophoresis
SELECTIVITY OF IMPRINTED POLYMERS:
AFFINITY SEPARATION
O. RamstroK m, ISIS } Universite& Louis Pasteur,
Strasbourg, France
Copyright ^ 2000 Academic Press
Ever since the discovery of antibodies and receptors,
and their remarkable selectivities for almost any given
chemical structure, scientists have been intrigued by
the quest of mimicking their properties in synthetic or
biologists and the scientiRc studies described
here clearly illustrate the applicability of CE in the
analysis of RNA. Through efRcient separations of
RNA molecules ranging from a few bases to several
kilobases, the speciRc and sensitive detection of
RNA sequences and the study of RNA reaction kinet-
ics, scientists have taken advantage of the prominent
characteristics of CE. Compared to the analysis of
DNA, additional challenges exist in the analysis of
RNA, challenges mainly related to RNA stability and
conformation. However, efforts should be made
to overcome problems related to inefRcient separ-
ation, identiRcation and detection of RNA in CE.
Extended insight into these phenomena will realize
the inherent potential of CE for a diversity of RNA
analyses.
Cellai L, Onori AM, Desiderio C and Fanali S (1998)
Electrophoresis 19: 3160}3165.
Dedonisio L, Raible AM and Gryaznov SM (1998) Elec-
trophoresis 19: 1265}1269.
Katsivela E and HoK Se MG (1995) Journal of Chromatogra-
phy A 717: 91}103.
Kolesar JM, Allen PG and Doran CM (1997) Journal of
Chromatography B 697: 189}194.
Saevels J, Schepdael AV and Hoogmartens J (1999) Journal
of Analytical Biochemistry 266: 93}101.
Skeidsvoll J and Ueland PM (1996) Electrophoresis 17:
1512}1517.
semisynthetic systems. A material carrying a selective
preference for one ligand in comparison with other
structurally similar compounds would be of out-
standing use in a wide variety of situations, extending
from molecular transportation, via analysis, to cata-
lysis and synthesis. A multitude of sophisticated ap-
proaches have also been developed over the years,
with the objective of controlling ligand binding to an
artiRcial receptor.
 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
Molecular imprinting technology (MIT) is an
emerging concept that meets the objective of creating
substrate selective materials in a fairly simple,
yet efRcient, manner. This technique is based on
the formation of binding sites, or imprints, in
a macromolecular matrix, or other suitable molecular
support, by a molecular casting procedure. The
ligand of interest is thus acting as an active guide,
positioning the formation of the binding site, and
the outcome is a two- or three-dimensional network
embracing the template. Using dynamic interactions
between the ligand (or its substitute) and selected
building blocks, materials carrying a memory of the
ligand structure can be formed. Once the formation
of the network is established, the ligand can be re-
moved, thus exposing the binding sites, which are
subsequently accessible for repeated binding. A more
general description of the molecular imprinting con-
cept is presented elsewhere in this Encyclopedia.
The ligand afRnities that can be obtained by this
concept are quite remarkable and are often very sim-
ilar to what can be achieved by the natural systems
they are mimicking. Binding constants in the
nanomolar range have been reported. Likewise, the
selectivities they display are conspicuous, such that
very small differences in ligand structure can be dis-
tinguished. Thus, the dynamic arrangement of inter-
acting groups in the binding site, moulded in place by
the imprinting process, can result in very speciRc
interactions between the imprinted material and the
targeted ligand. In this section, the selectivity of mo-
lecular imprinted polymers and other matrices will be
discussed, focusing on their use in afRnity separation.
Molecularly Imprinted Materials
Of all areas where molecularly imprinted materials
have found applications, such as diagnostic assays,
Figure 1 Reduction of nonspecific binding by chemical blocking.
4105
drug delivery, sensor technology, and catalytic proto-
cols, by far the most explored Reld is their use in
separation science. Such materials have been de-
veloped for use in liquid chromatography (molecular
imprinting chromatography, MIC), solid-phase ex-
traction (SPE), capillary electrophoresis, membrane
technology and library screening protocols. Especially,
chiral discrimination has been heavily investigated,
and a large number of impressive chiral separations
have been performed. More detailed overviews with
respect to chiral separation, as well as to SPE, by these
materials can be found elsewhere in this Encyclopedia.
Af\nity Separation
As is the case with all types of afRnity-based separ-
ation techniques, imprint-based protocols can be con-
sidered as a mixed-type separation process. The
recognition between the analytes and the matrix relies
on both afRnity-type interactions, exerted by the en-
semble of interactions between the analyte and the
matrix in the recognition site, and less selective pro-
cesses, such as general ion exchange, metal coordina-
tion and hydrophobic interactions. Although this
mixed-type separation can sometimes be advantage-
ous in as much as it offers a means for the analytical
chemist to control the retention, most often it is
a drawback, and attempts have been made to mask
unwanted recognition contributions. For example,
the nonselective ion exchange contribution of
charged moieties in the matrix can be chemically
blocked, resulting in a material where only the
charged groups in the sites are active (Figure 1).
The phenomenon of nonspeciRc binding can some-
times be difRcult to distinguish from the afRnity pro-
cess, since a comparison between a material that has
been imprinted to a nonimprinted blank can result in
differences for compounds that should not be recog-
nized. Since the physical properties of the material
4106 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
will inevitably change upon imprinting, a change in
retention will also be found for compounds not selec-
tively recognized by the material. For example, if
a material is prepared using methacrylic acid as func-
tional monomer, the resulting imprinted material
may show more pronounced retention than its
nonimprinted counterpart. This phenomenon can in
part be attributed to the formation of carboxylate
dimers in the blank material, thus displaying fewer
carboxylate groups free to interact with the analytes.
On the other hand, in the imprinted material the
carboxylate groups take part in the recognition pro-
cess and are more exposed to the analytes. Thus, in
order to estimate the selectivity of imprinted mate-
rials, an estimate of the behaviour of both imprinted
and nonimprinted matrices needs to be made. For
a full investigation, the behaviour of two or more
imprinted matrices is also a very useful comparison.
The nonimprinted matrix will address nonselective
contributions for a chosen eluent. A set of imprinted
matrices will provide information on both the selectiv-
ity accomplished and the physical change introduced
by the imprinting process. Of course, when chiral
separations are targeted, the chiral discrimination in
itself is proof of the accomplished selectivity, since
a nonimprinted matrix will normally never show enan-
tioselectivity (unless chiral building blocks are used).
Selectivity
As mentioned above, the selectivities that can be
accomplished with molecularly imprinted materials
are often on a par with natural binders such as anti-
bodies and biological receptors. In most cases, how-
ever, this can only be achieved by careful
consideration of the system chosen, with respect to
the ligand and the building block functionalities. Al-
though much of the beauty and attraction of the
entire concept lies in its seemingly compliant facility,
design is none the less imperative for the outcome of
an efRcient imprinting process. It is only after judi-
cious control of all system parameters that highly
selective materials can be produced. Rational design is
thus involved in using interactions as speciRc as pos-
sible from the very start. Most of the bond types used
are chosen in order to acquire a desired selectivity
directly from the resulting interaction with its corre-
sponding counterpart, and this selectivity can sub-
sequently be further accentuated by the imprinting
process.
Molecular Imprinting Protocols
The characteristics of the imprinting process depend
very much on the strength of the bonds used between
the ligand and the functional building blocks, ranging
Figure 2 The higher the bond energy used for point interac-
tions, the higher the compliance of the positioning of the groups in
the site, but the lower the applicability of the concept. For this
reason, most protocols are based on noncovalent systems.
from strong, albeit reversible, covalent bonds to
weak, noncovalent interactions. What can be gained
from an increased strength between the ligand and
the functional building blocks during the imprinting
process is often a drawback during the utilization of
the resulting materials, and vice versa if noncovalent
bond types are used. If too strong bonds are used,
such as carboxylic amides and esters, the ease with
which the extraction/rebinding process can procure is
strongly limited, requiring exceedingly harsh condi-
tions. Thus, weaker bond types have to be employed
(Figure 2). Amongst reversible covalent bond types
that have been used are boronic esters, imines and
disulRdes, all easily reversible under reasonably mild
conditions. However, only boronic acid esters have
been successfully used in chromatographic systems,
due to their faster exchange kinetics, which is a reSec-
tion of their very low bond strength in aqueous envi-
ronments (G&12 kJ mol\1 for the interaction
between phenylboronic acid and glucose).
The special characteristics of metal coordination
make these very useful for molecular imprinting. The
binding strengths lie between covalent and
noncovalent, and can be easily varied by exchange
of the metal ion used. The bond energy for a
commonly used building block, iminodiacetate (IDA)
complexed with Cu2# and imidazole, amounts
to &20 kJ mol\1 at room temperature in aqueous
solution.
Although the use of reversible covalent bonds, as
well as metal coordination, meets the prerequisite of
strong interactions prior to Rxation, these systems
suffer from one serious drawback } only a limited
number of ligands can be targeted using these bond
types. Noncovalent interactions on the other hand
permit a broader utility range that can be covered,
largely making them more advantageous. This is in
accordance with biological systems, where molecular
complexes are often formed by a plethora of non-
covalent interactions such as hydrogen bonds and ion
pairing. Although these interactions, when con-
sidered individually, are weak compared to covalent
bonds, the concerted action of several of these bond
types often leads to complexes with very high stabil-
ity. The high degree of speciRcity that can be
achieved, in combination with the dynamic proper-
ties of the interactions, makes these bond types a pre-
requisite for many biological processes, and has also
 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
been the preferred choice in many imprinting
protocols.
Attempts to overcome the drawbacks from the re-
spective protocols, but still be able to beneRt from
their advantages, can be made by combining the two
extremes. Thus, protocols based on covalent binding
during the Rxation process, but switching to non-
covalent interactions afterwards, have been designed,
for example, successful protocols based on cleavage
of esters, carbonates and amides.
Origin of Selectivity
The complex, normally macromolecular nature of
molecularly imprinted materials has precluded a full
understanding of the recognition mechanism between
the analyte and the matrix. Physical methods, such as
solid-state nuclear magnetic resonance, Fourier trans-
form infrared, atomic force microscopy (AFM) and
electron spin resonance (ESR), have revealed some of
the characteristics of the materials, but a more de-
tailed account of the site architecture remains to be
resolved. In contrast, the complexes between the
functional building blocks and the print species, for-
med in solution prior to any imprinting process, can
be studied more easily. Obviously, the use of covalent
interactions, and often metal coordination interac-
tions, results in structures that can be well character-
ized by common physical organic methods. In
noncovalent systems the situation is more complex,
but the use of solution-phase methods has allowed
a picture of the interactions to be drawn. Although
imprinting of these complexes/adducts may change
their structures considerably, the information that
can be retrieved from such studies is nevertheless
valuable.
The recognition of events taking place in encoun-
ters between ligands and receptors is highly depen-
dent on the additive effect of a number of
binding forces. An optimal combination of the poten-
tial binding forces may lead to strong binding. Thus,
in order to achieve proRcient complexation between
the host molecule and the guest species, several fac-
tors such as shape complementarity, functional
complementarity and contributions from the sur-
rounding pool of solvent have to be considered.
In all imprinting protocols, be they based upon
covalent or noncovalent interactions, the most impor-
tant factor governing the substrate selectivity is the
quality and number of point interactions used to form
bonds in solution prior to Rxation, gelation or polym-
erization. In self-assembly systems using noncovalent
interactions, the main factors responsible for selective
rebinding of ligands to the imprinted sites have been
shown to be strong noncovalent bonds such as
charge}charge interactions and hydrogen bonding.
4107
Other types of interactions, such as
}
stacking and
dipole interactions, may take part as well, but
normally to a lesser extent. The three-dimensional
arrangement of these interaction points, a reSection
of the solution-phase situation set in place by the
polymerization step, leads to an inherent speciRcity of
the formed sites. Likewise, in metal coordination sys-
tems, the major points of interaction occur between
the immobilized metal ions and the coordination sites
of the analyte, and the reversible covalent interactions
between the functional moieties of the matrix and
their counterparts of the analyte make up the major
interaction in covalent systems.
In addition to these point interactions, the architec-
ture of the surrounding matrix determines the steric
limits of the site, thus providing a more or less efR-
cient van der Waals surface to the analyte. Although
the arrangement of the functional moieties in the
matrix accounts for most of the selectivity, the shape
of the site also plays an important role. The conRg-
uration of the surrounding matrix backbone, as cast
in place around the print molecule, contributes to the
overall ligand speciRcity (Figure 3). This effect is
sometimes less pronounced and substantial freedom
in ligand structure can be observed, but often the
effect is considerable and minute differences in size
can be distinguished. In several cases, the position of
a single methyl group is crucial for recognition. In this
perspective, the shape effect of the site is particularly
pronounced for parts of an imprinted molecule in
close proximity to the point interactions. For struc-
tural features of the analyte more distal from such
bonds, the shape is less important.
Forms of Selectivity
The topic of chemical selectivity can be categorized
into three main groups:
1. chemoselectivity, i.e. differentiation among various
functional groups in a polyfunctional molecule
Figure 3 Schematic representation of interactions between
a chosen ligand (here captopril) and a molecularly imprinted
material. In addition to possible point interactions, whether
covalent (disulfide, acetal, ester) or noncovalent (hydrogen
bonds, coulombic electrostatic), the backbone of the imprinted
network may also contribute to overall selectivity.
4108 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
2. regioselectivity, referring to orientational control
of the interaction
3. stereoselectivity, specifying the control of stereo-
chemistry in the recognition site
Sometimes the last group is further subdivided with
reference to the control of relative stereochemistry
(diastereoselectivity) and absolute stereochemistry
(enantioselectivity). A further physical type of selec-
tivity is the discrimination of molecular size } an issue
that may be of substantial importance when dealing
with macromolecules and solid matrices.
Chemoselectivity In imprinting protocols based
upon reversible covalent bonds, such as boronate
esters, disulRdes and imines, the chemoselectivity is
principally a consequence of the properties of the
covalent bond type chosen. Thus, for boronate esters,
vic-diols are preferred as they give more stable esters
than, e.g., mono-alcohols or gem-diols, and disulRdes
are formed exclusively from thiols. Other bond types
of a slightly more general character, e.g. acetals,
selective for carbonyl groups/alcohols, and imines,
selective for carbonyl/amino moieties, are however
potentially useful in recognizing a wider variety of
ligands.
In metal coordination systems, the coordination
properties of the ligand-metal ion pair govern most of
the chemoselectivity that can be expected. This pro-
grammed selectivity can easily be changed, simply by
changing the metal ion. If such a molecularly im-
printed material is charged with different metal ions,
the coordination properties can be Rne-tuned. Such
a bait-and-switch strategy has been used for changing
the binding strength from high afRnity during the
imprinting process to low afRnity in the rebind-
ing/separation process. For example, Cu(II) can been
changed for Zn(II), since the latter displays a weaker
Figure 4 Metal ions can be very selectively recognized by imprinted matrices. For example, Hg(II) can be distinguished from Cd(II),
Pb(II) and Cu(II) by the imprinted structure (A). Similarly, material (B) could selectively separate Ca(II) and Mg(II).
afRnity for imidazole groups. When the ion itself is
the target species (ligand), coordinating building
blocks are normally chosen to occupy less than or
equal to half of the coordinating bonds of the ion,
allowing for adaptation during the imprinting process
(Figure 4). The ionic recognition that can be achieved
with such systems is often remarkable, and systems
that are able selectively to distinguish small differ-
ences in ionic size and coordination pattern have been
designed.
In noncovalent systems, the chemoselectivity is
much less obvious in the design of the interactions. If
a print molecule contains charges or hydrogen-bond-
ing moieties, and functional monomers are chosen
accordingly, the inherent chemoselectivity in the
complexes formed is less strong, and other ligands
can compete for the interactions as well. In addition,
the functional building blocks, as well as the ligands,
may self-interact to some extent. Nevertheless, a dra-
matic chemoselectivity can be accomplished when the
strength of the noncovalent interaction is altered, for
example, by introducing substituents which change
the electronic density of a compound, or substituting
a hydrogen taking part in a hydrogen bond. For
example, caffeine (1b) binds very poorly to a molecu-
larly imprinted material prepared against theophyl-
line (1a), only differing in the exchange of a hydrogen
for a methyl group, and the exchange of a hydroxyl
group for a keto group in cortisol/cortisone (2a/2b)
results in a remarkable loss in binding (Figure 5).
More elaborate systems, which can recognize
certain motives in a given ligand, can however be
designed. For example (Figure 6), the barbiturate
structure (3), displaying two acceptor}donor}acceptor
hydrogen bonding motifs, can be very selectively rec-
ognized in solution by a corresponding amidopyridine
(4) displaying a donor}acceptor}donor counterpart.
In comparison with monovalent functional mono-
 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
Figure 5 Examples of high chemoselectivity in noncovalent systems. Caffeine (1b) and theophylline (1a) differ only in one methyl
group. Nevertheless, caffeine does not bind to matrices prepared against theophylline. Similarly, cortisone (2b) binds poorly to an
anticortisol (2a) polymer.
mers, a high chemoselectivity can be designed for
a chosen structure prior to the imprinting process. On
the other hand, if a carefully designed multipoint
system does not allow for dynamic adjustment, not
much is gained from the imprinting process. In this
example, the general structure (3) is well recognized,
irrespective of R and R
.
In all these cases, however, the chemoselectivity
that is more or less rationally designed upon exam-
ination of the target print species can be radically
enhanced by the imprinting process. This is
particularly the case when dealing with less speciRc
noncovalent interactions. The positioning of the
functional groups in the three-dimensional network
of the matrix, together with the formation of the site
or imprint, often results in a tremendous increase in
speciRcity.
Regioselectivity Regioselectivity can often be nicely
demonstrated by molecularly imprinted materials us-
ing all types of imprinting protocols. In this case, the
imprinting process per se, rather than the bond type,
endows high regioselectivity. It is mainly due to
proper three-dimensional (or two-dimensional) posi-
tioning of the functional building blocks in the Rnally
produced matrix that high selectivity can be accomp-
lished. Sometimes, however, the selectivity is also
Figure 6 Multiple hydrogen bonding between cyclobarbital (3)
and bis-amidopyridines (4).
4109
a consequence of the architecture of the surrounding
backbone of the imprinted matrix in addition to
organized point interactions. Obviously, this works
concertedly, such that the build-up of the matrix
backbone may force the originally noninteracting
parts of the functional building blocks to accommo-
date a position, resulting in steric hindrance of differ-
ent analytes.
A number of systems have been designed where
bis-functional compounds displaying a difference in
distance between the point interactions can be efR-
ciently distinguished by the produced matrix. For
example, a range of either bis-aldehydes (5a/5b) or
diketones (6a/6b) could be selectively discriminated
by matrices produced using imine- and acetal-forma-
tions, respectively (Figure 7).
Another example is the imprinting of bis-imidazole
structures using Cu(II) coordination. A selectivity
between the structures could be recorded, largely
dependent upon the positioning of the Cu ions in the
material. In these examples, the distance between the
point interactions can be considered as being of major
importance for recognition.
Similar studies have been pursued using non-
covalent interactions, e.g. the interactions of bis-py-
ridyl and bis-aniline compounds with carboxylic
acids, which show high distance selectivities also.
Although investigations of such bis-compounds
can be considered as interesting model studies, prov-
ing the utility of the imprinting concept, many exam-
ples of compounds of more practical interest, e.g.
food additives or drugs, have been examined. Ster-
oids, for instance, are a class of compounds that lends
itself to being both interesting model systems as well
as commercially important, and several impressive
steroid separations have been demonstrated. For
example, -11-OH-progesterone (7a) can be most
selectively separated from -17-OH-progesterone
(7b) (Figure 8), in part due to the reasonably rigid
steroid framework, which helps to lower the entropy
loss upon steroid binding compared to a more Sexible
molecule.
4110 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION
Figure 7 Distance selectivity in reversible covalent systems. Bis-aldehydes 5a/5b, and bis-ketones 6a/6b could be distinguished by
matrices prepared against one of the structures using imine and ketal formation, respectively.
Stereoselectivity The physical properties of enantio-
mers, i.e. identical properties in a symmetrical envi-
ronment, render them ideal substrates for the study of
imprinting speciRcity. This is one of the reasons why
this aspect has been the most extensively studied in
MIT in general, and in MIC in particular. Indeed, this
quality offers the clearest evidence for the outcome of
a successful imprinting process. Obviously, if chiral
discrimination can be introduced in a matrix through
an imprinting process, based upon exclusively achiral
building blocks, an indisputable demonstration of the
concept is obtained.
A large number of investigations concerning chiral
separations by molecularly imprinted materials have
been performed, and hardly any type of compound
has been neglected from examination. Thus, com-
pounds such as amino acids, carbohydrates, nucleic
acids and pharmaceuticals have all been applied in
various imprinting protocols. The stereoselectivity
that can be achieved is also quite extraordinary, and
differences in binding of single hydroxyl groups (R-
and S-timolol), and even single methyl groups can
sometimes be observed (R- and S-naproxen). The
interested reader can Rnd a more detailed overview of
chiral separations using molecularly imprinted ma-
terials elsewhere in this Encyclopedia (Kempe M,
Chiral Separation: Molecular Imprints as Stationary
Phases).
Conclusions and Future Prospects
MIT is a technique that has substantially improved
over the last few years. Many of the drawbacks ini-
tially encountered have gradually been overcome,
and intense research is ongoing to resolve some of the
remaining challenges. As has been pointed out, the
technique is already competent at recognizing very
small structural differences, and the ultimate venture
that prevails in this respect is the separation of iso-
topes. Another task that needs further attention, al-
though some steps have already been taken in this
direction, is the selective recognition/separation of
biological macromolecules, and even whole cells.
For afRnity separation, however, high afRnities
may not always be the ultimate goal. High afRnities
sometimes lead to situations where elution from the
afRnity matrix requires harsh condition. For this rea-
son, a material providing a lower afRnity can some-
times be advantageous, and the process of
binding/elution can be accomplished using mild con-
ditions. Likewise, a high selectivity is not always
a goal in itself, and materials with lower selectivity
 III / SELECTIVITY OF IMPRINTED POLYMERS: AFFINITY SEPARATION 4111

Figure 8 Examples of the high regioselectivity that can be achieved by matrices produced against steroids. -11-OH-progesterone
(7a) and -17-OH-progesterone (7b) were efficiently separated by a molecularly imprinted chromatographic stationary phase. The
androstene-triol (8) could be selectively acetylated in the 11-position upon protection/binding to the matrix.

can also be preferable. This is the case when perform-
ing separations of groups of molecules rather than
a single species, such as the recognition of the -
lactam group as a whole, rather than individual peni-
cillins. Using materials that are apt at recognizing
a molecular chord, rather than single molecular
notes, can sometimes be beneRcial.
See also: I/Affinity Separation. II/Affinity Separation:
Immunoaffinity Chromatography; Imprint Polymers.
III/Chiral Separations: Molecular Imprints as Stationary
Phases. Immunoaffinity Extraction. Molecular
Imprints for Solid-Phase Extraction. Selectivity of
Imprinted Polymers: Affinity Separation.
Further Reading
Andersson LI (1998) Molecular imprinting as an aid to
drug bioanalysis. In: Reid E, Hill H and Wilson I (eds)
Drug Development Assay Approaches, Including Mo-
lecular Imprinting and Biomarkers, pp. 2}12. Royal
Society of Chemistry.
Bartsch RA and Maeda M (eds) (1998) Molecular and ionic
recognition with imprinted polymers. ACS Symposium
Series 703.
Mosbach K (1994) Molecular imprinting. Trends in Bio-
chemical Science 19: 9}14.
RamstroK m O and Ansell RJ (1998) Molecular imprinting
technology: challenges and prospects for the future.
Chirality 10: 195}209.
Remcho VT and Tan ZJ (1999) MIPs as chromatographic
stationary phases for molecular recognition. Analytical
Chemistry 71: A248}A255.
Sellergren B (1997) Noncovalent molecular imprinting:
antibody-like molecular recognition in polymeric net-
work materials. Trends in Analytical Chemistry 16:
310}320.
Shea KJ (1994) Molecular imprinting of synthetic network
polymers: the de novo synthesis of macromolecular
binding and catalytic sites. Trends in Polymer Science 2:
166}173.
Vidyasankar S and Arnold FH (1995) Molecular im-
printing: selective materials for separations, sensors and
catalysis. Current Opinion in Biotechnology 6:
218}224.
Whitcombe MJ, Alexander C and Vulfson EN (1997)
Smart polymers for the food industry. Trends in Food
Science and Technology 8: 140}145.
Wulff G (1995) Molecular imprinting in cross-linked ma-
terials with the aid of molecular templates } a way
towards artiRcial antibodies. Angew. Chem. Int. Ed.
Engl. 34: 1812.
Yano K and Karube I (1999) Molecularly imprinted poly-
mers for biosensor applications. Trends in Analytical
Chemistry 18: 199}204.
4112 III / SILVER ION / Liquid Chromatography
SILVER ION
Liquid Chromatography
W. W. Christie, Scottish Crop Research Institute,
Dundee, Scotland
Copyright ^ 2000 Academic Press
The Nature of the Technique
The technique of silver ion chromatography, which is
sometimes termed argentation chromatography,
was Rrst introduced for the separation of fatty acid
derivatives in two papers which appeared almost
simultaneously in Chemistry and Industry in 1962. In
its earliest manifestations, it was adapted to thin-
layer chromatography (TLC) and to low pressure
column chromatography. The former is still in wide-
spread use today, but high performance liquid
chromatography (HPLC) has more recently been
adapted to the purpose and may be expected to slow-
ly supplant TLC. Gas chromatography with station-
ary phases containing silver ions has been used to
separate mixtures of hydrocarbons containing low
molecular weight oleRns, but is not practical for
analytes of higher molecular weight. While lipid
chemists, especially those concerned with fatty acid
derivatives, have made most use of silver ion
chromatography, it has also been used for a wide
range of aliphatic and alicyclic compounds, including
terpenes, sterols, carotenoids, insect pheromones, etc.
However, applications only to separation of more
conventional lipids will be described here as examples
of what can be accomplished. A number of review
articles have appeared, that by Morris (1966) cover-
ing the early literature and those by Nikolova-
Damyanova and colleagues (Nikolova-Damyanova,
1992; Dobson et al., 1995) bringing the topic up to
date. Details of practical methods are given in a book
by Christie (1989).
Silver ion chromatography is based on a distinctive
property of unsaturated organic compounds, that is
the capacity to complex with transition metals in
general, and with silver ions in particular. The com-
plexes are of the charge-transfer type in which the
unsaturated compound acts as an electron donor and
the silver ion as an electron acceptor. In the accepted
model, it is believed that there is formation of a

-type bond between the occupied 2p orbitals of an
oleRnic double bond and the free 5s and 5p orbitals of
the silver ion, and a (probably weaker)
-acceptor
backbond between the occupied 4d orbitals of the
silver ion and the free antibonding 2p * orbitals of
the oleRnic bond. In chromatographic systems, com-
plexes are only formed transiently and are in kinetic
equilibrium with the free oleRn. The coordination
forces are weak, and IR spectra show very little shift
in frequencies from those of free double bonds, for
example. Stable silver ion}oleRn complexes have
been isolated in some circumstances, and then X-ray
crystallography has demonstrated that each silver ion
can interact with two double bonds simultaneously.
Other metals can form such complexes but none
other than silver has the correct combination of prop-
erties for general chromatographic use.
Until recently, the technique was very much an
ad hoc one that worked, but without a sound theoret-
ical basis. Silver ion TLC cannot be used to generate
reproducible chromatographic data, for example.
However, some useful qualitative data are available
from studies with simple model compounds and in
particular it is apparent that:
E Unsaturated aliphatic and alicyclic compounds
form more stable complexes than do aromatics.
E The stability decreases with increasing chain length
of the aliphatic substrate.
E The stability decreases with increasing numbers of
substituents of a double bond in the order
RCH"CH2'R2C"CH2' cis RCH"CHR'
trans RCH"CHR'R2C"CHR'R2C"CR2.
The greater stability of the cis-isomer may be due
either to the relief of strain when the complex is
formed or to steric hindrance by the two alkyl moie-
ties when they are in a transposition to each other.
E Conjugated polyenes form less stable complexes
than do those with methylene-interrupted double
bonds, and the greatest stability is when two
methylene groups separate double bonds, perhaps
because a chelate complex can then be formed.
E The stability of a silver ion complex increases when
a hydrogen atom from a molecule of the
RCH"CHR type, for example, is replaced with
deuterium or tritium.
E Monoenes form stronger complexes than mono-
ynes (one acetylenic bond).
E The strength of complexation increases as the tem-
perature is lowered.
As far as has been ascertained, these rules also
apply to larger molecules such as simply fatty acid

derivatives, and to all forms of silver ion chromato-
graphy. However, there are few quantitative experi-
mental or theoretical data on the mechanism of
complex formation between silver ions and compli-
cated unsaturated molecules like glycerolipids, which
have up to three unsaturated fatty acids per molecule,
although again, there is a substantial body of quali-
tative information.
A complicating factor, when considering silver ion
complexation in the context of a chromatographic
system, is the role of the support material. The most
widely used support for TLC, silica gel, possesses
appreciable polarity and absorptive activity. There-
fore, the elution order of lipids cannot always be
ascribed to the complexation reaction with silver ions
and double bonds only, although this is usually the
most important factor. A separate but related prob-
lem is the topology of silver ions on the surface of the
adsorbent. For example, in silver ion TLC it is appar-
ent that part of the silver nitrate remained in crystal-
line form Rlling the pores of the silica gel while
a further proportion remains dissolved in the water
which is always bound to silica gel. The aqueous
silver nitrate is assumed to be responsible for complex
formation, and some experience seems to conRrm this
observation. For example, Nikolova-Damyanova
considers that there is no need for incorporation of
excessive amounts of silver nitrate ('1}2%) into
TLC systems.
When discussing mixed retention mechanisms in
chromatography, it is also necessary to consider the
mobile phase. A proper choice of solvents determines
the selectivity of a separation to an appreciable ex-
tent. Again, there are no systematic data available,
but it has often been noted that better resolution is
achieved by using chlorinated solvents as major
components of the mobile phase in silver ion
chromatography.
The more recent marriage of silver ion chromato-
graphy with HPLC has given us a better understand-
ing of the mechanism of silver ion chromatography
and this is discussed below.
Silver Ion Thin-Layer Chromatography
Silver ion TLC uses simple equipment and can afford
excellent results in practice. Precoated TLC plates are
available commercially, although it is not difRcult to
prepare ones own (but wear gloves!). Thus, silver
nitrate is simply incorporated into the aqueous slurry
used to suspend the silica gel and the plates are spread
and activated in the usual way, though some care is
necessary to minimize exposure to light. Sometimes
10}20% of silver nitrate relative to silica gel is recom-
mended by authors, but 1}2% is generally sufRcient.
III / SILVER ION / Liquid Chromatography 4113
Alternatively, plates can be impregnated with silver
nitrate by careful immersion in a bath of a solution of
silver nitrate in methanol or acetonitrile, and this
option is often favoured with precoated TLC plates.
After the plates have been activated, they should be
stored in a desiccator in the dark.
Lipids are spotted on to the TLC plate and this is
usually developed in a closed tank (in the dark) con-
taining an appropriate mobile phase. However,
Nikolova-Damyanova recommends using open tanks
and carefully regulated volumes of solvent.
Chromatography is most often carried out at ambient
temperature, although temperatures as low as
!203C have on occasion been recommended to in-
crease the strength of complexation and improve the
separation. When the mobile phase nears the top of
the plate, the latter is removed from the tank, and
dried in a stream of air or nitrogen.
Various methods of detection and quantiRcation
are available. For example, one popular method con-
sists in spraying the plate with concentrated sulfuric
acid, and heating it at 1803C in an oven. The separ-
ated components are charred (converted to carbon)
and can be quantiRed directly by scanning den-
sitometry. A procedure of this kind is of course de-
structive to the sample, and has to be carried out with
great care to avoid hazard to the operator.
Alternatively, the developed plate can be sprayed
with a solution of 2,7-dichloroSuorescein in meth-
anol (0.1% w/v). After evaporation of the solvent, the
plate is viewed under a UV lamp; lipids appear as
yellow spots against a dark purple background. The
lipid/silica gel spots are scraped from the plate, and
lipids are recovered by extraction with an appropriate
solvent, though the extracts may have to be washed
with a solution of dilute buffer (pH 9) to eliminate
any silver nitrate and dye that co-elute. Commonly
components are identiRed and quantiRed by gas
chromatography of the fatty acid methyl esters, fol-
lowing transesteriRcation, in the presence of an added
internal standard, such as an odd-chain fatty acid.
As an example, Figure 1 illustrates the separation
of fatty acid methyl esters by silver ion TLC.
Saturated fatty acids do not form complexes with
silver ions, so migrate ahead of the other components
on the plate. Trans-monoenes form less stable
complexes than cis-monoenes, so the former migrate
faster. Dienoic fatty acids come next followed by
polyenes, which under these conditions remain near
the origin. A separation of this kind is the standard
method for reliable quantiRcation of fatty acids with
trans double bonds. If the polarity of the mobile
phase is increased, polyenoic fatty acid derivatives
can be resolved into fractions with three, four, Rve
and six double bonds, but saturated, and mono- and
4114 III / SILVER ION / Liquid Chromatography

benches, equipment and the Rngers of the analysts.

HPLC methods do not suffer from these difRculties.

Preparative Scale Column

Chromatography
By analogy with silver ion TLC it has proved possible
to impregnate silica adsorbents (or better, acid-
washed FlorisilTM) with silver nitrate and pack into
columns to enable separation of fatty acids by degree
of unsaturation. However, the technique suffers from
many of the problems associated with silver ion TLC.
As an alternative, macroreticular sulfonic acid ion
exchange resins have been utilized as adsorbents for
silver ion column chromatography. The resin is
loaded with silver ions by passing an aqueous solu-
tion of silver nitrate through a column of resin until
excess silver ions start to elute. The column is then
washed with water and methanol, and methanol is
used further as the mobile phase. Recently, Amberlyst
XE 284TM has been shown to give the best results, but
the separations improved signiRcantly only when the
mobile phase of methanol was modiRed with

Figure 1 Silver ion TLC (Kieselgel GTM containing 2% silver

nitrate) of fatty acid methyl esters, with hexane/diethyl ether (9 : 1,
v/v) as mobile phase.

dienoic fatty acids will then run together near the

solvent front.
Figure 2 is a schematic representation of a silver
ion TLC separation of triacylglycerols (which can
include all the common oils and fats of commerce),
which contain three fatty acids per molecule. In this
instance, trisaturated species elute Rrst followed by
disaturated monoenoic; the latter separating into two
fractions (more or less completely) according to
whether the unsaturated fatty acid is on position 2 of
the glycerol moiety or in one of the outer positions.
Then, the monosaturated dimonoenoic species is
eluted followed by a fraction containing one dienoic
fatty acid with two saturated, i.e. a dienoic fatty
acid is retained more strongly than two monoenes,
and so on.
Phospholipid molecular species have been resolved
in this way also, both in intact form (technically
difRcult) or after enzymatic conversion to non-
polar diacylglycerol derivatives.
Although the equipment is simple and inexpensive,
there are many drawbacks to silver ion TLC proced-
ures, not least that silver ions are eluted from TLC
Figure 2 Silver ion TLC (Kieselgel GTM containing 10% silver
plates and contaminate fractions in preparative ap- nitrate) of triacylglycerols, with chloroform/methanol (99 : 1 v/v)
plications, as do silica gel and dyes used for detection as mobile phase. Abbreviations: S"saturated, M"cis-mono-
purposes. Silver nitrate leaves indelible stains on enoic and D"dienoic fatty acyl residues.

acetonitrile, which enabled a good resolution of
monoenes, dienes, trienes and tetraenes (DeJarlais
et al., 1983). Further improvements were obtained by
grinding the resins Rrst to 270}350 mesh.
This technique permits fractionation of mixtures of
unsaturated fatty acids by degree of unsaturation on
the 10 to 20 gram scale. As the silver ions are held by
ionic bonds, they are not leached from the column
and clean fractions are obtained. However, the range
of mobile phases that can be employed is limited,
otherwise swelling and compaction of the resin will
occur.
Silver Ion High Performance Liquid
Chromatography
The approach to silver ion HPLC adopted by Christie
(1987) is to load a silica-based ion exchange medium
(chemically bonded phenyl sulfonic acid groups) with
silver ions. Again, the silver ions are held by ionic
bonds and are not leached from the column, while
rigidity of the silica matrix prevents compaction
of the packing material during gradient elution.
Preparation of the column involves merely taking
a standard prepacked column with the appropriate
stationary phase (NucleosilTM 5SA) and introducing
the silver ions via an injector while pumping water
through the column. Finally, the aqueous phase
is replaced with organic solvents. Only 50 to
80 mg of silver ions are bound to the stationary
phase, but this is quite sufRcient for very many useful
separations.
Lipids lack chromophores that facilitate UV detec-
tion, although it is possible to use UV detection with
appropriate fatty acid derivatives. Therefore, for
much of the work with silver ion HPLC columns,
evaporative light-scattering detectors have been em-
ployed as they permit the use of complex gradients
and mobile phases containing such solvents as di-
chloromethane and acetone. Although the detector is
destructive in that the sample is lost in the current of
air, it is possible to insert a stream splitter between
the end of the HPLC column and the detector to
divert much of the sample for collection.
Chlorinated solvents, such as dichloromethane or
dichloroethane, form the basis of the more useful
mobile phases, and the polarity can be increased to
elute highly unsaturated components by adding
acetone or especially acetonitrile, which has a high
afRnity for silver ions. Presumably the high dielectric
constant of the chlorinated solvents facilitates
the interaction between silver ions and the double
bonds. However, acceptable results can also be
obtained with hexane as the main component of
the mobile phase if a little acetonitrile is present.
III / SILVER ION / Liquid Chromatography 4115
Methyl esters are the most widely used fatty acid
derivative for chromatography, because of the ease of
preparation and their relatively low molecular
weight. One useful application of silver ion HPLC is
the separation of such derivatives from animal or Rsh
lipids, into fractions with zero to six double bonds.
SimpliRcation of complex mixtures by this means
makes the task of identiRcation by other chromato-
graphic means or by mass spectrometry much easier.
However, by an appropriate choice of solvents, it is
possible to separate positional and geometrical
isomers of unsaturated fatty acids on a micro-
preparative scale, a feat not readily achieved by other
chromatographic procedures. For example, the separ-
ation obtained with phenacyl esters of the three main
naturally occurring octadecenoic acid isomers is illus-
trated in Figure 3; all are clearly resolved to baseline.
It has become evident that the distance of the double
bond from the carboxyl group is more important in
governing the separation of positional isomers than is
the terminal region of the molecule. Phenacyl deriva-
tives of fatty acids were prepared at Rrst so that
quantiRcation with UV detection was possible, but
fortuitously, it has now become apparent that such
derivatives give especially favourable separations
(and this is also true for silver ion TLC). This tech-
nique can be used with equal facility for the separ-
ation of simple fatty acid derivatives with trans
double bonds, and will undoubtedly supplant TLC
for the purpose. Positional isomers of polyunsatu-
rated fatty acid derivatives can be resolved similarly
by increasing the polarity of the mobile phase.
Silver ion HPLC is also of great value for separ-
ation of molecular species of triacylglycerols.
The simplest elution scheme is a gradient of
acetone into dichloroethane}dichloromethane, which
serves for fats with low levels of linoleic acid, such
as sheep adipose tissue or bovine milk fat. This
Figure 3 Separation of phenacyl esters of the isomeric oc-
tadecenoic acids, petroselinic (6}18 : 1), oleic (9}18 : 1) and vac-
cenic (11}18 : 1), by HPLC on a NucleosilTM 5SA column in the
silver ion form eluted with dichloromethane/dichloro-
ethane/acetonitrile (50 : 50 : 0.25 by volume), with evaporative
light-scattering detection.
4116 III / SILVER ION / Liquid Chromatography
gives resolution of the main components with zero to
three double bonds in total in the fatty acyl chains,
including separation of fractions with trans- from
those with cis-monoenoic residues. Most triacyl-
glycerol samples are likely to contain appreciable
proportions of linoleic acid, and resolution into mo-
lecular species is then accomplished by ternary gradi-
ent elution, simply by introducing acetonitrile into
acetone after the Rrst fractions are recovered. Such
a separation of adipose tissue triacylglycerols is illus-
trated in Figure 4. One dienoic acyl moiety is retained
more than twice as strongly as a monoene, and one
triene (18 : 3 (n-3)) is retained by the same amount as
two dienoic residues in a molecule, so there is some
overlap of dienoic and trienoic species when -lin-
olenic acid is present in a sample. Otherwise, the basis
of the separation is similar to that described earlier
for TLC applications, in that trisaturated species elute
Rrst, followed by disaturated monoenoic and so forth.
Such highly unsaturated triacylglycerols as linseed
oil and Rsh oils have been resolved satisfactorily.
With the former, trilinolenin is the most abundant
single fraction, and a simple progression of fractions
with increasing numbers of double bonds are eluted
until this species is reached. When the more saturated
molecules of Rsh oils are eluted, resolution is excellent
and it is perhaps surprising to Rnd appreciable
amounts of trisaturated and disaturated monoene
species. Baseline resolution is no longer possible
when molecules containing polyunsaturated fatty
acids begin to elute, because the wide range of posi-
tional isomers causes similar components to overlap.
Nevertheless, valuable separations of species contain-
Figure 4 Separation of triacylglycerols from rat adipose tissue
by HPLC on a NucleosilTM 5SA column in the silver ion form, with
evaporative light-scattering detection. Abbreviations: S,
saturated, M, monoenoic; D, dienoic; T, trienoic fatty acyl resi-
dues. (Adapted from Christie, 1988.)
ing two saturated and/or monoenoic fatty acids and
one polyenoic fatty acid especially can be achieved.
In silver ion chromatography, the order of elution
of triacylglycerol species is easily understood because
only one property of the molecules is involved, i.e.
degree of unsaturation. The alternative technique
used for molecular species separations is reversed-
phase chromatography, with octadecylsilyl phases,
which effects separation both by chain length and
degree of unsaturation, each double bond reducing
the effective chain length by the equivalent of about
two methylene groups. When used in sequence the
two techniques make a much more powerful tool.
Fish oils give highly complex chromatograms with
reversed-phase HPLC, for example, and identiRca-
tion of individual components is impossible. On the
other hand, when fractions from silver ion HPLC are
collected and then subjected to reversed-phase HPLC,
separation is then, in effect, by chain length only and
the main peaks are easily identiRed. Each HPLC frac-
tion can be examined in turn in this way, and much
more information obtained in comparison to the use
of either technique on its own.
Some Mechanistic Considerations
In silver ion HPLC, we have a reasonable understand-
ing of how the silver ions are bound to the stationary
phase via the phenylsulfonic acid groups. There may
be some residual silanol groups on the surface of the
silica matrix, but these should not inSuence separ-
ations greatly when relatively polar chlorinated sol-
vents are used in the mobile phase. Also in HPLC, we
can control both the composition and Sow rate of the
mobile phase with a high degree of accuracy. Finally,
we can control the temperature of the column, an
important factor in the complexation reaction between
silver ions and double bonds. Accurate chromato-
graphic retention data can thereby be obtained for
a variety of lipid analytes of known structure.
It is known that silver ions can interact with two
double bonds in a fatty acyl residue at the same time,
but can they also react with one double bond and the
unpaired electrons on the carboxyl moiety as shown
schematically in Figure 5? This might explain how dif-
ferent positional isomers of fatty acids are separable by
this technique. For example, electron-rich esters, such as
the phenacyl derivatives illustrated, are held much more
strongly than are methyl esters when the double bond is
within about eight carbons of the carboxyl group, and
the elution patterns of series of isomers are very
different. From a 9}18 : 1 fatty acid derivative on-
wards, when the possibility of such a simultaneous
interaction would seem to be less likely, there is no
signiRcant difference between methyl and phenacyl
 III / SILVER ION / Thin-Layer (Planar) Chromatography
Figure 5 Schematic representation of the interaction of a silver
ion with the phenacyl ester derivative of petroselinic acid.
esters. Experiments with esters with a variety of differ-
ent electron-donating and electron-withdrawing sub-
stituents now provide Rrm evidence for this hypothesis.
An interaction between one silver ion and two
double bonds at the same time may explain the
chromatographic behaviour of fatty derivatives with
two or more double bonds in the acyl chain in silver
ion HPLC. When the distance between the double
bonds is optimum, i.e. with a 1,5-cis,cis-diene system,
fatty acids are very strongly retained, and the effect
diminishes as the number of methylene groups be-
tween the double bonds is varied. If the double bonds
interacted singly with silver ions, it might have been
anticipated that the kinetics of the system would be
such that retention would be comparable in magni-
tude to the sum of the individual parts, but this is
clearly not so. This theory of complexation between
silver ions and bis-double bond systems could poten-
tially be applied to polyenoic fatty acid derivatives. It
would predict that a triene would be held twice as
strongly as a diene, a tetraene three times as strongly
and so forth. Such a simple relationship is not found
in practice (the degree of complex formation is even
greater than anticipated), possibly because interac-
tions with the ester moiety have to be taken into
consideration and because the conformations of poly-
enes may permit some interactions between silver
ions and double bonds that are remote from each
other, via the formation of pseudo-cyclic structures.
Analogous physicochemical studies of the behaviour
of triacylglycerols on silver ion chromatography sug-
gests that a dual interaction is important in this in-
stance also. For example, it has been shown that highly
unsaturated triacylglycerols are retained especially
Thin-Layer (Planar) Chromatography
B. Nikolova-Damyanova, Bulgarian Academy of
Sciences, Sofia, Bulgaria
Copyright ^ 2000 Academic Press
4117
strongly; a species with nine double bonds is
held 10 000 times as strongly as one with a single double
bond. It is the strength of this interaction rather than
the efRciency of the column per se that is responsible for
the quality of the separations. However, here further work
is certainly necessary to conRrm the mechanism.
See also: III/Lipids: Liquid Chromatography; Thin-Layer
(Planar) Chromatography. Oils, Fats and Waxes: Super-
critical Fluid Chromatography. Silver Ion: Thin-Layer
(Planar) Chromatography.
Further Reading
Christie WW (1987) A stable silver-loaded column for the
separation of lipids by high-performance liquid
chromatography. Journal of High Resolution
Chromatography; Chromatography Communications
10: 148.
Christie WW (1988) Separation of molecular species of
triacylglycerols by high-performance liquid chromatog-
raphy with a silver ion column. Journal of Chromatog-
raphy 454: 273.
Christie WW (1989) Gas Chromatography and Lipids.
A Practical Guide. Dundee: The Oily Press.
DeJarlais WJ, Adlof RO and Emken EA (1983) Acetonitrile
as eluent in silver resin chromatography. Journal of the
American Oil Chemists Society 60: 975.
Dobson G, Christie WW and Nikolova-Damyanova
B (1995) Silver ion chromatography of lipids and fatty
acids. Journal of Chromatography B 671: 197.
Laakso P and Christie WW (1991) Combination of silver
ion and reversed-phase high-performance liquid
chromatography in the fractionation of herring oil
triacylglycerols. Journal of the American Oil Chemists
Society 68: 213}223.
Morris LJ (1966) Separation of lipids by silver ion chro-
matography. Journal of Lipid Research 7: 717.
Nikolova-Damyanova B (1992) Silver ion chromatography
and lipids. In: Christie WW (ed.) Advances in Lipid Meth-
odology } One, pp. 181}237. Dundee: The Oily Press.
Nikolova-Damyanova B, Christie WW and HersloK f BG
(1995) Retention properties of triacylglycerols on silver
ion high-performance liquid chromatography. Journal
of Chromatography A 694: 375.
Nikolova-Damyanova B, Christie WW and HersloK f BG
(1996) Mechanistic aspects of fatty acid retention in
silver ion chromatography. Journal of Chromatography
A 749: 47.
Introduction
Silver ions (Ag#), like the ions of many
other transition metals, interact speciRcally with
4118 III / SILVER ION / Thin-Layer (Planar) Chromatography
Table 1 List of compounds subjected to Ag-TLC
Class Compounds
Lipids Fatty acid mixtures
Isomeric octadecenoic and octadecadienoic
fatty acids
Furanoid fatty acids
Cyclopropene fatty acids
Oxygenated unsaturated acids
Triacylglycerol mixtures of natural fats and
oils of terrestrial and marine origin
Phospholipids
Sterol esters
Terpenes C10, C15, C20 terpenes
Terpene alcohols
Monoterpenes
Hydrocarbons Acyclic olefines
Alkylbenzenes
Alkylphenylsulfides
Miscellaneous Steroids
Sterols
Sterol acetates
Derivatized unsaturated aldehydes
and ketones
Prostaglandins
Hydroxyprogesterones
Estrogens
Mineral oils
unsaturated compounds to form complexes with
oleRnic double bonds. In 1938 Winstein and Lukas
and in 1952 Nichols demonstrated that the interac-
tion between Ag# and double bonds might be of
interest for chemical analysis. By applying
a liquid}liquid distribution system with Ag# present
in the aqueous phase it was possible to separate easily
unsaturated from saturated compounds and E- from
Z-monounsaturated oleRns. The great potential
of this interaction for separation of unsaturated
compounds was fully recognized when gas}
liquid chromatography (GLC) and thin-layer
chromatography (TLC) developed into routine
analytical techniques.
The chromatographic technique that utilizes the
interaction between Ag# and an oleRnic bond to
conduct the separation process is now called argenta-
tion (silver ion) chromatography. Argentation
chromatography was Rrst developed as a GLC tech-
nique. However, argentation TLC (Ag-TLC) soon
became a basic separation method for the analysis of
different types of unsaturated compounds. For many
years it has been a most valuable method in lipid
analysis, providing essential information about the
lipid structure and composition.
Compounds that have been most frequently exam-
ined by Ag-TLC are listed in Table 1.
Silver Ion Complexation with
Double Bonds
The model now considered to represent correctly the
bonding between Ag# and a double bond was sug-
gested by Dewar in 1951. It supposes the formation
of a
-type bond between the occupied 2p orbitals
of an oleRnic double bond and the free 5s and
5p orbitals of the Ag#, and a (probably weaker)
-acceptor backbond between the occupied 4d or-
bitals of the silver ion and the free antibonding 2pH
orbitals of the oleRnic bond (Figure 1).
It is suggested that a silver ion interacts with one
mono-oleRn molecule to give a planar complex with a
triangular structure. However, there is evidence that
a silver ion may interact with two ethylenic mol-
ecules. X-ray studies of crystalline silver ion com-
plexes with some short chain aliphatic dioleRns show
that the Ag# is coordinated with two double bonds
from different oleRnic molecules.
The stability of the silver ion}double bond complex
is inSuenced by the spatial arrangement of the over-
lapping orbitals, the basicity of and electronic effects
in the oleRnic molecule, and by solvent effects.
Quantitative data (for example equilibrium con-
stants) exist only for a number of short chain mono-
oleRns, dioleRns with accumulated, conjugated and
separated methylene-interrupted double bonds, and
for some cyclopenta- and cyclooctadienes. Most of
these data, as well as the estimation of the relative
strength of other complexes, are based on chromato-
graphic measurements.
The general conclusions about complex formation
reached so far are as follows:
1. Unsaturated acyclic and carbocyclic com-
pounds form more stable complexes than do
aromatics.
2. Carbocyclic compounds with a single exocyclic
double bond form stronger complexes than do
carbocyclic compounds with a single internal
double bond. Cyclopenta- and cyclooctadienes
Figure 1 Schematic presentation of the interaction between
a silver ion and a double bond.
 III / SILVER ION / Thin-Layer (Planar) Chromatography 4119

form very stable complexes, especially when hav-
ing a 1,5-diene system.
3. For acyclic compounds it has been found that:
} the stability decreases with the increasing chain
length;
} the stability decreases with an increasing number
of substituents at the double bond in the
order RCH"CH2'R2C"CH2'R2C"CHR'
R2C"CR2;
} the stability increases when a hydrogen atom
from a molecule of the RCH"CHR type is re-
placed with deuterium or tritium atom.
The effect has been ascribed to greater electron
release from a C}D than from a C}H bond, i.e. to
the higher basicity of the deuterated molecule;
} cis (Z)-isomers form stronger complex than do
trans (E)-isomers. The greater stability of the
cis-isomer is ascribed either to the relief of strain
when the complex is formed or, more probably,
to the steric hindrance of the double bond in
trans-isomers;
} dioleRns form more stable complexes than do
mono-oleRns, but conjugated dienes form less
stable complexes than do those with methylene-
interrupted double bonds;
} the stability increases with increasing distance
between the double bonds. The most stable diene
complexes are formed by the 1,5-diene system,
perhaps because a chelate complex can be formed
in the latter case.
Generally, the rate of complex formation for
most acyclic compounds is very rapid. The complexes
are usually unstable and exist in equilibrium with the
free form of the oleRn. The coordination forces seem
to be very weak. The IR spectrum, for example,
shows very little shift in frequencies from those of free
double bonds. These particular properties of com-
plexation between a double bond and a silver ion are
favourable for use in chromatography.
The Technique of Ag-TLC
The Layer
Argentation TLC is utilized in two modiRcations:
analytical and preparative. Both home-made and
precoated plates (glass only) are used. Plate dimen-
sions may vary from 5 cm;20 cm to 20 cm;20 cm.
High performance TLC plates (HPTLC) have re-
cently been found to be useful. Home-made layers,
despite their messy preparation, are more versatile
and often provide better separations than commer-
cially prepared ones. Common adsorbents are listed
in Table 2.
The thickness of the adsorbent layer ranges from
0.2 to 0.3 mm for analytical plates and from 0.5 to
1.0 mm for preparative plates. Fully automated
spreaders for home-made layers are commercially
available but simple spreaders are equally effective.
However, some practice is needed for layer prepara-
tion and precoated plates are now mostly preferred.
The Incorporation of Silver Ions
There are two general ways to perform silver ion TLC
of which by far the most common is to use a layer of
adsorbent impregnated with a silver salt. It is possible
as an alternative to add a silver salt to the mobile
phase when a reversed-phase TLC separation is per-
formed. This approach has found limited use only.
Although the inSuence of the salt anion has not been
studied systematically, there is evidence that its
nature may affect the resolution. A list of silver salts
used and their supposed effect is shown in Table 3.
Impregnation can be performed by incorporating
the silver salt into the slurry of adsorbent used to
make the layer. Also, the prepared plate can be im-
mersed in a methanol, acetone or acetonitrile solution
of the silver salt or sprayed with one of these solu-
tions. Only the Rrst approach affords proper control
of the silver content of the layer. However, it is
inconvenient and messy and is now rarely used. Since

Table 2 Adsorbents, frequently used with Ag-TLC

Adsorbent Compounds Comments

to be separated

Silica gel G (binder is
calcium sulfate)
Silica gel H (no binder)
Alumina
Fatty acids Fatty acids should be first converted into methyl esters.
Triacylglycerols Components are separated as intact compounds.
Diacylglycerols Compounds are separated after conversion into acetates.
Polar lipids Polar lipids should be converted into less polar derivatives, either
by removing or by derivatizing the polar head.
Terpenes Compounds are usually separated as intact compounds.
Polar lipids It is possible to separate polar lipids as intact compounds on this layer.
Fatty acids Species are separated as free fatty acids and as methyl esters.
Alumina as adsorbent should be used with care since it is known to react
with some analytes and solvents.
4120 III / SILVER ION / Thin-Layer (Planar) Chromatography

Table 3 List of silver salts used in Ag-TLC

Salts Components to be separated Comments

Silver nitrate Fatty acids derivatives The most frequently used silver salt with very
broad application for various separations.
Triacylglycerols
Polar lipids
Terpenes
Ammonia solution of silver nitrate Fatty acid methyl esters Silver salts other than silver nitrate are
supposed to improve the separation.
Silver sulfamate Fatty acid methyl esters
Silver benzenesulfonate Fatty acid methyl esters
Silver iodate Terpenes
Silver perchlorate Terpenes

it appears that the content of silver ions in the layer resolution of lipids, for example, without this
may not, in fact, be critical, immersion procedures are disadvantage.
mostly used. They can be standardized sufRciently
well to provide repeatable results. Spraying proced- Treatment of the Plates and Precautions
ures are less easily controlled. Spraying may have
Plates should be used immediately after being dried in
to be repeated from two to six times until
air (for 1 h). Many workers activate the plates before
the adsorbent layer is properly wetted. SufRcient
use by heating at 1103C for about 1 h. However,
spraying of the layer is somewhat arbitrary and de-
good results have been reported after activation for
pends on personal skill and on the samples to be
only 5 min. Thus, it seems that the necessity for ac-
examined. Immersion or spraying are only applicable
tivation is questionable, and the analyst must trust to
to precoated plates. These plates should be immersed
his or her own experience and the nature of the
in a silver salt solution (methanol or acetonitrile) for
samples that are being handled. Activation has been
not less than 15}20 min. A dynamic impregnation
found to be very important for silica gel H plates and
technique has also been proposed. The plate is de-
temperatures higher than 1103C for periods of much
veloped with a 10}20% solution of silver nitrate in
longer than 1 h have been recommended.
acetonitrile. It has been claimed that in this way
Atmospheric humidity has an appreciable effect on
a gradient of silver ions is formed in the development
separation, especially of highly unsaturated species. It
direction. The gradient is claimed to improve the
is recommended that activated TLC plates be kept in
separation of triacylglycerols. The approach has not
a desiccator over drying agents (ideally in the dark).
found wide application and its advantages cannot be
However, it is not easy to control humidity in prac-
estimated.
tice. This may be one of the reasons for the relatively
The silver content of the adsorbent layer varies
poor overall reproducibility of migration and resolu-
between rather broad limits and differs for analytical
tion in separations performed by Ag-TLC.
and preparative plates (Table 4). Layers contain-
ing a high percentage of silver nitrate were considered
Sample Preparation
necessary to achieve good analytical resolution.
This high percentage is, however, very inconvenient. Terpenes and triacylglycerols are applied as solutions
The plates are very sensitive to light and this of appropriate concentration in suitable solvents. Pre-
can greatly hamper detection and quantiRcation. Im- liminary fractionation and puriRcation from accom-
pregnation by immersion in 0.5% methanolic panying compounds is required. Fatty acids should be
silver nitrate provides excellent results in the converted into less polar derivatives, usually into
methyl esters. Recently, aromatic derivatives of fatty
Table 4 Concentration of the silver nitrate solutions and acids have proved to provide much better separation
methods for impregnation of a TLC layer of difRcult-to-resolve mixtures. Ag-TLC has not been
AgNO3 (%) Impregnation
very successful in separating intact complex lipids.
Most of the reported procedures employ preliminary
Analytical plates Preparative plates fractionation into speciRed classes and conversion of
the latter into less polar derivatives.
0.5}2 1}5 Immersion Sample size is an important factor since overload
5}30 5}20 Incorporation
10}40 40 Spraying
greatly worsens the resolution. Analytical Ag-TLC on
0.2 mm thick layers requires samples of a maximum
 III / SILVER ION / Thin-Layer (Planar) Chromatography 4121

of 30
g. For preparative separation (0.5}1.0 mm
layer thickness) the sample size can be scaled up to
80}100 mg, depending on the sample composition,
the quantitative ratio between components and the
required resolution.
Development
Most separation protocols recommend ascending de-
velopment in covered tanks in which the atmosphere
has been saturated with the vapour of the mobile
phase. The saturation is considered to shorten
the duration of development and often to improve the
reproducibility. There are no Rrm data to support this
conclusion and it might depend on the nature of the
analyte. Poor separation and tailing of zones have
also been reported under these conditions. Excellent
separations of fatty acids and triacylglycerols are ob-
tained without saturation of the atmosphere, or even
in an open container. The geometry and volume of
the developing container can also affect the separ-
ation. Narrow rectangular tanks and a moderate vol-
ume of the developing solvent provide better
resolution.
Ag-TLC plates are normally developed at ambient
temperature, but some improved resolution of fatty
acid isomers requires a temperature of !203C. It is
assumed that resolution improves because the stabil-
ity of the Ag# complexes increase when the temper-
ature decreases. This might be true, but the properties
of the sorbent and mobile phase also change at low
temperatures. The action of all three factors is prob-
ably responsible for the better separations.
Various solvents are used to give two or three
component mobile phases. Some of those most fre-
quently used for separation of fatty acids and triacyl-
glycerols are listed in Table 5. Plates are often de-
veloped more than once to improve resolution. The
separation should start with the most polar phase and
proceed, after drying between runs, with mobile
phases of gradually decreasing polarity. In this way,
highly unsaturated components are resolved Rrst and
do not move further with subsequent developments
when the more saturated components are separated.
Obviously, the separation will improve substantially
if a continuous development can be applied. In
a simple approach, which has been used for the analy-
sis of fatty acids and triacylglycerols, the development
proceeds in an open cylindrical tank where a Rxed
volume (4}15 mL) of the mobile phase has been
added. As the mobile phase is eluted through the
plate, it is permitted to evaporate from the upper
edge. Resolution was very effective and excellent re-
sults have been reported including the separation of
positional isomers of triacylglycerols. This open sys-
tem is quite sensitive towards the laboratory environ-
ment but operates very well in skilled hands.
Detection
All detection reagents that are suitable for visualizing
compounds separated on plain silica plates are, in
principle, suitable for use with Ag-TLC, particularly
when the percentage of silver ions in the layer is
below 2%. Destructive procedures are used for loca-
tion, for identiRcation and, to some extent, for quan-
tiRcation purposes. The reagent can be introduced by
spraying, by treatment of the plates with its vapours
or by incorporating into the layer. Some of the most
commonly used detecting reagents for lipids are listed
in Table 6. Solutions of chlorosulfonic acid in acetic
acid, ethanolic phosphomolibdic acid and antimony

Table 5 Examples of mobile phases used to separate fatty acids and triacylglycerols by Ag}TLC

Compound Mobile phase composition, by volume Development

Fatty acids with 0}3 double bonds
Fatty acids with 0}6 double bonds
Triacylglycerols with 0}6
double bonds
Hexane}diethyl ether, 90 : 10 or 80 : 20
Hexane}acetone, 100 : 4
First plate: hexane}diethyl ehter, 90 : 10
Second plate: hexane}diethyl
ether, 60 : 40
Hexane}diethyl ether}acetic
acid, 94 : 4 : 2
Benzene}ethyl acetate, 9 : 1
Benzene}diethyl ether, 85 : 15
Hexane}diethyl ether, 80 : 20
Chloroform}methanol, 96 : 4
Hexane}acetone, different proportions
One-fold development in closed tanks.
One-stage development in open cylindrical
tanks. E- and Z-monounsaturated fatty acids
are clearly separated.
Fatty acids are separated on two different
plates. Species with 0}3 double bonds are
separated on the first plate. Polyunsaturated
species are separated on the second plate.
All components are separated on a single
plate by one-stage development.
These mobile phases are used for both ana-
lytical and preparative separations in closed
tanks.
Development in open tanks with specified
volume of the mobile phase, see the example
in Figure 4.
4122 III / SILVER ION / Thin-Layer (Planar) Chromatography
Table 6 List of frequently used staining reagents for detecting lipids in Ag-TLC
Reagent
25}70% sulfuric acid
3% aqueous solution of copper acetate
in phosphomolybdic acid
Sulfuryl chloride
Rodamin 6G, 2,7-dichlorofluorescein
perchlorate in chloroform have been used to detect
terpenes separated by Ag-TLC.
Spraying is a rapid but inconvenient and quite
hazardous operation and should be avoided if and
when possible and replaced by treatment with va-
pours. Incorporation of the charring reagents into the
layer should be performed with circumspection, since
it may change the nature of the resolution.
For preparative purposes, after detection, the
separated zones are carefully scraped from the
plate and the compounds are extracted from
the sorbent with suitable polar solvents. As such
material is likely to be subjected to further analysis,
the extracts should be puriRed Rrst. For example,
in lipid analysis, excess silver ions and 2,7-dichlorof-
luorescein can be removed by passing the extract
through small silica columns or by washing
with bicarbonate, ammonia or sodium chloride
solutions.
Identi\cation
An advantage of Ag-TLC is the easy identiRcation of
the separated components with a substantial degree
of certainty. A reference compound, or reference mix-
ture of compounds, is usually applied beside the
sample. The reference and the sample are developed
simultaneously and this allows the migration dis-
tances (the RF values) to be compared. Fatty acids and
triacylglycerols form, for example, mixed zones with
the matching reference components. In case of ambi-
guity, preparative Ag-TLC is applied to isolate and
collect the component(s) in question for subsequent
spectral analysis.
Comments
A destructive reagent, usually applied by spraying as solution in
ethanol. Spraying must be very thorough and even, especially if the
plate is considered for densitometric quantification. Spots are
detected by heating the plate at temperatures of 150}2003C.
Advantageous in that precoated plates can be immersed in the
solution, thus providing an even staining. Spots are visualized by
heating.
A highly volatile destructive reagent that allows convenient treat-
ment of the plate in a closed container. This treatment provides
even staining when the plate is heated. Suitable for densitometric
quantification of lipids.
Nondestructive reagents used for preparative isolation of material
for further examination. Plates are sprayed with diluted ('1%)
solutions of the reagents in acetone. Spots are detected by viewing
under UV light. The isolated material is purified from the detecting
reagent by elution through a small silica gel column.
Quanti\cation
As in all TLC techniques indirect and direct ap-
proaches for quantiRcation have been employed.
The most widely used procedure involves scraping off
the detected zone and eluting the component(s) with
suitable solvent. Then any of the available chromato-
graphic or spectral techniques for quantiRcation can
be applied. Scanning densitometry can be used to
measure the quantities of the separated compounds
after a carefully chosen staining procedure. Staining is
required since even UV-absorbing or UV-tagged com-
pounds have either a very weak, or even no, signal in
the presence of Ag# in the layer. Procedures have
been reported for reliable densitometric quantiRca-
tion of fatty acids and triacylglycerols without the
need for calibration graphs and correction coefR-
cients, for example. Figure 2 presents the densitomet-
ric proRle of a triacylglycerol mixture separated by
Ag-TLC.
Interactions in Argentation TLC
The rules of complex formation, presented above,
have been found to be generally valid in the majority
of separations performed by Ag-TLC. In many cases
it is possible to predict the migration order. Interac-
tion of Ag# with compounds that have more than
two double bonds, however, seems to be less well
understood. This primarily concerns lipids. A mixture
of fatty acids, for example, may comprise compo-
nents of different chain length (from 12 up to 22
carbon atoms) and of zero to six double bonds in the
 III / SILVER ION / Thin-Layer (Planar) Chromatography
Figure 2 Densitogram of olive oil triacylglycerols separated by
Ag-TLC. Conditions: laboratory-made 5 cm;20 cm glass plate,
0.2 mm thick silica gel layer impregnated with 0.5% methanolic
silver nitrate (by dipping); sample size: 20
g; mobile phase: 6 mL
light petroleum}acetone}ethyl acetate, 100 : 3 : 2 (by
volume); one-stage development in an open cylindrical tank; de-
tection: successive treatment with bromine (30 min) and sulfuryl
chloride (30 min) vapours followed by carbonization by heating
at 180}2003C on a temperature-controlled hot plate; scann-
ing: Shimadzu CS-930 densitometer in zigzag mode; beam di-
mensions: 1.2 mm;1.2 mm; working wavelength: 450 nm.
Peak identity: Poly, minor components, containing octadeca-
trienoic fatty acid; the figures indicate the number of double
bonds in the acyl residues but not their position in the glycerol
backbone.
chain. Double bonds can be either cis (Z) or trans (E)
or both and can have different position in the
chain. Triacylglycerols and glycerophospholipids are
complex mixtures of species that differ in the type
and position of the acyl residues. So far, there are
no theoretical or experimental studies on the mecha-
nism of complex formation between Ag# and com-
plicated unsaturated molecules. Almost nothing
appears to be known about the electronic and steric
effects in these molecules and their possible inSuence
on complexation.
Attempts have been made to present the complexa-
tion of lipids in Ag-TLC in quantitative terms.
Estimation has been made on the basis of the
chromatographic retention of different molecular
types. For example, arbitrary values of 0, 1, 2#2a
and 4#4a, where a(1, have been proposed for the
complexing power of stearic, oleic, linoleic and lin-
olenic fatty acids (with 0, 1, 2 and 3 double bonds,
respectively). Evidently, the increase in the complex-
ing power values is greater than the increase in the
number of double bonds. More accurate equations
have been proposed but, in general, the values are
4123
close to those determined earlier and conRrm
the above conclusion. It has also been assumed that
the respective values for the tri- or diacylglycerols can
be expressed as a simple sum of the values for the
fatty acyl residues. The assumption considers that the
contribution of a fatty acid in a complex lipid mol-
ecule is not affected by the strength and properties of
the silver ion complexes with neighbouring fatty acids
in the same molecule. Such models are too simple and
do not take account of the steric factors that may
especially affect the complexation of a triacylglycerol
molecule.
The role of the support material must be also taken
into account. The most widely used support, silica
gel, possesses appreciable polarity and absorption
activity. Therefore, the retention of unsaturated com-
pounds cannot be ascribed to the complexation reac-
tion with Ag# and double bonds only, although
clearly that is a major factor. The retention of an
unsaturated molecule in any such system is the result
of a mixed retention mechanism. For example, an
unsaturated fatty acid methyl ester has been assumed
to complex with the silver ions through its double
bond(s) and to interact with the silanol moieties
through its methyl ester group. Depending on the
position of the double bond, the molecules will have
different conformations. Those molecules may be
held more strongly when the distance between the
double bond and the ester group has a better Rt with
the distance between the silver ion and the silanol
moiety. These two interactions have been suggested
in explanation of the speciRc migration patterns of
positionally isomeric fatty acid methyl esters.
A mixed retention mechanism should be taken into
account in the case of unsaturated compounds with
other polar functional groups which are subjected to
Ag-TLC.
A separate but related problem is the topology of
silver ions on the adsorbent surface. For example,
it was found that part of the silver nitrate remained in
crystalline form, Rlling the pores of the silica gel
after drying the plate to make the layer active.
However, a proportion of the silver nitrate remained
dissolved in the water, which is always bound to
silica gel. The aqueous silver nitrate was assumed
to be responsible for complex formation. Saturat-
ion of the silica layer with water before separation
has even been proposed in order to obtain a pure
complexation reaction. If the excess silver nitrate
does indeed remain in a crystalline form and does
not take part in complexation, impregnation of
the layer with a highly concentrated solution of
silver nitrate is of no practical value. Some of the
results obtained with Ag-TLC seem to conRrm this
observation.
4124 III / SILVER ION / Thin-Layer (Planar) Chromatography

Retention and Resolution

of Unsaturated Compounds
in Ag-TLC: Examples
In general terms, acyclic unsaturated compounds mi-
grate and can be resolved by Ag-TLC depending on
the number, conRguration and, occasionally, on the
position of the double bond in the molecule, the
number of the double bonds being the governing
feature. The separation of fatty acids and triacylgly-
cerols illustrates very well the resolution ability of
Ag-TLC.
The migration pattern of fatty acids with zero to
six double bonds is presented in Figure 3. For quali-
tative purposes it is possible to resolve fatty acids
with unsaturation in the above interval on a single
5 cm;20 cm plate using two solvent systems in se-
quence. Migration cannot always be predicted in the
case of a mixture of fatty acids of different chain
lengths. According to the general rules, when chain
length increases, stability of the complex with silver
ions decreases. This means that longer chain fatty
acids will migrate ahead of shorter chain species of
the same unsaturation (note the place of docosatet-
raenoic, 4a, and of octadecatetraenoic, 4b, fatty acids
on Figure 3). The migration order of triacylglycerols
follows the same rules, i.e. species with up to nine
double bonds and acyl residue chains of 16}18 car-
bon atoms are ordered according to the increasing
retention: 000, 001, 011, 002, 111, 012, 112, 003,
112, 013, 113, 222, 023, 123, 223, 133, 233, 333
(the Rgures indicate the number of double bonds in
the acyl moiety but not their position in the mol-
ecule). It should be noted that of two species with
an equal number of double bonds that in which all (or
most) of the double bonds are concentrated into one
fatty acyl moiety is held more Rrmly (011 and 002,
for example). The separation of natural triacyl-
glycerol mixtures depends strongly on the quantita-
tive proportions between the components. For
example, a mixture of components with relative high
saturation can be resolved on a single TLC plate by
the successive use of two mobile phases of decreasing
Figure 3 Migration pattern of fatty acid methyl esters with zero
polarity as illustrated in Figure 4. For satisfactory to six double bonds in Ag-TLC. Conditions: laboratory-made
resolution of plant triacylglycerols with up to nine 5 cm;20 cm glass plate, 0.2 mm thick silica gel layer impreg-
double bonds three different plates and two-stage nated with 0.5% methanolic silver nitrate (by dipping); mobile
development of each plate with mobile phases of phase: 5 mL light petroleum}acetone}formic acid, 97 : 2 : 1 (by
volume); one-stage development in an open cylindrical tank; de-
different polarity is required. The approach provides
tection: as in Figure 2. Spot identity: the figures indicate the
accurate identiRcation and densitometric quantiRca- number of double bonds; 4a, methyl docosatetraenoate; 4b,
tion of the separated species. methyl octadecatetraenoate.
E- and Z-isomers are normally easy to distinguish.
An example of the separation of a fatty acid mixture Under speciRed conditions Ag-TLC differentiates
is shown in Figure 5. Argentation TLC is may be the between mono- or diunsaturated fatty acids with dif-
easiest and cheapest way to determine trans-mono- ferent position of the double bond(s) in the carbon
enes in dietary fats. chain. In 1970 Gunstone and colleagues supposed
 III / SILVER ION / Thin-Layer (Planar) Chromatography 4125

was therefore of hardly any practical value until re-

cently, when it was shown that separation depends
strongly on the nature of the ester moiety. The effect
is demonstrated in Figure 6 and has been assigned
to the participation of the ester group and
the double bond in simultaneous complexation with a
silver ion to give a chelate-type complex.
Of practical value is the resolution of triacyl-
glycerols that differ by the position of the unsaturated
fatty acid residue in the triacylglycerol molecule
(Figure 7). Since the position of acyl residues is strict-
ly speciRc in natural triacylglycerol mixtures, Ag-
TLC provides an easy way to distinguish between
natural and modiRed edible fats and oils and the
approach is of value to the food industry.

Conclusion
Ag-TLC is a valuable qualitative and quantitative
method for the separation of unsaturated com-
pounds, lipids in particular. Ag-TLC has the
advantages of rapidity, simplicity and versatility
and does not require expensive instrumentation. The
information obtained reSects the whole sample, thus
helping the analyst to make rapid, correct and efRcient
judgements. It is widely used as a preliminary step
in combination with other chromatographic tech-
niques such as GC and HPLC. SufRciently pure

Figure 4 Migration pattern of coffee triacylglycerols. Condi-

tions: laboratory-made 5 cm;20 cm glass plate, 0.2 mm thick
silica gel layer impregnated with 0.5% methanolic silver nitrate (by
dipping); sample size 50
g; two-stage development with 4 mL
light petroleum}acetone, 100 : 4 (by volume) followed by 15 mL
light petroleum}acetone, 100 : 4 (by volume); detection: as in
Figure 2. Spot identity: the figures indicate the number of double
bonds in the acyl residues but not their position in the glycerol
backbone.
Figure 5 Migration pattern of E-, Z-fatty acid methyl esters.
Conditions: laboratory-made 5 cm;20 cm glass plate, 0.2 mm
that this is due to the participation of the fatty acid
thick silica gel layer impregnated with 0.5% methanolic silver
molecule in additional reaction(s) with either the sil- nitrate (by dipping); two-stage development with 2 mL light petro-
ver ions or with the adsorbent. Reliable resolution leum}acetone, 100 : 2 (by volume) followed by 3 mL light petro-
was achieved in isolated cases only. The approach leum}acetone, 100 : 0.7 (by volume); detection: as in Figure 2.
4126 III / SILVER ION / Thin-Layer (Planar) Chromatography
Figure 6 Migration pattern of mono-unsaturated octadecenoic
fatty acids differing in the position of the double bond in the carbon
chain. Sample: aniseed oil. Conditions: laboratory-made
5 cm;20 cm glass plate, 0.2 mm thick silica gel layer; detection:
as in Figure 2. (A) Separation of the fatty acid as methyl esters.
The layer is impregnated with 1% methanolic silver nitrate (by
dipping); development is in an open cylindrical tank with light
petroleum}acetone, 100 : 5 (by volume) at !203C. (B) Separ-
ation of the fatty acids as phenacyl esters by two-stage develop-
ment in a closed cylindrical tank (no preliminary saturation of the
atmosphere) with a mobile phase of chloroform}acetone,
100 : 0.25 (by volume) at ambient temperature. Note the im-
proved resolution achieved after conversion of the fatty acids into
phenacyl esters. Spot identity: S, saturated fatty acids, V, vac-
cenic acid, 11}18 : 1; O, oleic acid, 9}18 : 1; P, petroselinic acid,
6}18 : 1; D, linoleic acid, 9, 12}18 : 2 (position of double bond-
number of carbon atoms : number of double bonds).
components can be collected for further structural
elucidation by spectral methods. Combined
with scanning densitometry, Ag-TLC meets all re-
quirements of a reliable quantitative method for
determination of positional and conRgurational
fatty acid isomers and of triacylglycerols in natural
samples.
The resolution power of Ag-TLC will undoubtedly
increase if and when an automatically controlled
Figure 7 Migration pattern of positionally isomeric triacyl-
glycerols by Ag-TLC. Conditions: laboratory-made 5 cm;20 cm
glass plate, 0.2 mm thick silica gel layer. (A) Sample: randomized
lard; layer impregnated with 1% methanolic silver nitrate; one-
stage development in an open cylindrical tank with 12 mL chloro-
form}methanol, 95.5 : 0.5 (by volume). (B) Sample: randomized
sunflower oil; layer impregnated with 2% methanolic silver nitrate;
one-stage development in an open cylindrical tank with chloro-
form}methanol, 97.5 : 2.5 (by volume). Spot identity: the figures
indicate the number of the double bonds and the position of the
acyl residue in the glycerol backbone.
system for continuous development with a sequence
of different mobile phases becomes available.
See also: II / Chromatography: Thin-Layer (Planar): In-
strumentation; Layers; Modes of Development: Conven-
tional; Modes of Development: Forced Flow, Over-
pressured Layer Chromatography and Centrifugal; Spray
Reagents. III / Impregnation Techniques: Thin-Layer
(Planar) Chromatography. Lipids: Gas Chromatogra-
phy; Liquid Chromatography; Thin-Layer (Planar)
Chromatography. Silver Ion: Liquid Chromatography.
Further Reading
Ackman RG (1991) Application of thin layer chromatogra-
phy to lipid separation: neutral lipids. In: Perkins EG
(ed.) Analysis of Fats, Oils and Lipoproteins, pp. 60}82.
Champaign, IL: American Oil Chemists Society.

Cagniant D (ed.) (1992) Complexation Chromatography.
New York: Marcel Dekker.
Christie WW (1987) High Performance Liquid Chromatog-
raphy and Lipids. Oxford: Pergamon.
Christie WW (1982) Lipid Analysis, 2nd edn. Oxford:
Pergamon.
De Ligny CL (1976) Advances in Chromatography 14:
265}304.
Fried B and Sherma J (1996) Practical Thin-Layer
Chromatography } A Multidisciplinary Approach. Boca
Raton, FL: CRC.
Fried B and Sherma J (1998) Thin-Layer Chromatography
} Techniques and Application, 4th edn. New York:
Marcel Dekker.
SODIUM CHLORIDE: CRYSTALLIZATION
R. M. Geertman, Akzo Nobel Chemicals Research,
Arnhem, The Netherlands
Copyright ^ 2000 Academic Press
Introduction
Salt has been a part of human existence since time
immemorial. It was used for cooking wheat and barley
as early as 5000 BC. The Rrst salt was gathered from
shallow lagoons where seawater could evaporate. Later
rock salt was mined, and in the Alps, for instance, rock
salt is known to have been mined as early as 1400 BC.
Because of its importance to human life, salt has had an
inSuence on economy, history and culture. Many
sayings and words are derived from the use of salt:
e.g. to be worth ones salt is a compliment, the Bible tions can be divided into three major categories:
speaks of the salt of the earth and soldier is derived
from the Latin sal dare which means to give salt.
Indeed, in ancient times salt had a much greater
value than it has nowadays. It was traded weight by
weight with gold, and the salt trade was very proRt-
able. The Hanseatic League started by trading in salt.
Taxes on salt were very common, and in that sense
salt has played a role in many important historic
events. The French Revolution was partly in protest
against salt taxes. Gandhis campaign of civil dis-
obedience, which eventually led to the independence
of India, started when he evaded the British salt
monopoly by producing salt himself.
Uses of Salt
Before the industrial revolution and the discovery of
the electrolysis process, the uses of salt were limited.
III / SODIUM CHLORIDE: CRYSTALLIZATION 4127
Gmelin Handbuch der Anorganischen Chemie (1975)
Silver, vol. 61, TI.B5. Berlin: Springer-Verlag.
Morris LJ (1966) Separation of lipids by silver
ion chromatography. Journal of Lipid Research 7:
717}732.
Morris LJ and Nichols BW (1972) Argentation thin layer
chromatography of lipids. In: Niedewieser A (ed.)
Progress in Thin Layer Chromatography and Related
Methods, vol. 1, pp. 74}93. Ann-Arbor, MI: Ann-
Arbor-Humphrey Science Publishers.
Nikolova-Damyanova B (1992) Silver ion chromatography
and lipids. In: Christie WW (ed.) Advances in
Lipid Methodology } One, pp. 181}237. Ayr: The Oily
Press.
The main uses were for the cooking, preserving
and pickling of food and the tanning of hides. These
uses are still very important, as salt is essential for
the human body. With the development of chemical
processes the uses of salt have diversiRed enormously.
Apart from uses in the food industry, salt is,
for instance, used in dyeing, paper production,
highway de-icing, oil well drilling and the product-
ion of soda ash. Electrolysis of salt is the major
source of chlorine and sodium hydroxide for the
chemical industry. Chlorine is essential for the
production of a number of plastics, insecticides and
pharmaceutical compounds. Either directly, or in
the form of derivatives, salt Rnds application in
more than 14 000 ways. This multitude of applica-
chemical uses, highway de-icing, and food-related
uses. In the industrialized nations the chemical indus-
try accounts for approximately 50% of the salt con-
sumption, and highway de-icing for about 30% while
food applications make up the remainder. In develop-
ing countries most of the salt produced is used in
food.
Production of Salt
As salt is the most abundant nonmetallic mineral,
most countries have the ability to produce salt. It is so
abundant that it is hard to estimate salt reserves. In
the United States alone, reserves are estimated at 55
trillion tonnes. In 1996, 192 million tonnes of salt
were produced, approximately 55% in the industrial-
ized nations and about 45% in developing countries.
Details are given Table 1.
4128 III / SODIUM CHLORIDE: CRYSTALLIZATION
Table 1 World production of salt in 1996
Country Amount produced
(in millions of metric tons)
United States 42.9
Peoples Republic of China 28.9
Canada 12.3
Germany 10.9
India 9.5
Mexico 8.5
Australia 7.9
Other 71.8
Total 192.0
Salt is a cheap commodity. At a 1997 price level of
US $60 per tonne for chemical grade, merchant-de-
livered salt and US $10 per tonne for captive use, salt
is cheaper than any other reRned chemical. Because of
the low price of salt relative to the transportation
costs, most salt production plants are in the vicinity of
salt users. Producing an acceptable grade of purity at
the lowest deliverable costs is the most important
consideration for salt producers throughout the
world.
How the salt is produced depends very much on the
form in which the salt is available. In subtropic, arid
regions salt is mostly produced by evaporating sea-
water. Large production facilities for so-called solar
salt can be found in India, Australia and Mexico. In
temperate regions, where the climate is less favour-
able for the evaporation of seawater, rock salt is
mined. This can be done in two ways. Provided the
salt deposit is close to the surface, it can be mined in
the classical roof and pillar method. The rock salt is
crushed, sorted and sold as a low grade quality. For
more demanding applications further puriRcation is
needed. If the salt deposits are more deeply located,
the solution mining technique is used. This involves
pumping water down through a borehole to dissolve
the salt, and recovering the resulting brine. The brine
is then puriRed to remove foreign ions, and evapor-
ated. The salt produced in this way is known as
vacuum salt.
Fundamentals of Salt Crystallization
Salt or sodium chloride can occur in two forms. The
Rrst and best known form is the anhydrous form,
NaCl. The second form is the dihydrate which is
formed in a pure brine at temperatures below 0.13C.
The solubility of the dihydrate form is weakly tem-
perature dependent, whereas the solubility of the an-
hydrous form is nearly temperature independent. The
temperature dependence of the solubility of salt in
water is given in Figure 1.
The anhydrous form, NaCl ) 0H2O, crystallizes in
the Fm3m space group in which each sodium ion is
octahedrally coordinated by six chloride ions, and
vice versa. The
1 0 0 faces are the slowest growing
faces, resulting in the typical cube shape of salt crys-
tals. By adding additives the
1 1 1 faces can be
retarded, thus yielding octahedral shapes. Additives
such as Fe (CN)46\ or NTAA (nitrosyl triacetamide)
poison the
1 0 0 surfaces, resulting in preferential
growth along edges and on corners. These shapes are
given in Figure 2, together with a number of inter-
mediate shapes where both cubic and octahedral
faces are visible.
As mentioned before, salt is very soluble in water,
so the growth rate of salt crystals is diffusion control-
led. Like many other very soluble salts, the driving
force needed to obtain acceptable growth rates (in the
order of 10\8 m s\1) is low, typically of the order of
0.1% or lower. Salt crystals produced in a forced
circulation crystallizer (the most common type used
for salt crystallization) typically have a mean size of
350}400
m. In draft tube bafSed (DTB) type crystal-
lizers salt crystals can become larger, in the order of
500}1000
m. Apart from the inSuence on the mean
crystal size, the low driving force for crystallization
also strongly reduces agglomeration. Agglomerated
salt can only be obtained using techniques such as
antisolvent crystallization where high driving forces
are involved. There are three mechanisms for the
incorporation of impurities in the Rnal crystalline
product. The Rrst is direct incorporation of the impu-
rity in the crystal lattice, the second is the formation
of inclusions and the third is insufRcient washing of
the crystals.
In contrast to organic crystals, and to a lesser
extent hydrated salt crystals, the ions are densely
packed in the crystal lattice. This effectively prevents
the incorporation of larger molecules in the crystal
lattice as the lattice strain and the enthalpies involved
are extremely unfavourable. This applies to many
ions that have marked differences in ionic radius or
charge from either the chlorine or sodium ion. The
Figure 1 Solubility of NaCl in water.

Figure 2 Different crystal forms observed during the crystalliza-
tion of common salt. Reproduced with permission from Elsevier
Science.
tendency of salt to form solid solutions is therefore
very limited. The only notable exception is with the
incorporation of bromide, which is therefore very
hard to remove as an impurity, once incorporated.
The second mechanism, the formation of inclu-
sions, is much more common. In industrial crystalli-
zation the solid fraction in the slurry is high and
therefore collisions between crystals are frequent. If
the energy involved in a collision event is high enough
the corners of the cubic salt crystals will be damaged
and the crystals will be strained. Regrowth of these
damaged corners is often imperfect and inclusions are
formed. An effective method of reducing the impurity
uptake through this mechanism, though at the cost of
production capacity, is to lower the solids fraction of
the slurry. It should be noted that this mechanism
only occurs when large crystals are produced. If the
crystals are small the kinetic energy involved in the
collisions is not high enough to damage the crystals.
Good washing of the crystalline product is very
important for the Rnal purity. By evaporating water
not only is salt produced, but the impurities are also
concentrated. Any mother liquor that remains will
therefore have a profound detrimental effect on the
product purity. More impure mother liquors require
better washing. This is more important for solar salt
than for vacuum salt. Vacuum salt is produced from
puriRed brine that contains, apart from NaCl, few
very soluble salts, so despite incomplete washing the
amount of impurities will still be relatively low. Con-
centrated seawater, in contrast, also contains high
concentrations of very soluble salts, especially mag-
III / SODIUM CHLORIDE: CRYSTALLIZATION 4129
nesium salts. InsufRcient washing therefore has much
more inSuence on the product purity. Another impor-
tant aspect when washing solar salt is the agglomer-
ation.
A striking example of the inSuence of agglomer-
ation on the Rnal product purity is provided by the
antisolvent crystallization of NaCl. Though the crys-
tals are crystallized from a mother liquor containing
as much as 50% weight antisolvent (on a solvent
basis) the uptake of the antisolvent is as low as
30 ppm. Diluting the antisolvent stream with water,
which reduces the driving force for crystallization,
should result in a decrease in the impurity con-
centration. However this effect is offset by the in-
creased agglomeration of the system, so instead of the
expected lowering of concentration the concentration
of antisolvent is increased to 100 ppm.
Irrespective of the method by which salt is pro-
duced, seawater is the source of salt (salt deposits are
the result of natural solar salt production). The impu-
rities present in brine or rock salt are therefore the
same. These impurities are mainly Ca2#, Mg2#,
Sr2#, Br\ and SO24\. All these impurities have undes-
irable effects on the electrolysis of sodium chloride
and need to be removed during the production pro-
cess. During electrolysis calcium, strontium and mag-
nesium are deposited as hydroxides on the electrodes,
which is of course not desired. The presence of Br\
leads to the formation of ClBr, which is also un-
wanted. Finally the presence of sulfate increases the
cell potential needed for electrolysis, thus increasing
production costs. A further impurity is iron, which is
not only present in the salt, but is also added in the
form of ferrocyanide, an anticaking agent. Though
needed for proper salt handling, the presence of iron
interferes with the membrane electrolysis. The ferro-
cyanide must therefore be decomposed and the iron
precipitated as the hydroxide salt prior to electrolysis
of the sodium chloride solution.
The strategy for removing these impurities depends
on the production method. For solar salt production
fractional crystallization combined with careful
washing is employed, whereas in vacuum salt produc-
tion the brine is puriRed prior to crystallization of the
sodium chloride. In the rock salt production various
recrystallization methods are used.
Solar Salt
Crystallization sequence Seawater contains nearly
all elements of the periodic system in varying
amounts. The composition of seawater is given in
Table 2. When seawater is evaporated many different
salts will be formed, at different stages during
the evaporation. In that sense production of pure
sodium chloride from seawater somewhat resembles
4130 III / SODIUM CHLORIDE: CRYSTALLIZATION

Table 2 Composition of seawater is derived from the bitter taste of these salts. The
whole sequence is depicted graphically in Figure 3.
Component Amount present This Rgure shows the relationship between the
(g per 1000 g seawater)
density of the brine expressed as degrees Baume
Ca 0.408 (3Be"145!(145/speciRc density at 15.63C)) and
SO4 2.643 the crystallization of the various salts. The concentra-
Mg 1.265 tion factor can be deduced from the cumulative
Cl 18.95
amount of water evaporated, which is also given in
K 0.380
Na 10.48 the Rgure.
Br 0.065
Plant layout To produce pure salt, the crystalliza-
Total 34.19
tion of iron oxide, calcium carbonate and calcium
sulfate must be physically separated from the sodium
the distillation of crude oil, where one is interested in chloride crystallization. This is achieved in solar salt
obtaining well deRned fractions. When seawater is works by having two kinds of ponds: concentration
concentrated gradually iron oxide and calcium car- ponds and crystallizer ponds. Approximately 90% of
bonate start to crystallize Rrst, but the amount of iron the water must be evaporated before salt starts to
oxide produced is negligible. Then calcium sulfate crystallize, so the concentration ponds are much lar-
precipitates. It is important to note that when sodium ger than the crystallizer ponds. Though the water is
chloride is subsequently crystallized, the mother not completely evaporated in the crystallization sec-
liquor is concentrated with respect to both of the salts tion (the brine is discharged before the bitterns start
mentioned. After crystallizing about 75% of the to crystallize), the concentration pond/crystallizing
available sodium chloride (at which stage 97% per- pond area ratio is usually around 10 to 1. Evapor-
cent of the water has been evaporated), sodium bro- ation is a slow process, so solar plants must occupy
mide will start to crystallize as a solid solution with a large area. The solar salt plants in Australia
sodium chloride, and the so-called bitterns will also and Mexico, which supply the chemical industries
crystallize. The term bitterns is used for a collection in Japan and the United States, are tens of square
of magnesium, potassium, sulfate and chloride salts, kilometres in size. The seawater needs to be concen-
such as KCl, MgCl2, MgSO4 and double salts. It trated in stages so most solar salt plants have 5}10

Figure 3 Deposition of salts during the evaporation of sea water at 253C.


concentration ponds in series. These ponds are shal-
low to obtain the best surface area/volume ratio, with
the depth usually between 50 and 80 cm. Small, low
levees separate the different ponds. The crystallizer
ponds are smaller, ranging from several hundred
square metres in the case of manual harvesting to
several acres in the case of mechanical harvesting.
After harvesting the salt is transported to the washing
plant. An example of the layout of a solar salt plant is
given in Figure 4.
Solar salt production The need for large shallow
ponds in the vicinity of the sea determines where solar
salt plants can be operated. Small plants can be found
in all coastal areas near the tropics, large plants only
in India, Australia and Mexico. Such plants are oper-
ated all year round. Further north than the tropics, in
arid areas as far north as western France, the opera-
tion is seasonal; the salt is harvested before the winter
rains dissolve the salt produced.
Production of solar salt is started by taking in
seawater. The seawater is concentrated by evapor-
ation and the brine is reduced to about 60% of its
original volume. After the Rrst concentration stage
the brine is transferred to another area where calcium
carbonate starts to precipitate. Here a further 15% of
the original volume is evaporated and the brine is
Figure 4 Schematic diagram of Dampier salt field layout and location map of Western Australia. Reprinted from Garrett DE, with kind
permission from Elsevier Science Ltd.
III / SODIUM CHLORIDE: CRYSTALLIZATION 4131
transferred to a third type of pond, the gypsum pre-
cipitation pond. The brine is concentrated to the
point at which sodium chloride nearly starts to
crystallize before it is transferred to the pickling pond
where the brine, now saturated with sodium chloride,
is kept before being transferred to the crystallizer
pond. This pond is needed to reduce the gypsum
supersaturation in the brine. At this point about 90%
of the water has been evaporated. In the crystallizers,
depending on the required purity, 70}75% of the
available sodium chloride is crystallized before the
remaining bitterns are discharged. Production of high
quality solar salt is very much a question of knowing
when to start producing halite and when to stop.
A major concern is the sealing of the ponds to
prevent losses. It is impossible to treat the bottom of
the ponds because of their large size, so the ponds
must have a base such as clay, which is (fairly) imper-
vious to water. The precipitated calcium carbonate
(and also gypsum) will improve the sealing during
operation so the losses will go down with time. By
regularly changing the brine Sow through the plant,
all concentration ponds will be used as calcium
carbonate and gypsum precipitation ponds, thus
ensuring minimum brine losses (provided of course
that the ponds are equal in size). Note that this is only
possible in small plants.
4132 III / SODIUM CHLORIDE: CRYSTALLIZATION
The transmission of solar light in the brine is high,
which reduces the evaporation rate. To enhance sun-
light absorption dyes are added. The most common
practice nowadays is to add algae which reduce the
light transmission from 96% to 55}70%. An addi-
tional advantage is that the algae mats will also plug
the pond bottom. Care must be taken not to increase
the viscosity too much through abundant algae
growth, or other organisms have to be introduced to
keep the algae concentration within limits. Red
halophilic bacteria are added to the crystallizing
ponds for the same purposes.
The rate at which water evaporates depends on the
climate. The amount of sunshine, the mean temper-
ature, the relative humidity, the wind velocity and the
average amount of precipitation determine the net
evaporation rate. Accurate Rgures for the net evapor-
ation are needed both for design purposes and for
process control (read brine management). For design
purposes climate data (incorporating amount of
sunshine, air temperature and humidity and wind
velocity) and brine data are taken into account. The
gross evaporation rate can then be calculated by mul-
tiplying the difference in vapour pressure between air
and brine by a mass transfer coefRcient. In this mass
transfer coefRcient factors such as temperature, net
gain of radiant energy and the heat transfer coefRc-
ient are taken into account.
For production purposes another model is gener-
ally used. The evaporation of water from a fresh
water pan, situated on the site, is measured. This
Rgure is then corrected for the salinity, the size of the
pan and the rainfall.
Evaporation (pond)"(evaporation (pan)
;kscale;ksalinity!rainfall)
;area;w
Here kscale and ksalinity the scale factors and w is the
density of water. Using this method it is possible to
estimate how much has been evaporated from each
pond, to decide when to pump brine from one pond
to another, to take in sea water, etc.
In the crystallizer ponds a salt Soor is formed
during the crystallization of sodium chloride. The salt
is harvested by completely removing the salt Soor
(in seasonal operations) or scraping the top layer (in
year round operations). When the salt is mechanically
harvested great care should be taken that the Soor is
strong enough to accommodate heavy equipment.
Usually salt Soors for mechanical harvesting are at
least half a metre thick.
After harvesting the salt is washed by mixing the
crystals with fresh saturated brine, and transferred to
a mesh conveyor where the brine is drained off. Gener-
ally sea water sprays are then used to Rnish the wash-
ing process. Provided the washing is carried out very
thoroughly, and the crystals are crushed to remove the
mother liquor contained in cavities, solar salt can
reasonably pure, though not as pure as vacuum salt.
In most cases solar salt contains signiRcantly more
magnesium (300}500 ppm), calcium (200}300 ppm)
and sulfate (1000}1500 ppm) than vacuum salt.
The salt can be upgraded using the Salex process.
This process utilizes the difference in density and
morphology between sodium chloride and other min-
erals. The salt crystals containing the impurity crys-
tals are countercurrently washed with a saturated
brine. The salt crystals settle, and the impurity (main-
ly gypsum) crystals are carried away with the brine
and left to settle in a separate tank. The clariRed brine
can then be reused. By leaving the produced salt piles
exposed to rain, impurities will preferentially dis-
solve, thus improving the salt quality. This is called
the rain wash method.
Rock Salt
Provided the rock salt deposits are close to the sur-
face, rock salt can be mined in the classical way. First
a vertical shaft is dug until the salt bearing deposit has
been reached. Then horizontal shafts are blasted and
the salt thus produced is transported to the surface.
For de-icing use the salt is only crushed and sorted by
size. For chemical applications the rock salt must be
upgraded. This can be achieved in several ways. First
of all, the rock salt can be completely dissolved. The
resulting brine is then treated in the same manner as
brine obtained by solution mining. As this process is
more expensive than solution mining it is hardly ever
used. The second option is to recrystallize the brine
by making use of the fact that sodium chloride also
has a hydrated solid phase.
This process works as follows. First small, crushed
salt crystals are suspended in a brine at a temperature
of 53C. The brine is then cooled to a temperature
lower than 03C. At this temperature NaCl is
more soluble than the hydrated phase, NaCl ) 2H2O.
The anhydrous sodium chloride crystals will dissolve
and sodium chloride dihydrate crystals will be for-
med. When the slurry now containing the dihydrate
crystals is heated, the process is reversed. The dihyd-
rate crystals will dissolve and anhydrous crystals will
be formed. During the two consecutive crystallization
steps the impurities present in the dissolving crystals
will remain to a large extent in the brine, instead of
ending up in the new crystals formed. Using this
method, the purity of the salt can be markedly
improved without having to evaporate the brine.
After the recrystallization steps the now puriRed

crystals are separated from the brine and the brine
is reused after treatment.
Vacuum Salt
Brine puriVcation Vacuum salt is the name given to
salt produced by evaporative crystallization. Water is
injected into a salt deposit through a borehole, and
dissolves the salt so the resulting brine can be re-
covered. This brine is then puriRed before salt is
produced by evaporative crystallization. The deposits
used for this kind of salt production lie at depths
between a few hundred and three thousand metres.
Two pipes are used, one bearing fresh water to the
cavern, the other transporting brine to the surface.
Because of surface subsidence and ground move-
ments, the caverns cannot get too large. Normally the
size is smaller than 100 m in width. Furthermore,
above the cavern a layer of a few metres of salt must
be maintained to protect the overlying strata from
brine penetration. Once the cavern has reached its
maximum allowable size, solution mining is stopped
and a new borehole is drilled.
It is interesting to note that the salt mined in such
a way has already been crystallized once. All salt
deposits are remains from earlier lakes and seas,
which have been evaporated. Thus the salt is already
separated from the calcium sulfate and bitterns ori-
ginally present in the seawater.
Though already purer than the saturatd brine pro-
duced by seawater evaporation, the brine produced
still contains signiRcant amounts of calcium, magne-
sium, sulfate and bromide ions. Most of these ions are
removed in the brine puriRcation process. First the
sulfate is removed by adding calcium oxide, which
leads to the precipitation of gypsum. The oxide in-
creases the pH of the brine, which induces the precipi-
tation of magnesium hydroxide. Thus in the Rrst stage
sulfate and magnesium are removed. To reduce pro-
duction costs this process is carried out using very
simple equipment. Calcium oxide is simply mixed
with the brine in a very large tank. To decrease the
Figure 5 Scheme of a four effect evaporative crystallization plant.
III / SODIUM CHLORIDE: CRYSTALLIZATION 4133
supersaturation further tanks are used. The solids are
removed by draining the tanks and emptying them.
A drawback to this procedure is the increased cal-
cium level in the brine. To reduce this level gases from
an on-site power plant are usually used. The carbon
dioxide yields carbonate, due to the high pH of the
brine, which together with calcium forms the almost
insoluble calcium carbonate. Because of the high salt
concentration and high temperature, vaterite is for-
med rather than the usual calcite. This is an advant-
age because strontium, which is also present in the
brine, is also effectively removed as strontium car-
bonate forms a solid solution with vaterite.
Crystallization After puriRcation the water is evap-
orated and salt is crystallized. To conserve energy
this is done in several stages. Each stage is operated
under a lower pressure so the brine boils at a lower
temperature. The vapour produced by the previous
stage condenses in a heat exchanger and as this steam
was formed at a higher temperature, the brine starts
to boil. In this way the steam used in the primary
stage can be reused several times. This principle is
shown in Figure 5. Thus the total amount of energy
involved in the production of vacuum salt is greatly
reduced. In a typical four-effect installation (see
Table 3) the Rrst effect operates at slightly elevated
pressures while the other effects operate under re-
duced pressure, hence the name vacuum salt.
Generally salt crystallizers are of the forced circula-
tion type with external heat exchangers. For a
production of 1 million tonnes per year, a total cry-
stallizer capacity of 800}1000 m3 is needed. The sep-
arate crystallizers can be operated in different modes.
The salt produced can be separated from the mother
liquor separately for each crystallizer, or the slurry
can be transported from one effect to the other, thus
increasing the solids content of the slurry in each
successive stage. This has important implications for
the purity. In the Rrst case the salt produced in the
Rrst crystallizer is the purest. Little water has (yet)
4134 III / SODIUM CHLORIDE: CRYSTALLIZATION
Table 3 Temperatures and pressures in a four-effect evapor-
ative crystallization plant
Effect 1 2 3 4
Pressure (bar) 2.0 0.89 0.34 0.18
Temperature (3C) 120 96 72 58
been evaporated and the impurity concentration is low.
In the successive crystallizers the impurities are gradual-
ly concentrated, which has a detrimental effect on the
purity. Thus different grades of salt are produced.
When the slurry is transported from one effect to the
other, only one grade of salt is produced, which repres-
ents the mean purity when compared with the other
method. There will be local differences in purity, the
centres of the oldest crystals being the purest, where-
as the newer crystals and the collision prone corners
of the larger crystals will contain more impurities.
Brine recovery As with the production of solar salt,
the amount of impurities determines when the crys-
tallization is stopped. Again bromide is important in
that respect, as it is very difRcult to remove. One is
then left with a brine containing valuable salt that
unfortunately it is difRcult to recover because of the
high impurity content. Several procedures have been
devised to cope with this problem. The Rrst process,
used by Akzo Nobel, is the Bromin process. In this
process the remaining sulfate- and bromine-rich mother
liquor is further evaporated in a separate crystallizer,
where sodium sulfate is crystallized in addition to
sodium chloride. The remaining, greatly reduced
amount of mother liquor, now very rich in bromine, is
discharged in a nonproductive borehole. Sodium
chloride and sodium sulfate are added to the raw brine
entering the brine puriRcation. Sodium sulfate dis-
solves and the sulfate precipitates as calcium sulfate.
At the Salinen Austria GmbH production facility
nearly the same procedure is used, but instead of
crystallization of anhydrous sodium sulfate and
sodium chloride, the company claims that sodium
chloride and glaserite (Na2SO4 ) 3K2SO4) are formed.
After washing with raw brine, the glaserite is dis-
solved and the sulfate serves to precipitate calcium
while the sodium chloride remains behind.
Another method involves cooling the remaining
brine, thus producing Glauber salt, Na2SO4 ) 10H2O.
Subsequent evaporation of the mother liquor yields
sodium chloride. The remaining brine is then discarded.
SOLID-PHASE EXTRACTION OF DRUGS
See III / BIOANALYTICAL APPLICATIONS: SOLID-PHASE EXTRACTION
See Colour Plate 117.
Further Reading
Burnard E (1993) The use of computer models in solar salt
Reld process control. Seventh Symposium on Salt, vol. I,
pp. 499}505. Amsterdam: Elsevier Science Publishers.
Flachberger H and Krenn K (1999) Zum Stand der Technik
im Bereich der Aufbereitung von Salzmineralien. Berg-
und hu( ttenmannischen Monatshefte 144(6): 234.
Garrett DE (1969) Factors in the design and layout of solar
salt plants. Part I. Pond layout and construction. Pro-
ceedings of the Third International Symposium on Salt,
pp. 63}69. Northern Ohio Geological Society, Cleve-
land, Ohio, USA.
Jongema P (1983) Optimization of the fuel consumption of
an evaporation plant with the aid of the exergy concept.
Sixth International Symposium on Salt, pp. 463}496.
Salt Institute, Alexandria, Virginia, USA.
McArthur JN (1980) An approach to process and quality
control relevant to solar salt Reld operations in the
northwest of Western Australia. Proceedings of the Fifth
International Symposium on Salt, pp. 325}338. North-
ern Ohio Geological Society, Cleveland, Ohio, USA.
Mersmann A (1995) Crystallization Technology Hand-
book. New York: Marcel Dekker Inc.
Nielsen AE (1984) Electrolyte crystal growth mechanisms.
Journal of Crystal Growth 67: 289}310.
Ninane L, Craido Cl and Thomas L (2000) PuriRcation of
rocksalt by a new process at low temperature. Proceed-
ings of the Eighth International Symposium on Salt,
pp. 451}458. Amsterdam: Elsevier Science Ltd.
Sedivy VM (1996) PuriRcation of salt for chemical and
human consumption, Krebs Swiss, Zurich, Switzerland,
in Industrial Minerals, April.
Surdyk J, Leder AE and Ishikawa-Yamaki M (1998) CEH
Product Review } Sodium Chloride. Chemicals Econ-
omics Handbook 1998. SRI International, Menlo Park,
CA 94025-3477, USA.
US Geologic Survey (1996) U.S. Geologic Survey Minerals
Information 983 National Center, Reston, VA 20192,
USA.
Venkatesh Mannar MG and Bradley HL (1984) Guide-
lines for the Establishment of Solar Salt Facilities
from Seawater, Underground Brines and Salted
Lakes. Industrial and Technological Information Bank,
Industrial Information Section, United Nations
Industrial Development Organization, New York.
Venkatesh Mannar MG and Dunn JT (eds) Consumption
and uses of salt. In: Salt Iodization for the Elimination of
Iodine DeTciency. International Council for Control of
Iodine DeRciency Disorders.
 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES
SOLID-PHASE EXTRACTION WITH
CARTRIDGES
D. A. Wells, Sample Prep Solutions Company,
Maplewood, MN, USA
Copyright ^ 2000 Academic Press
Introduction
Sample preparation is an important component of an
analytical method. It is used to concentrate an analyte
to improve its limits of detection, as well as to isolate
an analyte from unwanted matrix components
that can cause interferences upon analysis. Solid-
phase extraction (SPE), as a tool for this sample
concentration and isolation, has gained acceptance
since its commercial introduction circa two decades
ago. SPE is performed using commercial packed car-
tridges (containing approximately 50}500 mg
packing material) as well as discs (containing from
4}500 mg). Many formats, chemistries and sizes
of SPE products are available to meet a range of
separation needs.
Solid-phase extraction is preferred to other types of
sample preparation techniques, such as liquid}
liquid extraction (LLE), for many reasons. SPE is
an efRcient technique, often achieving higher recov-
ery of analyte than other methods of sample prepara-
tion because of its selectivity. The chemistry of
attraction between an analyte and the solid sorbent
can be exploited by pH and solvent considerations
to allow interaction yet exclude interferences. SPE
is a less time-consuming and labour-intensive tech-
nique. Extraction typically involves adding different
liquids through SPE columns in parallel and collect-
ing the eluate at the Rnal step. Emulsion formation
is eliminated } in LLE an emulsion sometimes
forms between the aqueous and organic layers pre-
venting phase separation. Organic solvent consump-
tion is far less using SPE than typical LLE techniques,
saving money in terms of both purchase costs of
solvents and costs to dispose of these regulated
solvents. Reduced exposure of the analyst to organic
solvents also improves safety in the laboratory.
Unlike LLE the SPE procedure using columns can be
automated. There are several hardware choices avail-
able commercially that transform SPE from a manual
procedure into a fully automated one, allowing the
analyst to perform other tasks in the laboratory.
Batch procedures of automation are available in
which a number of samples are extracted to yield
the same number of eluates ready for analysis.
4135
On-line serial automation is common, in which
a sample is extracted then injected by the instrument
and, while analysis is ongoing, the next sample is
extracted. Miniaturization of SPE allows the conve-
nient use of smaller sample sizes and the ability to
physically work with eluates as small as 50
L when
using the disc format.
The SPE technique has been shown to be useful for
a variety of sample matrices, including (but not lim-
ited to): drinking water and river water, air, biolo-
gical Suids (e.g., blood, serum, plasma, urine), tissues,
peptides, drug formulations, microbial broths, animal
feed, beverages, fruits and vegetables, and soil. Thus,
the numbeBr of applications for this sample prepara-
tion technique in the literature is extensive and can be
found spanning the last 25 years. It is the goal of this
chapter to highlight many of the applications of SPE
for a variety of sample matrices and demonstrate the
versatility and usefulness of this sample preparation
technique (Table 1).
Table 1 Examples of classical analytical applications using
solid-phase extraction
Market Application
Environmental Trace enrichment of organic pollutants from
water
Organic acids, detergents and surfactants
from water
Insecticides and pesticides from soil
Explosives residues in groundwater
Oil and grease analysis
Food Pesticides in fruits and vegetables
Sodium benzoate in colas and fruit juice
Plant growth regulators in spinach juice
Toxic fungal metabolites in rodent feed
Vitamins in food
Cholesterol oxidation products in milkfat
Caffeine in beverages

-Agonists and antibiotics in meat products
Biotechnology Purification and fractionation of proteins and
peptides
Desalting of peptides
Purification of DNA from microbial broths
Pharmaceutical Antibiotic content in ointments
Aspirin content in tablets
Drugs in serum, plasma and urine
Clinical Catecholamines in plasma and urine
Lipids in serum
Drugs in tissues
Vitamins and steroids in serum
Cyclosporin in blood
4136 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES
Environmental Applications
Trace Enrichment from Environmental Samples
The gas and liquid chromatographic analyses of polar
pollutants in waters (e.g. for drinking, river and efSu-
ent) require a concentration step before analysis to
determine part-per-million levels and lower, as regu-
lations specify. Solid-phase extraction is the most
widely used technique for trace enrichment of polar
environmental pollutants since it uses low volumes of
hazardous organic solvent, can be automated using
cartridges and discs, and the analysis can be
done either off-line in batches or on-line with the
chromatographic system.
Typically, 1 L volumes of water samples are
required for analysis in the United States, as man-
dated by the Environmental Protection Agency
(EPA). Hydrophobic C18 and C8 sorbents are com-
monly used for the majority of trace enrichment
needs; the analyte structure dictates the optimal
sorbent chemistry. Practical considerations for pas-
sing 1 L of water through a SPE cartridge favour the
use of larger diameter (47 or 90 mm) discs (e.g. glass
Rbre and PTFE-based) for these applications. While
both discs and cartridges can be automated, discs are
preferred for their much larger cross-sectional surface
area. Using discs, liquid can be passed through at high
Sow rates without loss of analyte, thus reducing the
extraction time to about 10}15 min instead of about
1 h for narrower cartridges.
Aqueous samples (100 mL to 1 L) containing or-
ganochlorine pesticides (e.g. lindane, methoxychlor,
or endosulfan) can be concentrated from water,
made acidic by using C18 bonded silica in 47 mm
discs. Elution from the sorbent is accomplished
using 3}5 mL aliquots of ethyl acetate, from
which water is removed in a separate step using
anhydrous sodium sulfate, it is then concentrated
before analysis using gas chromatography (GC) with
electron capture detection (ECD). Polyaromatic hy-
drocarbons (e.g. phenanthrene, pyrene, anthracene),
organophosphorous pesticides (e.g. diazinon, methyl
parathion), and herbicides (e.g. atrazine, alachlor)
may be analysed in a similar manner using C18
sorbent and ethyl acetate elution from discs prior to
GC analysis.
Phenols and chlorinated phenols are moderately
polar compounds that can display ionic character at
pH values above 7. Another group of polar com-
pounds displaying ionic character is the acid herbi-
cides (e.g. 2,4-dichlorophenoxyacetic acid, 2,4,5-
trichlorophenoxyacetic acid and dicamba). Rather
than C18 bonded silica, a more efRcient sorbent for
extraction of these types of compounds is polystyrene
divinylbenzene (SDB), an organic polymer. SDB has
a slightly different selectivity than C18, owing to its
aromaticity, that allows it to extend its range of
attraction to include more polar species such as phen-
ols. Other advantages of SDB are that it is totally
organic, is stable across the entire pH range, and has
greater capacity per gram than comparable reversed-
phase bonded silica sorbents. Extraction of these
phenols is performed with a SDB disc (or cartridge).
Elution from the sorbent is accomplished using
3}5 mL aliquots of acetonitrile (methanol or acetone
may be substituted) before analysis.
Diquat and paraquat are examples of polar com-
pounds that are quaternary amines, thus always pos-
itively charged. These analytes are found only in very
small concentrations in water, since they more readily
attract to soils and plants via their cationic function-
ality. They can be concentrated from water on a cy-
ano sorbent or some types of C8 sorbent, those that
have a high degree of residual silanols available for
attraction of cationic species. The extraction method
for C8 includes an ion-pairing agent, to which para-
quat and diquat bind. Elution with 5 mL methanol
containing acid and diethylamine disrupts this bind-
ing; analysis is by HPLC. Table 2 lists examples of
USA EPA methods employing disc solid-phase
extraction.
On-Line Techniques
While SPE is a successful technique performed in
batch mode before the analysis step, there can be
drawbacks such as loss of sensitivity (only an aliquot
of the total mass isolated is used), losses due to
evaporation or during transfer and contamination
from external sources. Instrumentation has now ad-
vanced to allow for on-line trace enrichment, where
the sample eluent is injected onto a high-pressure
liquid chromatography (HPLC) apparatus. The
sample can be isolated on a guard column while the
HPLC is running the previous sample, so time is not
lost between isolation and analysis. The cartridge
performing the extraction on-line, coupled to a liquid
chromatographic system, can be commercially
bought, such as the PROSPEKT system (Spark
Holland) or constructed by hand using Empore
(3M Company) membrane extraction discs placed
into a holder (4.6 mm internal diameter). Multi-
residue methods that extract a variety of pesticides
(acidic, neutral and basic) from waters are
commonly used. In order to preconcentrate all these
compounds simultaneously, it is necessary in most
cases to acidify the sample and use a C18 bonded
silica or polymer-based SDB sorbent in series with
a cation exchanger. In order to avoid rapidly over-
loading the cation exchanger with samples of high
ionic strength, calcium ions are Rrst precipitated
 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES 4137

Table 2 United States Environmental Protection Agency EPA methods allowing the use of solid-phase extraction for sample
preparation

Method number Analytes Sorbent Analysis technique a

506 Phthalate and adipate esters in C18 GC/PID

drinking water
507 Nitrogen and phosphorous containing C18 GC/NPD
pesticides in water
508 Chlorinated herbicides and C18 GC/ECD
organochlorine pesticides in water
513 TCDD (2,3,7,8-tetrachlorodibenzo- C18 GC/MS
p-dioxin) in drinking water
515.2 Chlorinated acids in water SDB GC/ECD
525.1 Organic compounds in drinking water C18 GC/MS
548.1 Endothall Strong anion exchange GC/MS
549.1 Diquat and paraquat in drinking water C18, C8 or strong cation exchange HPLC/UV
550.1 Polycyclic aromatic hydrocarbons C18 HPLC/UV and flurorescence
(PAH) in drinking water
552.1 Haloacetic acids and dalapon in Strong anion exchange GC/ECD
drinking water
553 Benzidines and nitrogen containing C18 HPLC/MS
pesticides in water
554 Ozonation disinfection by-products C18 HPLC
(carbonyl compounds)
1613 Revision B Tetra- to octa-chlorinated dioxins and C18 HRGC/HRMS
furans
1664 Oil and grease C18 Gravimetric and infrared
3535 (SW846) Organochlorine pesticides and C18, SDB Various GC/ECD techniques
phthalate esters from groundwater,
wastewater and TCLP leachates

a
Abbreviations: GC, gas chromatography; PID, photoionization detector; ECD, electron-capture detector; MS, mass spectrometry;
HPLC, high-performance liquid chromatography; UV, ultraviolet, HRMS, high resolution mass spectrometry.

with oxalic acid and heavy metals are complexed
with ethylenediaminetetraacetic acid (EDTA) prior to
SPE. SPE cartridges on-line are sometimes preloaded
with sodium dodecyl sulfate (SDS) to improve reten-
tion of basic pollutants at low pH.
Organic Acids Found in the Environment
The trace determination of EDTA in environmental
water samples is an example of an analytical chal-
lenge } one in which the analyte readily chelates with
metals, is very water soluble and is an organic acid.
EDTA is commonly used in the clean-up of radioac-
tivity and heavy metal wastes, and is also found in the
environment as a detergent and water softening
agent. Chelation of EDTA with toxic metals facili-
tates the migration of these hazardous materials from
ground dumps into a water-soluble state where they
can be transported into lakes, rivers and streams. The
sample preparation technique of choice for EDTA is
SPE since it improves detection limits compared with
other techniques and can be fully automated. The
lowest detection limits (0.15
g L\1, Rve times lower
than previously reported methods using GC-mass
spectrometry (MS) and HPLC) have been obtained
using capillary electrophoresis (CE) with ion-spray
tandem mass spectrometry (MS-MS) for selective
detection.
The sample preparation of EDTA from water sam-
ples (5 mL) involves conversion of all free and
chelated EDTA present into the nickel EDTA chelate
by adding 100
L Ni(NO3)2 at a pH from 7}9. The
pH is then adjusted to about 3.0 using about 12
L of
9% formic acid. This sample is added to precon-
ditioned strong anion exchange solid-phase extrac-
tion cartridges (SPEC (Ansys Diagnostics glass
Rbre disc cartridges). Wash solvents used (in order)
are water (adjusted to pH 3.0 with formic acid),
water (neutral pH), and methanol. Finally,
the NiEDTA is eluted using a solution of 50 mM
triSuoroacetic acid, 1 mM bromothymol blue and
5% methanol. The eluate is evaporated to dryness,
reconstituted in 0.1% ammonium hydroxide,
evaporated to dryness again, and then reconstituted
in 30
L water for analysis. This extract is then
analysed by CE-MS. The strongly acidic elution
solvent dissociates the NiEDTA complex, while
bromothymol blue displaces the remaining NiEDTA
from the disc. Reconstitution in ammonium
hydroxide facilitates the re-complexation of the
NiEDTA.
4138 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES
Food Applications
Pesticides in Food
Food applications using SPE present complexities
that are not encountered in water extractions. Sub-
stances such as apples, lettuce, tomatoes and straw-
berries have tissue components that must be removed
before extraction, and the analytes within the tissue
Suids must be made available for extraction or re-
moval prior to analysis. Multiple pesticides are com-
monly analysed in food crops. One popular multi-
residue screening technique is the Luke II method, in
which a crop sample (100 g) is homogenized with
a water-miscible solvent (acetone). However, other
crop materials that have solubility in acetone are also
extracted. The solvent and water from the crop are
then Rltered and the Rltrate subjected to a series of
liquid}liquid partitioning extractions. The resulting
mixture is subjected to two or more SPE packed
cartridge (or disc) clean-up steps using sorbents with
varying selectivity to remove co-extracted materials
while pesticides pass through. Use of SPE techniques
within this method allows for reduced solvent use and
improved throughput.
A variation of this approach described uses SPE
discs with reversed-phase sorbents (SDB-RPS disc
stacked on top of a carbon disc; 3M Company) to
capture the pesticides, rather than the co-extracted
substances that often use normal-phase sorbents. This
disc procedure is as follows. A 100 g sample of each
crop material is mechanically blended with 100 mL
acetone. The puree is Rltered through a glass Rbre
Rlter and three 10-mL aliquots (10 g crop equivalent,
wet weight) of each Rltrate are transferred to centri-
fuge tubes, and the volume is reduced under nitrogen
to about 5 mL. Water is added to adjust volume to
15 mL. SPE discs are conditioned with acetone, fol-
lowed (in order) by ethyl acetate, methanol, then
water. Samples are Rltered through each disc. When
the entire sample has been extracted, the discs are
removed and inverted, so that SDB-RPS is on the
bottom and carbon on the top. Elution is accomp-
lished with 2 mL acetone, followed by two successive
5-mL aliquots of ethyl acetate. The eluent is dried
using anhydrous sodium sulfate, then concentrated
by evaporation to 5 mL volume, and analysed by
GC-ECD. The combination of SDB-RPS and carbon
sorbent chemistries for the extraction is superior to
reversed-phase bonded silica sorbents to extend the
range of attraction to the more polar pesticides with
high water solubility, namely dimethoate, o-methoate
and methamidophos. SDB-RPS contains the SDB
chemistry but because of the nature of sulfonic acid
groups bonded on the SDB surface it captures
cationic moieties also. Carbon is used to capture
analytes not retaining on SDB-RPS. By reversing the
order of sorbents for elution, the pesticides never
come in contact with the carbon and are quantitat-
ively recovered.

-Agonists in Cattle Meat

2-Agonists (e.g. clenbuterol, brombuterol, ma-
buterol and mapenterol), originally developed for
treatment of chronic obstructive pulmonary diseases
in humans, have been misused as a repartitioning
agent in the fattening of cattle. When cattle are
treated, residues may remain in the meat and liver. In
order to monitor regulatory bans on use of these
drugs in cattle, samples are removed at slaughter-
houses and analysed for the presence of these illegal
growth promoters. Urine is the matrix most com-
monly used for the analysis of these
-agonist drugs.
Solid-phase extraction has been shown to be an effec-
tive technique for these drugs, using reversed-phase
or mixed-mode sorbents (containing both reversed
phase and cation exchange functionalities). The SPE
procedure adds 1 mL 0.5 M potassium phosphate buf-
fer pH 4.0 to 5 mL urine, followed by centrifugation.
A mixed-mode sorbent bed is conditioned with meth-
anol, water, then 0.1 M potassium phosphate buffer
pH 4.0. The sample is loaded onto the cartridge,
followed by a wash solvent of 70% methanol
in water. After drying the cartridge, elution is ac-
complished with ethanol}n-hexane}ammonium
hydroxide (70 : 25 : 5, v/v/v) in two sequential por-
tions. Solvent is evaporated under nitrogen and heat
and reconstituted in 25% acetonitrile in water for
HPLC analysis.
Biotechnology Applications
Puri\cation and Fractionation of Proteins and
Peptides
Proteins are signiRcant components of most physio-
logical samples. It is often important to measure
very low concentrations of speciRc peptides in biolo-
gical Suids for diagnosis of disease states and to
investigate physiological roles of certain peptides.
Examples include examining the role of atrial nat-
riuretic peptide in cardiovascular disease, studying

-endorphins involved in the neurochemistry of the
brain, and isolating lymphokines to monitor their
effect on immune system regulation. The quantiRca-
tion of a peptide such as casein in milk products is an
application in the food area requiring isolation and
puriRcation. Solid-phase extraction is commonly used
as a preliminary puriRcation step to remove cross-
reacting or interfering materials in sample matrices
before analysis.
 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES
One common approach to purifying hydrophilic
proteins or peptides is to fractionate crude pro-
teinaceous extracts and remove hydrophobic pro-
teins. Proteins above 15}20 000 molecular weight are
usually too large and cannot easily enter the pores of
typical 60}100 A> bonded silica particles. Thus, these
large proteins pass unretained through reversed-
phase sorbents and can be effectively eliminated from
the analyte in this manner. The procedure is as fol-
lows. The sample is loaded onto the SPE column in an
aqueous buffer, then washed with dilute aqueous
acid (e.g. 0.1% triSuoroacetic acid, TFA) to remove
salts and low molecular weight contaminants. Pept-
ide analytes of interest are eluted with a mixture
of organic solvent (acetonitrile or propanol) in
water containing 0.1% TFA. The SPE sorbents
useful for peptide retention, in increasing order
of hydrophobicity, can generally be stated as cy-
ano(C2(phenyl(cyclohexyl(C8(C18. Very
polar peptides should be isolated using sorbents with
a high retention ability such as C8 or C18. Very
hydrophobic peptides could be isolated with a less
retentive sorbent such as cyano or C2. Medium and
highly hydrophobic peptides could be efRciently iso-
lated and fractionated with phenyl and cyclohexyl
sorbents.
SPE based on ionic interaction of proteins can
efRciently fractionate peptide mixtures into neutral,
acidic and basic pools. In ion exchange chromatogra-
phy, adsorption of proteins depends on the proteins
isoelectric point relative to the column pH. Proteins
with a high isoelectric point will bind tightly to a ca-
tion exchange column in the presence of a low pH
and a low salt concentration. Proteins with a high
isoelectric point will bind tightly to a cation exchange
column in the presence of a low pH and a low salt
concentration. Proteins with a low isoelectric point
will bind tightly to an anion exchange column in the
presence of a high pH and a low salt concentration.
Hydrophobic interaction chromatography uses a high
salt concentration to induce an interaction between
hydrophobic regions of a protein and a weakly hydro-
phobic column packing. In all three cases, elution
of the bound proteins can be achieved using a salt
gradient.
On-Line Preconcentration using SPE
The techniques of CE and on-line CE-MS have
been widely documented for the analysis of thera-
peutically important peptides of diverse nature.
A limitation of CE is that it works best for small
sample volumes (typically (50 nL for a 50
m inter-
nal diameter capillary). This volume restriction
leads to a poor concentration limit of detection
(CLOD) when compared with typical HPLC and
4139
LC-MS systems. The incorporation of a membrane
preconcentration cartridge (containing SPE in a mem-
brane format) in-line with the CE capillary has al-
lowed the introduction of much larger sample
volumes (e.g. 100
L), lowering the CLOD. Typical
materials used for the preconcentration are reversed-
phase sorbents, such as SDB and C18 bonded silica.
This technique has allowed the analysis of bio-
molecules present in complex matrices, such as pro-
teins in aqueous humour.
Pharmaceutical Applications
Bacitracin Extraction from a Pharmaceutical
Ointment
Bacitracin ointment, an oily pharmaceutical formula-
tion, is a mixture of at least nine antibiotic polypep-
tide complexes. These peptides are very polar and
soluble in water and ethanol, but not in acetone or
hexane. They can be separated from the ointment
base by adding chloroform to the sample matrix, and
the polar peptide antibiotics are adsorbed to a polar
diol SPE column. Upon addition of the matrix to the
column, the nonpolar solvent and ointment products
pass through. A wash of chloroform removes poten-
tially interfering components of the formulation.
Antibiotics are removed from the sorbent using 0.1N
HCl; protons from the acid displace the drugs from
the hydroxyl groups on bonded silica surface. This
technique can be useful for other drug substances by
optimizing the SPE conditions for the properties of
the drug and excipients, and selecting the appropriate
sorbent and eluent systems.
Analysis of Aspirin Content in Tablets
Aspirin can be analysed for content in tablets by using
a mixed mode sorbent containing both anion ex-
change and reversed-phase characteristics. Polysorb
MP-2 (Interaction) polymer is a cross-linked vinyl-
pyridine. At low pH the polymer is protonated and
exhibits anion exchange and reversed-phase proper-
ties. At high pH, the polymer is neutralized and ex-
hibits only reversed-phase properties. The sorbent is
conditioned with acid/organic 10/90 (v/v) to induce
polymer ionization. After sample loading, the sorbent
is washed with 20}50% acetonitrile/water to neutral-
ize the sorbent bed prior to elution of bound aspirin.
Elution is accomplished with acetonitrile 30%
NH4OH}30 mM diammonium sulfate monohydrate
(6 : 2 : 1, v/v/v). The polymer becomes neutral at this
basic pH'12.9 and aspirin remains ionized, disrupt-
ing its interaction with the sorbent and eluting. Salt
acts as counteranion to further assist in the elution of
aspirin.
4140 III / SOLID-PHASE EXTRACTION WITH CARTRIDGES
Clinical Applications
Catecholamines from Plasma and Urine
Catecholamines (e.g. dopamine, epinephrine and
norepinephrine) are of clinical interest for their role in
neurochemistry as diagnostic indicators of phaeo-
chromocytoma. These dihydroxylated amines are
commonly analysed by HPLC with electrochemical
detection. Many different SPE sorbents have been
reported for their sample preparation. Alumina par-
ticles (about 50 mg) are added to plasma and buffer
in a suspension, followed by centrifugation and sub-
sequent elution from alumina. Reversed-phase C18
has also been used, as well as phenylboronic acid and
strong cation exchange (SCX). The urine analysis of
catecholamines examines metabolites such as vanil-
lylmandelic acid and homovanillic acid. Typically,
solid-phase extraction (SPE) uses SCX sorbent to pro-
vide cleaner chromatograms, either alone or in addi-
tion to alumina or phenylboronic acid.
Lipids
The extraction of lipids, including phospholipids,
fatty acids, cholesterol, cholesteryl ester, and tri-
glycerides in serum has been accomplished using
polar SPE sorbents such as silica and aminopropyl.
The sample matrix is extracted with a nonpolar sol-
vent such as chloroform, and this extract is passed
through a preconditioned polar aminopropyl sorbent
for attraction of analytes by hydrogen bonding and
weak ion exchange mechanisms. Neutral lipids are
eluted with chloroform}propanol (2:1, v/v), fatty
acids are eluted with 2% acetic acid in diethyl ether,
and phospholipids are eluted with methanol. The
neutral lipid fraction is evaporated and reconstituted
in hexane. The hexane mixture is then passed through
a second amino SPE column. The cholesteryl esters
are eluted with hexane, with the second column in
series with the Rrst column to trap cholesterol, which
elutes with triglycerides. Triglycerides are eluted with
hexane}diethyl ether}methylene chloride (89 : 1 : 11,
v/v). The two amino columns are separated and cho-
lesterol is eluted from both } di- and mono-glycerides
elute from the upper amino column. Cholesterol is
eluted with 5% ethyl acetate in hexane, diglycerides
are eluted with 15% ethyl acetate in hexane and
monoglycerides are eluted with chloroform}meth-
anol (2 : 1, v/v).
Drugs in Tissues
The majority of reported methods for drug extraction
involve plasma, serum, urine or other polar Suids that
can easily pass through SPE cartridges. However,
blood and tissue extraction applications have not
been reported with much frequency in the published
literature. The extraction of drugs from tissues, such
as liver, kidney, intestine, brain, muscle and adipose
tissue is important for the forensic toxicologist, in
particular, since urine and blood are not available in
post-mortem cases. Concentrations of drugs in tissues
are also of great interest to researchers investigating
the deposition of drugs in certain tissues (e.g. oph-
thalmic drug delivery into the eye and delivery of
antidepressant drugs into the brain).
Drug extraction from tissues involves Rrst homo-
genization of the tissue with aqueous solution. After
homogenization, an enzyme digestion (e.g. Carlsberg
subtilisin, lipase or protease) step and/or protein pre-
cipitation can be used, followed by centrifugation.
Sometimes a small percentage (10}20%) of organic
solvent (e.g. methanol) is added to the water or buffer
for homogenization, and the solution is passed
through an SPE cartridge. Alternatively, 100% meth-
anol or acetonitrile can be used and, after centrifu-
gation, the solvent is evaporated, reconstituted in
aqueous solution or buffer, and passed through an
SPE cartridge. In addition to typical SPE sorbents,
diatomaceous earth is another choice for the analyst;
it facilitates a liquid}liquid extraction by attracting
analytes to the particles to increase surface area avail-
able for extraction when an organic solvent is passed
through the diatomaceous earth column. The use of
high-Sow SPE columns is now a reality owing to
larger particle-size sorbents in columns, typically
100}120
m particle sizes instead of 40}60
m.
Conclusion
Solid-phase extraction has been demonstrated to be
a reliable and cost-effective technique for the selective
isolation and concentration of a wide range of
analytes and sample matrices, and offers many im-
provements over traditional techniques such as
liquid}liquid extraction. Some of the classic applica-
tions for SPE include environmental trace enrichment
of organic pollutants, extraction of pesticides and
growth promoters from foods, puriRcation of pep-
tides, drug analysis in pharmaceutical dosage forms
and clinical applications for drugs in physiological
matrices. Its ability to solve sample preparation prob-
lems has been well documented in the literature over
the past two decades. There is a wide choice of
sorbents for SPE, including nonpolar, polar, ion ex-
change and mixed mode chemistries, providing the
analyst with the selectivity necessary to obtain clean
extracts for analysis. SPE can be used either manually
or with greater throughput using automated worksta-
tions. The introduction of new sorbents with more
selective modes of attraction, novel product formats

such as the SPE disc, and the proliferation of auto-
mated techniques for performing the extractions,
ensure that SPE will continue to be a preferred tech-
nique for sample preparation in many different ana-
lytical disciplines.
See also: II/Extraction: Solid-Phase Extraction. III/Solid-
Phase Extraction with Discs.
Further Reading
Berrueta LA, Gallo B and Vicente F (1995) A review of
solid-phase extraction: basic principles and new devel-
opments. Chromatographia 40(7/8): 474}483.
Guzman NA, Park SS, Schaufelberger D et al. (1997) Re-
view: new approaches in clinical chemistry: on-line
analyte concentration and microreaction capillary elec-
trophoresis for the determination of drugs, metabolic
SOLID-PHASE EXTRACTION WITH
DISCS
C. F. Poole, Wayne State University, Detroit,
MI, USA
Copyright ^ 2000 Academic Press
Introduction
Solid-phase extraction is a well-established technique
for the isolation, concentration and matrix simpliRca-
tion of analytes in samples with unfavorable proper-
ties for direct analysis by the best available approach.
Extraction is achieved using a particulate sorbent
packed into columns of short length (sometimes
called cartridges) or immobilized in the form of
a thin disc, referred to generically as disc techno-
logy. Since the same sorbent chemistry is used for the
extraction step and liquid desorption for the elution
step in both approaches, the two techniques differ
only in format. On an evolutionary scale, solid-phase
extraction using short columns was introduced as
a laboratory-scale technique in the late 1970s and
came to prominence in the 1980s. Disc technology,
by comparison, was Rrst introduced in 1989, and is
still evolving as a competitive technique to short
packed columns. Simply stated, disc technology
should be viewed as an alternative approach to per-
forming solid-phase extraction with additional bene-
Rts and capabilities derived from the difference in
format. Whereas packed columns are easily prepared
in the laboratory for evaluating new sorbent chemis-
tries and evaluating sampling properties, discs, so far,
III / SOLID-PHASE EXTRACTION WITH DISCS 4141
intermediates and biopolymers in biological Suids. Jour-
nal of Chromatography B 697: 37}66.
Hurst WJ (1996) Bonded solid}phase extraction for the
sample preparation of food materials. Seminars in Food
Analysis 1(1): 3}9.
Koester CJ and Clement RE (1993) Analysis of drinking
water for trace organics. Critical Reviews in Analytical
Chemistry 24(4): 263}316.
Krishnan TR and Ibraham I (1994) Solid phase extraction
technique for the analysis of biological samples. Journal
of Pharmaceutical and Biomedical Analysis 12(3):
287}294.
Scheurer J and Moore CM (1992) Solid-phase extraction of
drugs from biological tissues } a review. Journal of
Analytical Toxicology 16: 264}269.
Thurman EM and Mills MS (1998) Solid-Phase Extraction:
Principles in Practice, pp. 161}195. New York: John
Wiley & Sons.
have only been produced in a manufacturing setting.
Consequently, sorbent selection and device optimiza-
tion have been restricted by market-driven consider-
ations. As a consequence, the main applications of
disc technology are generally narrowly focused on the
needs of large volume users more so than is the case
for conventional short packed columns.
Disc Formats
Solid-phase extraction discs are available in different
styles and sizes. Particle-loaded membranes (Em-
poreTM discs) contain 8}12
m sorbent particles ho-
mogeneously distributed in a web of short poly
(tetraSuoroethylene) (PTFE) Rbrils. These are formed
into 0.5-mm thick discs with diameters from 4 to
96 mm. They are Sexible and superRcially resemble
Rlter paper discs. They are used with some supporting
structure such as a frited glass Rlter or porous plastic
support. The discs contain about 90% by weight of
sorbent with the balance being the PTFE microRbrils.
Some characteristic physical properties are indicated
in Table 1. Particle-loaded membranes are also avail-
able in a syringe barrel format similar to conventional
short packed column-sampling devices. In this case,
the sorbent bed contains particles of a larger dia-
meter, about 50
m, in thicker discs, about 1.0 mm,
sealed into the base of a 4 mm (1 mL), 7 mm (3 mL),
10 mm (6 mL) and 20 mm diameter (40 mL) open
syringe barrel. These discs have an integral preRlter
consisting of a graded density of poly(propylene)
4142 III / SOLID-PHASE EXTRACTION WITH DISCS
Table 1 Rough guide to the physical properties of solid-phase extraction disc according to disc diameter
Property 4 mm
Surface area (cm2) 0.13
Bed mass (mg)a 4 10
Flow rate (mL min\1) 0.5
Elution volume (mL) 0.15
Typical sample volume (mL) (1
a
Silica-based sorbents.
microRbres on the top sampling surface. They are
recommended as a replacement for short packed col-
umn-sampling devices as well as for processing small
volumes of viscous biological Suids. Particle-embed-
ded glass Rbre discs (SPECTM discs) contain particles
of a narrow size distribution, about 10}30-
m dia-
meter, woven into a glass Rbre-supporting matrix.
The smaller diameter discs are rigid and self-support-
ing but large-diameter discs require a supporting
structure similar to the particle-loaded membranes.
Particle-embedded glass Rbre discs are also available
with a depth Rlter region of 0.1}0.2 mm combined
with a sorbent extraction region of 0.8}0.9-mm
thickness. Laminar discs (SpeedisksT M) contain
10
m sorbent particles in a consolidated 0.5- or
1-mm thick bed (usually) retained by two glass-Rbre
Rlters held in place by screens and a retaining ring in
a preassembled cartridge with a 50-mm diameter
sampling area. This conRguration is designed to pro-
vide high sample Sow rates for extracting large vol-
umes of water.
Solid-phase extraction discs are available as loose
discs for use in Rltration-style apparatus for large
volume samples and in open syringe barrels and car-
tridges for extracting intermediate and small sample
volumes. These forms are compatible with sample
processing using suction, positive pressure, syringe
Rltration and centrifugation. Common sorbents in-
clude octadecylsiloxane- and octylsiloxane-bonded
silica particles, styrene-divinylbenzene porous poly-
mers, porous polymer and silica-based cation and
anion exchangers, mixed mode sorbents, activated
carbon and immobilized crown ether sorbents. Other
sorbents could easily be produced in a disc format if
a sufRcient market to support their production could
be identiRed. Solid-phase extraction discs have been
used primarily for sample preparation prior to
chromatographic analysis. The disc format supports
other, if minor applications at present, such as in situ
detection using radioactivity counting, phosphores-
cence, and matrix-induced laser desorption mass
spectrometry, etc. SPECTM discs can be inserted dir-
ectly into a hole in a glass Rbre thin-layer sheet and
developed in a conventional manner combining re-
covery and separation into a single step. This ap-
7 mm 10 mm 25 mm 47 mm 90 mm
0.38 0.80 4.9 17 64
25 140 500 1850
1.5 3 20 60 250
0.25 0.5 3 10 35
(5 (25 (250 (1000 (5000
proach is commonly used for toxicological screening
of biological Suids.
Advantages of Disc Technology
The change in format from a short packed column to
a disc has some attendant advantages for solid-phase
extraction. These can be brieSy summarized as fol-
lows. The larger cross-sectional area of discs and
decreased pressure drop compared to conventional
column-like sampling devices results in shorter
sample processing times and decreased plugging by
suspended particle matter. This is important in envir-
onmental surveillance programmes, such as the anal-
ysis of surface waters for persistent or toxic
substances, where large sample sizes are common to
obtain adequate detection limits and samples are of-
ten burdened by suspended particle matter. The large
surface area per unit bed mass of the discs facilitates
passive sampling approaches to solid-phase extrac-
tion and related uses as an indirect monitor of biocon-
centration (discussed later).
The use of smaller diameter sorbent particles and
the greater stability of the sorbent bed results in
improved kinetic performance and reduced channel-
ling. This allows the use of a smaller bed mass for
extraction and results in less variation between samp-
ling devices. The reduced bed mass provides cleaner
sample backgrounds and lower interferences by min-
imizing nonspeciRc matrix adsorption. Smaller bed
masses allow miniaturization of sampling devices for
convenient handling of small sample sizes together
with smaller elution volumes for analyte recovery.
They also facilitate novel sampling approaches such
as in-vial elution and on-disc derivatization (dis-
cussed later).
Kinetic Characteristics
Forced-Sow planar chromatography has been used to
study the kinetic properties of octadecylsiloxane-
bonded silica particle-loaded membranes and par-
ticle-embedded glass Rbre discs (Table 2). The total
porosity of the discs is about 0.50, comprised mainly
of interparticle porosity (about 0.40) with a large
 III / SOLID-PHASE EXTRACTION WITH DISCS 4143

Table 2 Kinetic properties of octadecylsiloxane-bonded silica solid-phase extraction discs

Property Particle-loaded membranes Particle-embedded glass fibre discs

Total porosity 0.52}0.54 0.51

Interparticle porosity 0.37}0.48 0.47
Intraparticle porosity 0.06}0.15 0.04
Specific permeability (10\14 m\2) 2.2}2.5 8.4
Flow resistance parameter 1000}1250 900}1000
Apparent particle size (
m) 5.8}7.7 15.3
Nominal pore diameter (nm) 6 8
Minimum plate height (
m) 56
Optimum mobile phase velocity (mm s\1) 0.13
Coefficients for Knox equation
A 3.75 (0.5}1.5 for columns)
B 1.72 (1.0}4.0 for columns)
C 1.54 (0.05}0.7 for columns)

fraction of the particle pore volume inaccessible to
the mobile phase. This is not unusual for porous silica
sorbents with a high loading of bonded phase restrict-
ing access to the pore volume. It is also compatible
with the desire for a high sorption capacity and is not
necessarily an undesirable feature. The speciRc per-
meability and Sow resistance parameter support the
hypothesis that the disc structure is homogeneous and
devoid of through pores (holes), an essential require-
ment for a thin sampling medium. The apparent par-
ticle size for the particle-loaded membranes, about
7
m, and particle-embedded glass Rbre discs, about
15
m, are in reasonable agreement with the manu-
fracturers claims given the assumptions used in calcu-
lations for converting the pressure}Sow relationships
to particle size values. The particle-loaded membrane
provides an optimum plate height of 56
m at a linear
velocity of 0.13 mm s\1. A 0.5-mm disc of 47-mm
diameter will provide between 4 and 9 theoretical
plates over the Sow rate range of 5}100 mL min\1
with a maximum value at 13 mL min\1. Compared
to typical slurry-packed columns, the contribution of
both Sow anisotropy and resistance to mass transfer
to the plate height are unusually large for the particle-
loaded membranes, and while the bed may have an
homogeneous structure, it does not have an ideal
kinetic structure. Fortunately, large plate numbers are
not required for efRcient extraction.
Disc Selection
The parameters of interest in selecting a disc for
a particular application are size, sorbent chemistry
and sample capacity. Referring to Table 2, large-dia-
meter discs are used for processing large sample vol-
umes to improve sample throughput. Small-
diameter discs are used for processing small sample
volumes and to recover analytes in a small solvent
volume to eliminate the need for solvent evaporation
prior to analysis. Small-diameter discs are frequently
used in clinical, forensic and pharmaceutical analysis
and large-diameter discs in environmental analysis.
The same sorbents used for conventional solid-phase
extraction are generally used for disc extraction, but
since discs are used for a narrower range of applica-
tions at present, the number of frequently used
sorbent types is smaller (Table 3). The majority of
applications proposed to date involve sampling of
aqueous solutions. Octadecylsiloxane-bonded silica,
poly(styrene-divinylbenzene) and activated carbon
sorbents are used for general reversed-phase samp-
ling; porous polymer and silica-based cation and an-
ion exchangers are used for isolating ionizable and
ionic compounds; and cheltaing ligands for the selec-
tive isolation of metals (particularly precious metals
and radionuclides). Discs with different sorbent
chemistries can be stacked on top of each other and
analytes recovered in a single elution or as groups after
physically separating the discs. Stacked discs of the
same or varied sorbent chemistry are useful for ex-
tracting complex samples containing compounds that
differ signiRcantly in polarity or ionization, for achiev-
ing larger breakthrough volumes, and as an approach
for reducing interferences by the selective sorption of
the matrix by one disc and the analytes by the other.
Stacked discs of a reversed-phase and cation ex-
change type are a suitable alternative to mixed-mode
sorbents for isolating drugs and their metabolites
from biological Suids. Discs containing porous poly-
mer sorbents have been proposed for air sampling,
particularly as a replacement for poly(urethane)
foams, but have not been widely adopted.
In recent years, disc-based solid-phase extraction
has become a popular technique for sample cleanup
in ion chromatography and capillary electrophoresis.
The interfering ions, often at relatively high concen-
trations, can mask, broaden or change the migration
time of the ions of interest. The Novo-CleanTM discs
4144 III / SOLID-PHASE EXTRACTION WITH DISCS
Table 3 Sorbent selection for disc applications
Sorbent
Octadecylsiloxane- and octylsiloxane-bonded silica
Poly(styrene-divinylbenzene)
Activated carbon
Mixed mode
Sulphonic acid and quaternary amine ion exchangers
Crown ethers
contain a sulfonic acid-functionalized resin in the
hydrogen, silver or barium forms housed in a car-
tridge adapted for syringe Rltration. The hydrogen
form is used to remove hydroxide and carbonate ions
from samples by neutralization. The silver form is
used to remove excess halide ions from samples by
formation of insoluble silver halide salts that precipi-
tate in the disc matrix. Figure 1 shows an example of
the detection of nitrate, Suoride and phosphate in
a hydrochloric acid digest of a paper coating by
capillary electrophoresis. The barium form is used to
remove sulfate through the formation of insoluble
barium sulfate.
Sample Processing
A generic outline for processing aqueous samples
using disc technology is presented in Table 4. This
General applications
Octadecylsiloxane-bonded sorbents are widely used for reversed-phase
sampling by non-specific sorption (and are more widely used than octyl-
siloxane-bonded sorbents). There are many applications to the extrac-
tion of non-polar and moderately polar pesticides (all classes),
herbicides, phthalate and adipate esters, polycyclic aromatic com-
pounds, food additives, pharmaceutical compounds, hydrocarbons and
grease, etc. from aqueous solution. A large number of approved
methods for water analysis.
Used for compounds poorly extracted by octadecylsiloxane-bonded sil-
ica sorbents because of high water solubility such as polar pesticides,
herbicides, phenols and pharmaceutical compounds. Lightly sulfonated,
acylated and hydroxymethylated polymers claimed to provide higher
recovery of neutral polar compounds because of better surface compati-
bility with aqueous samples.
Applications similar to poly(styrene-divinylbenzene) but less frequently
used. Methods proposed for triazine herbicides, some polar pesticides
and N-nitrosodialkylamines.
Usaully co-bonded octadecylsiloxane and benzenesulfonic acid groups
(or sorbent mixtures) used predominantly in clinical toxicology and
pharamaceutical analysis for the simultaneous isolation of drugs and
their metabolites. A number of established methods for drugs of abuse
(marijuana and cocaine metabolites, amphetamines, phencyclidine and
opiates) in biological fluids.
Selective isolation of ionic and easily ionizable compounds. Many
methods for acidic herbicides and pesticides in water and basic drugs in
biological fluids. Porous polymer sorbents are stable over the whole pH
range. Sulphonated cation exchange polymers in the hydrogen, silver or
barium forms used for cleanup of samples analysed by ion chromatogra-
phy and capillary electrophoresis.
Silica or other support with tethered mixed oxygen}nitrogen donor cryp-
tands used for the selective isolation of metals by molecular recognition
mechanisms. Commonly used for the isolation of precious metals (Pd,
Pt, Rh) and radionuclides (Cs, Sr, Pb) from high concentrations of other
metals to determine environmental burden and for dating geochemistry
samples. Other applications include the removal of base metal impurities
(Bi, Sb, Fe, Pb, Bi, Cu, Hg) from refinery streams, plating baths, etc.
can be rescaled for different disc and sample sizes
using the data in Table 1. The sampling process be-
gins with decontamination of the disc by rinsing with
organic solvent to remove impurities followed by
solvent conditioning to facilitate effective and repro-
ducible sorption of the analytes. Before the sample is
added to the disc the conditioning solvent is rinsed
from the disc with water to avoid premature break-
through of analytes. For large sample volumes,
a small amount of organic solvent (1}5% v/v) is
added to aqueous samples to maintain a constant
sample velocity. This usually has little inSuence
on the breakthrough volumes except for porous poly-
mer sorbents with a low degree of crosslinking.
The selective adsorption of organic solvent by the
polymer changes the sorbent volume and selectivity,
resulting in changes in the breakthrough volume of
up to an order of magnitude when either methanol,
 III / SOLID-PHASE EXTRACTION WITH DISCS 4145

Increasing the ionic strength of the sample (e.g.

Figure 1 Separation of trace concentrations of nitrate (1), fluor-
ide (2) and phosphate (3) anions (1}10 p.p.m.) from a hydrochlo-
ric acid digest of a paper coating without treatment (A) and after
disc cleanup using a Novo-Clean IC-Ag disc (B) by capillary
electrophoresis. (Reproduced with permission from Saari-Nord-
haus R and Anderson JM (1995) Membrane-based solid-phase
extraction as a sample clean-up technique for anion analysis by
capillary electrophoresis. Journal of Chromatography A 706:
563}569. Copyright Elsevier Science.)
2-propanol, acetonitrile or tetrahydrofuran are used
as sample processing solvents (Table 5). These results
are not in any way artifactual and are adequately
predicted by the solvation parameter model for the
breakthrough volumes. A sample processing solvent
is not usually required for small sample volumes.
Viscous samples may be easier to handle if diluted
with either water or an organic solvent to increase
their sample-processing rate. Samples containing
a signiRcant amount of particulate matter should be
either preRltered, processed with a Rlter aid (cover the
surface of the disc with 1 cm of 40-
m glass beads),
processed by using a combination of a disc with an
integral depth Rlter, or processed after adjusting the
pH to dissolve soluble particles. Weak acids and
bases are extracted by ion exchange or by ion sup-
pression reversed-phase sampling at a suitable pH.
adding sodium chloride) may improve the extraction
of neutral compounds in reversed-phase extractions.
The drying stage is important because water retained
by the disc (and the disc support structure) contami-
nates the elution solvent and may cause difRculty if
the solvent is to be reduced in volume or the analytes
analysed by gas chromatography. Water may also
reduce the efRciency of the elution step, particularly
for elution solvents only partially miscible with
water. Discs are usually dried by suction, but other
possibilities include freeze}drying or desiccation over
a drying agent with or without vacuum. For silica-
based sorbents, a masking agent such as triethylamine
(1% v/v) is sometimes added to the eluting solvent to
increase the recovery of basic compounds strongly
retained by interaction with silanol groups. Super-
critical Suid extraction with carbon dioxide has been
used to recover analytes isolated by disc extraction
from water. In this case, the discs were used for
preconcentration and media exchange since super-
critical Suid carbon dioxide provides a poor extrac-
tion medium for aqueous solutions.
Passive Sampling and
Bioconcentration
Passive sampling involves the extraction of organic
compounds by immersion of the sampling disc in
aqueous samples as opposed to Rltering samples
through the extraction disc. This process is conve-
nient but little explored. One reason is the slow equi-
librium of the extraction process even for stirred
solutions. The uptake of organic compounds in a stir-
red solution depends on the size of the disc, time and
the degree of mixing. Uptake by the disc may never be
complete in a reasonable time but greater than 80%

Table 4 Generic guide for sample processing using solid-phase extraction discs. In this example, a 47-mm disc and a 1-litre water
sample are used for illustration

Decontamination With disc installed in filtration apparatus (or cartridge) rinse with 10 mL of solvent (acetonitrile) by allowing the
solvent to soak into the membrane for a few minutes and then remove by suction.
Condition As above but using 10 mL of methanol. Before the last drop of methanol has been sucked through the disc, add
10 mL of deionized water. The disc should not be allowed to suck dry until after the sample has been
processed. It may be necessary to break the vacuum for ease of manipulating solutions.
Sample The sample containing 1% (v/v) methanol is passed through the disc with a suitable reservoir in place. Filter aid
or a prefilter may be required for samples with a heavy burden of particulate matter. The sample is processed at
a flow rate between about 50 and 100 mL min\1 using a vacuum of about 10}20 mmHg.
Drying Bulk water is removed from the disc and sampliing appratus by sucking air through the disc under full vacuum
for about 3 min. Volatile compounds may be lost.
Recovery The analytes are recovered by passing two 5-mL volumes of acetonitrile through the disc. The first volume of
acetonitrile is allowed to soak into the disc for a few minutes and then gently sucked through the disc. Without
letting the disc run dry, the second volume of solvent is added to the disc and sucked through the disc. The disc
is then allowed to suck dry.
4146 III / SOLID-PHASE EXTRACTION WITH DISCS

Table 5 Breakthrough volumes (cm3) of some organic compounds on porous polymer particle-loaded membrane discs with 1% (v/v)
organic solvent in water

Compound Acetonitrile Tetrahydrofuran 2-Propanol Methanol

Anisole 575 300 1750 1300

Acetanilide 95 25 100 75
2-Phenylethanol 150 25 100 200
Heptan-1-ol 2700 300 2700 1000
Propyl propanoate 1400 200 1250 750

extraction of pesticides in an agitated 500 mL sample
using a 47-mm diameter particle-loaded membrane in
24 h was reported. Octadecylsiloxane-bonded silica
discs have been explored as surrogate models for
bioconcentration. This process is referred to as bio-
mimetic extraction and is used to estimate the selec-
tive uptake of organic compounds by aquatic species.
Biomimetic extraction results in the preferential con-
centration of the more hydrophobic compounds from
water and is theorized to provide a more realistic
approach to assessing environmental effects of sol-
uble organic compounds than results obtained by
total extraction techniques. Measurements are made
using passive disc sampling under conditions that
avoid sample depletion to emulate the sorption of
organic compounds by aquatic species. This is a rela-
tively new approach to toxicity assessment and re-
quires further work to establish a satisfactory
relationship between the sorption properties of the
disc and those of target aquatic species.
exchange discs catalyse the rate of alkylation of acid
herbicides, surfactants and pesticides using a solution
of an alkyliodide as the derivatizing reagent. An
example is shown in Figure 2.
Preservation and Storage
Samples or their extracts frequently have to be stored
between sampling and analysis. Compounds differ in
their stability to various storage conditions. The
degradation or loss of pesticides and other organic
compounds stored in water is mainly due to hydroly-
sis, biodegradation, photolysis and evaporation.
Solid-phase extraction discs provide a convenient
storage medium for compounds recovered from natu-
ral waters. General results indicate that discs provide
equivalent or greater stability to the storage of pesti-
cides in water at 43C containing inhibitors to reduce

In-Vial Elution and Derivatization

The in-vial elution method reduces the amount of
organic solvent used for recovery and eliminates tedi-
ous sample-preparation steps compared to conven-
tional methods. This technique is an equilibrium-
based approach for the recovery of extracted
analytes. After extraction, the dried disc is placed in
an autosampler vial and covered with a few mL of
solvent. Recovery depends on the selection of the
solvent, temperature and equilibration time. An equi-
libration time of 4 h, or overnight, at room temper-
ature is sufRcient in many cases to provide acceptable
recovery ('90%) and method precision as well as
being compatible with automated sample analysis in Figure 2 Use of in-vial elution and derivatization for the deter-
mination of chlorinated acid herbicides in an extract from a spiked
gas and liquid chromatography. Addition of a de-
((10
g L\1 each) river water sample by gas chromatography.
rivatizing reagent to the solvent used for desorption A 500-mL sample was processed through a 25-mm strong anion-
allows both steps to be performed simultaneously in exchange disc, the disc dried by suction, sealed in an auto-
the same vial without removal of the disc. The vial sampler vial with 1 mL of acetonitrile and 0.2 mL of methyl iodide
can be sealed and heated to enhance the rate of and heated at 803C for 1 h, and excess methyl iodide removed by
evaporation. (Reproduced with permission from Field JA and
derivatization. Common reactions employed so far
Monohan K (1996) Chlorinated acid herbicides in water by strong
include trimethylsilylation and alkylation for gas anion-exchange disc extraction and in-vial elution and derivatiz-
chromatographic analysis but other reagents should ation. Journal of Chromatography A 741: 85}90. Copyright
be equally applicable. It was demonstrated that ion Elsevier Science.)

biodegradation. One of the primary mechanisms of
disc degradation was identiRed as hydrolysis resulting
from contact with water remaining in the disc after
drying by suction. Stability was increased by desicca-
tion of the discs. Stability has to be judged on an
individual compound basis but discs provide a more
efRcient and convenient medium for extract preserva-
tion. Their low weight and small size make them
a useful medium for shipping.
Automation and Multi-Well Formats
The introduction of solid-phase extraction discs oc-
curred at a time when the automation of common
laboratory process was very much in vogue for im-
proving the quality of laboratory information and to
reduce operating costs. Most disc extraction pro-
cesses can be automated. Series-coupled Rltration
devices are popular for parallel processing of environ-
mental water samples. Disc sampling devices in car-
tridge format are compatible with the automated and
robotic workstations in use for automated sample
preparation using conventional packed column for-
mats. Multi-well microtitre plates have been adapted
to disc extraction by incorporating the disc in the
bottom of the well and using automated liquid-hand-
ling systems for high throughput extractions. This is
achieved by a combination of automated parallel
processing and eluent collection without volume re-
duction in a matching multi-well plate compatible
with standard autoinjection devices. This approach
meets the high sample throughputs demanded of cur-
rent applications in bioanalysis and combinatorial
chemistry for drug discovery. Short precolumns con-
taining several discs stacked in series or packed with
a spiral of particle-loaded membrane have been
used for online extraction combined with liquid
chromatography for pesticide monitoring of drinking
and surface water samples. This is not an area
that extraction discs dominate, and although useful
and successful, clear advantages over short packed
columns have not been demonstrated.
Future Developments
Most developments in disc technology are recent
enough that they are still to some extent regarded as
novel rather than routine laboratory tools. They are
expected to continue to replace conventional short
packed columns in those applications where the disc
format has speciRc advantages. In addition, disc tech-
nology has already deRned new opportunities for
expansion of solid-phase extraction in passive samp-
ling, storage and preservation of extracts, Reld samp-
ling, and as an indirect indicator of bioconcentration.
III / SOLID-PHASE EXTRACTION WITH DISCS 4147
Since discs are not laboratory-made products, the
variety of disc chemistries available is linked to the
identiRcation of a viable market for a commercial
product. For the present, this has tended to decouple
efforts to promote disc technology from modern re-
search on sorbent chemistry for the next generation of
solid-phase extraction products.
See also: II/Extraction: Solid-Phase Extraction. III/Sorbent
Selection for Solid-Phase Extraction.
Further Reading
Barcelo D and Alpendurada MF (1996) A review of storage
and preservation of polar pesticides in water samples.
Chromatographia 42: 704}712.
Beals DM, Britt WG, Bibler JP and Brooks DA (1998)
Radionuclide analysis using solid-phase extraction discs.
Journal of Radioanalytical and Nuclear Chemistry 236:
187}191.
Degel F (1996) Comparison of new solid-phase extraction
methods for chromatographic identiRcation of drugs in
clinical toxicological analysis. Clinical Biochemistry 29:
529}540.
Fernando WPN, Larrivee ML and Poole CF (1993)
Investigation of the kinetic properties of particle-loaded
membranes for solid-phase extraction by forced Sow
planar chromatography. Analytical Chemistry 65:
588}595.
Izatt RM, Bradshaw JS and Bruening RL (1996) Accom-
plishment of difRcult chemical separations using solid
phase extraction. Pure and Applied Chemistry 68: 1237-
1241.
Lingeman H and Hoekstra-Oussoren SJF (1997) Particle-
loaded membranes for sample concentration and/or
clean-up in bioanalysis. Journal of Chromatography
B 689: 221}237.
Mayer ML, Poole CF and Henry MP (1995) Sampling
characteristics of octadecylsiloxane-bonded silica par-
ticle-embedded glass Rbre discs for solid-phase extrac-
tion. Journal of Chromatography A 695: 267}277.
Michor G, Carron J, Bruce S and Cancilla DA (1996)
Analysis of 23 polynuclear aromatic hydrocarbons from
natural water at the sub-ng L\1 level using solid-phase
extraction and mass-selective detection. Journal of
Chromatography A 732: 85}99.
Pichon V, Charpak M and Hennion M.-C (1998) Multi-
residue analysis of pesticides using new laminar extrac-
tion discs and liquid chromatography and application to
the French Priority List. Journal of Chromatography
A 795: 83}92.
Plumb RS, Gray RDM and Jones CM (1997) Use of re-
duced sorbent bed mass and disc membrane solid-phase
extraction for the analysis of pharmaceutical com-
pounds in biological Suids, with application in the 96-
well format. Journal of Chromatography B 694:
123}133.
Poole SK and Poole CF (1996) InSuence of solvent effects
on the breakthrough volume in solid-phase extraction
4148 III / SOLID-PHASE MATRIX DISPERSION: EXTRACTION
using porous polymer particle-loaded membranes. Ana-
lyst 120: 1733}1738.
Tomkins BA and Griest WH (1996) Determination of N-
nitrosodimethylamine at part-per-trillion concentration
in contaminated ground and drinking waters featuring
carbon-based membrane extraction discs. Analytical
Chemistry 68: 2533}2540.
SOLID-PHASE EXTRACTION: SORBENT
SELECTION
See III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION
SOLID-PHASE MATRIX DISPERSION:
EXTRACTION
S. A. Barker, Louisiana State University,
Baton Rouge, LA, USA
Introduction
Matrix solid-phase dispersion (MSPD) is a patented
analytical process for the preparation, extraction and
fractionation of solid and/or viscous biological sam-
ples prior to instrumental or other forms of analysis.
MSPD involves the direct mechanical blending of
samples with standard solid-phase extraction (SPE)
bonded-phase solid support materials. In this process,
the bonded-phase support acts as both an abrasive to
produce disruption of sample architecture and as
a bound solvent that assists in accomplishing sample
disruption. The sample is dispersed over the surface
of the bonded-phase support material, producing
a unique mixed-character phase for conducting target
analyte isolation. This process has been applied to the
isolation of a wide range of drugs, pollutants and
other compound classes from a variety of sample
matrices. The factors affecting the use of MSPD and
its applications for sample preparation, extraction
and fractionation are addressed.
Development of MSPD
The classical application of all forms of liquid
chromatography requires that the sample be applied
in a liquid state to the head of the column. Thus, in
order to accommodate the use of solid-phase extrac-
tion (SPE) columns and discs in the development of
an analytical procedure, methodology must also be
Verbruggen EMJ, Van Loon WMGM, Tonkes M, Van
Duijn P, Seinen W and Hermens JLM (1999) Biomimetic
extraction as a tool to identify chemicals with high
bioconcentration potential: an illustration by two fra-
grances in sewage treatment plant efSuents and surface
waters. Environmental Science and Technology 33:
801}806.
developed to render the sample and target analytes
contained therein into a liquid, relatively non-vis-
cous, particulate-free and homogeneous condition.
While some biological Suids are readily obtained in
this form, most biological samples do not start out
being directly applicable to SPE. This presents the
analyst with some rather unique opportunities to ap-
ply or develop the best process for rendering a sample
into a form compatible with liquid chromatography.
Indeed, the most difRcult and complex samples to
analyse are the solids and semi-solids that are derived
from a biological origin. Such samples may be ob-
tained from animal or vegetable material and consist
of a non-homogeneous array of fat and/or other
tissues, Rbre, pulp, etc.
For these and other reasons, the preparation of
biological samples for SPE requires an initial disrup-
tion of the gross architecture of the sample. This step
in the process assures access to all of the components
of a sample and initiates the necessary homogeneity
required for analysis. Disruption and homogenization
also provide a larger overall surface area, generating
greater access to solvents and reagents used for
analyte isolation, which is the next step in the pro-
cess. Samples that are by their nature already reason-
ably homogeneous, such as milk, fruit juices, plasma,
etc., are less complicated in this regard but may be
too viscous or contain particulates that could hinder
rapid SPE extraction and fractionation. For such
samples, dilution, Rltration and/or centrifugation
often solve these problems.
Classical processes for solid or semi-solid sample
disruption usually involve one or various combinations
 III / SOLID-PHASE MATRIX DISPERSION: EXTRACTION
of the following: mincing, shredding, grinding, pul-
verizing and/or pressurizing of the sample. All of
these approaches accomplish the basic requirement of
disrupting sample architecture. This initial disruption
may be followed or accompanied by the addition of
solvents, acids, bases, buffers, abrasives, salts, deter-
gents, chelators, etc., in an effort to more completely
disrupt cellular and architectural composition and
initiate the extraction and fractionation of various
sample components from the analyte(s) of choice.
Unfortunately, the creation of often intractable emul-
sions is often a consequence of these actions and
repeated centrifugation, re-extraction and sample
manipulation may be required to render the sample
suitable for application to an SPE column. Applica-
tion of this entire process strives to obtain the
analyte(s) in solution, free from solids, emulsions
or suspensions, reducing the solid sample to a
homogenous liquid extract.
In 1989, a technique that remedied many of the
complications of dealing with solid samples in their
subsequent extraction using solid-phase materials
was developed. This was accomplished by literally
combining the sample directly with the bonded-phase
solid support. This process, designated as matrix
solid-phase dispersion (MSPD), was observed to
simultaneously accomplish several steps in the more
classical approach to sample preparation and SPE
extraction/fractionation. A sample (tissue, fruit, etc.)
is placed in a glass mortar containing a bonded-phase
solid support material, such as octadecylsilyl silica
(C18). The solid support and sample are manually
blended together using a glass pestle. In this process,
the irregularly shaped silica-based solid support
serves as an abrasive that promotes disruption of the
samples general architecture, and also acts as
a bound solvent which appears to further disrupt the
sample by inducing lysis of cell membranes. The
blended material is packed into a column suitable for
conducting sequential elution with solvents and the
blended sample components, and their distribution in
the bonded-phase support provides a new phase that
exhibits a unique character for performing sample
fractionation (Figure 1).
Examination of blended tissues by scanning elec-
tron microscopy (SEM) showed that sample architec-
ture had been completely disrupted and that sample
matrix components had apparently been evenly dis-
tributed over the surface of the bonded phase/sup-
port, forming an observable layer. The thickness of
this new phase of dispersed matrix is approximately
100
m, similar to that of some micelle or membrane
bilayers. Indeed, it appears that this is what occurs in
the MSPD process. The sample is distributed over the
surface of the bonded phase as a function of interac-
4149
tions with the support and the bonded phase. The
tissue matrix components themselves, form a layered
phase consisting of support/lipophilic bonded-
phase/sample lipids and a further distribution of
sample-associated compounds arranged in and on
this new phase based on their own relative polarities.
MSPD Versus Other Forms
of Chromatography
MSPD is physically and functionally different from
classical SPE in the following ways: (1) MSPD is
a process that accomplishes sample disruption and
dispersal onto particles of very small size, providing
an enhanced surface area for subsequent extraction,
whereas sample disruption must be conducted as
a separate step in preparing samples for SPE; (2) SPE
samples must be in a liquid form, relatively free of
solids and of moderate viscosity before addition to
the column, while MSPD directly handles solid or
viscous liquid samples; and (3) The physical and
chemical interactions of the components of the sys-
tem are greater in MSPD and different, in some re-
spects, from those seen in classical SPE or other forms
of liquid chromatography.
In applying the MSPD process to a sample, the
interactions observed between the individual compo-
nents and the target analyte(s) involve (1) the sample
components with the solid support, (2) the sample
components with the bonded phase, (3) the
analyte with the solid support, (4) the analyte with
the bonded phase, (5) the analyte with the dispersed
sample components, (6) all of the above interacting
with the elution solvent(s) and their sequence of addi-
tion, and (7) the dynamic interactions of all of the
above occurring simultaneously. Nonetheless, gen-
eral chemical principles involved in conducting SPE
and other forms of chromatography are also operable
in applying MSPD. Thus, the chemical composition
and characteristics of the solid support and bonded-
phase are expected to affect the retention and elution
of the analytes. These same properties will also apply
to the dispersed sample components and the unique
phase that is created.
Factors Affecting MSPD
Extraction and Fractionation of
Samples
To date, only the use of silica-based support materials
has been reported for MSPD. The use and effect
of synthetic polymer-based solid supports is a
subject for further study, particularly supports that
possess unique surface and pore chemistries, such as
 4150
III / SOLID-PHASE MATRIX DISPERSION: EXTRACTION

Figure 1 A schematic representation of a MSPD procedure.

 III / SOLID-PHASE MATRIX DISPERSION: EXTRACTION
hydrophobic interaction supports. For silica-based
materials, however, studies have shown that the pore
size is of minor importance in MSPD. This effect
could vary with the sample and should be considered,
however.
The effect of average particle-size diameter has also
been examined. As may well be expected, the use of
very small particle sizes (3}10
m) leads to extended
solvent elution times for a MSPD column, requiring
excessive pressures to obtain adequate Sow. How-
ever, 40-
m particle size materials (60 As average pore
diameter) have been used most frequently and quite
successfully. It has been reported that a blend
of silicas possessing a range of particle sizes
(40}100
m) also work quite well, and such materials
also tend to be less expensive.
Depending on the application, non-endcapped ma-
terials or materials having a range of carbon loading
(8}18%) may also be used. It is a simple matter to
examine these variables for a given application and
should be considered for obtaining the best extraction
efRciency and the cleanest sample.
The bonded phase will, of course, play a pivotal
role. Depending on the polarity of the phase chosen,
rather dramatic effects on the results may be observed
and a range of available phases should be examined
for each application. It has been reported that in
applications requiring a lipophilic bonded phase,
C18 and C8 can be used interchangeably and that
the best ratio of sample to bonded-phase mat-
erial is 1 to 4. Most applications have employed
lipophilic bonded-phase (C18) materials, blending
2.0 g of solid support with 0.5 g of sample. The best
ratio is, of course, dependent on the application and
should be examined as a variable during method
development. Ratios of bonded-phase to sample less
than 4 : 1 have been used successfully and samples
have been scaled up to 2 g from the typical 0.5 g used
in most MSPD procedures, blended with a propor-
tionately greater amount of solid support.
In general, it has been observed that the isolation of
more polar analytes from biological samples is assist-
ed by the use of polar phases (cyanopropyl, for
example) and less polar analytes by less polar phases.
This would be expected based on retention character-
istics of compounds from classical SPE.
Preconditioning of the materials used for MSPD
greatly enhances analyte recovery, as has been estab-
lished with SPE. However, in MSPD it also appears to
speed the process of sample blending and dispersal by
breaking the surface tension differences that may
exist between the sample and bonded-phase solid
support. As with SPE, washing or rinsing the solid
support materials prior to use also eliminates con-
taminants from the Rnal eluates obtained for analysis.
4151
MSPD column character may also be altered by
modifying the matrix prior to or during the blending
step. Several extraction studies designed to isolate
a variety of different drugs have shown that addition
of chelating agents, acids, bases, etc. at the time of
blending affects the distribution and elution of target
analytes from the sample. The elution proRle of
matrix components is likewise affected. This effect
can be predicted from basic chemistry and applied in
MSPD during sample blending and/or by alteration
of the elution solvent composition. This effort to
increase or suppress ionization of analytes and
sample components greatly affects interactions of
speciRc analytes with the blended phase and the elut-
ing solvent(s). Thus, the use of matrix modiRcation
should be considered, as in SPE, as a possible variable
to be controlled for attaining reproducible and efR-
cient MSPD extraction.
The correct choice of elution solvents and the se-
quence of their application to a column is of utmost
importance to the success of MSPD or SPE fractiona-
tion of samples. Elution solvent sequence and com-
position can be varied to obtain the best analytical
results, when attempting to isolate the analyte or
further clean the column of interfering substances
with each solvent step. The nature of MSPD columns
and the enhanced degree of interaction permit isola-
tion of different polarity analytes or entire chemical
classes of compounds in a single solvent or in differ-
ing polarity solvents passed through the column. This
characteristic makes MSPD amenable to conducting
multiresidue isolation and analysis on a single
sample. In this regard, true gradient elution of
a MSPD column has not been reported to date but
should, nonetheless, prove applicable to the complete
fractionation of samples.
It has been observed that, in an 8-mL elution of
a 2-g C18 column blended with 0.5 g of sample, most
of the target analytes eluted in the Rrst 4 mL, or in
approximately one column volume. This will, of
course, vary with each application and with appropri-
ate solvent selection but should be examined to re-
duce the use of solvent and the elution of other
potentially interfering components.
The solid support, bonded phase and solvent elu-
tion sequence are all critical in performing MSPD, as
they are in SPE. However, they may prove less inSu-
ential overall than the effect of the sample matrix
itself. It should be kept in mind that in MSPD the
sample itself is dispersed throughout the column. In
contrast, much of the sample is retained only in the
Rrst few millimeters of the column bed in SPE. In
MSPD, the sample matrix components cover much of
the bonded-phase support surface, creating a new
phase that is dependent on their interactions with the
4152 III / SOLID-PHASE MATRIX DISPERSION: EXTRACTION
solid support and bonded phase, interactions that
give the MSPD column its character. This new phase,
in association with the analytes distribution and own
interactions with it, are perhaps the most important
controlling factors.
This matrix effect has been seen by anyone who
conducts chromatography, particularly on large num-
bers of samples that are less than pristine. Repeated
sample injection can create a build-up of non-volatil-
ized or non-eluting sample components at the head of
a gas chromatography (GC) or liquid chromatogra-
phy (LC) column, introducing a new phase. This
new phase may subsequently affect the stability, elu-
tion and retention character of target analytes as they
come into contact with it. The discontinuity of phases
within the analytical column may lead to peak tailing,
the formation of shoulders or multiple peaks for
a single analyte. It can lead, over time, to complete
loss of analytes that interact strongly with the new
phase. This matrix effect, or deposition of sample
components as an additional phase, is incorporated
throughout the column in MSPD. The dispersion of
components establishes a new level of, and consist-
ency in, equilibrium distribution of analytes that is
fundamentally different from that seen from limited
or discontinuous phases.
Another interesting aspect of this effect is that the
analytes tend to co-elute in fractions that are not
wholly consistent with predicted solubility behav-
iour. This observation may underscore a further
unique property of MSPD. Elution of a sample is
designed to remove the target analyte(s) but, even in
SPE, one simultaneously fractionates and co-elutes
some of the sample components. In MSPD the total
amount of sample components present is much
greater. In performing mass-balance experiments, it
has been observed that the entire sample, minus a few
per cent of what appeared to be denatured macro-
molecules and connective tissues, can be eluted from
an MSPD column. Thus, sequential elution of liver
tissue blended with C18 recovers 98% of sample tri-
glycerides in hexane and 98% of phospholipids and
steroids in dichloromethane. Sugars and polyols are
found in the acetonitrile fraction and phosphorylated
sugars in water. The presence and concentration
of eluted proteins follows the sequence methanol'
water'acetonitrile'ethyl acetate. Approximately
7% of the total mass of the sample remains on the
column, consisting of connective tissues and de-
natured macromolecules, including DNA and related
nucleotide polymers. Thus, the MSPD process has
been used to disrupt otherwise rugged and difRcult-
to-lyse bacteria, and to simultaneously perform frac-
tionation of the sample for subsequent analysis to
identify unique cellular components. These results
point to the fact that many of the unique elution
properties of MSPD are due, not only to interactions
of target analytes with the dispersed matrix but, also,
an association of the matrix with the target analyte as
speciRc classes of matrix components are eluted.
Therefore, co-elution of target analyte in association
with a particular class of matrix component, which is
simultaneously interacting with the other materials
remaining on the column, seems to be an important
factor in the overall chromatographic character of the
MSPD process.
The eluates obtained from an MSPD column are
often amenable to immediate instrumental analysis,
being adequately clean for direct injection. This is, of
course, dependent on the sensitivity of the method
and whether concentration is required before the
analysis. However, in the majority of cases additional
steps are required to address the removal of the co-
eluting matrix components described above. In
a number of cases this additional clean up of the
sample has been accomplished by the use of second-
ary solid-phase materials. For example, bonded-
phase material of the same polarity or even of
different character than that used in blending can be
packed at the bottom of the MSPD column (co-
column). Alternatively, the MSPD column may be
eluted directly onto a standard SPE column or disc
material for the purposes of conducting further
sample clean up and/or analyte concentration. Sim-
ilarly, the eluate may be evaporated and reconstituted
for application to another form of chromatography,
analysis by immunoassay, etc. Other techniques em-
ployed to conduct MSPD eluate clean up have in-
volved the use of classical liquid}liquid extraction,
conducted on a small scale, prior to analysis.
While the process of preparing a MSPD column is
currently a manual one, the steps of elution, collec-
tion and concentration of fractions, etc. are amenable
to automation.
Conclusions
A list of MSPD applications for the isolation of a range
of compounds from a variety of matrices is shown in
Table 1. This list illustrates the rather generic charac-
ter of MSPD for performing the extraction of a variety
of matrices for a number of compounds. In most cases,
MSPD has been found to provide equivalent results to
older ofRcial methods conducted by more classical
countercurrent and/or SPE techniques. Further, it has
been rather consistently observed that MSPD requires
95% less solvent and can be performed in 90% of the
time of such classical methods. The use of smaller
sample sizes, combined with lower solvent consump-
tion, purchase and disposal, make MSPD competitive
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4153

Table 1 Applications of MSPD to the analysis of residues in See also: II / Extraction: Solid-Phase Extraction; Solvent
various matrices Based Separation. III / Solid-Phase Extraction with Cart-
ridges. Sorbent Selection for Solid-Phase Extraction.
Analyte(s) Matrix

Aminoglycosides Bovine kidney Further Reading

Benzimidazoles Beef liver
Benzimidazoles Swine muscle Barker SA (1992) Application of matrix solid-phase
Beta-agonists Bovine liver dispersion (MSPD) to the extraction and subsequent
Carbofuran Corn analysis of drug residues in animal tissues. In: Agarwal
Chloramphenicol Milk VK (ed.) Analysis of Antibiotic Residues in Food Prod-
Chlorsulfuron Milk ucts of Animal Origin, pp. 119}132. New York: Plenum
Chlorsulon Milk Press.
Clenbuterol Bovine liver
Barker SA and Floyd ZE (1996) Matrix solid-phase disper-
Furazolidone Chicken muscle
Furazolidone Milk
sion (MSPD): implications for the design of new
Furazolidone Swine muscle bonded-phase surface chemistries. In: Pesek JJ, Matyska
Ivermectin Fish muscle MT and AbuelaRya RR (eds) Chemically ModiTed Sur-
Ivermectin Milk faces: Recent Developments, pp. 66}71. Cambridge,
Ivermectin Bovine liver UK: Royal Society of Chemistry.
Moxidectin Bovine tissues Barker SA and Long AR (1992) Tissue drug residue extrac-
Nicarbazin Animal tissues tion and monitoring by matrix solid-phase dispersion
Oxolinic acid Catfish (MSPD)-HPLC analysis. Journal of Liquid Chromatog-
Oxytetracycline Catfish muscle
raphy 15: 2071}2089.
PCBs Fish
Barker SA, Long AR and Hines ME (1993) The disruption
Pesticides Beef fat
Pesticides Catfish muscle and fractionation of biological materials by matrix
Pesticides Crayfish solid-phase dispersion. Journal of Chromatography
Pesticides Fish 629: 23}34.
Pesticides Fruit, vegetables Barker SA, Long AR and Short CR (1989) Isolation of drug
Pesticides Milk residues from tissues by solid-phase dispersion. Journal
Pesticides Oysters of Chromatography 475: 353}361.
Sulfa drugs Chicken tissues Crouch MC and Barker SA (1997) Analysis of toxic wastes
Sulfadimethoxine Catfish muscle in tissues from aquatic species: applications of matrix
Sulfadimethoxine Catfish, plasma solid-phase dispersion. Journal of Chromatography 774:
Sulfonamides Infant formula
287}309.
Sulfonamides Milk
Sulfonamides Salmon muscle US Patent C 5 272 094. Issued 21 December (1993)
Sulfonamides Swine muscle A bonded-phase matrix dispersion and extraction
Sulfonamides Eggs process. Isolation of drugs, and drug residues, from
Tetracyclines Milk biological specimens, and tissues (Dr Steven A Barker;
co-patent applicant, Dr Austin R Long, Louisiana State
University).
with such methods on several levels and should be
considered as an alternative when pursuing new ana-
lytical methodology. This is especially the case for
solid or semi-solid biological materials.

SOLID-PHASE MICROEXTRACTION

Analysis of drugs in biological samples is growing in

importance owing to the need to understand the
Biomedical Applications therapeutic and toxic effects of drugs and to continue
the development of more selective and effective
H. Kataoka, Okayama University, Tsushima, drugs. Furthermore, the screening and conRrmation
Okayama, Japan of abused drugs in body Suids is important for the
H. L. Lord and J. Pawliszyn, detection of potential users of drugs and the control
University of Waterloo, Ontario, Canada
of drug addicts following withdrawal therapy. Simul-
Copyright ^ 2000 Academic Press taneous analysis of these drugs in biological samples
4154 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications
is required in many circumstances, such as clinical
control for diagnosis and treatment of diseases, dop-
ing control, forensic analysis and toxicology. Al-
though high efRciency instruments have been
developed, most analytical instruments cannot
handle the sample matrix directly. Therefore, sample
preparation is very important to achieve a practical
and reliable method for the analysis of complex ma-
trices such as biological samples. In general, over
80% of analysis time is spent on sampling and sample
preparation steps such as extraction, concentration
and isolation of analytes. However, previous sample
preparation techniques, such as liquid}liquid extrac-
tion and solid-phase extraction, have their problems.
These techniques are generally time-consuming and
require large volumes of samples and solvents. For
example, a long sample preparation time limits the
number of samples that can be analysed, and multi-
step procedures are prone to loss of analytes. Further-
more, the use of a large amount of solvent inSuences
trace analysis, and also causes environmental pollution
and health concerns. Ideally, sample preparation tech-
niques should be fast, easy to use, inexpensive and
compatible with a range of analytical instruments.
Solid-phase microextraction (SPME), developed by
Pawliszyn and co-workers in 1990, is a new sample
preparation technique using a fused-silica Rbre that is
coated on the outside with an appropriate stationary
phase. The analyte in the sample is directly extracted
onto the Rbre coating. The method saves preparation
time, solvent purchase and disposal costs, and can
improve the detection limits. It has been used routine-
ly in combination with gas chromatography (GC) and
GC/mass spectrometry (GC/MS), and successfully
applied to a wide variety of compounds, especially for
the extraction of volatile and semi-volatile organic
pollutants from water samples. SPME was also intro-
duced for direct coupling with high performance
liquid chromatography (HPLC) and LC/MS in order
to analyse weakly volatile or thermally labile com-
pounds not amenable to GC or GC/MS. The
SPME/HPLC interface, equipped with a special de-
sorption chamber, is utilized for solvent desorption
prior to HPLC analysis, instead of thermal desorption
in the injection port of the GC. Moreover, a new
SPME/HPLC system known as in-tube SPME, was
recently developed using an open-tubular fused-silica
capillary column as the SPME device in place of
the SPME Rbre. In-tube SPME is suitable for automa-
tion, and automated sample handling procedures not
only shorten the total analysis time, but also usually
provide better accuracy and precision relative to
manual techniques.
In this article, we review SPME techniques
coupled with various analytical instruments and the
applications of these techniques to drug analysis. The
review consists of two main parts. In the Rrst part,
general aspects of SPME techniques are surveyed for
Rbre and in-tube SPME methods coupled with vari-
ous instruments. In the second part, applications of
the SPME methods in drug analysis are considered
according to the drug type.
SPME Techniques Coupled with
Various Analytical Instruments
Fibre SPME
The Rbre SPME device consists of a Rbre holder and
Rbre assembly with built-in Rbre inside the needle. In
Rbre SPME, analytes are extracted directly from the
sample onto a polymeric stationary phase coated on
the Rbre. When the Rbre is inserted into the sample,
the target analytes partition from the sample matrix
into the stationary phase until equilibrium is reached.
Two types of Rbre SPME techniques can be used to
extract analytes: headspace SPME and immersion
SPME. In headspace SPME, the Rbre is exposed in the
headspace of gaseous, liquid or solid samples. In
immersion SPME, the Rbre is directly immersed in
liquid samples. The Rbre with concentrated analytes
is then transferred to an instrument for desorption,
followed by separation and quantiRcation. Head-
space and immersion SPME techniques can be used in
combination with any GC, GC/MS, HPLC and
LC/MS system. The process of the Rbre SPME/GC
method is shown in Figure 1.
In Rbre SPME, the amount of analyte extracted
onto the Rbre depends on the polarity and thickness
of the stationary phase coating on the Rbre, extrac-
tion time, and the concentration of analyte in
a sample. In general, volatile compounds require
a thick polymer coat and a thin coat is effective for
semi-volatile compounds. Extraction of analytes is
also typically improved by agitation, addition of salt
to the sample, changing the pH, and temperature.
Although full equilibration is not necessary for accu-
rate and precise analysis by SPME, consistent extrac-
tion time and other SPME parameters are essential.
Furthermore, it is important to keep a consistent vial
size and sample volume. In general, immersion SPME
is more sensitive than headspace SPME for analytes
predominantly present in the liquid. On the other
hand, headspace SPME is suitable for the extraction
of more volatile compounds. Extractions from biolo-
gical samples by headspace SPME exhibit lower back-
ground than extractions obtained by immersion
SPME. Because the headspace and immersion SPME
techniques differ in kinetics, both approaches should
be evaluated to optimize Rbre SPME conditions for
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4155

Figure 1 Schematic illustration of headspace and immersion SPME/GC methods. (A) Headspace SPME; (B) direct immersion
SPME.

analytes. Fibre SPME techiques in combination with
GC or GC/MS are unsuitable for the extraction of less
volatile or thermally labile compounds. Thus derivat-
ization approaches are frequently used to extract po-
lar compounds from biological samples. Four types of
derivatization techniques in combination with SPME
are implemented. Direct derivatization in the sample
matrix is similar to well-established approaches used
in solvent extraction. Analytes are extracted by SPME
after derivatization in the vial. For in-coating derivat-
ization with the Rbre-doping method, simultaneous
derivatization and extraction are directly performed
in the Rbre coating by a two-step process: (1) dope
Rbre with derivatization agent and (2) expose doped
Rbre to sample for extraction. This technique can be
used for polar volatile compounds. Another in-coat-
ing derivatization technique is performed by the fol-
lowing two-step process: (1) dope Rbre to sample for
extraction and (2) expose doped Rbre in the head-
space of derivatizing agent. For derivatization in the
injection port, the analyte extracted by SPME is de-
sorbed in a GC injection port and then derivatized
with additional reagent.
The desorption of analyte from the Rbre coating is
performed by heating the Rbre in the injection port of
a GC or GC/MS, or by loading solvent into the
desorption chamber of the SPME/HPLC interface.
EfRcient thermal desorption of an analyte in a GC
4156 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

injection port is dependent on the injection depth,
injector temperature, and exposure time. A narrow-
bore GC injector insert is required to ensure high
linear Sow and the Rbre needs to be exposed immedi-
ately after the needle is introduced into the insert.
Needle exposure depth should be adjusted to place
the Rbre in the centre of the hot injector zone. Desorp-
tion time is determined by the injector temperature
and the linear Sow rate around the Rbre. The HPLC
interface, on the other hand, consists of a six-port
injection valve and a special desorption chamber, and
requires solvent desorption of the analyte prior to
HPLC or LC/MS analysis. A typical SPME/HPLC
interface is shown in Figure 2. The desorption cham-
ber is placed in the position of the injection loop.
After sample extraction, the Rbre is inserted into the
desorption chamber at the load position under am-
bient pressure. When the injector is changed to the
inject position, mobile phase contacts the Rbre, de-
sorbs the analytes, and delivers them to the HPLC
column for separation. Two desorption techniques
can be used to remove the analytes from the Rbre:
dynamic desorption and static desorption. In dy-
namic desorption, the analytes can be removed by
a moving stream of mobile phase. When the analytes
are more strongly adsorbed to the Rbre, the Rbre can
be soaked in mobile phase or other strong solvent for
a speciRed time by static desorption before injection
onto the HPLC column. In each desorption tech-
nique, rapid and complete desorption of analytes us-
ing minimal solvent is important for optimizing the
SPME/HPLC or SPME/LC/MS methods.
In-tube SPME
In-tube SPME using an open-tubular capillary col-
umn as the SPME device was developed for coupling
with HPLC or LC/MS. It is suitable for automation,
and can continuously perform extraction, desorption
and injection using a standard autosampler. With the
in-tube SPME technique, organic compounds in
aqueous samples are directly extracted from the
sample into the internally coated stationary phase of
a capillary column, and then desorbed by introducing
a moving stream of mobile phase or static desorption
solvent when the analytes are more strongly absorbed
to the capillary coating. A schematic diagram of the
automated in-tube SPME/LC/MS system is shown in
Figure 3. The capillaries selected have coatings
similar to those of commercially available SPME
Rbres. The capillary column is placed between the

Figure 2 Schematic of the SPME-HPLC system. (a) Stainless steel (SS) 1/16 inch tee joint; (b) 1/16 inch o.d., 0.02 inch i.d., SS
tubing; (c) 1/16 inch o.d. poly(ether ether ketone) (PEEK) tubing (0.02 inch i.d.); (d) two-piece finger-tight PEEK union; (e) PEEK
tubing (0.005 inch i.d.) with a one-piece PEEK union. (Reproduced with permission from Pawliszyn J (1997) Solid Phase Microextrac-
tion: Theory and Practice. Translated by permission of John Wiley & Sons, Inc. All rights reserved.)
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4157

Figure 3 Schematic of the in-tube SPME/LC/MS system. (A) Load position (extraction phase); (B) injection position (desorption
phase). (Reproduced with permission from Kataoka H, Narimatsu S, Lord HL and Pawliszyn J (1999) Analytical Chemistry 71: 4237.
Copyright American Chemical Society.)
injection loop and the injection needle of the HPLC
autosampler. While the injection syringe repeatedly
draws and ejects sample from the vial under com-
puter control, the analytes partition from the sample
matrix into the stationary phase until equilibrium is
reached. Subsequently, the extracted analytes are dir-
ectly desorbed from the capillary coating by mobile
phase Sow or by aspirating a desorption solvent. The
desorbed analytes are transported to the HPLC col-
umn for separation, and then detected with UV or
a mass selective detector (MSD).
In in-tube SPME, the amount of analyte extracted
by the stationary phase of the capillary column de-
pends on the polarity of capillary coating, number
and volume of draw/eject cycles and the sample pH.
A capillary column 50}60 cm long is optimal for
extraction. Below this level, extraction efRciency is
reduced, and above this level, peak broadening is
observed. In general, complete equilibrium extraction
is not obtained for any of the analytes, because the
analytes are partially desorbed into the mobile phase
during each eject step. The target analytes with higher
K-values need longer equilibration times. Although
an increase in number and volume of draw/eject
cycles can enhance the extraction efRciency, peak
broadening is often observed in this case. The optimal
Sow rate of draw/eject cycles is 50}100
L min\1.
Below this level, extraction requires an inconvenient-
ly long time, and above this level, bubbles form on the
inside of the capillary and extraction efRciency is
reduced. The in-tube SPME technique does not need
a special SPME/HPLC interface for desorption of
analytes. The analytes extracted onto the capillary
coating can be easily desorbed by a moving stream of
mobile phase or desorption solvent when the analytes
are more strongly adsorbed to the capillary coating.
Carryover in the in-tube SPME method is lower or
eliminated in comparison with the Rbre-SPME
method.
Although the theories of Rbre and in-tube SPME
methods are similar, the signiRcant difference be-
tween these methods is that the extraction of analytes
4158 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications
is performed on the outer surface of the Rbre for
Rbre-SPME and on the inner surface of the capillary
column for in-tube SPME. Therefore, with the in-tube
SPME method, it is necessary to prevent plugging of
the capillary column and Sow lines during extraction,
and typically particles must be removed from samples
by Rltration before extraction. On the other hand,
with the Rbre-SPME method, it is not necessary to
remove particles before extraction because they are
removed by washing the Rbre with water before inser-
tion into the desorption chamber of the SPME/HPLC
interface. Another signiRcant difference between in-
tube SPME and manual Rbre-SPME/HPLC is the pos-
sible decoupling of desorption and injection with the
in-tube SPME method. In the Rbre-SPME method,
analytes are desorbed during injection as the mobile
phase passes over the Rbre. On the other hand, in the
in-tube SPME method, analytes are desorbed by mo-
bile phase or by aspirating a desorption solvent from
a second vial, and then transferred to the HPLC
column by mobile-phase Sow. The Rbre-SPME/
HPLC method also has the advantage of eliminat-
ing the solvent peak from the chromatogram, but
peak broadening is sometimes observed because
analytes can be slow to desorb from the Rbre. With
the in-tube SPME method, peak broadening is not
observed because analytes are completely desorbed
before injection.
Biomedical Applications: Drug
Analysis
SPME methods applied to the analysis of various
abused and therapeutic drugs in biological samples
are listed in Table 1, according to the drug type,
sample type, extraction device, extraction mode, and
analytical technique. The SPME methods using 100-

m polydimethylsiloxane (PDMS) Rbres in combina-
tion with GC or GC/MS are widely used for the
analysis of various drugs. The SPME methods
coupled with HPLC or LC/MS are used for the analy-
sis of less volatile or thermally labile drugs. For recent
reviews of some of these methods for drug analysis
see Pawliszyn, Lord and Pawliszyn, Namera et al.,
Junting et al., Kataoka et al. and Sporkert and Pragst
in the Further Reading section.
Amphetamines and Related Compounds
Yashiki and co-workers developed a simple and rapid
method for analysing amphetamine (AM) and meth-
amphetamine (MA) in urine and blood samples by
headspace SPME and GC/MS-selected ion monitor-
ing (SIM). In order to move the analytes into the
headspace, the sample was heated at 803C for 20 min
under K2CO3 or NaOH alkaline conditions. Sub-
sequently, a 100-
m PDMS Rbre was exposed to the
headspace for 5 min, and then inserted into the injec-
tion port of GC/MS for desorption. The method was
twenty times more sensitive than the conventional
headspace method. Lord and Pawliszyn optimized
several extraction parameters for the analysis of AM
and MA in urine samples by headspace SPME/GC-
Same ionization detection (FID). Centini et al.
and Battu et al. reported simultaneous analysis
of amphetamines and their analogues, such as
3,4-methylenedioxyamphetamine (MDA), 3,4-methyl-
enedioxymethamphetamine (MDMA) and 3,4-methy-
lenedioxyethylamphetamine (MDEA), in urine sam-
ples by headspace SPME using a 100-
m PDMS Rbre.
As shown in Figure 4, a clean total-ion chromatogram
is obtained from a urine sample spiked with
100 ng mL\1 of each of the 21 central nervous system
stimulants and extracted by the headspace SPME
method. Koide et al. applied this technique to the
analysis of amphetamines in hair samples.
Degel, Penton, Ishii et al., Makino et al. and
Myung et al. used the direct immersion technique in
order to improve the extraction efRciency and sensi-
tivity. The extraction recoveries of AM and MA by
the immersion SPME method are several times higher
than those by the headspace SPME method. Ugland et
al. reported an SPME technique in combination with
derivatization. After derivatization with alkylchloro-
formate, amphetamines and their methylenedioxy
analogues were analysed by immersion-Rbre SPME/
GC-nitrogen-phosphorus detection (NPD) or GC/MS.
Kataoka et al. developed an in-tube SPME/LC/MS
method for the analysis of amphetamines and their
methylenedioxy analogues using Omegawax (Supelco,
Bellefonte, PA, USA) capillary as the extraction
device. As shown in Figure 5, these drugs spiked into
urine samples were selectively analysed without inter-
ference peaks by SIM-mode detection.
Anaesthetics
Kumazawa et al. developed headspace and direct-
immersion-SPME methods for the analysis of ten lo-
cal anaesthetics in blood samples. These drugs were
extracted with 100-
m PDMS Rbres after deprotein-
ization of the sample with perchloric acid. Heating in
a NaOH and (NH4)2SO4 solution during headspace
SPME gave the best recoveries of the drugs and the
cleanest backgrounds. The recoveries for all drugs in
the sample mixture at neutral pH in the presence of
NaCl were greater than for that of a sample at the
same pH without NaCl (see Figure 6). Although
a small amount of background noise appeared in the
direct immersion-SPME method, the advantage of
using immersion-SPME is that recovery is much bet-
ter than that of headspace-SPME.
Table 1 SPME methods for the analysis of drugs in biological samples

Drugs Specimen Extraction device Extraction Hyphenated Reference

modea analysis

Amphetamines and related compounds

AM, MA Urine 100-
m PDMS fibre HS GC/MS Forensic Sci. Int. (1995) 76: 169.
AM, MA Blood 100-
m PDMS fibre HS GC/MS Forensic Sci. Int. (1996) 78: 95.
AM, MA, MDMA, MDEA Urine 100-
m PDMS fibre HS GC/MS Forensic Sci. Int. (1996) 83: 161.
AM, MA Urine 100-
m PDMS fibre DI GC/MS Clin. Biochem. (1996) 29: 529.
AM, MA Urine 100-
m PDMS fibre DI GC/FID Can. Soc. Forensic Sci. J. (1996) 29: 43.
AM, MA Urine 65-
m PDMS/DVB fibre DI GC/NPD Jpn. J. Forensic Toxicol. (1996) 14: 228.
AM, MA Urine 100-
m PDMS fibre HS GC/FID Anal. Chem. (1997) 69: 3899.
AM, MA Urine 100-
m PDMS fibre D#DI GC/NPD J. Chromatogr. B (1997) 701: 29.
GC/MS
MA Urine 100-
m PDMS fibre DI GC/NPD Chromatography (1997) 18: 185.
GC/MS
AM, MA, MDA, MDMA, MDEA etc. Urine 100-
m PDMS fibre HS GC/MS J. Chromatogr. Sci. (1998) 36: 1.
AM, MA Hair 100-
m PDMS fibre HS GC/NPD J. Chromatogr. B (1998) 707: 99.
AM, MA, MDMA Urine 100-
m PDMS fibre DI GC/MS J. Chromatogr. B (1998) 716: 359.
AM, MA, MDA, MDMA, MDEA Urine 100-
m PDMS fibre D#DI GC/NPD J. Pharm. Biomed. Anal. (1999) 19: 463.
GC/MS
AM, MA, MDA, MDMA, MDEA Urine Omegawax 250 capillary IT LC/MS J. Chromatogr. B (submitted).
J. Anal. Toxicol. (2000) 24: in press.
Anaesthetics
Lidocaine etc. Blood 100-
m PDMS fibre HS GC/FID Jpn. J. Forensic Toxicol. (1995) 13: 182.
Lidocaine etc. Blood 100-
m PDMS fibre DI GC/FID Chromatographia (1996) 43: 59.
Phencyclidine Urine, blood 100-
m PDMS fibre HS GC/SID Chromatographia (1996) 43: 331.
Lidocaine etc. Blood 100-
m PDMS fibre HS GC/MS J. Chromatogr. B (1998) 709: 225.
Lidocaine Urine 100-
m PDMS fibre DI GC/FID Chromatographia (1998) 47: 678.
HPLC/UV

Anorectics
Fenfluramine Urine 30-
m PDMS fibre DI GC/MS J. Microcolumn Sep. (1997) 9: 249.

Antidepressants
Amitriptyline, imipramine etc. Urine 100-
m PDMS fibre HS GC/FID Jpn. J. Forensic Toxicol. (1995) 13: 25.
Amitriptyline Urine 100-
m PDMS fibre DI GC/MS Clin. Biochem. (1996) 29: 529.
Amitriptyline, imipramine etc. Plasma 100-
m PDMS fibre DI GC/NPD J. Chromatogr. B (1997) 696: 217.
Amitriptyline, imipramine etc. Blood 100-
m PDMS fibre HS GC/FID J. Chromatogr. Sci. (1997) 35: 302.
Maprotiline, mianseline, seliptiline Blood 100-
m PDMS fibre HS GC/MS J. Anal. Toxicol. (1998) 22: 396.
III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Antiepileptics
Valproic acid Plasma 100-
m PDMS fibre DI GC/FID J. Chromatogr. B (1995) 673: 299.
4159
Table 1 Continued
4160

Drugs Specimen Extraction device Extraction Hyphenated Reference

modea analysis

Antihistaminics
Diphenhydramine etc. Urine, blood 100-
m PDMS fibre HS GC/FID J. Chromatogr. Sci. (1997) 35: 275.
Ranitidine Urine Omegawax 250 capillary IT LC/MS J. Chromatogr. B (1999) 731: 353.

Antihypertensives
Propranolol etc. Urine, serum Omegawax 250 capillary IT LC/MS Anal. Chem. (1999) 71: 4237.

Antipsychotics
Promazine etc. Urine, blood 100-
m PDMS fibre HS GC/FID Jpn. J. Forensic Toxicol. (1996) 14: 30.

Barbiturates
Barbital etc. Urine 65-
m Carbowax/DVB fibre DI GC/MS J. Chromatogr. A (1997) 777: 275.

Benzodiazepines
Diazepam Urine 100-
m PDMS fibre DI GC/MS Clin. Biochem. (1996) 29: 529.
Diazepam Plasma Solvent-modified 85-
m PA fibre DI GC/NPD J. Chromatogr. B (1997) 689: 357.
Diazepam etc. Urine 65-
m PDMS/DVB fibre DI GC/FID Jpn. J. Forensic Toxicol. (1997) 15: 16.
Diazepam etc. Urine 85-
m PA fibre DI HPLC/UV Chromatography (1997) 18: 244.
Diazepam etc. Urine, serum 65-
m Carbowax/DVB fibre DI GC/FID J. Microcolumn Sep. (1998) 10: 193.
GC/MS
Benzodiazepine metabolites Urine 100-
m PDMS, 85-
m PD fibre DI GC/ECD J. Anal. Toxicol. (1999) 23: 54.

Narcotics and other illicit drugs

Cocaine Urine 100-
m PDMS fibre DI GC/NPD Jpn. J. Forensic Toxicol. (1995) 13: 207.
Meperidine Urine, blood 100-
m PDMS fibre HS GC/FID Jpn. J. Forensic Toxicol. (1995) 13: 211.
Cocaine Urine 100-
m PDMS fibre DI GC/NPD Chromatography (1997) 18: 185.
GC/MS
III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Morphine, heroin etc. Urine 65-
m PDMS/DVB 100-
m HS GC/FID Anal. Chem. (1997) 69: 3899.
PDMS fibre DI
Cannabinoids Saliva 100-
m PDMS fibre DI GC/MS Anal. Chem. (1998) 70: 1788.
Cannabinoids Hair 30-
m PDMS fibre DI GC/MS J. Anal. Toxicol. (1999) 23: 7.

Nicotine
Nicotine , Cotinine Urine 100-
m PDMS fibre HS GC/MS Jpn. J. Forensic Toxicol. (1995) 13: 17.
Nicotine, Cotinine Urine 100-
m PDMS fibre DI GC/FID Clin. Biochem. (1996) 29: 529.

Steroids
Estrone, estradiol etc. Serum 85-
m PA fibre DI#D GC/MS J. High Resolut. Chromatogr. (1997) 20: 171.
Corticosteroids Urine 65-
m Carbowax/ DVB fibre DI LC/MS Rapid Commun. Mass Spectrom. (1997) 11: 1926.

a
HS: headspace; DI: direct immersion; IT: in-tube; D: derivatization.
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4161

Figure 4 Total-ion chromatogram of a urine sample extract spiked with 21 central nervous system stimulants at 1000
g L\1. SPME
conditions: fibre, 100
m PDMS; extraction, at 803C headspace for 10 min with stirring; desorption, exposure for 10 min in GC injection
port. GC/MS conditions: column, PTA-5 (30 m;0.32 mm i.d., 0.5
m film thickness); injector, splitless mode at 2003C; split opening
time, 2 min; oven temperature, programme from 60 to 1203C at 303C min\1, then to 2103C at 53C min\1, and finally to 2803C at
303C min\1 and hold at 2803C for 5 min; transfer line and detector temperature, 2803C; helium flow-rate, 1.3 mL min \1, ionization,
70 eV. (Reproduced with permission from Battu C, Marquet P, Fauconnet AL, Lacassie E and Lacha( tre G (1998) Journal of
Chromatographic Science 36: 1, by permission of Preston Publications, A Division of Preston Industries, Inc.)

Furthermore, Kumazawa et al. reported a method
for analysis of phencyclidine in urine and whole
blood by headspace-SPME and GC with a surface
ionization detector (SID). Watanabe et al. developed
a simple method for analysis of Rve local anaesthetics
in blood samples by headspace SPME using a 100-
m
PDMS Rbre and GC/MS-SIM. Koster et al. reported
direct immersion-SPME methods coupled with GC-
FID and HPLC-UV for the determination of lidocaine
in urine samples. Desorption of the PDMS Rbre in
HPLC is more complicated than the desorption in
GC, because it is dependent on the composition of the
mobile phase or the desorption solvent.
Antidepressants
Kumazawa et al. developed a simple headspace-
SPME method for the analysis of four tricyclic antide-
pressants in urine and whole-blood samples. These
drugs were extracted with a 100-
m PDMS Rbre after
heating at 1003C in the presence of a NaOH solution.
Namera et al. reported a headspace-SPME/GC-MS
method for the analysis of three tetracyclic antide-
pressants in whole-blood samples, and its application
to a medicolegal case of setiptiline intoxication.
Ulrich and Martens developed a direct immersion-
SPME method for the simultaneous analysis of ten
antidepressant drugs and metabolites in plasma sam-
ples, and applied the method to toxicological analysis
after the accidental or suicidal intake of higher doses.
The sample was extracted with a 100-
m PDMS Rbre
for 10 min and the Rbre was exposed in the GC
injection port at 2603C for 1 min after washing in
50% methanol and subsequent drying at room tem-
perature. As shown in Figure 7, these drugs in plasma
samples were selectively analysed by NPD without
interference peaks. However, the recoveries of antide-
pressants from plasma samples were very low due to
the high protein binding of these drugs. The limits of
quantiRcation for these drugs in plasma samples were
90}200 ng mL\1. The sensitivity can be considerably
improved by increasing the extraction time and dilu-
tion of plasma samples with water.
Benzodiazepines
Krogh et al. developed a direct immersion-SPME
method in combination with GC-NPD for the analy-
sis of diazepam in plasma samples. The polyacrylate
(PA) Rbre doped with 1-octanol was used to extract
diazepam from the samples. The solvent-modiRed PA
Rbre was found to be more efRcient in the extraction
of diazepam than the untreated PA and PDMS Rbres.
This technique offers sufRcient enrichment for
bioanalysis, high selectivity, and short sample
preparation time. However, the potential of the
4162 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Figure 5 Total ion and SIM chromatograms obtained from urine samples spiked with amphetamines by in-tube SPME/LC/MS.
(A) Total ion chromatograms obtained from urine and spiked urine samples; (B) SIM chromatograms obtained from spiked urine
sample. Urine sample (10
L) was diluted ten times with water and used for analysis after filtration. Stimulants were spiked at
a concentration of 5 mg mL\1 urine. LC/MS conditions: column, Supelcosil LC-CN (3.3 cm;4.6 mm i.d., 3
m particle size); column
temperature, 253C; mobile phase, acetonitrile/50 mM ammonium acetate (15 : 85); flow-rate, 0.4 mL min\1; fragmentor voltage, 40 V;
ionization mode, positive ESI; SIM ion, m/z"136 (AM), 150 (MA), 180 (MDA), 194 (MDMA) and 208 (MDEA). In-tube SPME
conditions: capillary, Omegawax 250 (60 cm;0.25 mm i.d., 0.25
m film thickness); sample pH, 8.5; draw/eject cycles, 15; draw/eject
volume, 35
L; draw/eject flow-rate, 100
L min\1, desorption solvent, mobile phase. Peaks: 1, AM; 2, MDA; 3, MA; 4, MDMA; and 5,
MDEA. (Reproduced with permission from Kataoka H, Lord HL and Pawliszyn J (2000) Journal of Analytical Toxicology 24: 263, by
permission of Preston Publications, A Division of Preston Industries, Inc.)

solvent-modiRed SPME technique is limited by the Rve benzodiazepines in urine and serum samples.
incompatibility of the SPME coatings with most or- These drugs were efRciently extracted from these
ganic solvents. Luo et al. developed a direct immer- samples with a 65-
m Carbowax/divinylbenzene
sion-SPME method for the simultaneous analysis of (DVB) Rbre under conditions of saturated salt with
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4163

Figure 6 Capillary GC of ten local anaesthetics extracted from human whole blood by use of direct immersion-SPME. (A) The
authentic drugs (50 ng each on column); (B) a drug extract at pH 7 without salt; (C) a drug extract at pH 7 in the presence of 0.5 g NaCl;
(D) a blank extract at pH 7 in the presence of 0.5 g NaCl. The mixture of ten drugs (5
g each) was added to 1 mL of human whole blood.
SPME conditions: fibre, 100
m PDMS; extraction, at room temperature for 40 min with stirring; desorption, 1 min exposure in GC
injection port. GC conditions: column, DB-17 (30 m;0.25 mm i.d., 0.25
m film thickness); column temperature, initially hold at 1003C
for 1 min and increase to 2903C at 103C min\1; injector and detector temperatures, 2503C; He carrier gas flow-rate, 3 mL min\1;
injection, splitless; detector, FID. Peaks: 1, ethyl aminobenzoate; 2, prilocaine; 3, lidocaine; 4, procaine; 5, mepivacaine; 6, tetracaine;
7, bupivacaine; 8, p-(butylamino)benzoic acid-2-(diethylamino)ethyl ester; 9, benoximate; and 10, dibucaine. (Reproduced with
permission from Kumazawa T, Sato K, Seno H, Ishii A and Suzuki O (1996) Chromatographia 43: 59.)

pH 7 and sampling at 453C with agitation, and ana-
lysed by GC-MS.
Guan et al. analysed the metabolites of benzo-
diazepines from acid-hydrolysed urine samples using
a direct immersion-SPME method in combination
with GC-electron capture detection (ECD). The de-
tection limits were 2}20 ng mL\1 for most drugs tes-
ted. Jinno and Taniguchi, however, developed an
SPME method coupled with HPLC for the analysis of
six benzodiazepines in human urine samples. Sensitiv-
ity may be increased by the combination of saturated
salt and weakly alkaline conditions in the extraction
matrix. As shown in Figure 8, a 65-
m PA Rbre was
found to be more efRcient in the extraction of ben-
zodiazepines than a 100-
m PDMS Rbre.
Narcotics and Other Illicit Drugs
Kumazawa et al. and Makino et al. developed direct
immersion SPME methods in combination with
GC-NPD for the rapid analysis of cocaine in urine
4164 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Figure 7 Typical SPME-GLC-NPD chromatograms obtained from (A) blank plasma with internal standard, (B) plasma spiked with
ten antidepressant drugs and metabolites, each 375 ng mL\1, and (C) a sample of a patient after suicidal intoxication with amitriptyline
(amitriptyline, 766 ng mL \1; nortriptyline, 489 ng mL\1. SPME conditions: fibre, 100-
m PDMS; extraction, shaking at 700 rpm for
10 min at 223C; desorption, 1 min exposure in GC injection port. GC conditions: column, DB-1 (30 m;0.32 mm i.d., 0.25
m film
thickness); column temperature, programme from 1403C to 2203C at 203C min\1 and from 2203C to 2703C at 23C min\1; injector and
detector temperatures, 2603C and 3003C, respectively; N2 carrier gas flow-rate, 0.7 mL min\1; injection, splitless; detector, NPD.
Peaks: 1, amitriptyline; 2, trimipramine; 3, imipramine; 4a, cis-doxepine; 4b, trans-doxepine; 5, nortriptyline; 6, mianserine; 7,
desipramine; 8, maprotiline; 9, clomipramine; and 10, desmethylclomipramine. IS, internal standard (chloramitriptyline). (Reproduced
with permission from Ulrich S and Martens J (1997) Journal of Chromatography B 696: 217. Copyright Elsevier Science.)
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4165

Figure 8 Chromatograms of extracted drugs with (A) 100-
m PDMS and (B) 85-
m PA. SPME conditions: extraction, stirring at
840 rpm for 3 h at 603C; desorption, 30 min exposure in desorption chamber. HPLC conditions: column, Siperiorex ODS
(250 mm;1.5 mm i.d.); mobile phase, acetonitrile/water; flow-rate, 100
L min\1; detection, UV at 220 nm. Peaks: 1, nitrazepam; 2,
flunitrazepam; 3, fludiazepam; 4, diazepam; 5, clotiazepam; and 6, medazepam. (Reproduced with permission from Jinno K and
Taniguchi M (1997) Chromatography 18: 244.)

samples. Recovery of cocaine by this technique Other Drugs

using a 100-
m PDMS Rbre was 20%, and the
detection limit was about 12 ng mL\1. Lord
and Pawliszyn applied the SPME/GC-FID method
developed for amphetamines to the analysis of
meperidine, codeine, methadone, morphine and her-
oin in spiked urine samples. Furthermore, Hall et al.
applied an immersion SPME technique to the analysis
of four cannabinoids in human saliva. These drugs
were extracted with a 100-
m PDMS Rbre and ana-
lysed in the range from 5 to 500 ng mL\1 by GC/MS.
Using this method,
9-tetrahydrocannabinol (
9-
THA) was detected in a saliva sample collected
30 min after the subject had smoked marijuana
(Figure 9).
Strano-Rossi and Chiarotti reported an immers-
ion SPME method using a 30-
m PDMS Rbre in
combination with GC/MS for the analysis of canna-
binoids in alkaline hydrolysed hair samples. The
method is also applied to the analysis of other drugs
such as methadone, cocaine and cocaethylene in hair
samples.
Yashiki et al. developed a simple and rapid method
for the analysis of nicotine and its principal metab-
olite, cotinine, in urine samples using headspace
SPME and GC/MS-SIM. Krogh et al. applied a direct
immersion SPME technique to the analysis of the
antiepileptic drug valproic acid in plasma samples.
The drug was extracted with a 100-
m PDMS Rbre
after dialysis of plasma samples, and then analysed by
GC-FID. Seno et al. developed headspace SPME
methods for the simple analysis of Rve phenothiazine
drugs and thirteen diphenylmethane antihistaminic
drugs and their analogues in urine and whole blood
samples. A 100-
m PDMS Rbre was exposed in the
headspace of the sample vial after preheating of the
sample in the presence of NaOH, and the drugs extrac-
ted in the Rbre were analysed by GC-FID. The recove-
ries from blood extracts were lower than those from
urine extracts for all drugs. Hall and Brodbelt reported
a direct immersion SPME method coupled with ion-
trap GC/MS for the analysis of eight barbiturates in
4166 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Figure 9 Chromatograms after performing SPME on human saliva samples prior to and after marijuana smoking. (A) SIM
chromatogram of saliva sample before marijuana smoking; (B) total ion chromatogram of saliva sample after marijuana smoking;
(C) SIM chromatogram of saliva sample after marijuana smoking. SPME conditions: fibre, 100
m PDMS; extraction, immersion for
10 min with stirring; desorption, exposure for 12 min in GC injection port. GC/MS conditions: column, DB-5ms (30 m;0.25 mm i.d.,
0.5
m film thickness); oven temperature, initially hold at 503C for 0.2 min and increase to 2803C at 153C min\1, and finally hold at
2803C for 2 min; transfer line temperature, 2803C; detection, ion trap (electron ionization mode); SIM ion,
9 -THC (m/z"231, 299,
314). (Reproduced with permission from Hall BJ, Satterfield-Doerr M, Parikh AR and Brodbelt JS (1998) Analytical Chemistry 70: 1788.
Copyright American Chemical Society.)

urine samples. A 65-
m Carbowax/DVB Rbre was
suitable for the extraction of these drugs. The detec-
tion limits reached 1 ng mL\1. Okeyo et al. developed
a straightforward method for performing derivatizing
reactions of Rve steroids in situ in SPME Rbres. After
extraction of drugs from serum samples by direct im-
mersion SPME, the drugs extracted on 85-
m PA Rbre
were derivatized in the headspace of the silylating
reagent bis(trimethylsilyl)triSuoro-acetamide, and
then analysed by GC/MS. With derivatization, SPME
and GC analysis can be easily extended to the analysis
of semi- and non-volatile compounds.
Volmer and Hui developed a SPME/LC/MS
method for isolating and analysing eleven cor-
ticosteroids and two steroid conjugates from
urine samples. After extraction in the vial by direct
immersion SPME using 65-
m Carbowax/DVB
Rbre, the drugs extracted in the Rbre were desorbed
in the desorption chamber of the SPME/HPLC
interface, and then analysed by electrospray LC/MS.
As shown in Figure 10, several corticosteroids
and steroid sulfates spiked in urine samples were
selectively analysed, although a minor peak was
observed in the blank control urine in the SIM trace
for cortisone.
Furthermore, Kataoka et al. developed an auto-
mated in-tube SPME/LC/MS method for the deter-
mination of the histamine H2-receptor antagonist
ranitidine in urine samples. The ranitidine in
urine samples was directly extracted into Omegawax
250 capillary by 10 draw/eject cycles of 30
L of
sample at pH 8.5, desorbed from the capillary with
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications 4167

Figure 10 SPME/LC/MS analysis of several corticosteroids and steroid conjugates by time-scheduled SIM. The original urine
sample was spiked at the 20 mg mL\1 level. (A) Blank control urine; (B) spiked urine. LC/MS conditions: column, YMC ODS-AQ (50
mm ;4.0 mm i.d., 3
m particle size); column temperature, 253C; mobile phase, A"100 mM ammonium acetate and B"acetonit-
rile/methanol (50 : 50: #100 mM ammonium acetate), A : B was gradient programmed from 60 : 40 to 20 : 80 in 10 min; flow-rate,
1 mL min \1; fragmentor voltage, 40 V; ionization mode, negative ESI. SPME conditions: fibre, 65
m carbowax/DVB; sample pH, 8.5;
extraction, immersion for 15 min with stirring; desorption, methanol/water (50 : 50) for 5 min. Peaks: 1, estriol-3-sulfate; 2, cortisone; 3,
fludrocortisone; 4, estrone-3-sulfate; 5,6-methylprednisolone; 6, budesonide (epimer B); 7, budesonide (epimer A); IS"internal
standard (niflumic acid) at 20
g mL\1. (Reproduced with permission from Volmer DA and Hui JPM (1997) Rapid Communications in
Mass Spectrometry 11: 1926. Copyright John Wiley & Sons Limited.)

methanol, and then analysed by electrospray LC/MS. Prospective of SPME in Biomedical

Using this technique, nine beta-blockers and metab-
olites in urine and serum samples were also analysed.
These methods were simple, rapid, selective and sen-
sitive, and directly applied to urine samples and
serum samples after ultraRltration. Propranolol (PL)
and its metabolites were successfully detected in the
serum sample of a patient administrated PL (see
Figure 11).
Analysis
The main advantages of SPME are simplicity, rapid-
ity, solvent elimination, high sensitivity, small sample
volume, lower cost and simple automation. Since
1995, a number of SPME methods have been de-
veloped to extract drugs from various biological sam-
ples such as urine, serum, plasma, whole blood, saliva
4168 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications

Figure 11 Total ion and SIM chromatograms obtained from standard propranolol and its metabolites, and a clinical serum sample by
in-tube SPME/LC/MS. (A) Standard solution containing 200 ng mL\1 propranolol (PL), 50 ng mL\1 4-hydroxypropranolol (4-OH-PL)
and 7-hydroxypropranolol (7-OH-PL), 20 ng mL\1 5-hydroxypropranolol (5-OH-PL) and N-desisopropylpropranolol (NDP). (B) Clinical
serum sample (100
L). Serum sample was diluted five times with 1% acetic acid and used for analysis after ultrafiltration. LC/MS
conditions: column, Hypersil BDS C18 (5.0 cm;2.1 mm i.d., 3
m particle size); column temperature, 253C; mobile phase, acetonit-
rile/methanol/ water/acetic acid (15 : 15 : 70 : 1); flow-rate, programme from 0.25 to 0.45 mL min\1 for 20 min run; fragmentor voltage,
70 V; ionization mode, positive ESI; SIM ion, m/z "218 (NDP), 276 (hydroxypropranolols) and 260 (PL). In-tube SPME conditions:
capillary, Omegawax 250 (60 cm;0.25 mm i.d., 0.25
m film thickness); sample pH, 8.5; draw/eject cycles, 15; draw/eject volume,
35
L; draw/eject flow-rate, 100
L min\1, desorption solvent, mobile phase. Peaks: 1, 5-OH-PL; 2, 4-OH-PL; 3, 7-OH-PL; 4, NDP; and
5, PL. (Reproduced with permission from Kataoka H, Narimatsu S, Lord HL and Pawliszyn J (1999) Journal of Analytical Chemistry 71:
4237. Copyright American Chemical Society.)

and hair. The afRnity of the Rbre coating for an in biological samples were extracted with 100-
m
analyte is the most important factor in SPME. As PDMS for nonpolar drugs and 85-
m PA for polar
shown in Table 1, Rbre coatings of different polarity drugs. A solvent-modiRed Rbre can improve selectiv-
and thickness were selected for each drug. Most drugs ity and shorten extraction time. Although the theories
 III / SOLID-PHASE MICROEXTRACTION / Biomedical Applications
of Rbre and in-tube SPME methods are similar, there
are signiRcant differences between these methods.
The extraction of analytes is performed on the outer
surface of the Rbre for Rbre SPME and in the inner
surface of the capillary for in-tube SPME. Commer-
cially available SPME Rbres for drug analysis are
limited, by GC capillary columns with a vast array of
stationary phases are commercially available for in-
tube SPME. Headspace Rbre SPME is suitable for the
extraction of drugs in gaseous, liquid and solid sam-
ples, because of the avoidance of contact with an
aggressive matrix incompatible with the Rbre. Direct
immersion Rbre SPME can be used to extract drugs
from clear and cloudy liquid samples, however, in-
tube SPME is limited to the extraction of clear liquid
samples. The headspace SPME technique, therefore,
is suitable for direct extraction from whole blood
samples, while immersion Rbre SPME or in-tube
SPME methods require deproteinization or ultraRl-
tration of these samples prior to extraction. As men-
tioned above, the extraction efRciency of Rbre SPME
depends on extraction time, agitation, heating,
sample pH and salt concentration. For in-tube SPME,
number, volume and speed of draw/eject cycles, and
sample pH are important factors for efRcient extrac-
tion. On the other hand, the desorption of analyte
from a Rbre or capillary coating depends on the
temperature of the injection port and exposure time
in combination with GC or GC/MS, or component
and volume of solvent when used in combination
with HPLC or LC/MS. Therefore, these SPME para-
meters should be optimized when developing a new
SPME method for drug analysis.
With further development of new coating mater-
ials, such as afRnity coatings for target drugs and
chiral coatings for optically active drugs, the further
development of derivatization methods, further coup-
ling with different analytical instruments, such as
capillary electrophoresis, and improvement of the
extraction and desorption conditions, the SPME tech-
nique is expected to be widely applied in the future
for highly efRcient extraction of drugs from various
biological samples.
See also: II/Chromatography: Gas: Derivatization.
Extraction: Solid-Phase Microextraction.
Further Reading
Battu C, Marquet P, Fanconnet AL, Lacassie E and
Lachatre G (1998) Screening procedure of 21 ampheta-
4169
mine-related compounds in urine using solid-phase
microextraction and gas chromatography}mass spectro-
metry. Journal of Chromatographic Science 36: 1}7.
Degal F (1996) Comparison of new solid-phase extraction
methods for chromatographic identiRcation of drugs in
clinical toxicological analysis. Clinical Biochemistry 29:
529}540.
Eisert R and Levsen K (1996) Solid-phase microextraction
coupled to gas chromatography: a new method for the
analysis of organics in water. Journal of Chromato-
graphy A 733: 143}157.
Junting L, Peng C and Suzuki O (1998) Solid-phase micro-
extraction (SPME) of drugs and poisons from biological
samples. Forensic Science International 97: 93}100.
Kataoka H, Narimatsu S, Lord HL and Pawliszyn J (1999)
Development of on-line in-tube solid-phase microex-
traction/LC/MS system. Chromatography 20: 237}246.
Kroll C and Borchert HH (1998) Solid phase microextrac-
tion (SPME) for sample preparation during drug meta-
bolism studies. Pharmazie 53: 172}177.
Lord HL and Pawliszyn J (1998) Recent advances in solid-
phase microextraction. LC-GC S41}S46.
Namera A, Yashiki M, Kojima T and Fukunaga N (1998)
Solid phase microextraction in forensic toxicology.
Japanese Journal of Forensic Toxicology 16: 1}15.
Pawliszyn J (1995) New directions in sample preparation
for analysis of organic compounds. Trends in Analytical
Chemistry 14: 113}122.
Pawliszyn J (1997) Solid Phase Microextraction: Theory
and Practice. New York: John Wiley.
Pawliszyn J (1999) Applications of Solid Phase Microex-
traction. Cambridge, UK: The Royal Society of
Chemistry.
Penton ZE (1997) Sample preparation for gas chromato-
graphy with solid-phase extraction and solid-phase
microextraction. Advances in Chromatography 37:
205}236.
Sporkert F and Pragst F (2000) Use of headspace solid-
phase micro-extraction (HS-SPME) in hair analysis for
organic compounds. Forensic Science International 107:
129}148.
Ulrich S and Martens J (1997) Solid-phase microextraction
with capillary gas}liquid chromatography and nitro-
gen}phosphorus selective detection for the assay of anti-
depressant drugs in human plasma. Journal of
Chromatography B: Biomedical Applications 69:
217}234.
Volmer DA and Hui JPM (1997) Rapid determination of
corticosteroids in urine by combined solid-phase micro-
extraction/liquid chromatography/mass spectrometry.
Rapid Communications in Mass Spectrometry 11:
1926}1934.
Zhang Z, Yang MJ and Pawliszyn J (1994) Solid phase
microextraction: a new solvent-free alternative for
sample preparation. Analytical Chemistry 66:
844A}853A.
4170 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications
Environmental Applications
A. Andrews, Ohio University, Athens, OH, USA
Copyright ^ 2000 Academic Press
Introduction
Solid-phase microextraction (SPME) was Rrst de-
scribed by Pawliszyn and co-workers from the Uni-
versity of Waterloo in 1989 and has rapidly become
a popular extraction method in research.
BrieSy, SPME uses an immobilized liquid polymer
phase coated on the outside of a fused silica Rbre.
By dipping this Rbre into either a liquid (usually
aqueous) sample, or the headspace above a
liquid sample, absorption of the analytes from
the matrix into the polymer layer occurs. Analyte
desorption occurs in either the heated inlet port
of a gas chromatograph or a specially constructed
inlet loop in the case of liquid chromatographic
analysis.
Adjustment of the sample pH or ionic strength can
be used to enhance the analyte partitioning into the
Rbre coating. The adjustment of ionic strength is
commonly carried out by the addition of an inorganic
salt such as sodium chloride. To reduce the time
taken for equilibrium between the Rbre coating and
the sample agitation is used. This is usually achieved
with a magnetic stir bar system.
SPME has the advantage that it is a solvent-free
technique, which reduces both environmental pollu-
tion and sample preparation time, as no additional
concentration step is required. There is no packed
cartridge bed, as with solid-phase extraction (SPE),
and so SPME does not suffer from plugging and is
complete in two stages. This minimizes the chance of
analyte loss or operator error.
There are now many examples of the extraction of
analytes of environmental interest. This article will
cover the more recent examples from each class of
environmental pollutants. Articles listed in the Fur-
ther Reading cover these areas in more depth than is
possible here.
Volatile Organic Chemicals
One of the most common environmental applications
of SPME is the analysis of volatile organic compounds.
The chemicals frequently used as representative of
this class are benzene, toluene, ethylbenzene and the
xylene isomers (BTEX).
Most BTEX extractions have been carried out
using polydimethylsiloxane (PDMS) as the polymer
coating on the Rbre. Whilst PDMS works well for the
extraction of the BTEX compounds, the recoveries
from the sample are typically in the 0.1}10% range.
With such a low recovery, the limits of detection
(LOD) are often higher than those achieved with
conventional purge-and-trap analysis.
An interesting way to reduce LODs without resort-
ing to more expensive detectors is to use a Rbre
freshly coated with PDMS that is then dipped in
extra-Rne powdered activated charcoal. The total
thickness of this coating is approximately 100
m,
which is similar to the thickness of a commercially
available Rbre.
Using the PDMS/charcoal-coated Rbre, recoveries
of greater than 90% are achieved in about 15 min
extraction time at 253C, or in under 10 min at
753C using headspace analysis and salting out of
the sample, as illustrated in Figure 1. These high
recoveries give LODs of 421 pg mL\1 with GC
analysis using a Same ionization detector (FID). This
is over an order of magnitude better than the LODs
for the US Environmental Protection Agency method
524.2, and two orders better than conventional
PDMS SPME with gas chromatography (GC)-FID
analysis.
Surfactants
Alkylphenol surfactants are becoming of increasing
concern due to their possible role as endocrine disrup-
ters. SPME analysis of this class of surfactants elimin-
ates the need for the concentration step frequently
required with conventional liquid}liquid extraction
or SPE.
As alkylphenol ethoxylate surfactants are not
amenable to GC analysis without derivatization, and
SPME with derivatization is a much less well-
developed technique than classical SPME, analysis of
the surfactants has been accomplished by normal-
phase gradient high performance liquid chromatogra-
phy (HPLC) with UV detection.
This requires a speciRc interface for desorping the
surfactants from the Rbre after adsorption (Figure 2).
The Rbre desorption chamber is a three-way tee
in which two outlets are connected to the injection
valve and the third houses the SPME device. Flow to
the tee is via stainless steel tubing (d) and back to the
injection port is via poly(ether ether ketone) (PEEK)
tubing (e) connected via a Rnger-tight PEEK union (f).
The SPME device (g) is positioned in the top part of
the tee. One of the problems initially encountered
with this device was stripping of the Rbre coating by
 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications
Figure 1 Dependence of extraction efficiency on exposure time
and temperature. BTEX concentration, 500 pg mL\1 of each
compound; stirring speed, 50% of maximum; amount of NaCl,
15 g. (A) T"253C; (B) T"503C; (C) T"753C. Open circles,
benzene; open squares, toluene; filled circles, ethylbenzene;
filled squares, p-xylene; open triangles, m-xylene; filled triangles,
o-xylene. Reproduced with permission from Djozan DJ and
Assadi J (1997). Copyright Elsevier Science.
the narrow internal diameter stainless steel tubing (d).
This was because of swelling of the coating, which
occurred in the desorption solvents used. Increasing
the diameter of the tubing to account for this swelling
solved this problem. In order to be able to withstand
the high pressures used in HPLC, a slip-free super-
critical Suid extraction connector (i) was used to
provide a strong seal. A large diameter TeSon tube
(h) and a regular ferrule (j) complete the desorption
device.
When analysing for surfactants, it is important that
the extraction method does not discriminate against
any of the oligomers present in the sample. A Car-
bowax/template resin-coated SPME Rbre was found
to show good extraction of the various chain length
surfactants. To ensure that the extracted ethoxamer
4171
distribution was the same as the original sample,
salting out of the sample was carried out; without
this, a bias towards shorter chain ethoxamers is
obtained.
The Rnal analysis method has been successfully
applied to the determination of a number of
surfactants, as shown in Figure 3. LODs for the indi-
vidual alkylphenol ethoxamers are in the low p.p.b.
range.
Hetero-organic Pollutants
Compounds with nitrogen, sulfur or oxygen (NSO) in
the aromatic ring system are frequently encountered
as pollutants in groundwater. These hetero-organic
pollutants are often the result of creosote contamina-
tion. For NSO groundwater analysis, LODs in the
ng L\1 range are required. Currently this is only
achievable with liquid}liquid extraction using a con-
centration step.
For SPME analysis of NSOs, out of three Rbre
coatings investigated, PDMS, polyacrylate (PA) and
Carbowax, only the PA coating successfully extracted
all of the 15 NSO compounds investigated. However,
the amount extracted at equilibrium is small, ranging
from 0.4% for pyrrole to 57% for dibenzofuran.
Although pH over the range of 7}10 does not affect
the amount extracted, higher pH values degrade the
Rbre coating. Adjusting the sample ionic strength
by adding sodium chloride increases the amount
extracted.
The LODs obtained, distribution coefRcients and
aqueous solubility of the NSO compounds investi-
gated are shown in Table 1. In comparison to a con-
ventional analysis this SPME method showed
improved performance for water soluble and
semivolatile NSOs, but poorer performance for the
volatile NSOs. Carryover from one sample to the
next was also a problem and necessitated a 10 min
ofSine Rbre cleaning process between samples. Fibres
were found to degrade slowly with use, and were
discarded after 50 analyses.
Pesticides and Herbicides
The monitoring of groundwater for pesticide and
herbicide contamination is a common environmental
analysis. Accordingly, there have been a number of
reports on the use of SPME for pesticide extraction.
Many different classes of pesticides have been stud-
ied, including the organochlorine and organophos-
phorus pesticides.
With organophosphorus pesticides, the best Rbre coat-
ing is an XAD polymer (polystyrene-divinylbenzene)
4172 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications

Figure 2 Modified SPME-LC interface in (A) the fibre desorption and (B) insertion modes. Arrows indicate flow direction. Reproduced
with permission from Boyd-Boland and Pawliszyn (1996). Copyright American Chemical Society.

phase. The aromatic character of this material pro-
vides superior extraction for many of the or-
ganophosphorus pesticides in comparison to a PDMS
phase. This class of pesticides is one group of com-
pounds where salting out has no beneRcial effect.
The limitation of the SPME extraction was that the
relative standard deviation (RSD) values seen were
very large } up to 80% in some cases. In addition,
some carryover from run to run was experienced.
Until these problems can be solved, the method
is more suitable for screening than for routine
analysis.
 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications 4173

Figure 3 LC chromatograms of the extracted alkylphenol ethoxylates: (A) Triton X-100; (B) Rexol 25/4; (C) Rexol 25/5. Peak
assignment in (A) refers to the number of units in the ethoxylate chain. Reproduced with permission from Boyd-Boland and Pawliszyn
(1996). Copyright American Chemical Society.

One of the great areas of potential for SPME is
automated analysis. This has been partially demon-
strated in the Reld of pesticide analysis by coupling
SPME extraction from a Sow cell with GC-FID. The
online Sow-through cell used is shown in Figure 4.
The extraction is performed whilst pumping the
sample in a closed loop for 30 min. Changing the Sow
rate from 0.1 to 10 mL min\1 did not affect the
precision of the extraction, presumably because Sow
inside the extraction chamber where the Rbre is posi-
tioned is turbulent across this entire range. Method
precision was found to compare well with other
sample agitation methods, such as Rbre vibration or
magnetic stirring.
This method has been applied to the analysis
of the S-triazines (herbicides) and, although not
fully automated, as operator intervention was re-
quired to transfer the SPME Rbre from the extrac-
tion cell to the GC, it is clearly a step in that
direction.
Polycyclic Aromatic Hydrocarbons
and Polychlorinated Biphenyls
Many of the samples analysed for environmental con-
tamination are solids, in particular soils and sludges.
These present a challenge to SPME, particularly in the
case of semivolatiles, such as polycyclic aromatic
hydrocarbons (PAHs) and polychlorinated biphenyls
(PCBs), where heating and headspace extraction can-
not be used.
A solution to this problem is Rrst to extract the
analytes of interest with subcritical water and then
extract the resulting subcritical water solution via
SPME. In this way, no hazardous solvents of any kind
are used during the extraction procedure.
Extractions are carried out in a heavy-duty stain-
less steel pipe partially Rlled with solid sample and
approximately 3.5 mL of HPLC-grade water. Care
should be taken to avoid samples which might react
with water leading to pressures higher than expected.
4174 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications

Table 1 The measured distribution constants between polyacrylate fibre and water, the octanol}water partition coefficients (log Kow),
aqueous solubility (mg L\1), limit of detection of NSO by SPME-GC-FID and SPME-GC-ITMS in
g L\1 and %RSD

Compound K log Kow Sw LOD-FID LOD-MS % RSD ITMS

(mg L\1) (
g L\1) (
g L\1) (
g L\1)

Thiphene 0.34 1.81 3600 nd. 1.0 14

1-Methylpyrrole 0.04 ! Soluble nd. 2.5 13
Pyrrole 0.10 0.75 58 000 nd. 10 12
2-Methylpyridine 0.02 1.06 Soluble nd. (10) 5.9
2,4-Dimethylpyridine 0.09 ! Soluble nd. (10) 14
Benzofuran 3.1 2.67 ! 3 0.03 4.2
Benzothiphene 7.9 3.12 130 2 0.02 5.8
Quinoline 0.46 2.03 6500 15 0.3 10
Indole 3.2 2.00 1850 2 0.02 6.9
2-Methylquinoline 0.53 2.23 ! 10 0.2 3.3
Dibenzofuram 22 4.12 6.6 2 0.03 10
Dibenzothiophene 27 5.45 1.0 2 0.02 11
Acridine 7.7 3.50 46 0.5 0.02 9.2
Carbazole 29 3.71 1.2 0.5 0.02 10
DBT-sulfone 4.9 ! ! 0.5 0.04 9.1

!Unknown; n.d., not detected; LOD, determined by 100
g L\1 standard solutions. Reproduced with permission from Johansen and
Pawliszyn (1996) Copyright John Wiley & Sons, Inc.

The sealed pipe is then heated in a GC oven for an
initial extraction time period. After cooling, conven-
tional SPME extraction using a 100
m PDMS-
coated Rbre of 1.8 mL of the resulting supernatant
liquid is carried out in a 2 mL vial.
Increasing the temperature of the water extraction
to 2503C improved the extraction of PAHs but fur-
ther increases in temperature did not improve the
results. Increasing the water extraction time from 15
to 60 minutes also increased the amount of PAHs

Figure 4 Instrumental set-up for the online flow-through cell. The SPME fibre hosts in a Valco-tee unit sealed by a vespel ferrule. The
aqueous sample is pumped by a HPLC pump from the sample vial through the extraction cell and back to the reservoir (closed loop).
Reproduced with permission from Eisart and Pawliszyn (1997). Copyright Elsevier Science.
 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications 4175

extracted, but not by enough to justify the increased
time taken for the extraction procedure. One point of
note was the unusual conversion of d10-anthracene to
d8-anthracene during the extraction. This was unex-
pected, as the hot liquid water used is regarded as
being relatively unreactive and other deuterated stan-
dards did not show the same effect.
Partitioning of the higher molecular weight PAHs
back on to the soil during cooling of the water extract
was found to be a problem that affected quantiRca-
tion, unless deuterated standards for each compound
were added.
PCBs are another ubiquitous environmental pollu-
tant that have been determined by using SPME ex-
traction in a variety of water samples. Conventional
100
m PDMS Rbres were found to give sufRcient
extraction of PCBs from water samples in 15 min to
have a LOD of around 5 pg mL\1 for each congener
with GC and electron-capture detection. This would
allow reasonably fast screening of samples for the
presence of PCBs.
Carryover was found to be present on not only the
SPME Rbres, but also the TeSon coated stir-bars,
which were used to agitate the solution for more
efRcient extraction. New stir bars were required for
each sample to avoid this problem. Eliminating Rbre
carryover was more difRcult but ofSine desorption,
coupled with running blanks between every sample,
minimized the problem.
Old Rbres (used more than 30 times) were found to
show more severe carryover than new Rbres. This
degradation in use meant that even with the off-line
desorption, carryover could still be present, and
would necessitate regular changing of the Rbre.
In contrast to many other SPME studies, it was
found that no direct correlation existed between the
octanol}water partitioning coefRcient (Kow) and the
Rbre-water coefRcient (KSPME). The deviations begin
to occur with analytes of molecular mass greater than
200 and, hence, for these compounds, Kow cannot be
used to estimate KSPME.
Suspended solids in real water matrices were found
to reduce the aqueous PCB concentration by up to
50% after spiking within 24 h. This partitioning on
to suspended solids is similar to that seen for PAHs.
Chlorobenzenes
Chlorobenzenes are classiRed as priority pollutants in
both the US and the European Union. SPME has been
evaluated in both the headspace sampling and direct
sampling modes for the determination of chloro-
benzenes in soil samples. One of the problems of
using SPME with soil samples is that quantiRcation
problems occur with soil samples that have a high
organic content when using the external calibration
method. This has necessitated the use of the standard
addition method for reliable quantitation.
Another problem with heavily contaminated soils
is the overloading of the detector beyond the linear
dynamic range by the large amount of analyte extrac-
ted. This is a particular problem with detectors such

Figure 5 Effect of (A) (circles; 10%; triangles, 30%) methanol and (B) (circles, 20%; triangles, 30%) acetone on the absorption time
profile of pentachlorobenzene by direct SPME-GC-ITMS using a 100
m PDMS fibre with 0.030 g of soil, 40 mL of water}organic
solvent; stirring speed 1000 rpm, sampling temperature 303C, exposure time 25 min; splitless injection mode. Reproduced with
permission from SarrioH n et al. (1998). Copyright Elsevier Science.
4176 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications

as the ion trap mass spectrometer (ITMS). The addi- Table 2 Reproducibility and limits of detection for tin using
tion of a water-miscible solvent, such as acetone (up SPME-GC-ICPMS
to 30% v/v), to the water, which has previously been Component RSD (n"10) LOD (3 s, n"10)
added to the soil sample, reduces the amount extrac- (%) (ng L\1 as metal)
ted into the Rbre to within the liner dynamic range of
Monobutyltin 5.2 0.34
the ITMS. The organic solvent also reduces the time
Dibutyltin 8.9 2.1
required to reach equilibrium and allows shorter Tributyltin 14 1.1
extraction times to be used. This is illustrated in Methyl mercury 11 4.3
Figure 5. Trimethyl lead 8.2 0.19
Using a 100
m PDMS Rbre, there was no signiR-
Reproduced with permission from Moens et al. (1997). Copyright
cant difference seen in RSD values between head- American Chemical Society.
space and direct sampling. A 7
m PDMS Rbre did
show higher RSD values than the 100
m PDMS Rbre tin and selenium. Using this reagent, derivatization
for headspace sampling. Using GC-ITMS analysis, can be performed in aqueous solution simultaneously
the LODs were between 30 and 100 pg g\1 for the with the extraction.
100
m Rbre and headspace sampling. Headspace Optimal derivatization conditions were obtained at
sampling gave a cleaner extract and resulted in a lon- a pH of 5.3 using 1 mL of a 1% NaBEt4 solution with
ger Rbre life than the direct sampling method. 25 mL of sample. The rate-limiting step was the ex-
Comparison of SPME results obtained on a refer- traction into the Rbre coating in the headspace extrac-
ence soil (CRM-530 or industrially contaminated tion mode and not the derivatization.
clay soil) with results from other laboratories, mostly Compromise extraction conditions were used of
using Soxhlet extraction, showed good agreement for 10 min extraction time at 253C. Longer extraction
the mean values of the analytes. times increase the amount extracted, but 90% is
extracted within the Rrst 10 min. Increasing the ex-
traction temperature increased the amount extracted
Organometallics for some compounds, but decreased the amount for
Most SPME applications in the environmental Reld others. As a result, 253C was chosen, as it avoids the
have been with organic pollutants. With the use of need for sample heating.
derivatization SPME, this has recently been extended Using GC with inductively coupled plasma}mass
to include some organometallic pollutants. spectrometry (ICP-MS) detection, LODs in the low
For heavy metals there is a strong dependence of parts per trillion were obtained, as shown in Table 2.
the toxicity with the chemical form and the speciation The method has been applied to a standard reference
of an element in a sample. material (NRC PACS-1, a marine sediment) to deter-
Sodium tetraethylborate (NaBEt4) is a useful de- mine organotin content. A clean extract was ob-
rivatizing agent for a number of organometallic com- tained, as can be seen in Figure 6. The results for the
pounds, including those of lead, mercury, cadmium, dibutyl- and tributyltin showed good agreement with

Figure 6 LC chromatogram of the PACS-1 reference material. Identification of peaks, 1, tetraethyltin; 2, monobutyltin; 3, tripropyltin;
4, dibutyltin; 5, tributyltin. Reproduced with permission from Moens et al. (1997). Copyright American Chemical Society.
 III / SOLID-PHASE MICROEXTRACTION / Environmental Applications
Figure 7 LC chromatogram for SPME injection with UV detec-
tion at 275 nm. (a) Fibre blank for microporous hollow fibre with-
out dipping with DBC and HgCl2 solution. (b) DBC blank
microporous hollow fibre only dipping with 0.02 mol L\1 DBC
solution for 5 min. (c) Microporous hollow fibre dipping with DBC
solution for 5 min, then dipping with 0.02 mol L\1 HgCl2 aqueous
solution for 5 min. Reproduced with permission from Jia et al.
(1998). Copyright John Wiley & Sons, Inc.
the certiRed values and the values obtained via a clas-
sical liquid}liquid extraction.
The values for monobutyltin were signiRcantly
higher than the certiRed values. This has been re-
ported as being a problem with the derivatization
using NaBEt4 for monobutyltin and is not attributed
to any problems with the SPME part of the analysis.
Metal Ions
Inorganic metal ion analysis has also been achieved
with SPME. In order to extract metal ions from
aqueous solution, an unusual Rbre constructed of
a hydrophobic microporous polypropylene material
was used. This Rbre had a Rlm thickness of 30
m and
30% of the surface area was covered with pores
of dimensions 0.05;0.15
m. In order to extract
mercury(II) ions from aqueous solution, dibenzo-18-
crown-6 (DBC) was absorbed from solution into this
hollow Rbre and then this DBC-Rlled Rbre was used
for the extraction.
Separation of uncomplexed DBC from the DBC-
mercury(II) complex was achieved by normal-phase
HPLC with UV detection. The extraction equilibrium
between the Rbre and the solution with DBC was
reached in under 30 s. This is a very rapid equilibrium
for SPME. The time to reach equilibrium for the
mercury(II) ions between the treated Rbre and aque-
4177
ous solution was signiRcantly longer; equilibrium
was not reached after 1 h. In order to keep extract-
ion time closer to the HPLC analysis time, 10 min
extraction was used. With this extraction time
and UV detection, the LOD for mercury(II) ions
was around 500 p.p.b. The use of a more sensitive
detection technique should signiRcantly reduce this
LOD.
The effectiveness of the extraction can be seen in
Figure 7. No reports on the selectivity of this tech-
nique for metal ions have been published and this
remains a key area for the future.
Future Developments
SPME is currently poised to become one of the major
sample preparation methods for aqueous environ-
mental samples in the future. The advantages that it
offers, such as ease of use, no solvent and no plug-
ging, make it a potential replacement for many of the
liquid}liquid and SPE methods currently used. The
expansion of SPME to the analysis of other analytes
which are currently difRcult to partition into the Rbre
coatings is another expected trend. This may be
achieved by the development of new Rbre coatings or
by coupling existing coatings with derivatization or
complexation reactions.
The future for SPME in the analysis of solids is less
certain. Unless more robust SPME methods than
those currently described are found, the replacement
of Soxhlet extraction by SPME seems unlikely. Addi-
tionally, the problem of carryover is a cause for
concern with the analysis of higher molecular
weight analytes. Currently the cost of SPME Rbres
does not allow them to be used as single-use devices.
This may, of course, change in the future and single-
use Rbres are the surest way to ensure that there is no
carryover.
See also: II/Extraction: Analytical Extractions; Solid-
Phase Microextraction; Solid-Phase Extraction. III/Carba-
mate Insecticides in Foodstuffs: Chromatography
& Immunoassay. Surfactants: Liquid Chromatography.
Inclusion Complexation: Liquid Chromatography.
Pesticides: Extraction from Water. Polychlorinated
Biphenyls: Gas Chromatography. Superheated Water
Mobile Phases: Liquid Chromatography.
Further Reading
Boyd-Boland AA and Pawliszyn JB (1996) Solid-phase
microextraction coupled with high-performance liquid
chromatography for the determination of alkylphenol
ethoxylate surfactants in water. Analytical Chemistry
68: 1521}1529.
4178 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
Dean JR (1998) Solid phase microextraction. In: Extraction
Methods for Environmental Analysis, ch. 5. New York:
John Wiley.
Djozan DJ and Assadi J (1997) A new porous-layer ac-
tivated-charcoal-coated fused silica Rber: application for
determination of BTEX compounds in water samples
using headspace solid-phase microextraction and ca-
pillary gas chromatography. Chromatographia 45:
183}189.
Eisart R and Levsen K (1996) Solid-phase microextraction
coupled to gas chromatography: new method for the
analysis of organics in water. Journal of Chromatogra-
phy A 733: 143}157.
Eisart R and Pawliszyn J (1997) Design of automated solid-
phase microextraction for trace analysis of organic com-
pounds in aqueous samples. Journal of Chromatography
A 776: 293}303.
Hageman KJ, Mazeas L, Grabanski CB et al. Coupled
subcritical water extraction with solid phase micro-
extraction for determining semivolatile organics
in environmental solids. Analytical Chemistry 68:
3892}3898.
Jia C, Luo Y and Pawliszyn J (1998) Solid phase microex-
traction combined with HPLC for determination of
Food Technology Applications
R. Marsili, Dean Foods Technical Center,
Rockford, IL, USA
Copyright ^ 2000 Academic Press
Introduction
The chemicals responsible for off-Savours, mal-
odours and taints in foods and beverages can
originate from incidental contamination from envi-
ronmental (outside) sources (e.g. air, water, packag-
ing material, a contaminated ingredient) and from
chemical reactions occurring within the food material
itself (e.g. lipid oxidation, enzymatic action, microbi-
al metabolic reactions). In addition, imbalance off-
Savours can occur when certain ingredient compo-
nents that are normally present and often essential to
the product are present in abnormally high or low
concentrations.
When signiRcant off-Savour problems occur, one of
the Rrst priorities of the food chemist is to identify any
volatile or semivolatile organic chemicals that may be
responsible. Once the identity of the off-Savour chem-
ical(s) has been established, it is possible to speculate on
its mechanism of formation and then decide on what
corrective actions to implement to eliminate recurrence
of the problem in the future.
metal ions using crown ether as selective extracting
reagent. Journal of Microcolumn Separations 10:
167}173.
Johansen SS and Pawliszyn J (1996) Trace analysis of het-
ero aromatic compounds (NSO) in water and polluted
groundwater by solid-phase microextraction (SPME).
Journal of High Resolution Chromatography 19:
627}632.
Moens L, De Smaele T, Dams R et al. (1997) Sensitive,
simultaneous determination of organomercury, -lead
and -tin compounds with headspace solid phase micro-
extraction capillary gas chromatography combined
with inductively coupled plasma mass spectrometry.
Analytical Chemistry 69: 1604}1611.
Pawliszyn J (1997) Solid Phase Microextraction: Theory
and Practice. New York: Wiley-VCH.
SarrioH n MN, Santos FJ and Galceran MT (1998) Strategies
for the analysis of chlorobenzenes in soils using solid-
phase microextraction coupled with gas chromato-
graphy-ion trap mass spectrometry. Journal of
Chromatography A 819: 197}209.
Yang Y, Miller DJ and Hawthorne SB (1998) Solid-phase
microextraction of polychlorinated biphenyls. Journal
of Chromatography A 800: 257}266.
Analytical Strategy for Studying
Off-Flavours
The following steps are commonly used when trying
to determine which chemicals in a particular food or
beverage sample are the most important contributors
to off-Savours:
E Extraction of volatiles/semivolatiles. The chem-
icals responsible for the food taint must be extrac-
ted and usually concentrated from the food matrix.
This sample preparation step is critical to success.
To isolate and evaluate potential chemical compo-
nents that are responsible for the food taint,
analytes must be separated from interfering chem-
icals in the food matrix.
E Injection into the gas chromatograph (GC: with or
without cryofocusing).
E Separation of extracted volatiles on a GC capillary
column with a suitable liquid phase. It is not un-
common to miss important polar compounds be-
cause the chemicals do not chromatograph well on
nonpolar phases. Often the extraction technique is
blamed, but the problem could simply be that an
inappropriate analytical capillary column was used
for the separation. One example is not detecting
volatile fatty acids because separation was at-
tempted on a nonpolar column.
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications 4179

E Determination of peak odour by olfactometry. It is

Sample prep
often advantageous to sniff peaks as they elute

Table 1 General comparison of common analyte extraction techniques for studying food aromas (actual parameters vary depending on sample matrix, analyte, type of GC detector)

time (min)
from the GC column. The odour characteristics

10}30

10}60
'30
and intensities of the eluting peaks can help the

5}30

5}60

5}10
analyst determine if the chemical is a likely con-
tributor to the malodour or off-Savour. A variety

automation
of olfactometry detectors are commercially avail-

Sample
able; olfactometry detectors with heated transfer

Yes
Yes
Yes

Yes
lines are highly recommended.

No

No
E Determination of which volatiles/semivolatiles are
the most potent contributors to the products

Nonvolatile
odour. Gas chromatography}olfactometry (GCO)
analysis has evolved over time to include dilution
techniques (Aroma Extraction Dilution Analysis,

400
AEDA and CharmAnalysis), cross-modal matching
(Osme) and maximum perceived intensity. Of

Semivolatile

300
these three GCO modiRcations, extract dilution

Boiling point (3C)

techniques and cross-modal matching have become

200
the most common techniques used in analytical

Range of volatiles analysed

work on food Savours. Further discussion of the
various GCO techniques is beyond the scope of this

100
Volatile
article.

Capable of analysing liquids, but usually requires binding of liquid portion of sample with an inert matrix material.
Perhaps the most critical and challenging step in

0
the process of characterizing the Savour of foods is

!100
Gases
the sample preparation technique used to isolate/con-
centrate the Savour compounds from the food

G, Gas; L, liquid; S, solid; p.p.t., parts per trillion; p.p.b., parts per billion; p.p.m., parts per million.
matrix. Since it is not uncommon for the chemicals
responsible for food malodours to be present at p.p.b.
and even p.p.t. levels, the extraction technique must
Detection limit

p.p.b.}p.p.t.
p.p.b.}p.p.t.
collect as many molecules of off-Savour chemicals as
possible for GCO analysis. If the goal is to identify
p.p.m.

p.p.b.
p.p.b.
the chemicals responsible for an off-Savour, the p.p.b.
sample preparation method selected should extract
a representative proRle of as many organic vol-
Sample size

atiles/semivolatiles from the sample as possible. On

0.001}0.10

the other hand, it is also important that the extraction

1}1000
0.1}10

0.1}10
0.1}10
0.1}10

technique does not introduce or create volatiles that

(g )

are not in the food product. For example, sample

preparation techniques that involve heating the
Sample
matrix

sample (e.g. steam distillation) can generate artifact

G/L/S

G/L/S
L/S

L/S

peaks in sample chromatograms, and these odiferous

Sa
S

artifacts may be misinterpreted as the cause of the

malodour/off-Savour problem.
This article will discuss why solid-phase micro-
extraction (SPME) is such an excellent extraction/
Solid-phase microextraction (SPME)

Supercritical fluid extraction (SFE)

concentration technique for the study of food off-

Dynamic headspace (DH)/Tenax

Direct thermal desorption (DTD)

Savours and taints.

Solvent extraction (SE)
Static headspace (SH)

Advantages of SPME as an Extraction

Technique
Chemicals responsible for off-Savours can be polar,
semipolar and nonpolar and cover a wide range of
functional groups, boiling points and molecular
a
4180 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications

weights. As a result, no one analytical extrac-
tion/sample preparation method works in all cases. It
is not uncommon that multiple sample preparation
methods are required to identify the chemicals
responsible for off-Savours and malodours in a par-
ticular sample.
Each sample preparation technique has advantages
and disadvantages. The choice of a suitable sample
preparation technique depends on several factors, in-
cluding number of samples to be tested, how quickly
results are needed, type of sample (matrix effects),
the nature of the analytes of interest (i.e. functional
group, molecular weight, boiling point, thermal
stability, etc.), desired detection limits and required
accuracy.
Table 1 compares a few popular extraction tech-
niques used prior to GC analysis. Considering the
wide range of sample sizes that can be analysed by
SPME, the low detection limits, the wide range of
analyte boiling points that can be analysed, the fact
that SPME can be automated and the short sample
preparation time, it is no surprise that SPME is rap-
idly growing in popularity. The low cost of SPME
equipment is also an advantage.
One often overlooked beneRt of SPME is its high
precision and accuracy compared to other GC samp-
ling techniques. Studies comparing the precision and
accuracy of SPME to other GC sampling techniques
show that analytical results based on SPME extrac-
tion are often more precise and accurate than results
based on other sample preparation techniques.
Several polar and nonpolar Rbres with varying
afRnities for speciRc classes of compounds are
now available. As a result, SPME Rbre type can be
selected in order to optimize results for a particular
analyte class. Compounds that interfere with the

Table 2 SPME fibre selection guide

Analyte class Fibre type Linear range

Acids (C2}C8) Carboxen-PDMS 10 p.p.b.}1 p.p.m.

Acids (C2}C15) CW-DVB 50 p.p.b.}50 p.p.m.
Alcohols (C1}C8) Carboxen-PDMS 10 p.p.b.}1 p.p.m.
Alcohols (C1}C18) CW-DVB 50 p.p.b.}75 p.p.m.
Polyacrylate 100 p.p.b.}100 p.p.m.
Aldehydes (C2}C8) Carboxen-PDMS 1 p.p.b.}500 p.p.b.
Aldehydes (C3}C14) 100
m PDMS 50 p.p.b.}50 p.p.m.
Amines PDMS-DVB 50 p.p.b.}50 p.p.m.
Amphetamines 100
m PDMS 100 p.p.b.}100 p.p.m.
PDMS-DVB 50 p.p.b.}50 p.p.m.
Aromatic amines PDMS-DVB 5 p.p.b.}1 p.p.m.
Barbiturates PDMS-DVB 500 p.p.b.}100 p.p.m.
Benzidines CW-DVB 5 p.p.b.}500 p.p.b.
Benzodiazepines PDMS-DVB 100 p.p.b.}50 p.p.m.
Esters (C3}C15) 100
m PDMS 5 p.p.b.}10 p.p.m.
Esters (C6}C18) 30
m PDMS 5 p.p.b.}1 p.p.m.
Esters (C12}C30) 7
m PDMS 5 p.p.b.}1 p.p.m.
Ethers (C4}C12) Carboxen-PDMS 1 p.p.b.}500 p.p.m.
Explosives (nitroaromatics) PDMS-DVB 1 p.p.b.}1 p.p.m.
Hydrocarbons (C2}C10) Carboxen-PDMS 10 p.p.b.}10 p.p.m.
Hydrocarbons (C5}C20) 100
m PDMS 500 p.p.t.}1 p.p.m.
Hydrocarbons (C10}C30) 30
m PDMS 100 p.p.t.}500 p.p.b.
Hydrocarbons (C20}C40#) 7
m PDMS 5 p.p.b.}500 p.p.b.
Ketones (C3}C9) Carboxen-PDMS 5 p.p.b.}1 p.p.m.
Ketones (C5}C12) 100
m PDMS 5 p.p.b.}10 p.p.m.
Nitrosamines PDMS-DVB 1 p.p.b.}200 p.p.b.
Polyaromatic hydrocarbons 100
m PDMS 500 p.p.t.}1 p.p.m.
30
m PDMS 100 p.p.t.}500 p.p.b.
7
m PDMS 500 p.p.t.}500 p.p.b.
Polychlorinated biphenyls 30
m PDMS 50 p.p.t.}500 p.p.b.
Pesticides, chlorinated 100
m PDMS 50 p.p.t.}500 p.p.b.
30
m PDMS 25 p.p.b.}500 p.p.b.
Pesticides, nitrogen Polyacrylate 50 p.p.t.}500 p.p.b.
Pesticides, phosphorus 100
m PDMS 100 p.p.t.}1 p.p.m.
Polyacrylate 100 p.p.t.}500 p.p.b.
Phenols Polyacrylate 5 p.p.b.}500 p.p.b.
Surfactants CW-TPR 1 p.p.m.}100 p.p.m.
Sulfur gases Carboxen-PDMS 10 p.p.b.}10 p.p.m.
Terpenes 100
m PDMS 1 p.p.b.}10 p.p.m.
Volatile organic chemicals Carboxen-PDMS 100 p.p.t.}500 p.p.b.
100
m PDMS 20 p.p.b.}50 p.p.m.
30
m PDMS 100 p.p.b.}50 p.p.m.

Reproduced with permission from Scheppers-Wercinski (1999) by courtesy of Marcel Dekker Inc.
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
chromatography when the food extract is analysed by
GC can be eliminated or at least minimized. If lipid
oxidation is being studied, for example, the analyst
could choose a Carboxen-PDMS Rbre to measure
aldehydes in the 1}500 p.p.b. range. If concentrations
of aldehydes above 500 p.p.b. are present, the Car-
boxen-PDMS Rbre will become saturated and
a 100
m PDMS Rbre would be a better choice.
A SPME Rbre selection guide is shown in Table 2.
For some applications, the portability of SPME is
an important advantage. After analytes are adsorbed
on an SPME Rbre, they can be maintained on the Rbre
for an extended period of time by sealing the end of
the Rbre with a septum. This allows for convenient
Reld sampling. Perfumers have used this technique,
for example, to extract aroma chemicals from Sowers
in greenhouses, as well as the fragrant chemicals from
exotic Sowers found in the canopy of tropical rain-
forests. Another example is a food chemist who is
trying to determine if a malodour in a particular food
product is being absorbed by the product because it
has been stored near odiferous foods (e.g. spices) or
perhaps industrial solvents. The food chemist can
extract volatiles from the air in a warehouse or walk-
in cooler with SPME, transport the SPME device with
the trapped volatiles to the laboratory for GC analy-
sis, and see if the GC proRle matches the proRle of
a problem sample.
Retention characteristics are highly dependent on
the Rbre used and the volatility of the adsorbed
analytes. Studies have shown that even highly volatile
compounds can be stored on Carboxen-PDMS Rbres
for 3 days at room temperature without loss. The
pore dynamics of Carboxen 1006 make it a true
adsorbent. Retention of volatiles on 100
m PDMS
Rbres, however, is not nearly as good. Even when
Rbres are stored at !43C, only the least volatile
analytes will be retained.
Speci\c Applications of SPME
for Resolving Food Taints
The examples and case studies that follow illustrate
the advantages of SPME as a sample preparation tool
for the study of off-Savours and malodours in foods
and beverages.
Light-Induced Off-Flavours in Milk:
SPME vs. Headspace Analysis
Two types of light-induced oxidation reactions occur
in milk and dairy products. Initially, a burnt, oxidized
Savour develops and predominates for approximately
2}3 days. Dairy technologists refer to this off-Savour
note as light-activated Savour (LAF). Degradation of
sulfur-containing amino acids of the serum (whey)
4181
proteins is probably responsible for this reaction. The
exact reaction products for LAF have not been clearly
elucidated. Methional [(3-methylthio)propanal],
however, has been implicated as a possible contribu-
tor. Understanding the true impact that methional
has on LAF is difRcult to determine because it is
relatively unstable and breaks down into more stable
components, including mercaptans, sulRdes and dis-
ulRdes. Recently, researchers have postulated an al-
ternative mechanism for the formation of dimethyl
disulRde by singlet oxygen oxidation of methionine.
In addition to the poorly understood LAF
off-Savour, a second type of light-induced off-Savour
occurs in milk and is attributed to lipid oxidation.
This off-Savour, often characterized as metallic or
cardboard-like, usually develops after 2 days and
does not dissipate. Aldehydes (especially pentanal
and hexanal) and, to a lesser degree, ketones (e.g.
1-hexen-3-one and 1-nonen-3-one), alcohols and hy-
drocarbons have been observed to form in milk as
a result of light-induced lipid oxidation reactions.
When milk is exposed to light, various carbonyl com-
pounds form from the reaction of light and oxygen
with unsaturated fatty acids in the milk fat triglycer-
ides and other milk fat components. Autoxidation of
unsaturated fatty acids involves a free radical reac-
tion, forming fat hydroperoxides that degrade to vari-
ous malodorous compounds (e.g. hexanal, the
predominant lipid reaction by-product in light-ex-
posed milk in the case of linoleic acid).
In one recent study to quantitate pentanal and
hexanal in light-abused milk (skim milk and 2%
fat milk), a comparison was made using two
different sample preparation techniques: dynamic
headspace (DH) with a Tenax trap and SPME with
a Carboxen-PDMS Rbre. Results, which are sum-
marized in Table 3, show that standard calibrations
with SPME were more linear for both analytes in
both types of milk samples than with DH. (Calib-
ration was based on the method of additions tech-
nique using an internal standard of 4-methyl-2-
pentanone.) Furthermore, the SPME method had
about the same detection limit as the DH method. To
test the precision of each method, four replicates
spiked with 2 ng mL\1 of each aldehyde were com-
pared for both types of milk samples. When coefR-
cients of variations were calculated for this study,
SPME proved to be more precise than DH.
For these particular samples and these particular
analytes, SPME consistently demonstrated better pre-
cision without a sacriRce in sensitivity. Furthermore,
none of the problems with carryover, background
or artifact peaks that sometimes occur with DH sys-
tems were observed with the SPME experiments. No
carryover peaks were detected in milk samples, even
4182 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
Table 3 Comparison of the principal analytical parameters for pentanal and hexanal analysed by DH/GC-MS and SPME/GC-MS
Compound Sample Analytical
technique
Pentanal Skim DH
SPME
Hexanal Skim DH
SPME
Pentanal 2% Milk DH
SPME
Hexanal 2% Milk DH
SPME
a
For calibration curve of five standards ranging from 0.0 to 30.0 ng mL\1.
when injecting the SPME Rbre immediately after it
was used to analyse a milk sample spiked with high
levels (500 ng mL\1) of each of the following al-
dehydes: butanal, isopentanal, pentanal, hexanal,
heptanal and octanal.
Because so many different parameters need to be
optimized when performing DH and SPME experi-
ments, care must be taken when comparing SPME
and DH for precision, accuracy and sensitivity, and it
is probably an over-simpliRcation to say that one
method is better than another. None the less, this
work shows that SPME is a viable extraction tech-
nique for measuring oxidation products in milk and
dairy products.
Highly Volatile Malodorous Chemicals
Highly volatile compounds can be responsible for
off-Savours and malodours and can be difRcult to
trap and isolate. DH techniques with Tenax trapping
often fail to trap and detect low molecular weight
polar compounds. Static headspace works well for
highly volatile chemicals but may not be sensitive
enough for some applications.
SPME is an ideal extraction tool for highly volatile
analytes. Consider, for example, the analysis of acet-
aldehyde in buttermilk. Acetaldehyde has a boiling
point of 213C.
Acetaldehyde in buttermilk The delicate Savour
associated with high quality cultured buttermilk is
contributed by several bacterial metabolites, includ-
ing lactic acid, traces of acetic and formic acids,
ethanol, diacetyl and carbon dioxide. Two different
types of bacteria are used in buttermilk starter cul-
tures: the acid-producing types (usually strains of
Streptococcus lactis or S. cremoris) and the aroma
bacteria (usually Leuconostoc citravorum). Diacetyl,
the major Savour component of buttermilk, is pro-
duced by the fermentation of citric acid by the aroma-
producing bacteria.
Detection limit Repeatability of four Linear least-squares
(ng mL\1) replicates at 2 ng mL\1 correlation coefficientsa
(coefficient of
variation, %)
0.1 8.0 0.966
0.1 1.9 0.990
0.3 21.1 0.910
0.5 7.1 0.995
0.3 7.6 0.996
0.3 2.1 0.999
0.8 8.3 0.982
0.8 4.9 0.993
One common type of off-Savour in buttermilk is
called the green Savour defect. It is caused by the loss
of diacetyl (by conversion to acetyl methylcarbinol by
diacetyl reductase enzyme in the culture bacteria) and
an increase in acetaldehyde production. Measuring
the acetaldehyde to diacetyl ratio is a good way to
monitor this Savour defect.
As shown in Figure 1, SPME (e.g. Carboxen-
PDMS) is an excellent way to extract acetaldehyde,
diacetyl, acetic acid and other Savour-important
metabolites from buttermilk. Even with SPME, how-
ever, it is necessary to use cryofocusing (typically at
!1003C) after thermal desorption from the SPME
Rbre and prior to injection into the GC capillary
column. With cryofocusing, sharp GC peaks are ob-
tained for acetaldehyde; without cryofocusing, the
acetaldehyde peak may not be detected at all.
1,3-Pentadiene from sorbate degradation Testing
for 1,3-pentadiene in foods and beverages is another
example of how SPME can be used to quantitate
a highly volatile malodorous compound. Sorbic acid
(2,4-hexadienoic acid) and its water-soluble potassi-
um salt are commonly used as food preservatives to
prevent yeast and mould growth. Foods in which
sorbate has commercially useful antimicrobial activ-
ity include baked goods, cheeses and other dairy
products, confectionery products, dried fruits, Rsh
products, fruit juices, jellies (with artiRcial
sweeteners), syrup, vegetables and wine.
One problem with potassium sorbate is that some
moulds in the genus Penicillium can grow in the
presence of up to (approximately) 1.2% potassium
sorbate. Furthermore, some of these moulds have the
ability to decarboxylate sorbic acid, producing 1,3-
pentadiene, a highly volatile compound with an ex-
tremely strong hydrocarbon-like odour (typically
kerosene-like).
As in the case of testing for acetaldehyde in butter-
milk, using SPME with a Carboxen-PDMS Rbre and
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications 4183

Figure 1 Volatiles in buttermilk by SPME (Carboxen-PDMS) extraction followed by GC-MS. Sample preparation: 2 mL of buttermilk,
7
L of internal standard solution (54 p.p.m. 4-methyl-2-pentanone), and a small magnetic stirring bar were added to a 4 mL GC vial and
sealed. Headspace volatiles were extracted by SPME for 20 min at 503C. Peak identities: 1, acetaldehyde; 2, acetone; 3, dimethyl
sulfide; 4, diacetyl; 5, acetic acid; 6, 2-pentanone; 7, ethyl acetate; 8, internal standard; 9, butyric acid.

cryofocusing prior to release into the analytical col-
umn works well for measuring 1,3-pentadiene in
foods and beverages. A chromatogram showing 1,3-
pentadiene in a ready-to-drink refrigerated tea prod-
uct is shown in Figure 2. A consumer complained
that this particular tea sample had a kerosene odour.
High Boiling Point Compounds with Musty Odours
While extremely volatile compounds can be challeng-
ing to extract and isolate, so too are high boiling
point semivolatile chemicals. Sometimes it is neces-
sary to use combinations of sample preparation tech-
niques to extract and isolate sufRcient quantities of
this type of malodorous compound from foods to
achieve meaningful analytical results.
Algae, fungi, bacteria and Actinomycetes are
known to produce geosmin (GSM) and 2-methyl-
isoborneol (MIB). These semivolatile, lipophilic com-
pounds have a muddy, musty odour perceived as
disagreeable to consumers. Both compounds are rap-
idly absorbed from water into the lipid tissue of Rsh
and other aquatic organisms. When either compound
is present in tissue at concentrations exceeding
0.7
g kg\1, they render Rsh unRt for retail sale.
Current methods for quantifying the concentra-
tions of MIB and GSM in catRsh include: purge-
and-trap-solvent extraction (P&T-SE); microwave
distillation}solvent extraction (MD-SE) and micro-
wave distillation}solid phase extraction (MD-SPE).
These methods are time-consuming, labour-intensive
and require the use of small quantities of Sammable
and/or toxic solvents or expensive microwave equip-
ment. A faster and less expensive method could Rnd
broad application in catRsh Savour research, the cat-
Rsh-processing industry and other aquaculture indus-
tries plagued by this problem.
Lloyd and Grimm, USDA research chemists, have
developed a rapid and simple analytical procedure for
quantitating low levels of GSM and MIB in catRsh
tissue. Their method combines microwave distillation
(MD) with SPME. MD transfers lipophilic volatile
analytes from the lipid-rich matrix of catRsh tissue
into an aqueous matrix, and SPME is then used to
extract and concentrate the volatile organic com-
pounds from the aqueous solution. The technique is
a prime example of how combinations of two or more
sample preparation techniques can be a potent strat-
egy for resolving analytical problems that are inad-
equately addressed by a single sample preparation
technique.
While SPME has been shown to be a sensitive,
reproducible, quantitative sample preparation tool,
the direct analysis of p.p.b. levels of GSM and MIB in
Rsh tissue is not possible with SPME. Due to their
lipophilic nature, MIB and GSM partition from Rsh
tissue into the headspace in such low concentrations
that direct SPME is ineffective. Combining MD with
SPME yields a rapid, extremely sensitive technique
for the analysis of thermally stable volatile and
semivolatile compounds in complex matrixes.
Figure 3 is a schematic diagram of a typical MD-
SPME apparatus for analysing MIB and GSM in Rsh
tissue.
Mouldy/Musty Chemicals in Wine and Corks
Cork from Quercus suber has been used as a closure
for wine bottles since the 17th century. Cork offers
unique physical properties as a closure, including
4184 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications

Figure 2 Volatiles in tea with a kerosene-like off-flavour by SPME (Carboxen-PDMS) extraction followed by GC-MS. Sample
preparation: 2 mL of tea and a small magnetic stirring bar were added to a 4 mL GC vial and sealed. Headspace volatiles were
extracted by SPME for 20 min at 503C. Peak identities: 1, acetone; 2 and 3, 1,3-pentadiene isomers; 4, 2-butanone; 5, pentanal; 6,
2-pentanone; 7, hexanal; 8, 4-methyl-6-hepten-3-one; 9, 2,3-dehydro-1,8-cineole; 10, hexyl acetate; 11, 1,4-cineole; 12, 1,8-cineole;
13,
-terpineol.

long-lasting Sexibility, hydrophobicity and gas im-
permeability. Over the last two decades, the incidence
of mouldy and musty off-Savours in cork-sealed
wines has increased signiRcantly. 2,4,6-Trichloro-
anisole (TCA) has been identiRed as the primary
chemical responsible for cork taint. The human olfac-
tometry threshold for TCA is 4}10 ng L\1 in white
wine and 50 ng L\1 in red wine. In the case of wine,
a worldwide loss of roughly US$1 billion per year is
attributed to cork taint.
The use of SPME Rbres to extract TCA from the
headspace over an agitated wine and moistened cork
matrix is a short, inexpensive and solvent-free
method to determine TCA. Due to the efRcient ad-
sorption properties of PDMS SPME Rbres and the
high sensitivity of GC-MS, the limit of detection of
2.9 ng L\1 TCA is low enough to detect problem
wine and cork samples that exceed the olfactory
threshold range in wine of 4}50 ng L\1.
Immersion of the SPME Rbre into the wine was
found to give poorer sensitivity and can increase
contamination of the injector system and shorten the
lifetime of the SPME Rbre and analytical GC column.
Free Fatty Acids by Headspace and
Immersion Techniques
Free fatty acids (FFAs), even at relatively low concen-
trations, are critical to both desirable and undesirable
Savours in many types of food systems. Low levels of
FFAs are difRcult to detect in cheese and other food
samples by dynamic or static headspace methods.
SPME offers two alternative approaches to determine

Figure 3 MD-SPME apparatus for analysing 2-methylisoborneol and geosmin in fish tissue.
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
Table 4 Linearity of responses for free fatty acids using immer-
sion SPME (50 p.p.b.}25 p.p.m.)
Acid % RSD of response factor
Acetic 140
Propionic 16.1
Isobutyric 14.4
Butyric 18.9
Isovaleric 12.1
Valeric 14.2
Hexanoic 9.1
Courtesy of Dr Robert Shirey, Supelco Inc., Bellefonte, PA.
these compounds in cheese: the solid sample can be
warmed for headspace sampling, or the sample can be
liqueRed for sampling by immersion SPME. Shirey,
Supelcos SPME applications chemist, investigated
both approaches for monitoring FFAs in cheeses us-
ing varied extraction conditions.
The headspace SPME approach offered the greatest
sensitivity for these analytes, but immersion of the
Rbre into the liqueRed samples produced the widest
range of linear responses. Under all conditions, acetic
acid was particularly difRcult to quantify (Table 4).
The following conditions were used for the analysis
of Parmesan cheese for FFAs: sample: 100 mg cheese
in 40 mL vial; SPME Rbre: 65
m Carbowax/divinyl-
benzene StableFlexTM; extraction method: headspace
for 15 min at 653C; desorption: 1 min at 2503C.
Sanitizer Contamination in Milk
The food and beverage industry is now less dependent
on chlorine-based sanitizers for disinfecting process-
ing equipment. Because application does not lead to
toxic halogenated organic compounds, peroxyacetic
acid (PAA)-based sanitizers are now widely used for
disinfection in cleaning-in-place (CIP) systems in
breweries and dairies. One problem with PAA-based
sanitizers, however, is that even small amounts of
PAA contamination can lead to severe off-
Savours in milk. This problem can occur if sanitizers
are not completely rinsed from processing lines prior
to processing the next load of milk.
PAA, which can be quantitated in milk by HPLC
after derivatization with methyl p-tolylsulRde, has
a half-life in milk of approximately 20 min. As a re-
sult, PAA concentrations normally fall below thre-
shold taste limits after only a few hours, even in milk
contaminated with relatively large quantities of PAA.
Once milk is contaminated with PAA, however, there
is a signiRcant off-Savour that fails to dissipate over
time. The PAA-induced reactions that lead to this
off-Savour defect are not well understood but prob-
ably involve oxidation of the milk proteins by PAA
and/or hydrogen peroxide. To determine if an
4185
off-Savour in milk has occurred because of PAA con-
tamination, one approach is to check acetic acid
levels, since PAA degrades to water and acetic acid.
Headspace SPME with a Carboxen-PDMS or a Car-
bowax-divinylbenzene StableFlex Rbre is capable of
detecting p.p.b. levels of acetic acid in milk.
One popular sanitizer used by some dairies is
MatrixxTM (Ecolab, St Paul, MN). Matrixx has the
following composition (approximate): 4.4% PAA,
6.9% hydrogen peroxide and 3.4% octanoic
acid. Figure 4 shows chromatograms of a control milk
sample (no off-Savour) and a sample with a severe
off-Savour that was suspected to be caused by contami-
nation with Matrixx. Peaks for acetic and octanoic
acids are indicators that the sample is contaminated
with Matrixx sanitizer. The following conditions were
used for the analysis: sample: 2 mL of low fat
milk#1 mL 0.1-N phosphoric acid#1 g salt in
a 9 mL vial; SPME Rbre: 65
m Carboxen-PDMS; ex-
traction method: headspace (with stirring) for 12 min at
403C; desorption: 2 min at 2503C. The analytical capil-
lary column was FFAP2+ (Free Fatty Acid Phase).
Off-Flavours from Packaging Materials
Ironically, packaging materials, which are designed
to preserve the freshness and Savour of foods and
beverages, can be directly responsible for causing
off-Savour defects. Although plastic packaging ma-
terial consists primarily of nonvolatile high molecular
weight polymers, volatile low molecular weight com-
pounds are often added to improve functional prop-
erties of the materials: plasticizers to improve Sexibil-
ity, antioxidants to prevent oxidation of the plastic
polymers and the food inside the packaging and UV
blockers to prevent yellowing of polymeric material
when it is exposed to light. Additional additives in-
clude polymerization accelerators, cross-linking
agents, antistatic chemicals and lubricants.
Occasionally, packaging materials are not ad-
equately cured before they are used. As a result,
a small amount of solvent associated with the manu-
facturing of the packaging materials or from the inks
and dyes used on packaging graphics remains and is
absorbed by the food material inside the package.
Screening packaging material for undesirable resid-
ual solvents is a simple task with SPME. Figure 5
shows volatiles extracted from the headspace of
a closed, new (unused) cottage cheese carton (680 g
Rll weight). The lidstock is a linear low density poly-
ethylene (Dow 2503 resin), and the container body is
polypropylene (Montell copolymer). The volatiles
were sampled simply by poking a pinhole through the
top of the closed container and inserting an SPME
Rbre (Carboxen-PDMS) through the hole. A small
magnetic stirring bar was placed inside the carton to
4186 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications

Figure 4 (A) Low fat milk control and (B) complaint low fat milk with off-flavour. Peak identities are as follows: 1, acetic acid; 2,
internal standard (2-ethylhexanoic acid); 3, octanoic acid. Complaint sample is contaminated with 0.11% Matrixx sanitizer. Concentra-
tion of octanic acid is 37 p.p.m. See text for details of method.

facilitate air movement over the Rbre. The Rbre was detected. Nearly all peaks detected were hydrocar-
exposed to the atmosphere in the carton for 30 min at bons of various chain lengths. However, a signiRcant
room temperature. A large number of volatiles was amount of trichloroethylene was also detected.

Figure 5 Volatiles extracted from the headspace of a closed, new (unused) cottage cheese carton (680 g fill weight) by SPME. Peak
no. 1 is trichloroethylene; most of the other chromatographic peaks are alkanes. See text for details of method.
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
A few types of malodorous packaging solvents that
have been found to cause off-Savours in foods include
styrene, ethylstyrene, trimethylbenzene isomers and
propyl acetate.
Pesticides in Wine
Not all food taints involve odiferous chemicals that
contribute to off-Savours. Contamination of foods
with pesticides is another type of food taint of critical
concern. Wine is one type of beverage that can be
contaminated with pesticides.
Procymidone fungicide Procymidone is a fungicide
which is widely used against Botrytis cinerea on wine
grapes. If improperly applied, undesirable residues at
concentrations ranging from a few p.p.b. to several
hundred p.p.b. can be found in wine after fermenta-
tion and even in old bottles because of its well-known
persistence. The standard analytical sample prepara-
tion method for testing procymidone in wine is based
on time-consuming liquid}liquid extraction or solid-
phase extraction (SPE) using polymeric bonded silica
cartridges.
Urruty and co-workers at the UniversiteH de Bordeaux
(PeH rigueux, France) found that SPME (100
m PDMS)
results for procymidone in white and red wine corre-
lated very well to ELISA test results. SPME was as fast
as ELISA and offered slightly better precision.
Methyl isothiocyanate soil fumigant Another chem-
ical of concern to wine makers is methyl isothiocyan-
ate (MITC). It is used as a soil fumigant for
nematodes, fungi and other diseases in vegetables and
fruits. MITC is illegally employed as an antifermen-
tative substance in wines. The addition of antifermen-
tative agents in wines is controlled by EC and non-EC
regulations. In particular, the Italian legal system
does not allow the use of MITC in wines and requires
the control of all exported wines. Solvent}solvent
extraction is the traditional sample preparation
method for measuring MITC in wines.
Grandini and Riguzzi (Bologna, Italy) compared
SPME with the ofRcial Italian method. The SPME
Rbre used was Carbowax-divinylbenzene (65
m).
For SPME, headspace sampling of 5 mL of wine in
a 10 mL vial was conducted for 30 min; 1.25 g of
sodium chloride was added to the sample.
The lengthy standard sample preparation for
MITC in wine was as follows: a 100 mL sample of
wine was spiked with 100
L 4-ethylpyridine (inter-
nal standard). The pH of the wine was adjusted to
7 with sodium hydroxide. The sample was then ex-
tracted three times with 15 mL of pentane. Anhyd-
rous sodium sulfate was added to the solvent, which
was then concentrated to 0.3 mL with a rotary
4187
evaporator at 403C. No vacuum was applied, in order
to minimize MITC loss.
SPME-GC with a nitrogen-phosphorus detector
(NPD) gave a minimum detectable limit of 1 p.p.b.
and a linear detector response in the 1}200 p.p.b.
range. Although many methods use the NPD, includ-
ing the ofRcial method, they are not able to obtain
minimum detectable limits of less than 10 p.p.b.
Compared to the ofRcial method, SPME offered the
following advantages: low minimum detection limits,
wide linearity range, short analysis time and low
costs. Furthermore, sample pretreatment is elimi-
nated and solvents are not used.
Quality Control (QC) Applications:
SPME-MS-MVA as an Electronic Nose
The combination of SPME with GC and mass spec-
trometry}olfactometry detection is a potent tool for
understanding the causes of food off-Savours, mal-
odours and taints. However, the complexities in-
volved in performing capillary GC testing, as well as
the difRculties associated with the interpretation of
results, require highly trained chemists. Furthermore,
the technique is time-consuming and not
amenable to the rapid product evaluation and deci-
sion-making that is often required in quality control
situations. Even with assistance from peak recogni-
tion software that matches corresponding peaks in
different chromatograms, the large number of GC
peak data associated with Savour/off-Savour studies
of food systems is time-consuming and prone to er-
rors. As a result, SPME-GC-MS-OD is essentially
a tool for research and development chemists and
chromatographers.
Advantages of SPME-MS-MVA for QC Applications
There is, however, a relatively new SPME-based tech-
nique that has proved useful for food quality control
applications. The technique has been referred to as
SPME-MS-MVA (solid-phase microextraction}mass
spectrometry}multivariate analysis). Essentially, the
analytical system is an electronic nose (e-nose) in
which a mass spectrometer replaces the typical chem-
ical sensor array, and SPME replaces static or dynamic
headspace sampling as the extraction technique to
introduce volatiles/semivolatiles to the detector. The
GC is used, with the only modiRcation being the sub-
stitution of the typical 30 m coated capillary column
with a 1 m uncoated fused silica column.
The speed, simplicity, sensitivity, portability and
relatively low cost of SPME make it an ideal extrac-
tion technique for introducing volatiles and
semivolatiles to the e-nose detector. With multiple
manual SPME set-ups, it could be possible to analyse
4188 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications

one sample every 3 min using the same GC-MS sys-

tem. Another advantage of using SPME as a way of
introducing volatiles into the e-nose detector is that
different Rbres can be selected for different applica-
tions (see Table 2).
Using a mass spectrometer as a chemical sensor is
advantageous because it is sensitive and robust, does
not suffer from memory effects, and is not poisoned
by low levels of moisture injected from SPME extrac-
tions. Furthermore, unlike typical commercial e-nose
chemical sensors based on conducting polymers,
metal oxides, surface acoustic wave (SAW) devices,
quartz crystal microbalances (QCMs), or combina-
Figure 6 Principal component analysis scores plot of mass
tions of these devices, reliable easy-to-use benchtop intensity data for control and light-abused soybean oils as deter-
MS detectors have been in routine use for decades and mined by SPME-MS-MVA. Soybean oil: days of fluorescent light
have a proven track record. exposure (200 FC). 0, 0 days; 4, 4 days; 7, 7 days; D7, 7 days in
Another advantage of SPME-MS-MVA is that it the dark.
can easily be converted to SPME-GC-MS simply by
replacing the 1 m uncoated fused silica transfer line 4. The resulting mass intensity list provided the data
with an appropriate 30 m coated capillary GC col- used for PCA.
umn. Researchers can then perform more detailed
traditional analyses, including identiRcation and Two QA/QC examples of SPME-MS-MVA are pro-
quantitation of speciRc odour-active GC peaks. This vided below.
approach can be extremely helpful in determining
what masses to monitor (as well as what masses to Off-Wavour development in soybean oil exposed to
exclude) for speciRc e-nose application using MS as light Deodorized commercial soybean oil was ex-
the chemical sensor. posed to Suorescent light for different time periods
and analysed by SPME-MS-MVA. Prior to extrac-
Speci\c SPME-MS-MVA QC Applications tion, the soybean oil was placed in a 50 mL Nessler
With SPME-MS-MVA, the ability to identify indi- tube and exposed to 200 foot candles (FC) of Suor-
vidual chemical components is lost. However, the escent light. Four different types of samples were
trade-off is the gain in speed and simplicity of inter- analysed: control soybean oil (fresh oil, normal taste,
pretation of results. The technique is rapid and gener- no light exposure); control oil exposed to light
ally gives comparative rather than quantitative for 4 days; control oil exposed to light for 7 days;
information. It is ideally suited for quick quality as- and a Nessler tube Rlled with control oil, wrapped
surance (QA)/QC screening. in aluminium foil, and stored alongside the
SPME-MS-MVA generates mass intensity tables
for each sample tested. The mass intensity data used
to prepare the principal component analysis (PCA)
scores plots in Figures 6 and 7 were obtained in the
following manner:
1. Sample volatiles were extracted using SPME
(65
m Carboxen-PDMS) and desorbed from the
SPME Rbre by the heated GC injection port
(2503C) into a 1 m deactivated fused silica transfer
line heated to 503C.
2. Data acquisition (from m/z 50 to m/z 150) was
discontinued after 2 min.
3. The masses of the single resulting chromato-
graphic peak generated by the ion fragments
Figure 7 Principal component analysis scores plot of mass
from headspace volatiles of the sample were aver-
intensity data for fresh boiled beef and boiled beef refrigerated for
aged from 8 to 80 s, while masses from 0 to 4 days and 6 days and then reheated. Results generated by
7 s and from 81 to 100 s were subtracted as SPME-MS-MVA technique. 0, 0 days (freshly boiled); 4, 4 days
background. storage at 43C; 6, 6 days storage at 43C.
 III / SOLID-PHASE MICROEXTRACTION / Food Technology Applications
light-exposed oils for 7 days. All Nessler tubes were
sealed with ParaRlm and stored at 223C. Six samples
of each type were prepared and analysed, except for
the 7-day-old sample stored in the dark (i.e. wrapped
in foil); only three samples of this treatment were
analysed.
SPME procedure 2 g soybean oil )was added to
a 9 mL glass GC vial and capped with a polytetra-
Suoroethylene septum closure. Samples were heated
to 453C in a water bath and stirred vigorously with
a small stirring bar while the SPME Rbre was exposed
to the headspace vapours in the vial for 12 min.
Results The PCA scores plot for this set of samples
appears in Figure 6, which shows that SPME-MS-MVA
is capable of grouping together samples of soybean oil
that have been exposed to similar levels of light abuse.
Warmed-over Wavour (WOF) in boiled beef A beef
sample (500 g of chuck roast) was boiled for 60 min
in a water bath. The internal temperature of the beef
reached 923C. Immediately after boiling, the hot meat
was ground in a meat grinder, split into six separate
samples and analysed by SPME-MS-MVA. After stor-
age at 43C for 4 days, the samples were reheated to
503C in a convection oven for 30 min. Organoleptic
evaluation of the samples showed that their Savour
had changed from a typical beef Savour to an off-
Savour characterized as tallowy, green and metallic.
Samples were again refrigerated, stored for an addi-
tional 48 h, and re-analysed after warming to 503C.
Figure 8 Development of warmed-over flavour in cooked beef.
SPME-GC-MS chromatogram of boiled beef (A) at 0 days and (B)
after 6 days of storage and then reheated to 503C. Peak identities:
1, pentanal; 2, hexanal; 3, heptanal; 4, 2,4-nonadienal; 5, octanal;
6, 2,3-octanedione; 7, nonanal; 8, 1-octen-3-ol; 9, 2-heptenal.
4189
Samples after 6 days of storage developed even stron-
ger WOF notes.
SPME procedure 0.5 g boiled beef (ground) plus
2.5 mL water were added to a 9 mL glass GC vial. All
other conditions were the same as the soybean oil
SPME procedure given above.
Results The PCA scores plot for this set of samples
appears in Figure 7. SPME-MS-MVA is capable of
identifying groups of samples with similar levels of
WOF.
To ensure that SPME was measuring volatiles that
are known to contribute to WOF (e.g. aliphatic al-
dehydes, 2,4-nonadienal, etc.), a fresh boiled beef
sample (0 days) and a 6-day sample were analysed by
SPME-GC-MS. The resulting chromatograms, shown
in Figure 8, prove that SPME is extracting com-
pounds that have been identiRed as the source of
WOF by other researchers. The chromatogram was
generated using the identical method used for SPME-
MS-MVA, with the exception that the 1 m transfer
line was replaced with a 30 m FFAP capillary column.
Conclusion
As the numerous examples in this article illustrate,
SPME is one of the most potent extraction, isolation
and concentration techniques available for studying
off-Savour chemicals in foods and beverages. Im-
provements in SPME technology will probably be
made in the near future, making the technique even
more useful to Savour chemists. Important recent
developments in Rbre technology include:
1. StableFlexTM Rbres (which exhibit greater Sexibil-
ity and increased strength compared to previous
Rbres);
2. a highly cross-linked PDMS Rbre coating to min-
imize bleed and improve thermal stability;
3. coatings containing micro-adsorbent beads for re-
tention and selectivity for many polar and volatile
analytes;
4. dual-coated Rbres that have the ability to efRci-
ently extract low levels of both polar and nonpolar
analytes in the same sample.
See also: II/Chromatography: Gas: Headspace Gas
Chromatography. Extraction: Solid-Phase Microextrac-
tion. III/Airborne Samples: Solid Phase Extraction.
Fragrances: Gas Chromatography.
Further Reading
Charalambous G (ed.) (1978) Analysis of Foods and Bever-
ages. New York, NY: Academic Press.
4190 III / SOLID-PHASE MICROEXTRACTION / Overview
Charalambous G (ed.) (1992) Off-Flavors in Foods and
Beverages. Amsterdam: Elsevier Science.
Contis ET, Ho C-T, Mussinan CJ, Parliament TH, Shahidi
F and Spanier AM (eds) (1998) Food Flavors: Forma-
tion, Analysis and Packaging InUuences. Amsterdam:
Elsevier Science.
Elmore JS, Mehmet AE and Muttram DS (1997) Compari-
son of dynamic headspace concentration of tenax with
solid phase microextraction for the analysis of aroma
volatiles. Journal of Agriculture and Food Chemistry 45:
2638}2641.
Heath HB and Reineccius G (eds) (1986) Flavor Chem-
istry and Technology. New York: Van Nostrand
Reinhold.
Overview
J. R. Dean, University of Northumbria at Newcastle,
Newcastle upon Tyne, UK
Copyright ^ 2000 Academic Press
Introduction
Solid phase microextraction (SPME) has been applied
to a diverse range of analytes and sample types. The
growth in the application of SPME, since its inception
in 1990, can be seen in Figure 1 (information from
the Science Citation Index, February 1999). SPME is
used as both a method of preconcentration and as
a sampling device for (predominantly) chromato-
graphic analysis. SPME has been used in conjunction
with a range of other techniques, such as, ultraviolet
and infrared spectroscopy, Raman spectroscopy and
mass spectrometry, but it is its use in chromato-
graphic analysis which is the focus of this article.
SPME has most commonly been coupled to gas chro-
Figure 1 Frequency of SPME publications per year (informa-
tion from the Science Citation Index, February 1999, Copyright
International Scientific Communications, Inc.).
Ho C-T and Manley CH (eds) (1993) Flavor Measurement.
New York: Marcel Dekker.
Marsili RT (ed.) (1997) Techniques for Analyzing Food
Aromas, pp. 237}289. New York: Marcel Dekker.
Marsili RT (1999) Comparison of solid phase microextrac-
tion and dynamic headspace method for the GC-MS
analysis of light-induced lipid oxidation products in
milk. Journal of Chromatography Science 37: 17}23.
Marsili RT (1999) SPME-MS-MVA as an electronic nose
for the study of off-Savors in milk. Journal of Agricul-
ture and Food Chemistry 47: 548}654.
Scheppers-Wercinski SA (ed.) (1999) Solid Phase Microex-
traction: A Practical Approach. New York: Marcel
Dekker.
matography (GC), although some applications have
coupled it to high-performance liquid chromatogra-
phy (HPLC) (Figure 2). The following discussion will
concentrate primarily on the use of SPME coupled
with GC.
The SPME device consists of a fused silica Rbre,
coated with a stationary phase (Table 1) and moun-
ted in a syringe-type holder (Figure 3). The SPME
holder has two functions: to provide protection for
the Rbre and allow insertion into the hot environment
of the GC injector using a needle. As samples and
standards are normally introduced into a GC via
a syringe the use of this device offers no additional
complexity.
At rest the fused silica-coated Rbre is retracted
within the protective needle of the SPME holder. In
operation however, the Rbre is exposed to the analyte
within its matrix (air, water, solid) for a predeter-
mined amount of time. The active length of the Rbre
is typically 1 cm. Two common approaches for
sample extraction are employed; direct and head-
space (Figure 4). The Rrst involves direct contact be-
tween the coated Rbre and the sample matrix; in this
way analytes within the sample are able to be trans-
ported to the Rbre coating. This transportation can be
achieved by several means. In the case of liquid (or
solid samples that have been mixed with an aqueous
solution, i.e. a slurry), transportation is achieved by
agitation of the sample vial, agitation of the Rbre,
stirring or sonication of the sample solution. For
gaseous samples, natural convection is usually sufR-
cient. In the headspace mode, the process relies on the
release of volatile compounds from the sample
matrix. This may be achieved by heat, chemical modi-
Rcation or the inherent volatility of the analyte.
After sampling, the Rbre is retracted within its
holder for protection until inserted in the hot injector
 III / SOLID-PHASE MICROEXTRACTION / Overview 4191

Table 1 Commercially available fibre coatings

z 7
m Polydimethylsiloxane (bonded)
z 30
m Polydimethylsiloxane (non-bonded)
z 100
m Polydimethylsiloxane (non-bonded)
z 85
m Polyacrylate (partially crosslinked)
z 60
m Polydimethylsiloxane/divinylbenzene
(partially crosslinked)
z 65
m Polydimethylsiloxane/divinylbenzene
(partially crosslinked)
z 75
m Polydimethylsiloxane/Carboxen
(partially crosslinked)
z 65
m Carbowax/divinylbenzene (partially crosslinked)
z 50
m Carbowax/Template resin (partially crosslinked)

Quantitation in SPME is achievable in much the

same way as for any other sample analysis. For
example, in GC a series of standard solutions are
prepared in organic solvent over the appropriate con-
centration range for the analytes under investigation.
From the results obtained a calibration graph can be
constructed [a plot of signal intensity (area or peak
height) versus concentration]. Then, an organic sol-
vent extract of the unknown is injected into the GC
and its response compared to the calibration graph. In
the same manner for SPME, a series of standard
Figure 2 Solid phase microextraction}high-performance liquid
chromatography interface (reproduced with permission, from solutions need to be prepared in aqueous solution or
Analytical Chemistry 67: 2530, 1995, Copyright American soil slurry form. The Rbre is then exposed to the
Chemical Society). solution (or soil slurry) for a prespeciRed time and
then introduced into the hot injector of the GC. In
of the GC or mobile phase of the HPLC; desorption
of analytes occurs due to the inSuence of temperature
(GC) or organic solvent (HPLC). In either case the
Rbre is exposed for a particular time to allow for
effective desorption of the analytes. As the coating on
the Rbre is selective towards the analyte, it is common
to Rnd that no solvent peaks are present in the sub-
sequent chromatograms. As the Rbre coating is selec-
tive towards the target analytes it is important to
select the most appropriate Rbre coating for the
sampling process. Figure 5 compares the inSuence of
three Rbre coatings, i.e. polystyrene}divinylbenzene
(XAD), polyacrylate, and polydimethylsiloxane
(PDMS) for the extraction of 49 organophosphorus
pesticides from a water sample. The selectivity of
each Rbre coating is evident from the chromatograms
(Figure 5).
It is important to note that the Rbre can equally
adsorb analytes from the atmosphere as well as the
sample (in some cases the atmosphere may be the
sample). Extreme caution should be taken Rrst of all
to clean the Rbre. This can be done, for example, by
exposing the Rbre to the hot injector of the GC before
sampling. Also, it is important to minimize the time Figure 3 Solid phase microextraction device (reproduced with
between the sorption step and the subsequent desorp- permission from Analytical Chemistry 66: 844A, 1994, Copyright
tion and analysis step. American Chemical Society).
4192 III / SOLID-PHASE MICROEXTRACTION / Overview
Figure 4 Common approaches for SPME. (A) Direct SPME and (B) headspace SPME.
this manner a calibration graph can be constructed.
Similarly, an aqueous SPME extract (or slurry ex-
tract) of an unknown sample is injected in the GC and
its signal response compared with the calibration
graph. Calibration is also done in this manner for
headspace SPME, the difference being that the Rbre is
exposed to the headspace above the sample only and
not placed in the solution or soil slurry itself. It is
common practice to utilize an internal standard for
all quantitative analysis. Calibration is also possible
using the method of standard additions. For further
information on quantitative headspace methods see
the book by Kolb and Ettre listed in the Further
Reading section.
The diversity of applications of SPME is continual-
ly expanding, limited only by peoples ingenuity, so it
is not unfamiliar to Rnd applications of SPME in such
diverse areas as environmental and clinical, food and
pharmaceutical, forensic and military use. However,
the most popular application area is environmental
analysis (water and soil). In order to provide exam-
ples of the diversity of applications, selected areas
have been considered. For further information, the
reader is recommended to consult the Further Read-
ing Section or the current scientiRc literature.
Extraction of Analytes from Aqueous
Matrices
Analysis of polar and labile analytes in aqueous
matrices usually involves extraction and preconcen-
tration. This has traditionally been based on
liquid}liquid extraction (LLE). In this context,
a small volume of organic solvent is added to a larger
volume of the aqueous sample and shaken (it may be
necessary to salt-out the analytes, this is done by
saturating the aqueous sample with an inorganic
salt). The organic phase containing the analytes is
then analysed. [Note: additional preconcentration
may be required using evaporation in a stream of
inert gas (manual or automated) or vacuum evapor-
ation.] However, if the analytes are sufRciently vol-
atile they can be purged from an aqueous sample
using a gas, such as nitrogen, preconcentrated by
trapping on a suitable sorbent, e.g. Tenax, at low
temperature and eluted by rapidly heating the trap.
The analytes are then directly transferred into a gas
chromatograph for separation and detection. This
procedure, known as dynamic headspace or purge
and trap sampling is an effective procedure for vol-
atile analytes. An alternative to the requirements for
extraction and preconcentration of non-volatiles is
solid phase extraction (SPE).
SPE uses a stationary phase, such as C18-silica, to
adsorb analytes from a large volume of sample solu-
tion. Elution of analytes is then achieved by using
a small volume of organic solvent. In this manner,
effective extraction and preconcentration is achieved.
The use of SPME takes this method a stage further in
miniaturization.
Effective extraction and preconcentration of
analytes in aqueous matrices can be achieved using
SPME. Two approaches are commonly used. In the
Rrst approach, the Rbre is inserted directly into
an aqueous sample for a prespeciRed time, with or
 III / SOLID-PHASE MICROEXTRACTION / Overview 4193

Figure 5 SPME of 3
g L\1 organophosphorus pesticides. (A) 15
m XAD polystyrene}divinylbenzene)-coated fibre, (B) 85
m
polyacrylate-coated fibre, and (C) 30
m polydimethylsiloxane-coated fibre. (Reproduced with permission from Journal of High
Resolution Chromatography 20: 487, 1997, Copyright John Wiley & Sons Limited.) GC conditions: column 30 m length ;0.25 mm
internal diameter ;0.25
m film PTE-5 fused silica open tubular; temperature programme 603C (4 min hold) to 1503C at 303C min\1
and from 150 to 3003C at 53C min\1 (hold for 3 min). SPME conditions: 15
m XAD coated fibre; absorption time, 30 min; desorption
time, 20 min at 2703C. Eighty-five
m polyacrylate coated fibre; adsorption time, 30 min; desorption time, 20 min at 2803C. Thirty
m
polydimethylsiloxane coated fibre; adsorption time, 30 min; desorption time, 20 min at 3003C. Spiking level was 3
g L\1 per
compound; sample volume was 1.5 mL. Peak identification: 1"aspon; 2"azinphos-ethyl; 3"azinphos-methyl; 4"bolstar;
5/6"carbophenothion/famphur; 7"chlorfenvinphos; 8/9"chlorpyrifos-methyl/parathion-methyl; 10/11"chlorpyrifos/parathion-
ethyl; 12"coumaphos; 13"crotoxyphos; 14"demeton-O; 15"demeton-S; 16"diazinon; 17"dichlorfenthion; 18"dichlorvos;
19"dicrotophos; 20"dimethoate; 21"dioxathion; 22/23"disulfoton/phosphamidon; 24"O-ethyl-O-(4-nitrophenyl)phenylphos-
phono-thioate (EPN); 25"ethion; 26"ethoprop; 27"fenitrothion; 28"fensulfothion; 29"fenthion; 30"fonophos; 31"hexa-
methylphosporamide (HMPA); 32"leptophos; 33"malathion; 34"merphos; 35"mevinphos; 36/37"monocrotophos/sulfotepp;
38"naled; 39"phorate; 40"phosmet; 41"ronnel; 42"stirophos; 43"tetraethylpyrophosphate (TEPP); 44"terbufos;
45"thionazin; 46"tri-O-cresylphosphate; 47"tokuthion; 48"trichlorfon; and 49"trichloronate.

without stirring and with or without the addition of
salt. The Rbre is then retracted into its protective
holder and the adsorbed analytes desorbed in either
the hot injector of the GC or in the mobile phase of an
HPLC system. This approach is to be favoured for the
more non-volatile, labile type of analytes. The alter-
native approach is to place a small volume of the
liquid sample in a sealed vial and to insert the Rbre
into the headspace above the sample for a prespeci-
Red time. Again, stirring may be beneRcial as well as
the addition of salt. In addition, warming the sample
vial may prove to be beneRcial by increasing the
concentration of volatile analytes in the headspace
above the sample.
4194 III / SOLID-PHASE MICROEXTRACTION / Overview

Table 2 Limits of detection (ng L\1) for selected pesticides of 20 organochlorine pesticides extracted from
from water using a 95
m polyacrylate coated fibre a groundwater sample by both SPME and LLE are
shown in Figure 7. In the case of SPME, a 30
m
Compound FID a NPD b MS c MS d
polydimethylsiloxane Rbre was inserted in a sample
EPTC 2000 50 0.8 16 volume of 1.5 mL for 20 min. Desorption
Butylate 1000 20 0.1 1 was achieved by insertion into the GC injector for
Vernolate 1000 20 0.5 2 10 min at 2603C. The spiking level was 1
g L\1.
Pebulate 1000 20 1 19
For LLE a 100 mL sample spiked at the 0.5
g L\1
Molinate 2000 60 0.3 12
Propachlor 6000 800 15 16 level was extracted with 20 mL, then 10 mL of
Cycloate 800 20 0.05 1 hexane. The combined extracts were dried with an-
Trifluralin 400 30 0.02 1 hydrous sodium sulfate and concentrated to l mL
Benfluralin 300 30 0.4 1 using a stream of nitrogen prior to analysis. In most
Simazine 1000 70 1 15
cases similar results were obtained by SPME and
Atrazine 7000 40 3 11
Propazine 10 000 50 0.3 6 LLE. Anomalous results for endosulfan I and II were
Profluralin 200 30 0.1 1 reported.
Terbacil 15 000 200 1 9 Examples of the direct approach for non-volatile
Metribuzin 14 000 200 3 19 compounds, using SPME-HPLC, are shown in Fig-
Bromacil 19 000 400 0.1 8
ures 8 and 9. In Figure 8, a comparison is made
Metolachlor 1000 200 0.01 8
Isopropalin 300 10 0.1 1 between SPME and a 1
L loop injection for the
Pendimethalin 200 20 0.1 1 analysis of polycyclic aromatic hydrocarbons (PAHs)
Oxadiazon 300 30 0.01 1 using reversed phase HPLC. Using the SPME-HPLC
Oxyfluorofen 200 300 6 1 interface, as shown in Figure 2, thirteen PAHs have
Hexazinone 2000 6000 1 15
been analysed after sampling for 30 min using a 7
m
a
Determined from 100
g L\1 solutions. PDMS-coated Rbre. Some differences, in terms of
b
Determined from 10
g L\1 solutions. peak height, are noted (Figure 8) for peaks 1}4 when
c
Determined from 0.01
g L\1 solutions. SPME is compared with direct injection. These differ-
d
Calculated for the line of best fit with a zero intercept, over the ences are attributable to the selectivity of sampling
range 0.1}100
g L\1(n"3). Values (a)}(c) are from Boyd-Bo-
associated with SPME. The versatility of the SPME-
land AA and Pawliszyn J (1995) Journal of Chromatography 704:
163. Values for (d) are from Boyd-Boland AA et al. (1996) Analyst HPLC approach is further highlighted in Figure 9. In
121: 929. this case, an alkylphenol ethoxylate (Triton X-100) in
the aqueous phase is sampled for 60 min with stirring

Direct Extraction Table 3 Limits of detection (ng L\1) for selected pesticides
from water using a 100
m polydimethylsiloxane-coated fibre
Examples of the direct approach have allowed mul-
tiple analytes, e.g. pesticides, to be determined in Compound NPD a MS a MS b
aqueous samples. For example, limits of detection for
Dichlorvos 1500 80 30
the determination of pesticides in water by GC with EPTC 20 10 2
Same ionization detector (FID), nitrogen-phosphor- Butylate 50 20 1
ous detector (NPD) or mass spectrometer (MS), using Vernolate 100 20 1
a 95
m polyacrylate-coated Rbre, are shown in Pebulate 40 10 14
Molinate 110 20 4
Table 2. Other SPME conditions are as follows:
Cycloate 130 30 1
a 50 min equilibration time with stirring at room Simazine 360 10 18
temperature; and desorption by inserting the Rbre Atrazine 110 30 23
into the hot GC injector (2503C) for 5 min. Similarly, Propazine 40 10 5
selected detection limits for a 100
m polydimethyl- Diazinon 60 10 1
Disulfoton 40 10 0.7
siloxane Rbre are shown in Table 3. A typical
Metolachlor 220 20 9
SPME}GC}NPD chromatograph for the analysis of
drinking water spiked with 36 pesticides (EPA a
Determined from 100
g L\1 solutions. Other SPME conditions:
Method 507) at the 10
g L\1 is shown in Figure 6. 20 min equilibriation time from a saturated sodium chloride solu-
In addition, to evaluating the sensitivity of SPME tion at room temperature and pH 7. From Choudhury TK et al.
(1996) Environmental Science Technology 30: 3259.
by determining detection limits, an alternative ap- b
Calculated for the line of best fit with a zero intercept, over the
proach is to evaluate the performance of SPME range 0.1}100
g L\1(n"3). Other SPME conditions: 50 min
against a traditional aqueous extraction procedure equilibriation time with stirring at room temperature. From Boyd-
(liquid}liquid extraction). Results for the extraction Boland AA et al. (1996) Analyst 121: 929.
 III / SOLID-PHASE MICROEXTRACTION / Overview 4195

Figure 6 SPME}GC}NPD analysis of drinking water spiked with 10
g L \1 each pesticide (36). (Reproduced with permission from
American Chemical Society, Environmental Science and Technology, 30(11): 3259, 1996.) SPME conditions: 100
m polydimethyl-
siloxane fibre; adsorption time, 20 min; desorption time, 5 min at 2203C. Samples were extracted with stirring at ambient temperature,
at pH 7.0 and with a final 4.0 mL saturated sodium chloride solution. GC conditions: 30 m length;0.32 mm internal diameter
;0.25 mm film 5% phenyl/95% dimethylsilicone fused silica open tubular column; temperature programme 1003C to 3003C at
43C min\1. 1"dichlorvos, 2"EPTC, 3"butylate, 4"vernolate, 5"pebulate, 6"molinate, 7"cycloate, 8"ethoprop, 9"chlor-
propham, 10"simazine, 11"atraton, 12"prometon, 13"atrazine, 14"propazine, 15"terbufos, 16"pronamide, 17"dia-
zinon, 18"disulfoton, 19"disulfoton sulfone, 20"simetryn, 21"alachlor, 22"ametryn, 23"prometryn, 24"terbutryn,
25"metolachlor, 26"triademoton, 27"MGK 264, 28"diphenamid, 29"butachlor, 30"carboxin, 31"stirofos,
32"fenamiphos, 33"napropamide, 34"merphos, 35"norflurazon and 36"fenarimol.

and at room temperature. Desorption is achieved Headspace SPME from Water

by exposing the Rbre for l min to the mobile
In headspace SPME, the Rbre is exposed to the air
phase. Separation is achieved using normal phase
above an aqueous sample, which is in equilibrium with
HPLC.

Figure 7 Extraction of 20 organochlorine pesticides from groundwater: Comparison between SPME and LLE. (Adapted from Journal
of High Resolution Chromatography 19: 247, 1996.) SPME conditions: 30
m polydimethylsiloxane fibre; adsorption time, 20 min;
desorption time, 10 min at 2603C. Spiking level was 1 g L\1. Sample volume was 1.5 mL. There were three determinations. GC
conditions: 30 m length ;0.25 mm internal diameter ;0.25 mm film SPB-608 fused silica open tubular column; temperature
programme 1003C (4 min hold) to 1503C at 303C min\1 then to 3003C (8.6 min hold) at 83C min\1. LLE conditions: 100 mL sample
extracted with 20 mL, then 10 mL hexane. Extracts were then combined, dried with anhydrous sodium sulfate and concentrated to 1 mL
under a stream of nitrogen. Spiking level was 0.5 g L\1. There were three determinations. GC conditions: 15 m length ;0.53 mm
internal diameter ;0.88 mm film HP-5 fused silica open tubular column; temperature programme 1503C (0.5 min hold) to 2753C (5 min
hold) at 53C min\1.
4196 III / SOLID-PHASE MICROEXTRACTION / Overview
Figure 8 Separation of polycyclic aromatic hydrocarbons by
(A) 1
L loop injection, and (B) SPME using a 7
m PDMS-coated
fibre for 30 min from 100 ppb of each compound spiked into water.
(American Chemical Society, Analytical Chemistry, 67: 2530,
1995.) HPLC conditions: column, 25 cm;2.1 mm internal dia-
meter, 5
m ODS; flow rate, 0.2 mL min\1; (detection, UV
254 nm; solvent programme, acetonitrile}water (80 : 20, v/v) lin-
ear gradient to 100% acetonitrile in 15 min. SPME conditions:
7
m polydimethylsiloxane fibre; adsorption time, 30 min with stir-
ring. Spiking level was 100 ppb. Peak identification: 1"acenaph-
thylene, 2"fluorene, 3"phenanthrene, 4"anthracene,
5"pyrene, 6"benz[a]anthracene, 7"chrysene, 8"benzo[b]-
fluoranthene, 9"benzo[k ]fluoranthene, 10"benzo[a]pyrene,
11"dibenzo[ah]anthracene, 12"indeno[1,2,3-cd ]pyrene, and
13"benzo[ghi ]perylene.
the aqueous phase. For this approach to be useful the
analytes of interest must partition favourably into
the vapour phase. Therefore, the approach is
useful for volatile organic compounds in aqueous
samples. Most work has been done with the BTEX
compounds, i.e. benzene, toluene, ethylbenzene and
the xylene isomers.
Samples and standards are introduced into glass
vials, e.g. 40 mL volume, with TeSon-lined septa. It is
beneRcial for the speed of extraction to add a (TeSon-
coated) stirring bar and/or salt for salting-out to
improve sensitivity. Then, each vial is capped. The
septum is then pierced and the SPME device inserted.
The exposed coated-silica Rbre is positioned approx-
imately l cm above the surface of the aqueous sample.
The entire assembly is mounted on a magnetic stirring
plate. Care is required during the stirring process that
the vortex generated is not so vigorous so that the
aqueous sample comes into contact with the exposed
Rbre (a vortex of depth 1 cm is adequate). In addition,
the sample vial may be heated, by placing it in a
temperature controlled water bath at temperatures in
the range 40}803C. The extraction time can be varied
between 5 and 50 min, as desired. After a suitable
exposure time, the Rbre is retracted into its holder,
withdrawn from the vial and immediately inserted
into the hot injector of the GC for subsequent separ-
ation and detection. The typical performance of this
type of headspace SPME is summarized in Table 4. The
results in Table 4 compare the statistical detection limits
obtained by both the headspace SPME and purge and
trap approaches. In both cases, the statistical detection
limits were approximately an order of magnitude higher
than those required for the analysis of drinking water
(US EPA Method 524.2). The use of a more sensitive
detector, for instance an ion trap mass spectrometer,
could lower the detection limits achievable.
Extraction of Analytes from Solid
Matrices
Traditional approaches for the extraction of analytes
include Soxhlet extraction (and its variants), shake
Sask extraction and sonication. Soxhlet extraction is
frequently referred to as the benchmark technique, so
it is not suprising to Rnd that results obtained with
newer extraction techniques are compared to data
obtained by Soxhlet extraction. While Soxhlet is used
as the method of choice for many people for extract-
ing analytes from solid matrices, it is a time-consum-
ing process and uses relatively large volumes of
organic solvent. Alternatives have therefore been
sought to produce analytical data more rapidly and
that use smaller amounts of organic solvent (or none
at all). In this context, alternatives that have been
proposed include supercritical Suid extraction,
microwave-assisted extraction and pressurized Suid
extraction. However, the high capital cost of all these
alternatives and in some cases the level of expertise
required to operate the instruments effectively has
precluded their wide acceptance. In this context,
the use of SPME has been proposed. However, in
order for SPME to be of any use, the analytes must be
released from the solid matrix and enter either
a liquid phase or the gaseous phase. Variants on these
themes for SPME are now considered.
 III / SOLID-PHASE MICROEXTRACTION / Overview 4197

Figure 9 Normal phase HPLC chromatogram of extracted Triton X-100. Peak assignment refers to the number of units in the
ethoxylate chain. (American Chemical Society, Analytical Chemistry 68: 1521, 1996.) HPLC conditions: column, 25 cm;4.6 mm
internal diameter, 5
m Supelcosil LC-NH2; flow rate, 1.5 mL min\1; detection, UV 220 nm; solvent programme, 3}53%B, where A is
90 : 10, v/v hexane}2-propanol and B is 90 : 10, v/v 2-propanol}water. SPME conditions: carbowax/template resin fibre; adsorption
time, 60 min with stirring at room temperature; desorption time, 1 min. Spiking level was 100 ppm. Sample volume was 4 mL.

Several approaches can be adopted for the extrac- Direct (Slurry) SPME
tion of analytes from solid matrices using SPME.
For slurry extraction, a known quantity of sample,
These include direct extraction of the analytes from
e.g. 10 mg to l g of soil, is mixed with a solvent
a soil}water suspension or slurry; extraction of the
(water) and stirred. It may be necessary to adjust the
analyte from the sample matrix using hot water; or,
pH of the solution (to convert all compounds to
headspace extraction. In the Rrst two approaches, it is
a non-ionized form) and add salt to improve the
assumed that the analytes are highly soluble in water
extraction efRciency. The SPME Rbre is then exposed
and that water is a suitable solvent to liberate the
directly to the resultant suspension or slurry for
analyte from its matrix. The latter scenario assumes
a prespeciRed time (1}60 min) and then analysed. In
that the analytes of interest are volatile or semi-
addition, it also assumes that the matrix itself will not
volatile so that they are available in the headspace
interfere with the extraction process. If this is the
above the sample.
case, the SPME Rbre can be placed inside a protective
membrane in the slurry. The major limitation of this
Table 4 Analysis of BTEX compounds from aqueous samples: approach is that the membrane itself does not pre-
determination of statistical method detection limits (
g L\1)a clude any of the analytes of interest. However, this
approach has not yet been fully tested and further
Compound Headspace Purge and trap c EPA 524.2 d evaluation is necessary. Typical results for the analy-
SPME b
sis of chlorophenols from a contaminated land site
Benzene 0.70 0.38 0.03 are shown in Table 5 using the slurry SPME ap-
Toluene 0.30 0.37 0.05 proach and two methods of quantitation (direct calib-
Ethylbenzene 0.35 0.43 0.03 ration using an internal standard and the method of
m-/p-xylene 0.23 0.72 0.05
standard addition). The results are compared with
o-xylene 0.19 0.30 0.06
those obtained by Soxhlet extraction. A typical
a
Data from MacGillivray B et al. (1994) Journal of Chromato- SPME}GC}MS chromatogram of the soil sample is
graphic Sciences 32: 317. shown in Figure 10.
b
Headspace SPME conditions: 100
m polydimethylsiloxane
fibre was used to extract BTEX compounds from a 25 mL of water
Combined Hot Water Extraction}SPME
containing 10.0 g of NaCl. The sample was stirred and the tem-
perature maintained at 403C. The extraction time was 50 min. The An alternative to the slurry method is to extract
fibre was desorbed for 2 min at 1803C. Analysis was by GC-FID.
c
the solid sample with hot water and then isolate the
Purge and trap conditions: 5 mL samples were purged for 10 min
using a helium flow rate of 40 mL min\1 and a sample temper- analytes from the water using SPME prior to
ature of 403C. The compounds were trapped on a Tenax-charcoal chromatographic separation and detection. This
trap. Analysis was by GC-MSD in the full scan mode. is a relatively new approach with few relevant publi-
d
US Environmental Protection Agency guidelines for BTEX cations to date. The basis of the approach, however,
in drinking water (method 524.2). Reference: Measurement of
is that hot, pressurized water can selectively
Purgeable Organic Compounds in Water by Capillary Column
Gas Chromatography/Mass Spectrometry, Revision 3.0, US EPA leach analytes from the solid matrix. Early work has
Office of Research and Development, Cincinnati, OH, 1989, suggested that the water temperature needs to be
EPA Document EPA/600/4-88/039. above 2003C and a pressure of 50 atm for effective
4198 III / SOLID-PHASE MICROEXTRACTION / Overview

Table 5 Slurry analysis using SPME of a soil sample: comparison with Soxhlet extraction a

Compound SPME/internal standard SPME/standard addition Soxhlet extraction

(
g g\1) b (
g g\1) b (
g g\1) c

2,4-Dichlorophenol 0.9 1.9 2.0

2,4,6-Trichlorophenol 2.4 3.7 4.4
2,3,4,6-Tetrachlorophenol 7.6 8.4 12.8
Pentachlorophenol 533.8 562.2 642.4

a
Data from Lee MR et al. (1998) Journal of Chromatography 806: 317.
b
SPME: 40 mg of soil in 12.5 mL of a 20
g L\1 internal standard (2,4,6-tribromophenol) solution and, then solution was diluted to
50 mL with pH 1 buffer solution and 5 M KCl added.
c
Soxhlet: 2 g soil extracted with 150 mL of n-hexane}acetone (1 : 1) for 8 h. Analysis using GC-SIM-MS.

extraction of semi-volatile compounds of environ-
mental interest including polycyclic aromatic hydro-
carbons (PAHs). It is also important in this type of
work to be vigilant for analyte degradation, which
obviously might result in lower recoveries than ex-
pected (but also not to neglect the possibilities of
formation of compounds of interest). The dynamic
extraction of organic pollutants from solid matrices
using water is possible using apparatus designed for
supercritical Suid extraction (Figure 11A). By placing
the soil sample in the extraction cell of the SFE appar-
atus, effective extraction using water can be accomp-
lished at elevated temperature ('2003C) and
pressure (50 atm). Quantitation is then achieved us-
ing SPME by inserting the Rbre in the water extract
(Figure 11B) followed by chromatographic analysis.
Preliminary results using this type of approach are
shown in Table 6.
Headspace^SPME
Instead of using SPME to extract from the aqueous
extract or slurried sample an alternative strategy uses
headspace}SPME. In this approach SPME is used to
extract volatile or semi-volatile analytes from the
headspace above a solid sample. A soil sample (l0 mg
to 1 g) is placed in a headspace vial and the vial is

Figure 10 SPME-GC-MS chromatogram of a real soil sample

contaminated with chlorophenols. (Journal of Chromatography A,
806: 317, 1998, Copyright Elsevier Science.) GC conditions:
column, 30 m length;0.25 mm internal diameter ;0.5
m film
DB-5.625 fused silica; injection, splitless mode with an injector
temperature of 2903C, and a splitless time of 1 min; temperature
programme 60}1903C at 303C min\1 and from 190 to 3103C at
103C min\1. Slurry preparation: 40 mg of sieved soil (mesh size
1.981 mm and 2.000 mm) was prepared in 12.5 mL, of 20
g mL\1
internal standard solution and, then, the solution was diluted to
50 mL with pH 1 buffer solution and 5 M KCl added. SPME condi-
tions: 85
m polyacrylate coated fibre; adsorption time, 40 min with
stirring at 1000 rpm; desorption time, 2 min at 2903C. Peak identi-
fication: 2,4-DCP"2,4-dichlorophenol; 2,4,6-TCP"2,4,6-trich-
lorophenol; 2,3,4,6-TeCP"2,3,4,6-tetrachlorophenol;IS"2,4, Figure 11 Combined hot water extraction}SPME: (A) Appa-
6-tribromophenol; PCP"pentachlorophenol. ratus for hot water extraction and (B) quantitation using SPME.

Table 6 Dynamic high temperature water extraction of selected polycyclic aromatic hydrocarbons from an urban air particulate
reference material (NIST 1649) a
Compound Certified concentration
g g\1 (%RSD)
Fluoranthene 7.0 (7)
Pyrene 7.2 (7)
Benzo[a ]pyrene 2.9 (17)
a
Reference: Daimon H and Pawliszyn J (1996) Analytical Communications 33: 421.
b
High temperature water extraction: 2503C and 50 atm.
sealed by crimping with an appropriate cap, e.g. an
open-centred aluminium cap containing a PTFE/grey-
butyl moulded septum. In order to promote the re-
lease of volatiles a small quantity of water (10}30%)
may be added to the soil sample. In addition, the
volatility of an analyte can be increased by heating
the sample. This can simply be done by placing the
sealed sample vial in a thermostatically-controlled
water bath. It has been suggested that at ambient
temperature this headspace SPME approach can be
effective for three ring PAH compounds or more
volatile compounds.
Conclusion
While SPME has been applied to a wide range of
application areas, it is those with an environmental
theme that have been mainly used to date. The main
focus of this article is on the method of operation
for a range of sample types. SpeciRc examples have
been provided as to the application of SPME for
extraction of analytes from aqueous and solid
SOLVENTS: DISTILLATION
B. Buszewski, Nicholas Copernicus University,
Torun, Poland
This article is reproduced from Encyclopedia of Analytical
Science, Copyright ^ 1995 Academic Press
Copyright ^ 2000 Academic Press
Introduction
In modern chemical analysis various physicochemical
methods are used to achieve high detection sensitiv-
ity. In trace analysis, reported detection levels are
often measured in
g mL\1 (ppm) ng mL\1 (ppb), as
well as in pg mL\1 (ppt). Achievement of such low
detection levels has been made possible by the use of
modern analytical instruments equipped with new
types of detectors and by improved sample prepara-
III / SOLVENTS: DISTILLATION 4199
Estimated concentration as % of certified
concentration (%RSD) (n"3) b
134.0 (16)
87.5 (15)
72.0 (29)
matrices. In addition, the different forms of analysis
from aqueous samples are considered, i.e. direct and
headspace sampling while for solid samples, the use
of a slurry technique, prior to hot water extraction
and headspace SPME is considered. The experimental
data provided should act only as a guide to the poten-
tial and diverse applications of SPME.
See also: II/Chromatography: Gas: Headspace Gas
Chromatography. III/Environmental Applications: Soxh-
let Extraction.
Further Reading
Dean JR (1998) Extraction Methods for Environmental
Analysis. Chichester: John Wiley.
Kolb B and Ettre LS (1997) Static Headspace-Gas
Chromatography. Theory and Practice. New York:
Wiley-Interscience.
Pawliszyn J (1997) Solid-Phase Microextraction. Theory
and Practice. New York: John Wiley.
Pawliszyn J (ed.) (1999) Applications of Solid-Phase Micro-
extraction. Cambridge: Royal Society of Chemistry.
tion methods. Both factors are closely related to the
purity of the solvents used as the mobile phases
in different variants of liquid chromatography (LC),
capillary zone electrophoresis (CZE), liquid}liquid
(LLE) and solid-phase (SPE) extraction, Rltration and
Sotation. Moreover, high-purity solvents are also
employed to dilute samples investigated using
chromatography, spectral and electrochemical analy-
sis. Thus, solvents used in chemical analysis must
fulRl many physicochemical requirements.
High-purity solvents, for example for liquid
chromatography (LC) and/or for spectroscopy, are pro-
duced by many manufacturers. However, puriRcation
and quality testing of solvents are often necessary before
use, particularly in the above-mentioned techniques
4200 III / SOLVENTS: DISTILLATION

applicable to multicomponent systems. For instance,
the mobile phases in LC, CZE and SPE are often
modiRed by different organic and/or inorganic salts.
In this article the most important physicochemical
properties of commonly used solvents are brieSy
reviewed, including methods of their puriRcation,
recovery and quality testing.
Solvents: Properties and their Usage
In many cases the purity of the solvents used has
a great inSuence on whether the molecular processes
occurring in a multicomponent system proceed in the
desired manner. The quality of measured data can
often be improved by using high-purity solvents. For
chromatography or spectroscopy a good solvent is
characterized by high purity, and in consequence, by
low values characterizing the transparency (cut-off)
and refraction (refractive index). Parameters such as
reactivity and good miscibility are also important
criteria for solvent selection. In LC investigations,
multicomponent solvents which have low boiling
points (between 20 and 603C) and low viscosity
((50 cP) are preferred. A good solvent should be
able to dissolve the sample, although this is seldom
a problem in analytical separations when the mobile
phase has the correct strength.
The important physical parameters characterizing
the most commonly used solvents in many analytical
techniques are listed in Table 1. On the basis of this
table it can be seen that LC is the most frequently
applied technique and consequently it consumes
a large amount of the various solvents. A high usage
of solvents in LC results from the various separation
modes of this technique.
In normal-phase (adsorption) liquid chromatogra-
phy (NPLC) aliphatic hydrocarbons (e.g. n-hexane,
n-heptane) are usually used as the mobile phase. The
elution strength of these solvents is often modiRed by
other organic compounds. A fundamental problem
with NPLC eluents is the presence of dissolved water
and trace amounts of the higher-fraction oleRnes.
These contaminations can cause changes in the cut-
off values (UV detection, spectrophotometry), base-
line perturbation and poor reproducibility of
retention data. Halogenated solvents such as
dichloromethane can react with other reactive or-
ganic solvents (e.g. acetonitrile) and form crystalline
products.
In reversed-phase liquid chromatography (RP-LC)
aqueous solutions of methanol, acetonitrile, tetrahyd-
rofuran and dioxan are used as eluents. In this mode
of LC the most serious problem, besides the purity of
the organic modiRers in the mobile phase, is the

Table 1 Properties of solvents used in various analytical techniques

Solvent UV-cut-off Refractive index Boiling point  (253C)  (203C)

(nm) (25 3C) ( 3C) (cP)

Acetone 330 1.356 56 0.30 20.70

Acetonitrile 190 1.341 82 0.34 37.50
Benzene 280 1.498 80 0.60 2.30
n-Butanol 210 1.397 118 2.60 17.52
Carbon tetrachloride 265 1.457 77 0.90 2.24
Chloroform 245 1.443 61 0.53 4.80
Cyclohexane 200 1.423 81 0.90 2.02
Cyclopentane 200 1.404 78 0.42 1.97
Diethyl ether 218 1.350 35 0.24 4.30
Dimethylformamide 268 1.428 153 0.80 36.73
Dimethylsulfoxide 268 1.477 189 2.00 4.70
Dioxane 215 1.420 101 1.22 2.21
Ethanol 210 1.359 78 1.08 24.60
n-Heptane 195 1.385 98 0.40 1.92
n-Hexane 190 1.372 69 0.30 1.88
Isooctane 197 1.389 99 0.47 1.94
Methanol 205 1.326 65 0.54 32.71
Methylene chloride 233 1.421 40 0.41 8.93
n-Octanol 205 1.427 195 7.31 10.30
n-Pentane 195 1.355 36 0.22 1.84
i -Propanol 205 1.384 84 1.90 20.30
n-Propanol 240 1.385 97 1.90 20.30
Tetrahydrofuran 212 1.405 66 0.46 7.56
Toluene 285 1.494 110 0.55 2.40
Water 1.333 100 0.89 80.00

, viscosity; , dielectric constant.


purity of the water, which may contain trace impu-
rities of phenols, hydrocarbons, etc. Tetrahydrofuran
is frequently used as the solvent in gel permeation
chromatography (GPC). It must be stabilized by
butylate hydroxytoluene (BHT), which is used as an
antioxidant. However, in ion suppression and ion
pair RP-LC the mobile phases are often modiRed
by the addition of compounds which can change
the dissociation constants of analytes, such as
inorganic and organic acids or ionic substances (e.g.
ammonium chloride, cetyl chloride). All these com-
ponents may contain inconvenient impurities which
can also crystallize, causing some perturbation in
detection during the elution process. As a result of
these impurities extra peaks may appear in the
chromatograms, making identiRcation of the sub-
stances analysed more difRcult. The impurities may
interfere with the analytes retention even if a UV-
transparent mobile phase (e.g. methanol}water,
acetonitrile}water and/or ethanol}water) is used.
Figure 1 shows the peak heights of different com-
positions of solvent gradually decreasing from their
cut-off limits to become very low or almost negligible
at wavelengths larger than 320 nm.
Another factor which inSuences the separation
process and which may result from solvent impurities
during chromatographic elution is peak tailing and/or
peak splitting. In solid-phase extraction (SPE) irre-
versible sorption of solvent impurities on active sites
of packing materials (e.g. surface silanols or chemic-
ally bounded ligands with polar groups such as }OH,
}NH2, }CN, etc.) can be manifested by relatively high
differences in reproducibility of recoveries.
Therefore, before chromatographic or spectro-
scopic measurements are made, a special preparation
Figure 1 Plot of peak heights versus wavelength of radiation in
the UV spectrophotometer for the mobile phase methanol (ana-
lytical grade) } water (home purified) at compositions 30 : 70%
(v/v) (*), 50 : 50% (v/v) () and 70 : 30% (v/v) (), and com-
position 50 : 50% (v/v) () prepared from LC-grade solvents
(J. T. Baker, Deventer, The Netherlands). (Reproduced with per-
mission from Buszewski, Bleha and Berek (1985).)
III / SOLVENTS: DISTILLATION 4201
(e.g. isolation, puriRcation and preconcentration) of
the sample is recommended. For this purpose tech-
niques such as LLE, SPE, Rltration and membrane
separation are used. It should be mentioned that high-
purity solvents are required to prepare analytical
samples.
Methods of Solvent Puri\cation
There are many methods of solvent puriRcation, but
the most common are simple distillation, fractional
distillation and steam distillation. A recent method of
solvent puriRcation using solid-state packing mate-
rials is becoming more popular. It can be carried out
by LC, SPE, Rltration, and ultraRltration utilizing
membranes.
Distillation
Distillation is the oldest and simplest procedure for
solvent puriRcation, and it is also inexpensive. It is
based on Raoults law, which states that the partial
vapour pressure of a solvent is proportional to its
mole fraction. The physical foundations of separation
by distillation depend on the distribution of constitu-
ents between the liquid and the vapour phases being
at equilibrium. In general, the composition of the
vapour is different from the composition of the dis-
tillate. Only azeotropic mixtures distill without
a change in their composition. However, during dis-
tillation of a normal mixture the principal component
with the lowest boiling point distils Rrst, followed by
compounds with higher boiling points. The effec-
tiveness of distillation depends on the physical prop-
erties of the components in the mixture, the
equipment used and the method chosen.
Simple distillation Simple distillation refers to the
process in which molecules transferred from the
liquid phase to the vapour phase are not subjected to
partial condensation or contact with the condensed
liquid prior to reaching the condenser. The composi-
tion of the vapour near the liquid phase does not
change as it moves toward the condenser. In this
technique, equipment requirements are minimal; usu-
ally a Sask Rtted with a condenser and a product
receiver are sufRcient.
Fractional distillation Fractional distillation is used
when a more efRcient separation process than simple
distillation is required. This type of distillation is an
equilibrium process in which the composition of the
distillate is constantly changing as the distillation
proceeds. The main element of the apparatus is the
distillation column, which consists of a series of
4202 III / SOLVENTS: DISTILLATION
plates placed one above the other in a suitable tube.
The column is placed under the receiver. Liquid evap-
orating from one (lower) plate condenses on the next
(higher) plate, where it again evaporates. On each
plate an equilibrium between the liquid and the va-
pour is established. Vapour enriched in more volatile
components Sows upwards, whereas vapour enriched
in less volatile components Sows downwards. The
performance of the column increases with an increas-
ing number of plates.
Steam distillation Steam distillation is a simple dis-
tillation procedure in which evaporation of the mix-
ture is achieved either by continuously blowing steam
through the mixture or by boiling water and the
sample together. If the sample contains both hydro-
phobic and hydrophilic components, two layers of
distillate develop. In typical steam distillation two
layers can be recovered individually. Aqueous distil-
lation seems to provide the best compromise between
time, cost and effort.
In many cases satisfactory results are achieved by
using distillation for solvent puriRcation. Water used
in LC investigations should normally not be pur-
chased, but should be prepared by the user. An inex-
pensive and easy laboratory procedure for water
puriRcation is as follows: deionized water with added
alkaline solution of KMnO4 is left for a few days, and
then the solution is distilled twice in an apparatus
made of hard glass. The conductivity of water puri-
Red in this fashion is 10\6 S m\1.
In Figure 2, plots of absorbance for three combina-
tions of water and methanol mixtures of various
purities are shown. The main source of UV absorp-
tion by the binary mobile phase was evidently impu-
rities in the methanol. A comparison of the two
bottom curves shown in Figure 2 proves the effec-
tiveness of water puriRcation by distillation.
Other solvents, such as aliphatic and aromatic
hydrocarbons, alcohols, ethers, tetrahydrofuran and
halogenated solvents, can also be puriRed using distil-
lation methods. However, in many cases the solvent
impurities, e.g. water in aliphatic hydrocarbons or
halogenated solvents, can make distillation less efR-
cient. In these cases it is necessary to use more effec-
tive methods for solvent puriRcation.
Adsorption Methods
In adsorption chromatography (NPLC), control of
the water content in the solvents is important. In
some cases it is preferred to mix known amounts of
dry and water-saturated solvents together in order to
know the percentage of water saturation. Generally,
analytical grade purity solvents should not contain
Figure 2 Plot of absorbance for mixtures of water and meth-
anol at 254 nm versus the composition of the solution obtained on
the basis of spectrophotometric data using a 3 cm quartz cuvette.
Key: , methanol (analytical grade) and water (home purified);
, methanol (LC grade, J. T. Baker, Deventer, The Netherlands)
and water (home purified); and *, water and methanol (LC grade,
J. T. Baker). (Reproduced with permission from Buszewski, Bleha
and Berek (1985).)
impurities. On the other hand, the addition of ac-
tivated molecular sieve beads (0.4 or 0.5 nm in dia-
meter) to the Sask with the solvents clearly improves
their purity and reduces the water content. Various
impurities, in addition to water, can be often removed
by adsorption methods, particularly frontal analysis,
which is often utilized in LC.
In this method a glass column packed with small
adsorbant particles (0.15-0.2 mm in diameter), usu-
ally dried silica gel or alumina oxide, is used (Fig-
ure 3). Before solvent puriRcation the column is
heated at 473 K for 8 h under vacuum conditions
(10\3 Torr) to remove physically adsorbed water.
After this operation, unpuriRed solvent is injected
into the column using a syphon-type injector. The
puriRed solvent is collected in a solvent receiver
equipped with a moisture trap.
During LC investigations puriRcation and stabiliz-
ation of the mobile phase is frequently carried out on
precolumns. These columns are located between the
solvent reservoir and the injector valve. SPE, in con-
junction with frontal analysis, is also used for solvent
clear-up. It is mainly used for purifying small
amounts of solvents, particularly for sample dilution.
Filtration and Membrane Techniques
A reliable and easy laboratory technique for solvent
puriRcation is Rltration. The mobile phase contain-
ing added buffers or reagents may be Rltered through
a Rlter (0.5
m mesh or smaller) to remove particulate
matter which can damage the system. The equipment
for Rltration is very simple, generally consisting
of an Erlenmayer Sask connected to vacuum and
 III / SOLVENTS: DISTILLATION 4203

(hydrophobic or hydrophilic) of Rlters and/or mem-

branes determine the degree of puriRcation.
Water may be puriRed satisfactorily using a
compact water puriRcation system that combines
Rltration, deionization and charcoal treatment in a
convenient, high-volume unit.

UltraVltration UltraRltration is the other alternative

to Rltration, where large molecules are separated
from solution by using membranes. Membranes are
commercially available for separation molecules with
relative molecular masses in the range 103}106. Ultra-
Rltration is primarily used for the isolation of low- or
high-molecular mass substances from different sol-
vents. Figure 4 shows a comparison of the efRciency
of acetonitrile puriRcation utilizing frontal analysis
combined with ultraRltration.

Reverse osmosis Reverse osmosis is similar to ultra-

Rltration except that membranes of a much smaller
pore diameter are employed and the operating pres-
sure is much higher. The operating pressure must
exceed the natural osmotic pressure. This technique
has been successfully applied to the puriRcation of
water and organic solvents.

Dialysis In dialysis, solvents are puriRed by differ-

ential diffusion through membranes under a concen-
tration gradient. The overall efRciency of this process
is controlled by the ratio of the Sow rates and the
properties of membrane, Suid channel and local Suid
velocity.

Recovery and Quality Control

After the puriRcation process or before using solvents
in various analytical techniques, quality control is
necessary. For this reason, many different analytical

Figure 3 Apparatus for solvent purification. 1, glass column

with cooling jacket; 2, glass sinter; 3, silica gel or alumina adsor-
bent; 4, syphon-type injector; 5, container for unpurified solvent;
6, PTFE tube; 7, moisture trap; 8, valve; 9, receiver for purified
solvent. (Reproduced with permission from Buszewski, Lod-
kowski and Trocewicz (1987).)

a reservoir, in which a porous Rlter disc or membrane

is placed. The porous discs are usually made from
nonporous spherical glass beads (1}2
m in diameter)
and/or PTFE. Membranes are usually made from
PTFE, cellulose or nylon. To improve the efRciency
of the separation process, the surface of the Rlter discs
or membrane is often modiRed chemically, as are the Figure 4 Liquid chromatograms with UV detection using
packing materials, with chemically bonded phases in acetonitrile as the mobile phase purified through (1) frontal analy-
RP-LC and/or SPE. In this case the surface properties sis and (2) ultrafiltration.
4204 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION

Table 2 Concentration of water and difference in absorbance Table 2 summarizes the results of solvent puriRca-
for cut-off (
A) of solvents before (C b) and after (Ca) purification tion by frontal analysis. In each case, puriRcation of
the solvent improves its absorbance at the cut-off
Solvent
A Cb Ca
( % v/v) (mg L\1) (mg L\1) point (
A).
The small improvement found in the case of the LC
Acetonitrile A 69.7 2216.0 167.0 grade methanol B is as expected for this high-purity
Acetonitrile B 10.2 297.0 143.0 solvent. The high impurity (absorbance difference,
Benzene 24.6 308.0 27.5

A"77.5%) of analytical grade methanol A pre-
Cyclohexane 16.8 49.0 4.5
n-Heptane 21.3 18.8 7.7 cludes its use in LC investigations. Similarly, tol-
n-Hexane 17.2 19.6 8.6 uene, benzene, tetrahydrofuran and acetonitrile
Methanol A 77.05 650.0 134.0 A cannot be used in LC measurements without prior
Methanol B 3.6 120.0 114.0 puriRcation.
Tetrahydrofuran 61.3 1081.0 32.6
Recovery values for the puriRcation of water,
Toluene 23.1 167.0 10.0
acetonitrile, alcohols, ketones, aliphatic and aromatic
Reproduced with permission from Buszewski, Lodkowski and hydrocarbons obtained in distillation methods are
Trocewicz (1987). usually in the range 85}95% (v/v). In the case of
halogenated solvents this range is narrower, i.e.
methods have been utilized for measuring the charac- 75}80% (v/v). Utilizing membrane techniques for
teristic parameters. Generally, chromatographic solvent clear-up, it is possible to obtain recovery in
techniques such as gas chromatography (GC), LC, the range 90}97% (v/v).
thin-layer chromatography (TLC) have been used,
but ultraviolet (UV), infra-red (IR) and nuclear See also: II/Distillation: Laboratory Scale Distillation.
magnetic resonance (NMR) spectroscopy can also be Membrane Separations: Filtration. III/Flash Chromato-
applied. Water in many organic solvents is usually graphy.
determined by Karl Fischer titration. On the basis of
experimental data obtained before and after puriRca- Further Reading
tion the efRciency of the clean-up procedure is
Brock TD (1983) Membrane Filtration. Madison: Science
determined. In general, the efRciency of puriRcation Technology.
process, e.g. the recovery, is expressed by the Buszewski B, Bleha T and Berek D (1985) UV detection
coefRcient R. This parameter is deRned as the ratio of solvent peaks in liquid chromatography with
of the volume or concentration of removed impurities mixed eluents. Journal of High Resolution Chroma-
to the volume or concentration of solvent before tography and Chromatography Communications 8:
puriRcation: 860}862.
Buszewski B, Lodkowski R and Trocewicz J (1987) PuriR-
R$"(Va$a)(Vb$b);100% (v/v) [1] cation of solvents for liquid chromatography. Journal of
High Resolution Chromatography and Chromatogra-
or: phy Communications 10: 527}528.
Hampel CA and Hawley GG (eds) (1973/74) Handbook of
Chemistry and Physics, 54th edn. Cleveland: CRC Press.
R$"(Ca$a)(Cb$b);100% (v/v) [2] Minear RA and Keith LH (eds) (1984) Water Analysis, vol.
III. Orlando: Academic Press.
where Va, Ca and Vb, Cb denote volume or concentra- Poole CF and Poole SK (1991) Chromatography Today.
tion of the removed impurities and solvent samples, Amsterdam: Elsevier.
respectively, and , a, b are individual standard Riddick JA and Bunger WB (1970) Organic Solvents, 3rd
deviations. edn. New York: Wiley Interscience.

SORBENT SELECTION FOR

SOLID-PHASE EXTRACTION
E. M. Thurman, US Geological Survey, Lawrence, Solid-phase extraction (SPE) is a method of sample
KS, USA preparation that concentrates and puriRes analytes
Copyright ^ 2000 Academic Press from solution by sorption onto a disposable solid-
phase cartridge, followed by elution of the analyte
 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION
with a solvent appropriate for instrumental analysis.
The mechanisms of retention include reversed phase,
normal phase, size exclusion, and ion exchange.
Solid-phase extraction was invented in the mid 1970s
as an alternative approach to liquid}liquid extraction
for sample preparation.
Initially, SPE was based on the use of polymeric
sorbents, such as XAD resins, which were packed in
small disposable columns for use on drug analysis.
The early environmental applications consisted of
both XAD resins and bonded-phase sorbents, such as
C18. These precolumns were used for sample trace
enrichment prior to liquid chromatography and were
often done on line, which means at the same time as
liquid chromatography. However, these Rrst, steel,
on-line precolumns were quickly replaced with an
off-line column made of plastic, in order to be both
inexpensive and disposable. Eventually, the term
solid-phase extraction was coined for these low-
pressure extraction columns.
SPE columns are now typically constructed of poly-
propylene or polyethylene and Rlled with 40-
m
packing material with different functional groups.
A 20-
m polypropylene frit is used to contain from
50 mg to 10 g of packing material. A liquid sample is
passed through the column and analytes are concen-
trated and puriRed. The sample volume that can be
applied ranges from 1 mL to over 1 L. The sample
may be applied to the column by positive pressure or
by vacuum manifold. After quantitative sorption of
the analyte, it is removed with an appropriate elution
solvent.
Therefore, SPE is a form of digital liquid
chromatography that removes the solute onto a solid-
phase sorbent by various sorption mechanisms. The
term digital refers to the on/off mechanism of sorp-
tion and desorption. The goal of SPE is to quan-
titatively remove the analyte from solution and
completely recover it in an appropriate solvent.
PuriRcation consists of removing the analyte from
interfering compounds and concentrating the analyte
in a small volume of solvent. For example, pesticides
are concentrated from a water sample by SPE into
a small volume of organic solvent for analysis by gas
chromatography/mass spectrometry. Interfering sub-
stances, such as humic and fulvic acids, ionic metab-
olites and salts are removed.
Typically, SPE replaces liquid}liquid extraction as
a sample preparation tool and provides a method that
is simple and safe to use. The beneRts of SPE include:
high recoveries of analytes; puriRed extracts; ease of
automation; compatibility with chromatographic
analysis; and reduction in the consumption of organic
solvents. As a result of the Sexibility that SPE offers,
it has found application in the preparation of environ-
4205
mental, clinical, and pharmaceutical samples. The
simplicity of the SPE procedure and the use of dispos-
able SPE supplies have encouraged the design of auto-
mated sample preparation stations, which decrease
the time and cost of sample preparation. Finally,
recent advances in on-line methods of SPE allow
automation of sample preparation directly to both
liquid and gas chromatography.
How to do SPE
Figure 1 illustrates the four-step process of SPE. First
the solid-phase sorbent is conditioned (step 1). This
simply means that a solvent is passed through the
sorbent to wet the packing material and to solvate the
functional groups of the sorbent. Furthermore, the air
that is present in the column is removed and the void
spaces are Rlled with solvent. Typically, the condi-
tioning solvent is methanol, which is then followed
with water or an aqueous buffer. The methanol fol-
lowed by water or buffer activates the column in
order for the sorption mechanism to work properly
for aqueous samples. Care must be taken not to allow
the bonded-silica packing or the polymeric sorbent to
go dry. In fact, if the sorbent dries for more than
several minutes under vacuum, the sorbent must be
reconditioned. If it is not reconditioned the mech-
anism of sorption will not work effectively and re-
coveries will be poor for the analyte.
Another cleaning step of the sorbent may also be
added during conditioning, if necessary. Simply, the
eluting solvent is passed through the column after the
methanol wetting step to remove any impurities that
may be present in the packing material. This cleaning
step would then be followed by methanol and aque-
ous buffer, which prepares the column for sample
addition.
Next, the sample and analyte are applied to the
column (step 2, Figure 1). This is the retention or
loading step. Depending on the type of sample, from
1 mL to 1 L of sample may be applied to the column
either by gravity feed, pumping, aspirated by vacuum,
or by an automated system. It is important that the
mechanism of retention holds the analyte on the col-
umn while the sample is added. The mechanisms of
retention include Van der Waals (also called non-
polar, hydrophobic, partitioning, or reversed-phase)
interaction, hydrogen bonding, dipole}dipole forces,
size exclusion, and cation and anion exchange. Dur-
ing this retention step, the analyte is concentrated on
the sorbent. Some of the matrix components may also
be retained and others may pass through, which gives
some puriRcation of the analyte.
Step 3 (Figure 1) is to rinse the column of inter-
ferences and to retain the analyte. This rinse will
4206 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION

Figure 1 The four-step process of solid-phase extraction. Step 1, condition sorbent; step 2, apply sample and analyte; step 3, wash;
step 4, analyte elution. (Reprinted from Thurman and Mills (1998) Copyright John Wiley & Sons, Inc.)

remove the sample matrix from the interstitial spaces
of the column, while retaining the analyte. If the
sample matrix is aqueous, an aqueous buffer or
a water/organic-solvent mixture may be used. If the
sample is dissolved in an organic solvent, the rinse
solvent could be the same solvent.
Finally, in step 4 an appropriate solvent is used that
is speciRcally chosen to disrupt the analyte}sorbent
interaction, resulting in elution of the analyte from
the sorbent. The eluting solvent should remove as
little as possible of the other substances sorbed on
the column. This is the basic method of solid-phase
extraction.
Columns and Apparatus for SPE
The sorbents used for SPE are packaged in three basic
formats. There are discs, cartridges, and syringe
barrels. Figure 2 shows the different types of pre-
sentation of SPE products. The discs are available in
different diameters from 4 to 90 mm, the standard
disc size being 47 mm. Cartridges vary from as little
as 100 mg to 1 g or more. Syringe barrels are avail-
able in different volumes and with different masses of
packing material. Syringe barrels range in size from
1 to 25 mL and packing weights from 50 mg to 10 g.
These various sorbents allow for the effective treat-

Figure 2 The three formats of SPE: (A) discs; (B) cartridges; and (C) syringe barrels. (Reprinted from Thurman and Mills (1998)
Copyright John Wiley & Sons, Inc.)
 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION 4207

ment of different types of sample and different sample

volumes.
Currently, the most commonly used format for SPE
consists of a syringe barrel that contains 40-
m
sorbent material, with a 20-
m polypropylene frit at
the bottom and a 20-
m polypropylene frit at the top
of the syringe. The syringe barrel is typically poly-
propylene with a male Luer-tip Rtting and is dispos-
able. Some vendors do make glass syringe barrels and
TeSon frits, but these conRgurations are used less
frequently. The glass and TeSon system is used when
one is interested in the analysis of plasticizers or is
concerned with the potential sorption of speciRc
analytes onto the polyethylene tube.
Solvent reservoirs may be used to increase the vol-
ume of the syringe barrel. Reservoirs are typically
50}100 mL in volume. Coupling Rttings are used to
join the reservoirs and syringe barrels between the
Luer Rtting and the opening of the syringe barrel
(Figure 3).
The barrel of the syringe terminates in a male Luer
tip. The male Luer tip is the standard Rtting on SPE
cartridges, so that they are interchangeable with dif-
ferent SPE vacuum manifolds. The vacuum manifold
is used to draw the sample and eluting solvents
through the syringe barrel under negative pressure by
applying a vacuum to the manifold. Figure 4A shows
a typical vacuum-manifold system, which is Rtted
with a small vacuum pump and a waste receiver.
Stopcock valves are available to control the vacuum
applied to each column. Other types of sample
processing that may be used include centrifugation
(Figure 4C) and positive pressure (Figure 4D), which
forces the sample through the syringe barrel from
above. Simple gravity Sow through the syringe barrel
or cartridge may also be used (Figure 4B).
A typical solid-phase extraction cartridge (see Fig-
ure 2) consists of a polyethylene body with both a
female and male Luer tip for positive pressure from
a syringe, or negative pressure from a vacuum mani-
fold. Polyethylene frits measuring 20
m are placed at
either end of the cartridge to hold the packing material
in place. The packing material is packed and com-
pressed to improve or optimize Sow characteristics.
The third type of SPE format is the disc, which is
available in several styles by different manufacturers. Figure 3 Disposable column and reservoir. (Reprinted from
One of the most popular extraction discs is the Em- Thurman and Mills (1998) Copyright John Wiley & Sons, Inc.)
pore extraction disc, which consists of 8}12
m
particles of packing material embedded in an inert a membrane format as loose discs, or are placed in
matrix of polytetraSuoroethylene (PTFE) Rbrils. Be- a syringe-barrel format called an extraction disc
cause the particles are suspended in PTFE, no binder cartridge. The syringe-barrel format consists of a
is required to give structure to the disc and the matrix standard polyethylene syringe that is Rtted with a 20-
is essentially inert. The discs are not coated with
m TeSon frit, an Empore disc, and a preRlter of
PTFE so that they can interact with the solvent and glass Rbre. This arrangement allows for micro-scale
sample during extraction. The discs are available in work using the disc. Discs are conditioned and used in
4208 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION

Figure 4 Techniques for processing SPE cartridges: (A) vacuum manifold; (B) gravity; (C) centrifugation; and (D) positive pressure.
(Reprinted from Thurman and Mills (1998) Copyright John Wiley & Sons, Inc.)

a similar fashion to the packed columns, with Sow of
sample by negative pressure by vacuum.
A major advantage of the disc format is rapid mass
transfer because of the greater surface area of the
8}12
m particles, which results in high Sow rates for
large volume samples. This rapid Sow rate is espe-
cially useful for environmental samples where 1 L of
water may be processed in as little as 15 min. Rapid
mass transfer owing to embedding of small particles
into the disc also means that channelling is reduced
and small volumes of conditioning and elution sol-
vents may be used. For example, a 4-mm disc in
syringe format (extraction disk cartridge) requires
only 100
L of elution solvent, and a 7-mm disc uses
only 250
L. This volume is small compared with the
millilitre amounts applied to a 3}5-mL syringe barrel
that contains loose packing.
Another type of disc called SPEC manufactured
by Ansys, Inc., uses a glass-Rbre matrix rather than
TeSon to hold the sorbent particles. This disc has
a somewhat more rapid Sow rate and is more rigid
and thicker than the TeSon disc. There is another disc
called the Speedisk manufactured by J. T. Baker,
which consists of 10-
m packing material that is
sandwiched between two glass-Rbre Rlters without
any type of TeSon binder.
 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION 4209

Sorbents and Modes of Interaction
The sorbents used for SPE are similar to those used in
liquid chromatography, including normal phase, re-
versed phase, size exclusion, and ion exchange. Nor-
mal-phase sorbents consist of a stationary phase that
is more polar than the solvent or sample matrix that is
applied to the SPE sorbent. This means that water is
not usually a solvent in normal-phase SPE because
it is too polar. Normal-phase sorbents, therefore,
are used in SPE when the sample is an organic solvent
containing an analyte of interest. Polar interactions,
such as hydrogen bonding and dipole}dipole
interactions, are the primary mechanisms for solute
retention.
Reversed-phase sorbents are packing materials that
are more hydrophobic than the sample. Reversed-
phase sorbents are commonly used in SPE when
aqueous samples are involved. The mechanism of
interaction is Van der Waals forces (also called non-
polar, hydrophobic, or reversed-phase interactions)
and occasionally secondary interactions such as hy-
drogen bonding and dipole}dipole interactions. Size-
exclusion sorbents utilize a separation mechanism
based on the molecular size of the analyte. It is
a method only recently being used in SPE, usually in
conjunction with reversed phase or ion exchange.
Ion-exchange sorbents isolate analytes based on the
ionic state of the molecule, either cationic or anionic,
where the charged analyte exchanges for another
charged analyte that is already sorbed to the ion-
exchange resin. SPE applications in this case are es-
sentially identical to classical ion exchange.
Thus, the mechanisms of interaction include:
hydrogen bonding and dipole}dipole forces (polar
interactions); Van der Waals forces (non-polar or
hydrophobic interactions); size exclusion; and cation
and anion exchange. Some sorbents combine several
interactions for greater selectivity. The extensive line
of sorbent chemical structures facilitates one of the
most powerful aspects of SPE, which is selectivity.
Selectivity is the degree to which an extraction tech-
nique can separate the analyte from interferences in
the original sample. The number of possible interac-
tions between the analyte and the solid phase facili-
tates this selectivity.
Table 1 lists the common sorbents that are avail-
able for SPE and their mode of action (i.e. reversed
phase, normal phase, ion exchange, and size exclu-
sion). Typically the sorbents consist of 40-
m silica
gel with approximately 60-A> -pore diameters. Chem-
ically bonded to the silica gel are the phases for each
mode of action. For reversed-phase sorbents, an
octadecyl (C18), octyl (C8), ethyl (C2), cyclohexyl, and
phenyl functional groups are bonded to the silica.
Typical loading of reversed-phase sorbents varies
from approximately 5% for the C2 phase to as much
as 17% for the C18 phase. The per cent loading is the

Table 1 Common sorbents available for SPE

Sorbent Structure Typical loading

Reversed phase
Octadecyl (C18) }(CH2)17CH3 17%C
Octyl (C8) }(CH2)7CH3 14%C
Ethyl (C2) }CH2-CH3 4.8%C
Cyclohexyl }CH2CH2-cyclohexyl 12%C
Phenyl }CH2CH2CH2-phenyl 10.6%C
Graphitized carbon Aromatic carbon throughout
Copolymers Styrene divinylbenzene
Normal phase
Cyano (CN) }(CH2)3CN 10.5%C, 2.4%N
Amino (NH2) }(CH2)3NH2 6.4%C, 2.2%N
Diol (COHCOH) }(CH2)3OCH2CH(OH)CH2(OH) 8.6%C
Silica gel }SiOH *
Florisil Mg2SiO3 *
Alumina Al2O3 *
Ion exchangers
Amino (NH2) }(CH2)3NH2 1.6 meq g\1
Quaternary amine }(CH2)3N#(CH3)3 0.7 meq g\1
Carboxylic acid }(CH2)2COOH 0.4 meq g\1
Aromatic sulfonic acid }(CH2)3-phenyl-SO3H 1.0 meq g\1
Size exclusion
Wide-pore hydrophobic (butyl) I(CH2)3CH3 5.9%C
Wide-pore ion exchangers}COOH 12.2%C
4210 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION
amount of C2 or C18 phase that is present by weight of
carbon. The capacity of the sorbent in mg g\1 of
analyte that may be sorbed is related to both the
chemistry of the phase and the loading weight of
carbon. Polymeric sorbents, such as styrene divinyl-
benzene and carbon, also are used for reversed-phase
SPE. These sorbents were some of the classical rever-
sed-phase sorbents introduced in the 1960s. They are
currently produced in puriRed form and are useful for
the isolation of more polar solutes that have low
capacities on the C18 reversed-phase sorbents.
For normal-phase SPE, cyanopropyl (CN),
aminopropyl (NH2), and diol functional groups are
chemically bonded to the silica gel. The loading on
the cyano, amino, and diol columns are sufRciently
large (&6}10% as carbon) in that they may some-
times be used for reversed-phase applications, espe-
cially for the removal of hydrophobic solutes from
water or other polar solvents. These hydrophobic
solutes would otherwise sorb too strongly to a more
hydrophobic C8 or C18 sorbent and would be difRcult
to elute. Straight silica gel is also used for normal-
phase SPE along with Florisil (magnesium silicate)
and alumina (aluminum oxide in neutral, basic, and
acidic forms).
Ion-exchange sorbents usually contain both weak
and strong cation and anion functional groups
bonded to the silica gel (Table 1). Strong cation-
exchange sorbents contain ion-exchange sites consist-
ing of sulfonic acid groups, and weak cation-ex-
change sorbents contain sites consisting of carboxylic
acid groups. Strong anion-exchange sites are quater-
nary amines, and weak anion-exchange sites are pri-
mary, secondary, and tertiary amines. Strong and
weak refers to the fact that strong sites are always
present as ion-exchange sites at any pH, while weak
sites are only ion-exchange sites at pH values greater
or less than the pKa, which determines whether a site
contains a proton or not. The typical loading for an
ion-exchange sorbent is expressed in meq g\1 of
sorbent, which is called the exchange capacity of the
sorbent. The values vary from &0.5 meq g\1 to
1.5 meq g\1. These exchange capacities are some-
what less than a typical ion-exchange resin, which
will have from 2 to 5 meq g\1 because of a higher
density of ion-exchange sites. Also these SPE ion-
exchange sorbents are not as rugged as the polymeric
ion-exchange resins because of the silica matrix of the
SPE sorbent, which is susceptible to dissolution by
strong acid or base. The typical ion-exchange resin,
however, consists of a cross-linked styrene-divinyl-
benzene polymer.
Size-exclusion sorbents, called wide-pore sorbents
(Table 1), use a silica-gel matrix with a large pore size
(approximately 275}300 A> ) rather than the 60-A>
pores of most bonded-phase silicas. The advantage
of the larger pore size is that molecules of larger
molecular weight ('2000 daltons) may enter the
pore of the sorbent and sorb by hydrophobic, polar,
or ion exchange interactions. Two examples are
shown in Table 1. One is a hydrophobic sorbent of
C4 (butyl) with a carbon loading of almost 6%, and
the other is a weak cation sorbent using the carboxyl
exchange site.
Another packing material, which is not listed in
Table 1, was recently introduced for drug analysis. It
is a mixed-mode resin. This packing material contains
both a bonded reversed-phase group (typically a C8)
and a cation-exchange group on a silica gel or a poly-
meric matrix. The combination of bonded groups is
used so that both types of mechanisms retain the
analyte at different times, or simultaneously, in the
clean up of complex samples of urine and blood. The
principle of the mixed-mode resin is that different
wash solvents may be used to remove interferences,
but that the solute is always retained by one or both
of the interactions.
Applications of SPE
Table 2 shows a general application guide for the use
of SPE sorbents. The C18 reversed-phase sorbent has
historically been the most popular packing material
and has been used most frequently. The surface of the
sorbent is one of the most hydrophobic and has
a large capacity. Capacity is the amount of analyte
sorbed (usually expressed in mg g\1) before break-
through occurs. Applications of C18 reversed phase
include: isolation of hydrophobic species from aque-
ous solutions, such as drugs and metabolites from
urine, serum, plasma, and other biological Suids;
desalting of peptides and oligonucleotides; isolation
of pigments from wine and beverages; and trace en-
richment of pesticides from water for analysis by gas
chromatography/mass spectrometry or high pressure
liquid chromatography.
Graphitized carbon and reversed-phase polymeric
sorbents are also frequently used in environmental
applications, such as trace enrichment, for soluble
molecules that are not isolated by reversed-phase
sorbents, such as C18. Water soluble analytes require
a more hydrophobic sorbent with greater surface area
per gram for complete retention. Carbon and poly-
meric sorbents may also be used for polar metabolites
of drugs and pharmaceuticals that are poorly retained
on C18. Another advantage of the aromatic sorbents is
their selective interaction with the aromatic rings of
analytes. Because both the graphitized carbon and the
styrene divinylbenzene structures contain aromatic
rings, they have the ability to sorb analytes by a
 III / SORBENT SELECTION FOR SOLID-PHASE EXTRACTION 4211

Table 2 Selected application guide for SPE

Sorbent Application

C18 Reversed phase application one of the most hydrophobic phases

Drugs in serum, plasma, and urine
Organic acids in wine
Pesticides in water by trace enrichment

Graphitized carbon Reversed phase application one of the most hydrophobic phases
polymeric sorbents Trace enrichment of polar pesticides from water
(styrene divinylbenzene) Isolation of polar drug metabolites

C8 Reversed phase application}hydrophobic phase

Drugs from serum, urine, and plasma
Peptides in serum and plasma

Silica Normal phase application}polar neutral phase

Isolation of low to moderate polarity species from non-aqueous solution
Lipid classification

Florisil Normal phase application}polar slightly basic phase

Isolation of low to moderate polarity species from non-aqueous solution
Pesticides in food and feeds
Polychlorinated biphenyls is transformer oil

Alumina A Normal phase application}acidic polar phase

Isolation of hydrophilic species in non-aqueous solution
Low capacity cation exchange

Cation exchange Cation exchange phase

Isolation of cationic analytes in aqueous or non-aqueous solutions
Fractionation of weakly basic proteins and enzymes

Anion exchange Anion exchange phase

Isolation of anionic analytes in aqueous or non-aqueous solutions
Extraction of acidic and weakly acidic proteins and enzymes

Mixed mode Reverse phase (C8) and cation exchange phase

Isolation of basic and amphoteric drugs from serum, plasma, and urine

Aminopropyl NH2 Normal phase, reverse phase, and weak cation exchange
Low capacity weak anion exchanger
Drugs and metabolites from body fluids
Petroleum and oil fractionation

Cyanopropyl Normal phase and reversed phase

CN Analytes in aqueous or organic solvents
Drugs and metabolites in physiological fluids
Diol Normal phase and reversed phase
OH Analytes in aqueous or organic solvents
Drugs and metabolites in physiological fluids

speciRc pi-pi interaction. This sorption mechanism
may selectively isolate aromatic compounds.
The C8 reversed-phase sorbents (Table 2) are often
the most popular sorbent for drug analysis because of
a shorter hydrocarbon chain than a C18 sorbent. The
shorter chain length makes it much more easy for
secondary interactions between the analyte and the
silica gel which enhances retention of the analyte.
This added interaction is useful in the puriRcation of
drugs and metabolites from blood and urine because
they contain basic nitrogen atoms that may hydrogen
bond to the silica gel.
Normal-phase sorbents such as silica and Florisil
are used to isolate low to moderate polarity species
from non-aqueous solutions. Examples of applica-
tions include lipid classiRcation, plant-pigment separ-
ations, and separations of fat-soluble vitamins from
lipid extracts, as well as the clean up of organic
solvent concentrates obtained from a previous SPE
method or liquid}liquid extraction. Alumina is
4212 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY
used to remove polar species from non-aqueous
solutions. Examples include vitamins in feeds and
food, and antibiotics and other additives from feed.
Normal-phase chromatography has been used for
a number of years and most applications for normal-
phase column chromatography may be easily trans-
ferred over to normal-phase SPE.
Cation and anion exchange is used to isolate ionic
compounds from either aqueous or non-aqueous
solutions. Examples of applications are: isolation of
weakly basic proteins; removal of acidic pigments
from wines and fruit juices; and the removal of or-
ganic acids from water. Many of the applications of
classical ion exchange may be used in ion-exchange
SPE; however, care must be exercised in the use of
strong acids and bases with SPE ion-exchange
sorbents that are based on a silica matrix. Further-
more, care must be taken not to exceed the ion-
exchange capacity of the sorbent.
Finally, sorbents such as aminopropyl, cyano-
propyl, and diol can be used for both reversed-phase
and normal-phase separations. Many manufacturers
supply their sorbents in variety packs, which may be
used for methods development. Also quality assur-
ance reports are commonly available for the various
sorbents, which is a good indication of their repro-
ducibility.
Automation of SPE
Automation of a manual SPE method can provide
many beneRts, which include safety, improved re-
sults, and cost savings. Because automated work-
stations are mechanical they can operate in environ-
ments that are hostile, for example, noisy production
locations or a refrigerated room. The use of automa-
tion results in improved precision because of reduced
operator errors compared with manual methods of
SPE. For these reasons automation provides for better
utilization of resources.
SPACE EXPLORATION:
GAS CHROMATOGRAPHY
R. Sternberg and F. Raulin,
Universite& s Paris 12 et 7, CNRS UMR 7583,
Cre& teil cedex, France
C. Vidal-Madjar, CNRS UMR 7581,
Thiais, France
Copyright ^ 2000 Academic Press
There are many types of automation equipment
for SPE. They include semi-automated instruments,
workstations that carry out the entire SPE operation
without intervention, and robotic systems that carry
out many activities besides SPE and are specially
customized for the user. Finally, there are on-line
SPE}HPLC systems that allow the user to merely add
the sample to the autosampler and analyse the sample
directly. The concept of on-line SPE is that a sample is
pumped and processed onto the SPE cartridge while
the liquid chromatograph or gas chromatograph is
processing the preceding sample.
See also: II/Chromatography: Liquid: Column Techno-
logy. Extraction: Solid-Phase Extraction. III/Solid-Phase
Extraction with Cartridges. Solid-Phase Extraction
with Discs.
Further Reading
Fritz JS (1999) Analytical Solid-Phase Extraction. New
York: John Wiley.
Hennion M and Pichon V (1994) Solid-phase extraction of
polar organic pollutants from water. Environmental
Science and Technology 28: 576A}583A.
Horack J and Majors RE (1993) Perspectives from the lead-
ing edge in solid-phase extraction. LC-GC 11: 74}90.
McDonald PD and Bouvier ESP (1995) Solid Phase Extrac-
tion Applications Guide and Bibliography, a Resource
for Sample Preparation Methods Development, 6th
Edition. Milford, Massachusetts: Waters.
Poole SK, Dean TA, Oudesma JW and Poole CF (1990)
Sample preparation for chromatographic separations
and overview. Analytical Chimica Acta 236: 3}42.
Simpson N and Van Horne KC (1993) Sorbent Extraction
Technology Handbook. Harbor City, CA: Varian.
Thurman EM and Mills MS (1998) Solid-Phase Extraction:
Principles and Practice. New York: John Wiley.
Varian Sample Preparation Products (1992) Applications
Bibliography. Harbor City, CA: Varian.
Zief M and Kiser R (1988) Sorbent Extraction for Sample
Preparation. Phillipsburg, NJ: J. T. Baker.
The development in the past 40 years of space explor-
ation has brought important information about the
formation and evolution of the solar system and has
opened a broad study of organic matter and its con-
tinuous chemical evolution which led to the appear-
ance of life.
 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY
The study of the physics and chemistry of planetary
environments (Table 1) has provided important in-
formation about the origin of Earths early atmos-
phere and comparative planetology gives a better
understanding of our own planet. Titan, the largest
satellite of Saturn, because of the composition and
density of its atmosphere, is of particular interest for
the understanding of the prebiotic chemistry on
primitive Earth. Comets are also of interest since they
contain very large amounts of organic material and
they are considered as the most primitive objects
in the solar system, retaining traces of its early
evolution.
Since the beginning of space exploration, most of
the many probes which have been sent to explore
other planetary atmospheres and surfaces have
carried analytical instruments to determine the el-
emental, isotopic and molecular (inorganic and or-
ganic) compositions of extraterrestrial environments.
Severe constraints are required in space instrumenta-
tion such as low weight and small size, low power
consumption, high mechanical strength and resist-
ance to deep space conditions. Gas chromatography
(GC) fulRls these requirements and is one of the most
frequently used technique for in situ analysis in space
missions.
Chemical sensors based on GC and mass spectro-
metry (MS) instrumentation have already been used
in atmospheric probes or surface landings for analys-
ing extraterrestrial environments, including the analy-
sis of surface materials from Venus and Mars. Until
recently, the equipment was only packed column GC
with mainly magnetic, then quadrupole mass spec-
trometers (MS).
Since the Titan atmospheric probe of the
Cassini}Huygens mission, launched in 1997 to reach
the vicinity of Saturn in 2004, highly sensitive pyroly-
sis}GC}MS instruments have been developed using
capillary columns.
The Rosetta comet exploratory mission, to be laun-
ched in 2003, will use, in two GC-based experiments,
miniature thermal conductivity detectors (TCDs) and
Table 1 Extraterrestrial environments already analysed or planned to be analysed by gas chromatography
Distance Surface
to sun temperature
(AU) (3C)
Venus 0.7 460
Earth 1 22
Mars 1.5 !70 to 20
Titan 9.6 !180
Comets 1.1}5.1 !200
4213
a new design of time of Sight mass spectrometer
(TOF), based on a speciRc geometry of the MS
allowing analysis of low mass compounds and good
resolution.
This article reviews the different chromatographic
instruments used in previous missions to Mars, Venus
and Titan as well as those which are being developed,
in particular for the forthcoming cometary mission.
Missions to Mars
Mars is presently the most likely planet on which
there is a possibility of Rnding past, or even present
extraterrestrial life. The average atmospheric pres-
sure on its surface is extremely low, about 7 mbar.
The primary atmospheric constituent, CO2, produces
a small warming of the surface above radiation tem-
perature (Table 1).
One of the main objectives of the Viking mission to
Mars was the search for Martian life. The US Nation-
al Aeronautics Space Administration (NASA) sent
two identical spacecraft to Mars in 1976. Each
Viking lander, carrying scientiRc instruments, was
successfully placed on the surface of Mars. Biemann
was responsible for the MS instrument designed
mainly for the detection of organic compounds in the
GC}MS mode, but also used to determine indepen-
dently the composition of the minor constituents of
the lower atmosphere. In addition, biological investi-
gations were carried out on board the landers.
Oyama and Berdahl used GC in a gas exchange ex-
periment (GEX) to determine the gas composition
changes above a soil sample humidiRed or incubated
in the presence of an aqueous nutrient.
Viking GC+MS Experiments
The GC}MS system (Table 2) was designed to analyse
the organic compounds released from a heated surface
sample. It consists of different sub-systems: (1) three
sample ovens mounted in a sample holder, (2) a GC,
(3) an efSuent divider to protect the MS, (4) a carrier
gas separator and (5) the MS itself.
Surface Atmosphere
pressure
(bar) Major constituents Organic compounds
90 CO2, N2 None
1 N2, O2 Many
0.007 CO2, N2 Not detected
1.5 N2, CH4 Several hydrocar-
bons and nitriles
(0.001 H2O, CO, CO2, Many
H2CO, CH3OH
4214 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY

Table 2 An overview of gas chromatography instruments of in situ planetary missions

Mission Launch Experiment Analytical columns Temperature carrier Detectors

arrival sample size gas

NASA Viking}Mars 1975/1976 GEX 0.1 cm3 (gas) One pair of Porapak 243C, He 1 Thermistor TCD
Q (7.6 m;1 mm) (323C)

GC}MS 0.06 cm3 One Tenax coated with 503C (12min)# MS

(soil) polymetaphenoxylene linear prog. to 2003C,
(2 m;0.76 mm) H2

NASA Pioneer- 1978/1978 LGC 0.35 cm3 (gas) In parallel: one pair of 183C, He 623C, He In parallel: two
Venus}Venus Porapak N thermistor TCDs
(15.85 m;1.1 mm);
one pair of PDVB
(2.13 m;1.1 mm)

USSR 1978/1978 SIGMA GC (gas) In series: one Polysorb 703C, Ne In series: three Ne
Venera12}Venus (2 m) 1 molecular sieve ionization
(2.5 m); one reduced
manganese

USSR Vega}Venus 1984/1985 SIGMA-3 GC (gas In parallel: one Porapak 703C, He, N2, N2 He ionization
and aerosol) QS#N; one Porapak # TCD; ECD; ECD
QS#N; one Porapak T

NASA/ESA Cassini} 1997/2004 GC}MS (gas and In parallel: one carbon H2 isothermal Five MS sources
Huygens}Titan aerosol) molecular sieve 30}603C (three connected to
(2 m;0.75 mm); one each column)
glassy carbon WCOT
(14 m;1.8 mm);
1 CPP}DMPS WCOT
(10 m;1.8 mm)

ESA Rosetta}comet 2003/2011 COSAC/GC}MS In parallel: six WCOT He 30}603C One MS, eight nano
P/Wirtanen (nucleus) capillaries; two PLOT TCDs
capillaries
PTOLEMY/GC}MS Two WCOT capillary; one He One MS
(nucleus) chemical trap; one cold trap

For both instruments on the Viking Lander-1
(VL-1) and Viking Lander-2 (VL-2), one sample oven
could not operate and the analyses were limited to
four samples from the Martian surface. Two were
from the Chryse Planitia region (VL-1) and other two
from Utopia Planitia (VL-2). For each sample, a num-
ber of analyses were performed with various GC oven
temperature (50, 200, 350 or 5003C). The GC}MS
operated successfully, as contaminant peaks (methyl
chloride, Suorocarbon) were detected. The analysis of
Martian soil samples demonstrated the absence of
organic compounds above the detection limit of the
GC}MS instrument (a few ppb for the more volatile
organic compounds).
Viking Gas Exchange Experiments (GEX)
The biology instrument system had three different
experiments integrated in the same package: the pyro-
lytic release, the label release and the gas exchange
(GEX) experiments. In the GEX experiments Martian
soil samples were introduced, and a GC, with thermal
conductivity detection (TCD), measured ppm con-
centrations of metabolic gases such as methane. An
incubation gas (a mixture of CO2, He and Kr) was
introduced into the test cell and the biological activity
was stimulated either by humidifying the soil or by
adding a nutrient solution in the incubation chamber
(temperature between 5 and 273C). The changes in
the composition of the incubation gas were measured
periodically with a miniaturized GC.
The GC instrument (Table 2) was designed to
analyse light gases such as N2, O2, CH4, Kr, Ne and
CO2 at detection limits ranging from 20 to 60 ppm.
Ar and CO were co-eluted on the Porapak Q column
used. The composition of the Martian atmosphere
was determined by GC at both landing sites (four
analyses). The mean abundances were: CO2 96.2%,
N2 2.3%, O2(2.3% and Ar 1.5%, assuming that Ar
abundance is an order of magnitude larger than for CO.
 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY
The gas changes that occurred in the incubation
chamber of the GEX have raised much debate. The
decrease of CO2 just after wetting the sample material
has been explained by pH changes. The signiRcant
amount of O2 and its increase by humidifying the
sample could be due to the decomposition of inor-
ganic oxidants in the Martian soil.
Missions to Venus
Extreme temperatures (up to 4603C) and pressures
(up to 90 atm) are encountered during the descent,
with many reactive materials in a mainly CO2 atmos-
phere (Table 1). The clouds of the lower atmosphere
are composed of droplets of sulfuric acid. In a number
of experiments, the very short time of descent of the
probe through the Venusian atmosphere limited the
time available for GC analysis.
Pioneer Venus Gas Chromatograph
The GC on board the sounder probe of the NASA
Pioneer Venus mission was designed by Oyama and
co-workers for the in situ measurement of the com-
position of the lower atmosphere of Venus. It is
a modiRed version of the GC used in the Viking
GEX (Table 2). The separation was performed on the
two different analytical columns, each connected
to a TCD. The analysis of light gases (mainly Ne and
CO) was performed at 163C, using a long column
packed with Porapak N. The short column packed
with a polydivinyl benzene (PDVB) porous polymer
was used for separating gases from CO2 to SO2 at
623C.
The Pioneer Venus probe entered the Venusian
atmosphere on December 1978. During the time for
the probe to reach the surface (54 min), the GC ana-
lysed three atmospheric samples at 52, 42 and 22 km
altitudes. ChloroSuorocarbons were added to the
third sample in order to calibrate the instrument.
Table 3 gives the Venus atmospheric composition at
Table 3 Gas chromatography measurements of the lower atmospheric composition of Venus
Mission
Pioneer-Venus
Altitude (km) 52 42
CO2 (% by volume) 95.4 95.9
N2 (% by volume) 4.60$0.14 3.54$0.04
H2O (% by volume) (0.06 0.52$0.07
Ne (ppm by volume) (8 11
O2 (ppm by volume) 44$25 16$7
Ar (ppm by volume) 60.5 63.8$13.6
CO (ppm by volume) 32 30$18
SO2 (ppm by volume) (600 176
H2SO4 (mg m\3) aerosol
4215
different altitudes from the Pioneer Venus GC
measurements. The water result is consistent with the
value of the vapour pressure in presence of sulfuric
acid solutions.
For Ar, the lower abundance, as published in an
earlier paper, was due to an incorrect identiRcation
(Ar was identiRed as O2 and CO as Ar) as the assign-
ment of the peaks was made on the basis of absolute
retention times. Later, a correction accounting for
Sow rate variations and relying on retentions relative
to those of Freon internal standards was published.
Venera 12 Gas Chromatograph
In December 1978 the in situ analysis of the chem-
ical composition of the Venusian atmosphere was
performed by Gelman and co-workers with a GC on
board the USSR Venera 12 lander. The SIGMA in-
strument (Table 2) consists of three GC units ar-
ranged in series, each with a column connected to
a pure neon ionization detector operating in the cur-
rent-saturation mode with
-sources of different ac-
tivities.
A column packed with a modiRed sorbent was used
to separate H2S, COS, SO2 and H2O, in the presence
of CO2. The low boiling point gases (O2, N2, Kr, CH4
and CO) were analysed with a column packed with
molecular sieves. The third column (a chemical
reactor packed with reduced manganese) was used to
obtain the Ar content. The columns and detectors
were thermostatted at 703C.
Since the ionization detectors are sensitive to
carrier gas (neon) contamination, the whole GC sys-
tem was pressurized during storage and Sight. Eight
samples were analysed from a 42 km altitude to the
surface, with 18 chromatograms for the separation of
sulfur compounds (Rrst detector) and 27 for light
gases (second detector). The GC of the Venera
mission could not analyse for Ne, as this was used as
the carrier gas. The Ar abundance (Table 3) was
determined from its response as a negative peak. This
Venera 12 Vega
22 0}42 60}55
96.4$0.1
3.41$0.01 2.5$0.5
0.135$0.015 (0.01
4.3$0.7
(20
67.2$2.3 40$20
19.9$3.1 28$14
185$43 130$60 (100
&1
4216 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY
behaviour enabled estimation of the O2 mixing ratio
((2.10}3.00% by volume) as the GC of Venera
instrument did not directly measure O2 concentration
(O2 coelutes with Ar). The values of Ar and O2 mix-
ing ratios lie in the same range of experimental error
as the revised data of the Pioneer Venus GC.
Measuring atmospheric composition with different
instruments is clearly advantageous, allowing cross-
checks and preventing measurement errors. The quite
good agreement of the data in Table 3 validates the
results and gives a reasonable basis for models of
Venusian atmospheric chemistry.
Vega GC Experiments
The two spacecraft of the Vega mission reached
Venus in June 1985. The atmospheric probes (Vega
1 and 2) used a balloon to allow an hours duration
for the descent into the Venusian atmosphere. The
SIGMA-3 GC on board each probe was designed by
Mukhin and co-workers for the analysis of the gases
and aerosols of the Venus cloud layer (60}55 km
altitude). Three GC sub-units were arranged in paral-
lel, each having a column connected to a different
detector. Three different detectors were used: a he-
lium ionization detector, a TCD and an electron-cap-
ture detector (ECD). The carrier gas was helium
except for the sub-unit employing the ECD (carrier
gas: ultrapure N2). The separation of H2S, COS and
SO2 in the presence of CO2 and water vapour was
performed at 703C, with a column packed with
a mixture of Porapak QS and Porapak N.
In the gas analysis mode, the sample was heated at
803C and directly injected on to the column. At the
sampling altitude (60}55 km) the experiment demon-
strated the absence of H2S, COS and SO2 down to the
detection limit of the GC instrument (10}100 ppm,
depending on the substance). In the pyrolytic mode,
the cell containing a carbonized Rbre-glass Rlter was
heated at 3503C. At this temperature H2SO4 is broken
down into CO2, H2O and SO2. The comparison of
Sight experiments with simulation data enabled es-
timation of the concentration of H2SO4 in the
Venusian atmosphere to be about 1 mg m\3 for the
60}55 km altitude range.
Missions to Titan
Titan, a giant satellite of Saturn, has a dense atmos-
phere (Table 1). As with the Earths atmosphere the
main constituent is N2. CH4 is present at a low per-
centage. Traces of other organic compounds were
revealed by Voyagers infrared spectrometer. The
presence of these compounds was also predicted by
Raulin and co-workers from the results of laboratory
simulations. In addition, the atmosphere contains
aerosols and cloud droplets that obscure the surface
of the satellite.
One of most important goals of the Cassini}
Huygens mission to Titan is the in situ analysis of the
composition of Titans atmosphere. Successfully
launched on October 1997, the NASA spacecraft
(Cassini) carries a probe (Huygens) provided by the
European Space Agency (ESA). After release from the
orbiter in November 2004, the probe will slowly
descend to Titan by deploying three parachutes. The
six scientiRc experiments on the probe are designed to
determine the physical and chemical properties of the
atmosphere and the surface of Titan. Among these
instruments are the GC}MS and the aerosol collector
and pyrolyser (ACP).
Huygens GC+MS Experiments
The main objective of the GC}MS designed by
Niemann and co-workers is to measure the chemical
composition of the stratosphere and troposphere
(from 170 km to the surface) during the 2.5 h descent.
The GC}MS connected to the ACP will determine the
nature and the abundance of the organic and inor-
ganic compounds present in the atmosphere, both in
the gaseous phase and in the aerosols themselves.
Three columns operating in parallel will be used to
separate the expected species of Titans atmosphere
(Table 2). The identiRcation and detection are
achieved by connecting each column to an indepen-
dent MS ion source. The MS (quadrupole, range
2}150 amu) will operate in two modes, either
coupled to the GC or independently, by direct sample
injection (Figure 1). For many substances (noble
gases and many organics) mixing ratios as low as
0.1 ppb will be detected.
Capillary columns will be used for the Rrst time for
the in situ analysis of an extraterrestrial planetary
atmosphere. Sternberg and co-workers selected the
columns of the Cassini}Huygens mission for their
compatibility with the severe constraints imposed by
the experiment: fast analysis time, stability of the
stationary phase (vacuum, cosmic rays, high energy
electronic bombardment, inlet carrier gas pressure,
1.4}1.9 bar, outlet Sow-rate of (1 mL min\1,
isothermal analysis in the range of 30}603C. The Rrst
column, a carbon molecular sieve micro-packed
column, will be used to separate light gases such
as N2 to CH4. The second, a wall-coated open-
tubular (WCOT) capillary column of glassy carbon
will be used to separate low molecular mass hydro-
carbons (C1}C3). The analysis of the saturated and
unsaturated hydrocarbons (C4}C8) and the nitriles
(up to C4) will be achieved using a silicosteel
WCOT capillary column having a slightly polar
 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY 4217

Figure 1 Schematic diagram of the GC}MS instrument on board Huygens. Abbreviations: ACP, aerosol collector pyrolyser; BD,
burst diaphragm; CL, column; CR, column restrictor; CGR, carrier gas reservoir; FR, flow restrictor; G, getter pump; HA, heater (inlet
ACP); HI, heater (inlet atmosphere); HS; heater (sample volume); IV, isolation valve (H2 system); IVA, isolation valve (inlet ACP); PR,
pressure reducer; PSH, pressure sensor (H2 tank); PSC, pressure sensor (column); RL, restrictor leak; SV, sample volume; VC, valve
(column); VD descent/analysis control valve; VG, sample/carrier gas valve; VS, sample valve. (Reproduced from Sternberg R, Szopa
C, Coscia D, Zubrzycki S, Raulin F, Vidal-Madjar C, Niemann H and Israel G (1999) Gas chromatography in space exploration.
Capillary and micropacked columns for in-situ analysis of Titans atmosphere. Journal of Chromatography A 846: 307 with permission
of Elsevier Science.)

stationary phase: cyanopropyl phenyl (CPP) dimethyl- Mission to Comets

polysiloxane (DMPS) (Figure 2).
Aerosol Collector and Pyrolyser (ACP) Instrument
The ACP instrument was designed by Israel and co-
workers to sample and collect the aerosols of Titans
atmosphere, and then to transfer the products from
evaporation or pyrolysis to the Huygens GC}MS, for
analysis. The aerosols are collected on a multilayered
stainless Rlter by direct impaction for the Rrst samp-
ling (135}80 km). A pump is used at lower altitudes
(80}32.5 km) and (22}17 km) to draw the atmos-
phere through the Rlters for the two other samplings.
The Rlter is moved to an oven and heated at 2503 or
6003C. Labelled nitrogen is used to transfer the gas
and pyrolytic products to the GC}MS.
Each sample will be analysed using the direct MS
mode. The GC}MS analysis will be performed once
with the 6003C pyrolysis sequence. The tests for
validating the ACP}GC}MS experiment were made
by analysing the products synthesized by Raulin, Coll
and co-workers when the photolytic and radiolytic
processes expected in Titans atmosphere were
simulated (Figure 3).
It is generally believed that cometary nuclei, due to
their formation in the outer solar system at very low
temperatures, should retain the composition of the
solar nebula and thus the average composition of the
solar system. Considered as the most primitive bodies
of the solar system, comets are believed to have
seeded Earth with organic matter and water (Table 1)
through numerous impacts on the surface of the
primitive Earth. Therefore, cometary exploration is
of primary importance for a better understanding of
the solar system, as well as the origin of life on Earth.
Following several cometary Sy-by missions (e.g.
GIOTTO, VEGA) which provided the Rrst images of
a cometary nucleus and in situ measurements of the
composition of gas and dust released from the sur-
face, the ESA Rosetta mission will explore the nucleus
of P/Wirtanen comet. The ESA Rosetta mission will
be launched in 2003 and after two gravity-assisted
Sy-bys of Earth and Mars, it will reach the comet in
2011. The Rosetta mission will include a spacecraft
and a landing probe for the in situ analysis of the
cometary nucleus and its environment. The scientiRc
4218 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY

Figure 2 Chromatogram of a mixture of hydrocarbons and nitriles with the CPP-DMPS (14 : 86) WCOT capillary column for in situ
analysis of Titans atmosphere. Capillary column MXT 1701 (10 m;0.18 mm). Carrier gas, He; temperature, 303C, pressure drop, 0.3
bar. (1) Methane, (2) 1-butene, (3) n-pentane#1-pentene, (4) 2-methyl-2-butene, (5) cyclopentane#3-methylpentane, (6) n-
hexane#1-hexene, (7) acetonitrile, (8) acrylonitrile, (9) n-heptane#cyclohexene, (10) benzene#methacrylonitrile, (11) propionitrile,
(12) iso-butyronitrile, (13) cis- or trans-crotonitrile, (14) n-octane, (15) butyronitrile, (16) toluene, (17) cis- or trans-crotonitrile.
(Reproduced from Sternberg R, Szopa C, Coscia D, Zubrzycki S, Raulin F, Vidal-Madjar C, Niemann H and Israel G (1999) Gas
chromatography in space exploration. Capillary and micropacked columns for in situ analysis of Titans atmosphere. Journal of
Chromatography A 846: 307 with permission from Elsevier Science.)

payload of the cometary lander includes two instru-
ments for chemically analysing the surface.
The Rrst experiment, named COSAC (cometary
sampling and composition experiment), has been
built by Rosenbauer and co-workers at the Max-
Planck-Institut fuK r Aeronomie (Lindau, Germany)
and the second, Modulus (method of determining and
Figure 3 GC}MS analysis of a gaseous sample obtained after
irradiation (5 h spark discharge) of a gas mixture of N2 (800 mbar)
and CH4 (13 mbar) at 100}150 K. Capillary column CP}Sil}5CB
(25 m;0.15 mm). Carrier gas, He; temperature, 203C, then pro-
grammed at low temperature ((1503C). (Adapted from Coll P,
Guillemin JC, Gazeau MC and Raulin F (1999) Report and im-
plications of the first observation of C4N2 in laboratory simulations
of Titans atmosphere. Planetary Space Science 47: 1433}1440.)
understanding light elements from unequivocal stable
isotope compositions), has been built by Pillinger and
co-workers at the Planetary Science Research Insti-
tute, Open University (Milton Keynes, UK). These
two instruments will use state-of-the-art GC tech-
niques, involving pyrolysis and MS.
Cometary Sampling and Composition (COSAC)
Experiments
The COSAC experiments by Py}GC}MS are designed
to analyse gases either sampled directly from the
atmosphere around the nucleus, or provided by the
heating of nucleus material collected by the landers
sampler which can drill to a depth of at least 20 cm.
The pyrolyser consists of micro-ovens, mounted on
a carousel, which allow vaporization by stepwise
heating of the cometary solid sample. The GC sub-
system contains eight capillary columns divided into
two packages of four sharing a common injector. Up
to four columns, which can be selected individually,
can be operated in parallel in the temperature range
0}2003C. GC detection is performed by miniature
solid-state thermal conductivity detectors. COSAC
can be used as a stand-alone instrument or can be
coupled to the time-of-Sight (TOF) MS. Five of the
GC columns are WCOT and PLOT columns dedi-
cated to general chemical composition analysis.
In term of speed, efRciency, weight and carrier
 III / SPACE EXPLORATION: GAS CHROMATOGRAPHY
gas Sow-rate, these PLOT capillary columns will ad-
vantageously replace packed columns. The three
other columns will be dedicated to the measure of
chirality. Using chiral stationary phases they will be
able to separate enantiomers, and thus determine the
eventual presence of an enantiomeric excess.
Due to the large fraction (50%) of water vapour
which is expected in the cometary sample, a single
chemical water trap will be placed ahead of the
columns. The mass of the COSAC experiment is
constrained to 4.38 kg and the average power con-
sumption during operation should not exceed 15 W.
The Modulus GC+MS Experiments
The Modulus experiments will determine the abund-
ance and isotopic composition of major, minor and
trace constituents of the cometary nucleus. It uses
several analytical trains in parallel, each set com-
posed of: chemical reactors to quantitatively chemic-
ally transform the cometary samples into very light
volatile compounds, GC columns to purify and separ-
ate the resulting gases and detectors, including an ion
trap MS, to quantitatively analyse these gases.
By converting the elements of interest into speciRc
gases of low molecular weight such as O2, CO2,
N2 and CH4, the Modulus experiment only requires
a MS of low mass range with limited resolution.
Thus, it uses an ion trap MS with a mass of 10}20 g
(not including its power supply). The entire experi-
ment could require less than 3 kg weight and 5 W of
power. Two WCOT capillary columns, one of which
has a ceramic coating stationary phase, will be used.
Highly speciRc to volatiles (including permanent
gases and water) this stationary phase is robust and
withstands space constraints. It has to be noted that
a variant of this experiment will equip the Orbiter
craft to enable a comparative study between the
chemical composition of the coma and the nucleus to
be carried out.
Future Developments
Mars is the most interesting planet to study because it
may once have had an atmosphere similar to that of
the primitive Earth. In the next decade, an extensive
space programme will be devoted to the exploration
of the planet with the purpose of comparing its evolu-
tion with that of the Earth. The most consistent
explanation for the Viking failure to detect organic
molecules lies in photochemically produced oxidants
(such as H2O2) which originated in the atmosphere
and diffused into the soil, and are potential sources
of degradation of organic compounds. Missions, such
as the Mars Sample Return (to be launched in
2007) are now being planned with a landing probe
4219
including an experiment for exobiological character-
ization of the Martian surface material. The objec-
tives of these missions, in the frame of a large interna-
tional programme involving NASA, ESA, CNES and
other national space agencies, are to search for sub-
surface water as well as for traces of life (past and
present), organic compounds and oxidants. Several
instruments on board the lander, among them
a Py}GC}MS, will perform an in situ analysis of the
subsurface, at depths where the effects of ultraviolet
radiation and oxidizing agents are negligible.
Space instrumentation, because of its many
constraints, has brought about several technological
developments in the Reld of chromatography and has
opened the way for the chemical analysis of more
complex compounds in extraterrestrial environ-
ments. But there is a need of new instrumentation for
the analysis of non-volatile and/or thermally fragile
organic compounds, such as amino acids, incompat-
ible with pyrolysis techniques. The adaptation of
high-performance liquid chromatography (HPLC)
and supercritical Suid chromatography (SFC) to
space conditions seems difRcult. Chemical derivatiz-
ation coupled to GC (CD}GC) might be the solution.
The development of an automated chemical derivat-
ization process is under investigation and could be
used for the in situ analysis of the Martian soil in
forthcoming missions to Mars.
See also: II/Chromatography: Gas: Detectors: General
(Flame Ionization Detectors and Thermal Conductivity
Detectors); Detectors: Mass Spectrometry; Detectors:
Selective; Pyrolysis Gas Chromatography; Sampling
Systems. III/Atmospheric Analysis: Gas Chromato-
graphy. Chiral Separations: Gas Chromatography. Gas
Analysis: Gas Chromatography.
Further Reading
Biemann K, Oro J, Toulmin III P, Orgel LE, Nier AO, Ander-
son DM, Simmonds PG, Flory D, Diaz AV, Rushneck DR,
Biller JE and LaSeur AL (1977) The search for organic
substances and inorganic volatile compounds in the surface
of Mars. Journal of Geophysical Research 82: 4641.
Coll P, Guillemin JC, Gazeau MC and Raulin F (1999)
Report and implications of the Rrst observation of C4N2
in laboratory simulations of Titans atmosphere. Planet-
ary Space Science 47: 1433}1440.
Gelmen BG, Zolotukhin VG, Lamonov NI, Levchuk BV,
Mukhin LM, Nenarokov DF, Okhotnikov BP, Rotin VA
and Lipatov AI (1979) Venera 12 analysis of Venus
atmospheric composition by gas chromatography.
Prisma v Astronomicheskii Zhurnal 5: 217.
Israel G, Cabane M, Coll P, Coscia D, Raulin and Niemann
H (1999) The Cassini}Huygens ACP experiment and
exobiological implications. Advances in Space Research
23: 319.
4220 III / STEROIDS / Gas Chromatography
Mahaffy PR, AHearn MF, Atreya SK, Bar-Nun A, Bruston
P, Cabane M, Carignan GR, Coll P, Crifo JF, Ehren-
freund P, Harpold D, Gorevan S, Israel G, Kasprzak W,
Mumma MJ, Niemann HB, Owen T, Raulin F, Riedler
W, Schutte W, Sternberg R and Strazzulla G (1999) The
Champollion cometary molecular analysis experiment.
Advances in Space Research 23: 349.
Mukhin LM, Nenarokov DF, Porschnev NV, Bondarev VB,
Gelman BG, Israel G, Raulin G, Runavot J and Thomas
R (1987) Preliminary calibration results of Vega 1 and
2 SIGMA-3 gas chromatograph. Advances in Space
Research 7: 329}335.
Niemann H, Atreya S, Bauer SJ, Biemann K, Block B,
Carignan G, Donahue T, Frost S, Gautier D, Harpold D,
Hunten D, Israel G, Lunine J, Mauersberger K, Owen T,
Raulin F, Richards J and Way S (1997) The gas
chromatograph mass spectrometer aboard Huygens.
European Space Agency (ESA)-SP-1177: 85}107.
Oyama VI and Berdahl BJ (1977) The Viking gas exchange
experiment results from Chryse and Utopia surface sam-
ples. Journal of Geophysical Research 82: 4669.
STEROIDS
Gas Chromatography
H. L. J. Makin, St Bartholomews and the Royal London
School of Medicine and Dentistry, London, UK
Copyright ^ 2000 Academic Press
Introduction
This review aims to summarize the application of gas
chromatography (GC) to the analysis of steroids. The
review concentrates mainly on hyphenated GC}mass
spectrometry technology as the use of GC linked to
detectors other than mass spectrometry (MS) is now
decreasing. A survey of literature using MEDLINE
indicated that in the period 1990 to date, more than
90% of around 400 references used GC}MS, as might
be expected as the mass spectrometer is now the most
effective detector for GC and simple, cheap and sensi-
tive bench-top GC}MS systems are now widely avail-
able. Use of MS can often compensate for poor GC
resolution or peak shape, but use of GC}MS still
requires that attention is paid to optimization of both
GC and MS behaviour, if maximum sensitivity is
required. The MS, of course, has the added advantage
that it can provide structural data and can be used to
conRrm that a GC peak is indeed a steroid. By com-
Oyama VI, Carle GC, Woeller F, Pollack JB, Reynolds RT
and Craig RA (1980) Pioneer Venus gas chromatogra-
phy of the lower atmosphere of Venus. Journal of Geo-
physical Research 85: 7891.
Rosenbauer H, Fuselier SA, Ghielmetti A, Greenberg JM,
Gosemann F, Ulamec S, Israel G, Livi S, MacDermott
JA, Matsuo T, Pillinger CT, Raulin F, Roll R and
Thiemann W (1999) The COSAC experiment on the
lander of the ROSETTA mission. Advances in Space
Research 23: 333}340.
Sternberg R, Szopa C, Coscia D, Zubrzycki S, Raulin F,
Vidal-Madjar C, Niemann H and Israel G (1999) Gas
chromatography in space exploration. Capillary and
micropacked columns for in-situ analysis of Titans
atmosphere. Journal of Chromatography A 846:
307}315.
Wright IP and Pillinger CT (1998) Modulus } an experi-
ment to measure precise stable isotope ratios on
cometary materials. Advances in Space Research 21:
1537}1545.
paring the mass spectrum obtained with those in
a library can often identify the steroid. Retention time
data, on their own, are not a satisfactory criterion for
identiRcation but can be considerable value when
combined with MS data.
Steroids range from the C18 oestrogens to C27 ster-
ols such as cholecalciferol (vitamin D) and include
androgens, progestagens, corticosteroids and bile
acids as well as a large number of synthetic steroids,
some of which may be used therapeutically. The for-
mulae of some of these steroid types are illustrated in
Figure 1 in III/STEROIDS/Liquid Chromatography
and Thin-Layer (Planar) Chromatography. Alterna-
tively readers can consult the Dictionary of Steroids,
which lists some 10 000 steroids together with
their formulae, trivial and systematic names and
other useful information.
Derivatization
Most steroids have melting points in excess of 1503C
(estradiol-17
, the female sex hormone, for example,
has a melting point of 1763C). It is therefore often
necessary to derivatize steroids of interest in order to
optimize their GC performance. Derivatization im-
proves volatility, a necessary characteristic as the
analyte in GC must be in the vapour phase. High
injection (around 350}4003C) and column temper-
atures (up to 3503C) may also be necessary to achieve
 III / STEROIDS / Gas Chromatography 4221

separation, especially of higher molecular weight

steroids and their derivatives. The raised temperature
necessary to achieve satisfactory separation, also
brings with it the problem of analyte decomposition,
although decomposition should not be taken to mean
destruction. It is, for example, possible to separate
oestrogens and androgens and some progestagens
without derivatization, but 17-hydroxylated C21 ster-
oids (such as cortisol) undergo thermal side-chain
cleavage and cholecalciferol (vitamin D3) and its
metabolites all undergo B-ring closures, giving rise to
two isomers, even when derivatized. Such character-
istic reactions may often have useful diagnostic fea-
tures. It is also possible to enhance particular thermal
reactions such as dehydration by the use of catalysts
in order to obtain quantitative conversion in the in-
jection port to dehydration products, which may have
improved MS characteristics. 25-Hydroxyvitamin
D3 can be analysed in this way using aluminium
powder in the injection port and the dehydration
product has intense high mass ions which improve
sensitivity of MS detection and of course time-con-
suming derivatization is avoided. Figure 1 illustrates
this particular example.
Derivatization also improves GC peak shape as the
presence of hydroxyls increases adsorption during
Figure 1 On-column quantitative dehydration of underivatized
chromatography and at very low concentrations this 25-hydroxyvitamin D3 using aluminium powder in the injection
adsorption may give rise to a nonlinear response. In liner. (Upper panel) EI(#) mass spectrum of the dehydration
addition, if GC}MS is to be used, the appropriate product(s) } there are at least two dehydration products, which do
choice of derivative may also have a profound inSu- not separate, but only one is illustrated. Note the greatly in-
creased intensity of the molecular ion } m/z 364. (Lower panel)
ence on sensitivity and/or speciRcity of detection. An
Single ion monitoring of m/z 364, indicates only a single peak.
example of this is given in Figure 2, where the mass (From G Jones et al. In: Modern Chromatographic Analysis of
spectrum of the 3,17
-di(trimethylsilyl) ether of 19- Vitamins (eds A DeLeenheer, WE Lambert and HJ Nellis), 2nd
nor-androsterone is compared with the spectrum edn. New York: Marcel Dekker, 1992, with permission of authors
from the 17
-trimethylsilyl (TMS) ether. In this and publisher.)
example (this is the urinary metabolite which is mea-
sured in order to conRrm abuse of the anabolic ster- pact ionization (EI), halogenated derivatives are not
oid nandrolone), it can be seen that the intensity of necessary and hydroxyl groups are usually derivatized
the two high mass ions in the di-TMS ether (which is as TMS ethers and oxo groups as O-methyloximes (or
a 3-enol ether) are considerably greater than those for enolized to give enol-TMS ethers). Mixed derivatives
the mono-TMS ether, allowing greater sensitivity and are also used (e.g. O-methyloxime-TMS derivatives)
speciRcity of measurement. Table 1 lists some of the and in this example the oxime is formed Rrst and
common methods of derivatization, which protect protects the oxo group against subsequent enoliz-
against adsorption and decomposition and at the ation by the silylating reagent.
same time improve MS characteristics. Negative-ion 17-Hydroxylated C21 steroids are thermostable
chemical ionization (CI) techniques, which use soft when derivatized as 17-TMS-ethers-20-oximes and
ionization and yield predominately the molecular or can thus be analysed without degradation. Steroid
pseudo-molecular ion, can provide very sensitive as- carboxylic acids (e.g. bile acids) will not run in GC
say methods but require the presence of electron- systems except as aliphatic esters (usually this means
capturing moieties. Most steroids do not possess formation of methyl esters as otherwise molecular
these and derivatization is often used in this context weight and thus retention time increases). Other es-
to provide steroid derivatives containing the neces- ters have, however, been used for GC of faecal ex-
sary chlorine, iodine or bromine atoms (e.g. per- tracts to separate the bile acids from the neutral
Suoroacyl or chloro- or iodomethyldimethylsilyl sterols, which are insufRciently resolved as methyl-
ether derivatives). For the higher energy electron im- TMS ethers. Use of n-butyl-TMS ethers increases the
4222 III / STEROIDS / Gas Chromatography

Figure 2 Enhanced sensitivity of detection of the anabolic steroid, nandrolone, by formation of different derivatives. (Top) EI(#)
mass spectrum of the 17
-trimethylsilyl ether and (bottom) EI(#) mass spectrum of the 3-enol,17
-di(trimethylsilyl) ether. It can easily
be seen that the two ions at m/z 405 and m/z 420 of the di-TMSI carry more of the total ion current than the corresponding ions (m/z 333
and m/z 348) of the mono-TMS. These mass spectra were produced using equal amounts of nandrolone and the ion at m/z 91 offers
a useful index for comparison. (With permission of Mrs J Nolan.)

retention time of the bile acids sufRciently to separate are selective for particular parts of the steroid struc-
them from the sterol-TMS ethers. This is illustrated in ture, such as formation of cyclic boronates across
Figure 3. Other derivatives have also been used which vicinal hydroxyls. Such derivatives being selective
 III / STEROIDS / Gas Chromatography 4223

Table 1 Some derivatization procedures used for the GC and GC}MS analysis of steroids. This list is not comprehensive but includes
the majority of the most popular derivatives

Steroid group Derivative Formula*

Hydroxyl Trimethylsilyl ether (TMS) (CH3)3Si}O}St

t-Butyldimethylsilyl ether (TBDMS) (CH3)(CH3)2Si}O}St
Chloromethyldimethylsilyl ether (CH2Cl)(CH3)2Si}O}St
Dimethylethylsilyl ether (CH3CH2)(CH3)2Si}O}St
Pentafluorophenyldimethylsilyl ether (C6F5)(CH3)2Si}O}St
Acetate ester CH3CO}O}St
Formate ester HCO}O}St
Hepta- and pentafluorobutyrate ester CF3CF2CH2CO}O}St
CF3CF2CF2CO}O}St
Dimethylisopropylsilyl ether (CH3)2(CH3CHCH3)Si}O}St

Vicinal hydroxyls n-Butylboronate ester CH3(CH2)3B}(O)2}St

Oxo groups O-Methyloxime (St}C)"N}O}CH3

Enol-TMS ether (St}C"C)}O}Si(CH3)
O-perfluorobenzyloxime (St}C)"N}O-C6F5

Carboxylic acids Methyl ester (St}CO)OCH3

Isobutyl ester (St}CO)O(CH2)CH(CH3)2
n-Butyl ester (St}CO)O(CH2)3CH3

*St"steroid.

are diagnostic of structure and may also have the
advantage of improving sensitivity and speciRcity of
measurement.
Column Performance
For good GC performance, the intention is to obtain
symmetrical peaks with retention times as short as
necessary to achieve the desired separation. In the
past considerable attention was paid to the develop-
ment of different stationary phases in order to opti-
mize resolution but the advent of capillary column
and their linkage to MS systems has reduced the need
for new stationary phases. Although capillary col-
umns with a variety of bonded stationary phases are
available, most GC}MS systems for steroids use non-
selective (nonpolar) methylsilicone phases (e.g. HP1
columns from Hewlett-Packard), although more po-
lar phases may be necessary for particular separations
(i.e. C20 steroid carboxylic acids). Columns are usu-
ally around 15}30-m long (i.d. 0.2}0.4 mm with Rlm
thickness from around 0.1
m upwards) and carrier
gas Sow rates are 1}2 mL min\1, allowing direct in-
sertion of the column exit into the ion source of the
mass spectrometer. There are numerous means of
sample injection but we have found the easiest to be
direct on-column splitless injection using a syringe.
For optimum chromatographic performance, we have
found that the injection temperature is best kept at
4003C and that the choice of solvent can also have
inSuence. This high temperature causes considerable
problems in that most injection port septa are not
suitable and breakdown products cause interference.
This has been overcome by use of a septumless injec-
tion system (JADE injector) in which the syringe
needle injects onto the column through two stainless-
steel ball-bearings, which form the back-pressure
seal. Other injection procedures have found favour in
the steroids Reld, all of which strive to inject as much
of the extract as possible. These systems include
a dropping glass needle in which the sample is loaded
into small glass capillaries which can be automati-
cally loaded sequentially into the heated zone of the
injector. Cold trapping splitless injection has also
proved useful in that it allows for the on-column
injection of relatively large volumes of solvent into
silanized glass liners. Injection systems which load the
whole of the extract onto the top of the column
necessarily shorten column life and for quantitative
work, column deterioration must be monitored to
achieve consistent and high sensitivity. When deterio-
ration is detected, the column can be regenerated by
removal of the top 10 cm or so but this may lead to
alteration in retention characteristics of steroid deriv-
atives.
Prior to GC analysis, steroids must be extracted
and puriRed, the degree of puriRcation depending
upon the speciRcity of the detector system employed.
SpeciRc GC}MS systems require less pre-puriRcation
but the possible contamination of the MS ion source
must always be considered. Extended column life and
increased periods between ion source cleaning are
4224 III / STEROIDS / Gas Chromatography
Figure 3 GC chromatogram of sterols and bile acids present in
stool from a healthy control. 10 mg of freeze-dried stool contain-
ing 20
g nor-cholic acid was subjected to derivatization. After
dissolving in 200
L hexane, 1
L was injected into the GC col-
umn. Chromatographic and derivatization details can be found by
consulting the original paper. Peak identification: 1, nor-cholic
acid; 2, lithocholic acid; 3, iso-deoxycholic acid; 4, deoxycholic
acid; 5, chenodeoxycholic acid; 6, cholic acid; 9, 3-oxo,12-hy-
droxy-5
-cholanoic acid; 10, 12-oxo-lithocholic acid; a, copros-
tanol; b, cholesterol; c, 24-methyl-coprostanol; d, campesterol; e,
24-ethylcoprostanol; f, stigmasterol; g, sitosterol; h, sitostanol.
(From AK Batta et al (1999) Journal of Lipid Research 40:
1148}1154, with permission of authors and FASEB.)
obtained if attention is paid to pre-column puriRca-
tion. Silylating reagents should also be removed prior
to injection by use of small Lipidex 5000 columns,
unless they are sufRciently volatile not to cause
a problem. Trimethylsilylimidazole, a valuable re-
agent for the formation of TMS ethers on sterically
hindered hydroxyls (e.g. at positions C11
, C17,
C25), must be removed before GC}MS, whereas N-
methyl-N-trimethylsilyltriSuoroacetamide (MSTFA)
can be injected directly. Steroid glucuronides and
sulfates must be hydrolysed prior to GC as they do
not run in GC systems unless special derivatization
methods are adopted. While we have found
trimethylsilyl ethers to be stable, others have not. It is
advisable therefore to store and inject steroid TMS
ethers in MSTFA.
Mass Spectrometry and
Other Detectors
The GC of steroids can be carried out with a variety
of detectors, Same ionization (FID) being the most
widely used today. Electron-capture detectors (ECD)
which were commonly used in the past to improve
sensitivity of detection, have now largely been re-
placed with negative ion chemical ionization (CI)
mass spectrometry. Selective detection of steroid
oximes can be accomplished using nitrogen-phos-
phorus detectors. Today, however, the mass spec-
trometer in various forms offers the most versatile
detection system for GC, providing improved selec-
tivity and sensitivity in comparison to other detectors.
Because of the successful development of immunoas-
says for most of the clinically important steroids, GC
has not in recent times found much application for
individual steroid analysis, although occasional pub-
lications can still be found. However, the advent of
capillary columns with immense resolving power sug-
gested the possibility of utilizing GC as a means of
examining in a quick and simple way, the complex
patterns of steroids in human urine and how they
change in disease states. In the late 1970s Shackleton,
utilizing the pioneering work of Gardiner and Horn-
ing of ten years before, introduced the concept of
urinary steroid proRling. Urinary steroid extracts
(with or without
-glucuronidase hydrolysis) were
derivatized to form steroid O-methyloxime-TMS
ether derivatives and analysed using capillary open-
tubular columns monitoring the analytes by Same
ionization detection. Use of two internal standards
allowed the quantiRcation of 23 different steroids in
children with various steroid abnormalities. These
robust techniques are still in use today and provide
valuable information to assist clinical diagnosis and
monitoring of treatment and modern data handling
technology has greatly eased the task of interpreting
these complex proRles. The urinary proRling tech-
nique also allows identiRcation of unknown peaks in
the extract, when the original Same ionization de-
tector is replaced with a mass spectrometer. An
example of this methodology is illustrated in
Figure 4. Further information about this valuable
approach to urinary steroid analysis by GC}MS and
its application in the diagnosis of steroid related dis-
orders can be found in Shackletons article in the
Further Reading section.
The necessary process of puriRcation and derivatiz-
ation means that for quantitative work, suitable inter-
nal standards must be used. For GC}MS the best
internal standards are of course stable isotope
(deuterium or carbon-13) labelled analogues of the
analyte. In these situation at least three isotopic
atoms must be incorporated and the percentage of the
triply labelled standard (i.e. in the case of deuterium
labelled, d3) should be greater than 99%. Deuterium
labels are usually introduced by acid-catalysed
deuterium exchange and thus the label may not be

Figure 4 Steroid profiles by gas}liquid chromatography ob-
tained from urine samples from (upper trace) a normal adult and
(lower trace) a 16-year-old male with congenital adrenal hyper-
plasia (21-hydroxylase deficiency). Steroids were extracted with
Sep-Pak C18 cartridges and after hydrolysis of glucuronide and
sulfate conjugates, re-extracted and O-methyloxime-trimethylsilyl
ether derivatives were formed. These were analysed by
GLC using an OV1 capillary column. The major metabolites of 17-
hydroxyprogesterone (the substrate of the 21-hydroxylase en-
zyme) are named in the lower trace. Other peaks are as follows:
A, B and C: internal standards, androstanediol, stigmasterol and
cholesteryl butyrate; 1: androsterone; 2: aetiocholanolone; 3:
dehydroepiandrosterone (DHEA); 4: 11-oxo-androsterone; 5:
11
-hydroxy-androsterone; 6: 11
-hydroxy-aetiocholanolone; 7:
16-hydroxy-DHEA; 8: pregnanediol; 9: pregnanetriol; 10:
androstenetriol; 11: tetrahydrocortisone; 12: tetrahydro-11-dehyd-
rocorticosterone; 13: tetrahydrocorticosterone; 14: allo-tetrahyd-
rocorticosterone; 14: tetrahydrocortisol; 15: allo-tetrahydrocortisol;
16: -cortolone; 17:
-cortolone#
-cortol; 18: -cortol. (Kindly
provided by Dr Norman Taylor, Kings College School of Medicine
and Dentistry.)
stable in acid conditions. Ideally 13C-labelled stan-
dards are to be preferred but this requires incorpora-
tion into the nucleus of the steroid which can only be
achieved by extensive synthetic chemistry. All ster-
III / STEROIDS / Gas Chromatography 4225
oids are analysed by GC}MS in the same way and the
criteria used to ensure speciRcity/accuracy are those
adopted by the Substance Abuse and Mental Health
Services Administration (SAMHSA) for drug con-
Rrmation in employee drug-screening programmes
} two, but preferably three, speciRc ions (with as high
a mass : charge ratio as possible) must be monitored
and the results derived from each ion must not deviate
by more than 10% from the mean. Figure 5 illus-
trates the chromatograms obtained by multiple ion
detection, monitoring two of the relevant ions of the
analyte (25-hydroxyvitamin D3) and the equivalent
two ions from the hexadeuterated internal standard
present in a plasma sample extract. In this example
the standard curve relating peak height ratio
(analyte : internal standard against mass of standard
analyte) was linear and the intercept was not signiR-
cantly different from zero.
Isotope dilution GC}MS is widely acknowledged
as the gold standard of steroid analysis and is used as
a means of providing target values for external qual-
ity-assurance schemes and for the conRrmation of
immunoassay screening procedures for drugs of
abuse. Table 2 gives brief details of the application of
this methodology to the analysis of steroids in body
Suids, which are taken from papers in the literature
published between 1998}1999 and use both stable
isotope-labelled and unlabelled internal standards.
The availability of accurate and precise methods of
steroid analysis by GC}MS is becoming of increasing
public interest as the number of sportsmen and
women in whose urine metabolites of anabolic ster-
oids are found, continues to increase. It is clear that it
is important for steroid (and other) drug testing that
proper methodology for both qualitative detection
and quantitation is available and this methodology
can withstand public scrutiny. GC}MS provides pre-
cisely this. Excellent and up to date reviews of the
application of GC and GC}MS to the analysis of
steroids can be found in the Further Reading.
Mass Spectrometry for Structural
Analysis
The other important aspect of GC}MS, apart from
providing a high speciRcity method of steroid assay, is
the role of MS as a means of identifying both known
and unknown steroids. The use of the GC in this
context is simply a means of delivering a relatively
puriRed steroid derivative to the MS. The present
author and his colleagues have successfully used
GC}MS as a means of studying the metabolism of
calcitriol analogues in target tissues and while the
illustrations given are derived from these studies, they
have wider application and the methodology used can
4226 III / STEROIDS / Gas Chromatography

Figure 5 Isotope dilution mass fragmentography of 25-hydroxyvitamin D3. GC was carried out after formation of per-trimethylsilyl
ether derivatives using a non-selective OV1 column. An internal standard, [25,26-2H6]25-hydroxyvitamin D3, was added to the plasma
sample prior to extraction and purification. The GC column was inserted into the ion source of the mass spectrometer and four ions
were monitored (m/z 413 and 439 from the analyte and the corresponding ions, m/z 419 and m/z 445, from the internal standard). It will
be noted that the hexadeuterated internal standard runs slightly earlier than the non-deuterated analyte. In this case the ratio of peak
areas of analyte to internal standard gave a straight-line response which went through zero. Only two ions were monitored in this
example whereas increased specificity can be obtained if three are monitored. The major peaks, the pyro-isomer and the isopyro-
isomer, can be seen at approximately 11.20 min. Both peaks have a cyclized B-ring.
 III / STEROIDS / Gas Chromatography 4227

Table 2 Some examples of methods for the measurement of steroids by gas}liquid chromatography, published in 1998}1999

Subject Column details Derivatives used Internal standard Detection

Urinary steroid metabolite Non-selective O-methyloxime- Androstanediol, Flame ionization

analysis methylsiloxane and 5% trimethylsilyl ethers stigmasterol and detection* or mass
phenylmethylsiloxane 17 cholesteryl butyrate spectrometry (EI#)
and 25 m;0.2 mm
Ovarian steroids in the 15 m DB1 column O-methyloxime- None given Mass spectrometry
catfish trimethylsilyl ethers (EI#)
3-Reduced neuroactive 30 m;0.25 mm with O-methyloxime- None given Mass spectrometry
steroids in human plasma 0.2
m film thickness heptafluorobutyrate (EI#)
} HP5 esters
Anabolic steroid 30 m;0.2 mm with Pentafluoropropionates [1,2-2H2]-Testosterone Mass spectrometry
metabolites in urine 0.33
m film thickness. (CI!) using methane
5% phenylmethylsiloxane as reagent gas)
(ultra-2)
Detection of exogenous DB7 (50% Acetates Not relevant as only Mass spectrometry
testosterone phenylmethylsiloxane). 13
C/12C ratios } combustion isotope
administration 30 m;0.25 mm with being measured ratio
0.15
m film thickness
Endogenous HP1 30 m;0.25 mm with Trimethylsilyl ethers Trideuterated Mass spectrometry
19-nor-androsterone and 0.25
m and HP5 (5% (enols) and t-butyl- 19-nor-aetiocholanolone (EI#)
aeticholanolone in human phenylmethylsiloxane) dimethylsilyl ethers
urine 25 m;0.2 mm with
0.33
m
Serum DHA and DHA DB5 30 m;0.25 mm with No derivatization Androst-5-en-3
-ol- Ion trap mass
sulfate 0.25
m film thickness 16-one methyl ester spectrometry (EI#)
Testosterone : 17 m;0.2 mm with 3-Trimethylsilyl Not applicable as ratio Mass spectrometry
epitestosterone in equine 0.11
m film thickness ether-17-pentafluoro- being measured (EI#)
urine (5% phenyl- phenyldimethylsilyl
methylsiloxane) ether
Testosterone in hair Optima 1 25 m;0.2 mm Heptafluorobutyrate Trideuterated Mass spectrometry
with 0.1
m film thickness testosterone (EI#)
Urinary 3-oxo-4-bile 30 m;0.2 mm Carboxylic acid methyl 3,7-Dihydroxy-24- Mass spectrometry
acids methylsiloxane ester-dimethylethylsilyl nor-5
-cholanic acid (EI#)
ether and O-methyloximes
Biliary elimination of None given Heptafluorobutyrates Trideuterated High-resolution mass
endogenous 19-nortestosterone spectrometry (EI#)
19-nortestosterone

*See Figure 4 which illustrates the application of GC}FID for urinary steroid analyses.
DHA, dehydroepiandrosterone.

be applied to all steroids and their metabolites.
Figure 6 shows the GC trace of calcitriol (1,25-dihyd-
roxyvitamin D3) as the per-trimethylsilyl derivative.
Two peaks are always seen, as B-ring cyclization
which occurs at the high temperature of the injection
port quantitatively produces pyro- and isopyro-
isomers which are always formed in the same ratio.
Thus for every vitamin D-like compound, two GC
peaks are observed. It is the pyro-peak which pre-
dominates and Figure 6 also shows the EI(#) mass
spectrum derived at the apex of the pyro-peak after
background subtraction. For the purpose of struc-
tural analysis, EI(#) spectra are preferable to CI
spectra as CI is a much softer technique giving less
useful fragmentation } a similar objection applies to
LC}MS, which also uses soft ionization. It is of
course also possible to obtain a spectrum of the un-
derivatized, albeit cyclized, calcitriol by ignoring the
GC and inserting the calcitriol into the ion source of
the MS by direct probe. This gives a molecular ion
(M#) of 416. Examination of the mass spectrum of
the per-TMS derivative shown in Figure 6 indicates
a molecular ion of 732. Knowing that each TMS
formed increases the molecular weight by 72 amu, it
is possible to calculate the number of hydroxyl groups
in an unknown compound (632!416"216 and
216/72"3). Metabolism of calcitriol and its ana-
logues usually involves cytochrome P450-catalysed
4228 III / STEROIDS / Gas Chromatography
Figure 6 EI(#) mass spectrum of 1,25-dihydroxyvitamin D3. The total ion current is shown in the upper panel indicating the two
cyclized isomers (pyro- at 10.90 min and isopyro- at 11.64 min) which are formed. In the lower panel is the mass spectrum of the
per-trimethylsilyl ether of the pyro-isomer.
hydroxylation(s). The number of hydroxylations can
be determined by the same procedure described above
and if the MS of the per-TMS of the substrate is
known, direct probe MS is not necessary. However
further interpretation of the MS becomes necessary in
order to decide where on the steroid molecule the
hydroxylation has occurred. It is clearly also possible
to deduce the presence of an oxo group as this in-
creases the molecular ion of the substrate by 14 amu
but again knowledge of the presence of this group
does not determine its position. To carry out these
calculations, it is necessary to be able to determine the
molecular ion value. It is not always possible to do
this directly as the mass spectra of the steroids usually
have very low intensity molecular ions. However, as
can be seen in Figure 6, all these cyclized steroid-TMS
ethers have a prominent (M-131)# ion, which is
usually derived from A-ring cleavage, as well as (M-
90)# ions, derived by successive loss of silanols. It is
therefore possible even in the absence of discernible
M# ions in the spectrum, to determine the m/z value
of the molecular ion.
For the identiRcation of the position of extra hy-
droxyls and oxo groups or even truncation, where
cytochrome P450 lyases have cleaved the side chain,
GC retention time data can prove extremely useful.
Hydroxylation increases retention time but the fur-
ther out along the side chain (distal) the hydroxyla-
tion is, the longer the retention time. Truncation, by
reducing molecular weight, clearly decreases reten-
tion time. Retention time, although useful is not
sufRcient on its own and further study of the frag-
mentation data has to be made. Further examples can
be obtained by consultation of the texts listed in the
Further Reading section. Consideration should also
be given to the use of chemical reactions which mod-
ify the molecule under investigation. Reduction of
oxo groups with sodium borohydride and subsequent
derivatization as TMS ethers and GC}MS provides
further evidence of structure. Cleavage of car-
bon}carbon bonds between vicinal hydroxyl groups
with periodate can also provide valuable information
about the site of hydroxylation if the reaction product
is subsequently derivatized and subjected to GC}MS.
A very good example of the interpretation of mass
spectra obtained from GC}MS of per-TMS deriva-
tives is given in Figure 7. The metabolites illustrated
here are all mono-hydroxylated metabolites of 1-
hydroxyvitamin D3 and thus give the same value of
632 amu for their molecular ion. All four metabolites
show the characteristic (M-131)# ion at m/z 501 as
well as (M-90)#, 542 and (M-90-90)#, 452. The
abundance of the M# ion is, as usual, very low but it
can easily be conRrmed as being the ion at m/z 632 by
 III / STEROIDS / Gas Chromatography 4229

tic for the position of the hydroxyl on the side chain

and the derivation of these ions is shown in the
fragmentation patterns illustrated in Figure 7. Many
other examples of this sort of elucidation of seco-
steroid structure can be given, all of which rely on the
same sort of approach.
Routine steroid analysis at ng mL\1 concentrations
by GC}MS utilizes low-resolution mass spectrometry
but there are occasions when increased sensitivity is
required for the detection/measurement of steroids at
concentrations in the pg mL\1 range. This can be
achieved by using high-resolution (double-focusing)

Figure 7 The EI(#) mass spectra of metabolites of 1-hy-

droxyvitamin D3 (1-OHD3). GC}MS was carried out after de-
rivatization to form the per-trimethylsilyl ethers. Both pyro- and
isopyro-isomers of each metabolite were observed but the mass
spectrum of the pyro-isomer (the major peak) is shown in each
case. The major ions (m/z 632 (M#), m/z 542, 432 and 362 (not
highlighted) (M# losing successive silanols) and m/z 501 (M#
losing 131 by A-ring cleavage) are the same in all the spectra. m/z
217 is the characteristic ion always seen in these 1,25-dihyd-
roxylated steroids and m/z 251 (not highlighted) arises by side-
chain cleavage and subsequent loss of three silanols. It is how-
ever possible to distinguish each isomer from the characteristic
fragmentation patterns illustrated for each above the appropriate
spectrum. (From G Jones and HLJ Makin (2000) In: Modern
Chromatographic Analysis of Vitamins (eds A DeLeenheer, WE Figure 8 High-resolution mass fragmentography of an extract
Lambert and HJ Nellis), 3rd edn. New York, Marcel Dekker, to be of serum from a patient taking vitamin D2, showing ion chromato-
published, with permission of authors and publisher.) grams of per-trimethylsilylated (TMS) ether of putative 1,24-
(OH)2D2, monitoring three separate ions, m/z 513.3584 (A), m/z
consideration of the origin of the more abundant 554.3975 (B), and m/z 601.3929 (C), showing the trace between
9 and 14 min. The peaks from the pyro-isomer of 1,24-(OH)2D2-
ions. Although not shown here, the retention times
TMS are shaded. The ion ratios in this extract are the same as
increase as the hydroxylation position moves distally those in the mass spectrum of the authentic compound. (From EB
along the side chain. It is the presence of other less Mawer et al. (1998) Journal of Clinical Endocrinology and Meta-
abundant ions of lower m/z value, which are diagnos- bolism 83: 2156}2166, with permission of authors and publisher.)
4230 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography
instruments which, although they increase speciRcity,
reduce overall sensitivity but paradoxically allow in-
creased sensitivity of measurement by increasing the
signal : noise ratio. GC}HRMS has successfully been
used for the measurement of a calcitriol analogue,
hexaSuorocalcitriol, with a minimum detectable limit
of 2 pg mL\1, which gives this assay the sensitivity to
measure plasma calcitriol itself, which circulates at
concentrations around 30 pg mL\1. This principle is,
of course, generally applicable and most steroids can
be detected at lower concentrations by the use of
GC}HRMS. This technique has been used mainly by
laboratories interested in the detection of anabolic
steroids in athletes urine (e.g. metandienone,
stanozolol and clostebol) as a means of detection of
drug abuse in sport but also as a means of detecting
illicit steroid administration to cattle (4-chlorotestos-
terone). It has occasionally been suggested, as in
the case of nandrolone, that metabolites observed
have arisen de novo by in vivo metabolism from
other steroids rather than from exogenous sources.
Use of GC}combustion-MS (isotope ratio mass spec-
trometry) has been shown, by measuring the 12C : 13C
ratios, to have considerable potential as a means of
distinguishing between exogenous and endogenous
sources. Figure 8 gives a further example of the sensi-
tivity of GC}HRMS which was used to demonstrate
the presence of 1,24-dihydroxyvitamin D2 in human
plasma by focusing on three speciRc ions and demon-
strating that they had a retention time the same as the
standard and were present in the same ratio and as
they were in the MS of the pure standard. Similar
studies with low-resolution MS detection were un-
able to demonstrate the presence of this steroid.
Conclusion
GC}FID of steroids is today primarily conRned to the
analysis of urinary steroid proRles, a technique intro-
duced in the 1980s but, as a brief examination of the
recent literature will show, still produces valuable
Liquid Chromatography and Thin-Layer (Planar) Chromatography
H. L. J. Makin, St Bartholomews and the Royal London
School of Medicine and Dentistry, London, UK
Copyright ^ 2000 Academic Press
Introduction
This review aims to summarize the application of
liquid chromatography (LC) in all its forms, including
thin-layer chromatography (TLC), for the analysis of
clinical information today. Much improved data are
obtained when the GC is interfaced with the a mass
spectrometer, allowing greater sensitivity and speciR-
city of detection with the added beneRt of structural
information about unknown steroids. It is interesting
to note that C21 steroids are usually analysed by
immunoassay or LC}MS whereas GC}MS is still
widely used for the speciRc analysis of oestrogens and
androgens, particularly in the sports area where the
deRnitive detection of anabolic steroids is required.
GC}MS, particularly when high-resolution MS is
used, is still more sensitive than LC}MS for steroid
assay and EI(#) ionization methodology provides
more useful structural information than can be
achieved with LC}MS or even LC}MS}MS. It will be
interesting to see whether GC}MS will hold its own
against LC}MS over the next ten years.
See also: II/Chromatography: Gas: Derivatization; De-
tectors: Mass Spectrometry; High Temperature Gas
Chromatography. III/Steroids: Liquid Chromatography
and Thin-Layer (Planar) Chromatography; Supercritical
Fluid Chromatography.
Further Reading
Hill RA, Kirk DN, Makin HLJ and Murphy GM (eds)
(1991) Dictionary of Steroids. London: Chapman and
Hall.
Makin HLJ, Gower DB, and Kirk DN (eds) (1995) Steroid
Analysis. London and Glasgow: Blackie Academic and
Professional.
Makin HLJ, Trafford DJH and Nolan J (1998) Mass
Spectra and GC Data of Steroids: Androgens and Estro-
gens. Weinheim: Wiley-VCH.
Shackleton CHL (1993) Mass spectrometry in the diagnosis
of steroid-related disorders and hypertension research.
Journal of Steroid Biochemistry 45: 127}140.
Wolthers BG and Kraan GPB (1999) Clinical applications
of gas chromatography and gas chromatography}mass
spectrometry of steroids. Journal of Chromatography
A 843: 247}274.
steroids. As LC relies on either adsorption or parti-
tion, extraction of the analyte from the matrix, a sim-
ilar process, has been considered, as has the necessary
Rnal step of LC-quantitation. Readers who seek fur-
ther information are encouraged to use the texts given
in the Further Reading section, which are valuable
sources of information from which original research
references can be obtained as well as information
about alternative means of steroid analysis.
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography 4231

Steroids comprise a large group of compounds
which occur naturally in both plants and animals.
Their structures are all based upon the cyclopen-
tanoperhydrophenanthrene nucleus and all the nat-
urally occurring steroid hormones are synthesized in
humans in vivo from cholesterol. Some steroid hor-
mones } those derived from vitamin D3 which are
derived from cholesterol precursors } have a broken
B-ring and are described as secosteroids. Various
chemical modiRcations of the nucleus can be made by
increasing the size of the rings or modifying them in
some way to produce large numbers of synthetic
steroids. As an illustration of the wide variety of
steroids which are available today, the Dictionary of
Steroids lists around 10 000 compounds. Steroids
have a wide spectrum of therapeutic uses and this has
encouraged the synthesis of large numbers of syn-
thetic steroids in an attempt to enhance or depress

Figure 1 Formulae of some naturally occurring steroids. The numbering system used to identify the individual carbon atoms in the
steroid skeleton is also illustrated.
4232 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography
particular physiological responses. From the point
of view of a person working in a biomedical environ-
ment, the naturally occurring steroids are of particu-
lar interest and these include the gluco- and
mineralo-corticoids secreted by the adrenal cortex,
the sex hormones produced in the gonads, proges-
terone synthesized in the placenta and the bile acids
which aid the digestion of fats. The parent compound
of all these naturally occurring human steroids, cho-
lesterol, is an integral part of the structure of cell
membranes. The nomenclature of steroids is complic-
ated by the fact that trivial names of many important
steroids (see, for example, cortisol and testosterone in
Figure 1) are still widely used. There are agreed
IUPAC rules for the nomenclature of steroids but
application of these rules gives rise to long and cum-
bersome names. Readers who are unfamiliar with
steroid nomenclature are referred to the Dictionary of
Steroids which contains a very useful summary.
Figure 1 also illustrates the numbering of some of the
important carbons in the steroid nucleus.
It has been estimated that, of the armoury of thera-
peutic drugs available for prescription in the UK,
around 25% either are or contain steroids in their
formulation. Because of the physiological and thera-
peutic importance of steroids and the huge number of
different steroids which one may encounter, they rep-
resent a considerable analytical challenge. In this
short summary of the liquid chromatographic
methods for the separation of steroids, attention will
be devoted in the main to the separation and quanti-
tation of the naturally occurring steroid hormones
and bile acids. Figure 1 gives the structures of some of
the important steroids which are found in human
serum and examples of their conjugates. Readers who
wish to learn more about the inRnite variety of ster-
oids are referred to the classic organic chemistry text
by Fieser and Fieser and the Dictionary of Steroids.
A text on the Biochemistry of Steroid Hormones is
given in the section on Further Reading.
Steroids are in the main hydrophobic, a property
conferred by the nucleus, and this hydrophobicity is
modiRed by hydroxyls and oxo groups on the periph-
ery of the nucleus. Steroids are often conjugated with
glucuronic and sulfuric acids, particularly through
the hydroxyl at carbon 3. These conjugates are of
course more water-soluble than the unconjugated
steroid. The side chain attached to carbon 17 of the
nucleus in cholesterol contains a further 10 carbons
and this side chain is in vivo enzymatically cleaved
between C24 and C25 to produce the C24 bile acids and
between C20 and C22 to produce steroid hormones.
The C24 carboxyl can also be conjugated with glycine
and taurine which again increase the water-solubility
of these molecules. There is therefore quite a wide
variation in hydrophobicity between different classes
of steroids and within these classes, which can be
further modiRed by conjugation. Most steroids are
neutral but the phenolic A ring of the oestrogens and
the C24-carboxyl in the bile acids render them acidic
and this property can be used for the differential
extraction of these two classes of steroids.
In any analytical system there are three interdepen-
dent steps: extraction (removal of the analyte from
the matrix), separation of the analyte from other
compounds, which may interfere in the Rnal step
} quantitation. Separation and quantitation are
clearly very closely linked in that a quantitation
procedure of high speciRcity may well not require
such intensive separation as would be required with
a low speciRcity quantitation. Because of the chem-
ical similarity of the many steroids with each other
and, in general, the lack of highly speciRc quantitat-
ion procedures, separation of steroids prior to quanti-
tation is still extremely important. Each of these three
stages will be dealt with individually but it must be
remembered that, when an analytical procedure is
being put together, one stage cannot be viewed in
isolation from the others.
Extraction
Unconjugated steroids are hydrophobic and are rela-
tively easy to extract from the aqueous matrices in
which they are often found. The apparent dichotomy
of hydrophobicity and the presence of unconjugated
steroid hormones in human plasma is resolved when
one recognizes the presence of speciRc binding
globulins. The main glucocorticoid, cortisol, has
a speciRc binding globulin (transcortin) and the sex
hormones also have a speciRc globulin which trans-
port these steroids in human blood. To extract ster-
oids therefore from serum or plasma, it is necessary to
disrupt the steroid}protein binding. Some steroids,
such as cholesterol or vitamin D3, are particularly
difRcult to extract and it is thought that this occurs
because they become involved in lipoprotein struc-
ture. It is possible to overcome this difRculty by ex-
tracting with ethanol}ammonium sulfate or
pentylamine. Most steroids however can be extracted
from plasma/serum or incubation medium with
a simple Bligh & Dyer extraction which utilizes meth-
anol}chloroform (2 : 1, v/v). A simple wash of the
organic extract with alkaline buffer will remove fatty
acids which are also extracted but may interfere
in subsequent analysis. However, washing with
alkaline buffer may also remove substantial quanti-
ties of acidic steroids such as bile acids and oestro-
gens. In the past, ether was a common solvent for
steroid extraction as it is less dense than water and the
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography
aqueous layer can be frozen with solid CO2 and the
organic extract poured off. However, in more safety-
conscious times, the Sammability of ether has re-
duced its use.
The extraction of steroids using solvents is dis-
cussed in more detail in texts cited in the Further
Reading section. Such procedures should not be
viewed solely as a means of extraction as judicious
choice of solvents can give a surprising degree of
selectivity and particular steroid groups can be prefer-
entially extracted. Steroid conjugates, which are more
difRcult to remove from aqueous media, can also be
extracted from, for example, human urine using
ether}isopropanol after saturation of the urine with
ammonium sulfate}so-called forced extraction. The
conjugates can then be hydrolysed using enzymes
(-glucuronidase or sulfatase) or, in the case of sul-
fate, acid solvolysis can be utilized. The need for
hydrolysis of steroid conjugates depends upon the
subsequent separation and quantitation techniques.
Clearly hydrolysis of conjugates loses information
which may or may not be of importance. As will be
seen later, modern methods of analysis using
LC}mass spectrometry (LC-MS) allow for the separ-
ation and quantitation of intact conjugates and it may
therefore be unnecessary to hydrolyse before proceed-
ing to the separation or steps.
Solvent extraction leads to the generation of rela-
tively large volumes of potentially hazardous solvents
which need to be removed, usually using a rotary
evaporator or simply blowing nitrogen onto the sol-
vent while heating it not higher than about 403C.
Solvents which have high boiling points or solvent
mixtures containing water are particularly difRcult to
remove. Because of these problems, other methods of
extraction have been investigated and, in the case of
steroids, major advances have been made, parti-
cularly in the Reld of solid-phase and immuno-
afRnity extraction. As examples of solid-phase extrac-
tion (SPE), one can consider the use of microparticu-
late silica for the extraction of steroids and vitamin
D metabolites. There are a wide variety of such ma-
terials which are all based upon microparticulate sil-
ica, modiRed by derivatizing the polar groups with
silanes (i.e. octadecylsilane (ODS) C18, is widely
used). Structures and performance of the solid-phase
materials can most readily be obtained by looking at
the catalogues of manufacturers of these materials.
In the UK a very useful source of information is
the catalogue of International Sorbent Technology,
a major supplier of such materials (e.g. Bond-Elut).
Sep-Pak is another useful proprietary brand, manu-
factured and marketed by Waters. As an example of
the use of these materials, vitamin D3 metabolites in
plasma, although not vitamin D3 itself, can be extrac-
4233
ted with acetonitrile, which disrupts the protein bind-
ing. After centrifugation to remove the precipitated
protein, the extract is then poured through an ODS-
silica column or cartridge (Sep-Pak C18 or Bond-Elut
C18) and the metabolites of interest can be eluted,
after washing, with methanol. A similar procedure
can be used for other steroids in plasma or urine and
often their conjugates as well.
SPE techniques for steroid extraction, although not
speciRc, are increasingly used in preference to solvent
extraction. The SPE material can often be reused
many times, if satisfactory washing procedures are
applied between each use. Highly speciRc extraction
can be achieved using immunoafRnity columns where
antibodies to speciRc steroids or groups of steroids
are immobilized by linking to Sepharose. Aqueous
mixtures of steroids can then be passed down the
column: steroids of interest are bound to the antibody
and after the unwanted steroids have passed through
the column the steroid antibody binding can be dis-
rupted and the steroid(s) of interest eluted. In ideal
cases using highly speciRc antibodies and a relatively
speciRc quantitation, it may not be necessary to carry
out any further separation procedures. Sometimes
simple procedures can be extremely effective. As an
example, the binding of some plasma steroids to
speciRc globulins can allow selective extraction as
ammonium sulfate can sometimes be used to precipi-
tate the speciRc globulin, which brings the steroid of
interest with it.
Separation
Today high performance liquid chromatography
(HPLC) is widely used for steroid separation because
this technique can be directly linked to quantitation.
This is, however, not to imply that other methods of
separation may not Rnd use in particular applica-
tions. Open-column chromatography (either adsorp-
tion or partition) is still used with advantage on
occasions. A major and very useful separatory tech-
nique is TLC and, if microparticulate material is used,
it becomes high performance TLC (HPTLC). TLC is
particularly advantageous in that numbers of separ-
ations can be carried out at the same time and the
apparatus required is inexpensive. For these reasons
and because TLC is relatively easy to carry out, it is
still quite widely used and scrutiny of recent papers
on steroid separation conRrms this. It is however true
to say that very little development of TLC systems has
occurred in the last 20 years and most systems are
based upon methods described prior to 1980. TLC
can also be used as a preliminary means of investigat-
ing new solvent systems for the separation of steroids
by HPLC.
4234 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography

Column Chromatography ModiRed cross-linked dextran columns (i.e.

This procedure utilizes adsorbent materials such as
alumina, Florisil (magnesium silicate) or silica. Ster-
oids are adsorbed to these materials and are eluted by
solvents of increasing polarity. The order of elution
depends upon the solvent and the differing polarity of
the steroids under consideration. Clearly, the more
polar a steroid (in general, this means the more hy-
droxyls it contains), the more hydrophilic the steroid
becomes and the longer it takes to be eluted. These
adsorbent materials are usually packed into small
columns (for example, Pasteur pipettes can be used)
and exquisite separations can now be achieved by the
use of microparticulate silica. These columns are
simple to use, provide rapid separations and have the
advantage that after washing they can be reused
many times. Classical steroid separations using col-
umns can be achieved by partition chromatography
using biphasic solvent systems. The stationary aque-
ous phase is mixed with celite (a diatomaceous earth)
and this material is packed into the column; the
steroid mixture is applied and eluted with the organic
mobile phase. This is rather a cumbersome procedure
but does offer a considerable degree of separation. It
is not widely used today, although examples of parti-
tion chromatography used in this way can still be
found. An example of this technique for the separ-
ation of some progesterone metabolites is illustrated
in Figure 2.
Lipidex) have been used to provide steroid separ-
ations and can Rnd more mundane uses such as the
removal of excess trimethylsilylimidazole reagent
when forming steroid trimethylsilyl ethers. A similar
material, Sephadex LH-20, has also been used to
fractionate steroids into free steroids, glucuronides,
mono- and disulfates. Sephadex can also be modiRed
by linking it to form, for example, diethylaminoethyl-
substituted material which can act as an ion ex-
changer as well as a size exclusion material. These ion
exchange/gel Rltration columns are particularly use-
ful for the separation of steroid conjugates; for
example, bile acids can be separated by the use of
another modiRed Sephadex, PHP-LH20. Figure 3
shows such a system for the separation of steroids and
their conjugates from human urine prior to gas
chromatography}MS (GC-MS).
Thin-Layer Chromatography
TLC is in effect very similar to column chromatogra-
phy and is based on the same principles. A thin layer
of adsorbent or inert material is spread on a glass,
plastic, or aluminium sheet. For the separation of
steroids using organic solvents, the use of thin layers
on plastic sheets is not recommended and either glass
or aluminium should be used. After separation,
steroids may be recovered from the plate by scraping
off the thin layer and eluting the steroid of interest.

Figure 2 Chromatogram of selected C21 steroids in urine. Radiolabelled and nonradiolabelled authentic steroid standards were
applied to a 40 g column of celite}propylene glycol and eluted with iso-octane (e.g. Iso-8, 40 fractions) and thence a linear gradient of
iso-octane}ethyl acetate (e.g. Iso-8, 140 fractions). Steroids are identified as follows: 5
DHP, 5
-pregnane-3,20-dione; 5
3
, 5
-
pregnan-3
-ol-20-one; 5
20
, 5
-pregnan-20
-ol-3-one; 5
3, 5
-pregnan-3-ol-20-one; 20
DHP, 4-pregnen-20
-ol-3,20-dione;
DOC, 4-pregnen-21-ol-3,20-dione; 6OH-P, 4-pregnen-6-ol-3,20-dione; 21OH-P5, 5-pregnen-3,21-diol-20-one. (Reproduced
with permission from Dombroski RA, Casey ML and MacDonald PC (1997) Journal of Steroid Biochemistry 63: 155}163.)
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography 4235

Figure 3 Use of solid-phase extraction and Sephadex-based packing materials for the extraction and separation of steroids and their
glucuronide, sulfate and N-acetylglucosaminide conjugates from human pregnancy urine prior to analysis by gas chromato-
graphy}mass spectrometry (GC-MS). (Reproduced with permission from Meng LJ, Griffiths WJ and SjoK vall J (1996) Journal of Steroid
Biochemistry and Molecular Biology 58: 585}598.)

The advantage of thin layers on aluminium foil is that
the area of interest can be cut out and the whole
area of the plate can be eluted without removing the
adsorbent.
The commonest forms of TLC for the separation of
steroids use adsorbent thin layers of either alumina or
silica gel, although there are descriptions of reversed-
phase partition TLC. Use of microparticulate silica
(HPTLC) confers a slightly enhanced selectivity in
separation but HPTLC is not widely used. The steroid
extract of interest is applied to the bottom of the plate
and eluted with various solvents. Because of the abil-
ity of such separation systems to deal with a number
of separations at once, they are still quite widely used
4236 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography

today. The vast majority of separation systems use
silica gel but alumina has been used to separate
C19 androgens. The resolution is achieved by judi-
cious choice of solvents as the number of adsorbents
available is limited.
Very little development of this separation tech-
nique for steroids has been carried out over the last 20
years as attention has been diverted towards the de-
velopment of HPLC. Recent TLC systems using silica
gel and eluting with ethyl acetate : petroleum ether
have enabled hydroxylated metabolites of proges-
terone to be resolved and again using silica gel with
benzene}heptane}ethyl acetate or chloroform}ethyl
acetate has enabled resolution of catechol oestrogens.
In many instances, however, suitable solvents can no
longer be used as they pose unacceptable risks of
Sammability or toxicity (e.g. benzene). A study of
papers published in the last 10 years involving the
use of TLC for the separation of steroid metabolites
indicates that most procedures were published
many years ago, although the recent use of cyclo-
dextrin in the TLC of bile acids (cyclodextrin
in the mobile phase) and steroidal drugs (cyclodex-
trin polymer-coated silica) has been an interesting
innovation.
Table 1 lists some examples of the recent use of
TLC for the separation of steroids and clearly illus-
trates the increasing reliance on silica gel as the adsor-
bent, using the differing polarity of the elution solvent
mixture for resolution. Most of the development
systems are modiRcations of previously published sol-
vent mixtures. Table 1 also illustrates the use of
two-dimensional TLC and multiple development in
the same direction. It is also possible to utilize one
solvent mixture for de-fatting, followed by another to
separate the steroids of interest or sequential solvent
systems, separating Rrst one group of steroids
followed by a second solvent mixture to separate
another steroid group.
The problem associated with the use of TLC for the
separation of steroids is to locate the position of the
steroids after chromatography and some methods

Table 1 Some examples of recent use of TLC for steroid separation

Steroids Adsorbent Development solvent Detection

Side chain cleavage products of Silica gel G Di-isopropyl ether}hexane} Radioactivity by

14
C-cholesterol acetic acid (70 : 30 : 2) autoradiography
Metabolites of 7
-OH- Silica gel Ethyl acetate}hexane (3 : 7)
androstene-3,17-dione
7-Hydroxylated DHEA Silica gel F254 Ethyl acetate UV absorption
Oestrogens Keiselgel 60 F254 Toulene}acetone (4 : 1) UV absorption and
iodine vapour
3
H corticosteroids Silica gel 60 F254 Choloroform}acetone (23 : 2) Autoradiography
3
H testosterone metabolites Polygram SIL G Methylene chloride}diethyl ether Autoradiography
(4 : 1)
Oestrogens Whatman LK6DF Silica gel 60 Ether}chloroform (6 : 4) Iodine vapour
Chloroform}ethyl acetate (4 : 1)
Corticosteroids Silica gel Ethyl acetate}isooctane (7 : 3)
Corticosteroid sulfates Silica gel Chloroform}methanol}NH4OH
(65 : 25 : 0.1)
Ethyl acetate}methanol}NH4OH
(75 : 25 : 2)
Progesterone metabolites Fisherbrand F254 Two-dimensional TLC;2 with Iodine vapour and
chloroform}ether (10 : 3) then UV absorption
;2 in hexane}ethyl acetate (5 : 2)
Androst-16-ene biosynthesis PE-SIL-G Silica gel Chloroform}acetone (9 : 1) and Iodine vapour and
hexane}ethyl acetate (5 : 3.5) UV absorption
Run ;2
Metabolites of 3H-progesterone, Fisher silica gel 60 F254 Two-dimensional TLC firstly to Autoradiography
-pregnenolone and -DHA de-fat in cyclohexane}ethyl acetate
(95 : 5);5}7. Then toluene}acetone
(8 : 2);2, finally ;2 in first direction
with cyclohexane}ethyl acetate (1 : 1)

Reprinted from Journal of Steriod Biochemistry Molecular Biology Copyright (1996) with permission from Elsevier Science.
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography 4237

used for this purpose are also summarized in Table 1. ally based on silica modiRed with silanes of various
For steroids containing UV-absorbing groups, such as chain lengths } C18 (ODS) silica is the most widely
the -4-3-oxo group in the A-ring of most active used material for this purpose. Column packings
steroid hormones and the aromatic ring in the oestro- based on synthetic material are now being made
gens, visualization can be achieved by examining the available and may in the future replace silica. Con-
plate under UV light. To improve the detection of siderable advances in the production and quality of
steroids absorbing at around 240 nm, most commer- these packing materials have been made over the past
cially available TLC plates have a Suorescent mater- 10 years and thus reproducibility of separations has
ial incorporated which improves detection of the improved.
absorption of UV light at around 254 nm. Other Most steroid separations today use reversed-phase
techniques are often destructive and require reagents systems with C18 or C8 silica, although an exception
to be sprayed on to the plate. Thus, to avoid destroy- to this general rule is the separation of metabolites of
ing the steroid of interest, it is necessary to have vitamin D which use normal-phase systems eluting
standards run on the same plate at the side so that the with hexane}isopropanol}methanol or hexane}meth-
position of the steroids of interest can be gauged. anol}chloroform. Binary hexane}isopropanol sys-
There are other nondestructive means of visualiz- tems give signiRcant tailing and resolution problems
ation, such as the use of iodine vapour but these are which are improved by the use of ternary solvent
not always satisfactory, particularly at low concen- mixtures. These normal-phase systems give excellent
trations. One advantage of TLC is in the separation of resolution of the metabolites of vitamins D2 and
radiolabelled steroids which can be visualized by D3 but do not separate the vitamins themselves
autoradiography. (Figure 4), which can be achieved using reversed-
phase ODS}silica eluting with methanol}water. If
High Performance Liquid Chromatography
steroids are to be recovered from the eluting solvent,
The application of HPLC to the separation of steroids it is clearly advantageous if normal-phase systems
has been extensively studied over the last 20 years. with sufRcient resolving power can be devel-
A detailed and comprehensive review of the HPLC of oped as the removal of aqueous solvents used in
steroid hormones was published in 1988 and updated
in 1995 (see Further Reading). HPLC is in essence no
different to the column and thin-layer systems dis-
cussed above, although the resolving power of mod-
ern HPLC columns is signiRcantly greater: some
reversed-phase columns achieving 60 000}80 000
theoretical plates per metre. New solvent systems for
normal-phase HPLC can be investigated using TLC.
Columns, because of the high resolving power which
can now be achieved, are usually quite short
(100}300 mm long with an internal diameter of
4.6 mm). Microbore columns ((2 mm in diameter)
can be used to reduce mobile phase consumption but
may present practical problems because of the limita-
tion in sample capacity. However, microbore col-
umns may have considerable application in LC-MS,
because of the low solvent Sow rates. Silica contains
free OH groups and these can be modiRed, replacing
the OH for example with cyanopropyl (CN, used
for the separation of corticosteroids) or amino-
propyl (NH2 , used for the separation of oestrogens).
Improved resolution of particular steroids can some-
times be achieved using these modiRed silicas and, for
example, silica-CN gives selective retention of ster-
oids containing oxo groups, and has been particularly
Figure 4 LC separation of vitamins D2 and D3, 24-OH-D2, 24-
useful in the separation of the 25-hydroxyvitamin
OH-D3, 25-OH-D2 and 25-OH-D3, using Zorbax-SIL and eluting
D3-23,26-lactone which is difRcult to resolve from with 2.5% isopropanol in hexane. (Reproduced with permission
24,25-dihydroxyvitamin D3 in conventional normal- from Jones G and DeLuca HF (1975) Journal of Lipid Research
phase HPLC systems. Reversed-phase HPLC is usu- 16: 448}453. Copyright 1975 FASEB.)
4238 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography

Table 2 Examples of some recent HPLC systems for the separation of steroid hormones and bile acids

Steroid group Carbon atoms Stationary phase Mobile phase Comments

Vatamin D C27 (D3 series) Zorbax ODS MeOH (MeCN)}H2O (acid) Separation of D3 and D2
Secosteroids C28 (D2 series) Zorbax CN Hx}IPA}MeOH Retards metabolites containing oxo groups
Zorbax SIL Hx}IPA}MeOH Usual system for metabolite resolution
Bile acids C24 YMC GEL C8 MeCN}H2O#cyclodextrin Bile acids and their conjugates as
bromopyrenacyl esters
Bile Pak II Gradient with Bile acid conjugates using post-column
MeCN}MeOH}PO4 buffer immobilized 3
OHSDH and fluorescence
detection
CorticosteroidsC21 Cosmosil C18 MeOH}H2O gradients Corticosterone and DOC
Novapak C18 MeOH}buffer Electrospray}MS of DHE and DHF
Nucleosil C8 MeCN}H2O#HCOOH Dexamethasone metabolites
Progesterone C21 Finepak SIL-C18 Tetrahydrofuran or Pregnane- and pregnene-diols
metabolites MeCN}imidazole buffer

Androgens C19 NovaPak C18 Gradient of MecN}H2O DHEA and 7-OH-DHA in newborn foals blood
Hypersil BDS-C8 Gradient of 7.5 mmol L\1 LC-MS of testo. and epi-testo.
NH4Ac}MeOH glucuronides/sulfates
Hibar
Lichrosorb-DIOL Hx}IPA Metabolites of DHT and androstanediol
Oestrogens C18 Wakosil C18 MeOH}H2O Plasma oestrogens}pre-column derivatization as
benzimidazoles
Beckman ODS MeCN}H2O#cyclodextrin Urinary oestrogens

reversed-phase separations causes considerable difR-
culties. Corticosteroids have been successfully separ-
ated using normal-phase systems based on DIOL-
silica sorbents (-Si-2,3-dihydroxypropoxypropyl) and
ion exchange HPLC has also been used for the separ-
ation of polar oestrogens. Some recent examples of
typical HPLC systems used for the separation of ster-
oids are given in Table 2, which illustrates the fact
that most methods in use today are reversed-phase
systems using ODS/C18 packings.
All steroids are susceptible to permanent absorp-
tion and/or chemical destruction or modiRcation by
unsilanized glass surfaces, exposed metal and by non-
inert supports and, for quantitative HPLC, great care
must be taken to remove or reduce such mate-
rials. Sorbents containing accessible hydroxyl groups
should not be used for the separation of 18-hy-
droxylated steroids as chemical modiRcation of these
steroids may occur during chromatography. Al-
though not relevant to HPLC, it should be pointed
out that partially end-capped ODS}silica can still be
used with advantage in particular situations for rapid
separations after SPE. As an example, the use of Bond-
Elut C18-OH allowed both extraction and subsequent
separation on the same cartridge in a method for the
assay of calcitriol (1,25-dihydroxyvitamin D3).
One particular advantage of HPLC is that it is not
destructive and can thus be used for the separation
and quantitation of intact steroid conjugates. These
polar molecules are not susceptible to analysis by GC
as the high temperature necessary to maintain ster-
oids in the vapour phase causes hydrolysis of the
conjugate. Examples of the application of HPLC to
the separation of androgen and oestrogen glucuron-
ides and the application of LC-MS(MS) to the separ-
ation of intact conjugates and steroid fatty acid esters
are given in the Further Reading section. An example
of such a separation is given in Figure 5. Until the
advent of HPLC, steroid conjugates had to be hydro-
lysed prior to resolution and important information
was thus lost. The use of LC, particularly when
coupled to MS(MS), has allowed resolution and
quantitation of intact conjugates together with struc-
tural information. HPLC systems for steroid conju-
gates are usually but not exclusively reversed-phase
primarily based on ODS-silica eluting with meth-
anol}, tetrahydrofuran}, acetonitrile}water or buffer
solvent systems. The conjugates can be detected in the
same way as nonconjugated steroids and this is dis-
cussed below.
Detection/Quantitation
This is the Rnal and perhaps most important step and
there are a variety of methods which can be used for
the detection or quantitation of steroids. These
methods are usually considered only in conjunction
with HPLC as they are usually insufRciently speciRc
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography 4239

Figure 5 Analysis of maternal urine for the detection of placental sulfatase deficiency (PSD). The microbore HPLC}ES}MS
chromatograms represent selected ions for detection of pregnanediol glucuronide, oestriol glucuronide and 16
-hydroxy-DHEA sulfate.
The selected-ion chromatograms on the left of the figure are from a normal individual and those on the right are from a patient with PSD.
Oestriol and its glucuronide cannot be synthesized by women with PSD, and the precursor 16
-hydroxy-DHEA sulfate is a dominant
steroid in urine. The amount injected into the microbore column was equivalent to 25
L of urine, the eluate being split 10 : 1 prior to the
mass spectrometer. The column used was a Reliasil number 9, 1;100 mm. Solvent A was 98% 10 mmol L\1 ammonium acetate in
H2O, 2% acetonitrile; solvent B was 100% acetonitrile. The gradient was 2% B to 30% over 20 min, and a flow rate of 40
L min\1 was
used. From Makin HLJ, Gower DB and Kirk DN (eds) Steroid Analysis (1995) Reproduced with permission from Kluwer Academic
Publishers.

to be used without rigorous prior separation of inter-
fering steroids } the exception to this being im-
munoassay which can, depending upon the antibody,
be sufRciently selective and sensitive for use directly
on plasma/serum without extraction or prior puriR-
cation. There are a number of such assays available
today but their uncritical use can lead to problems.
An example of this is an immunoassay for 17-
hydroxyprogesterone, developed for use with adults.
This was applied to the diagnosis of a genetic defect in
cortisol biosynthesis, congenital adrenal hyperplasia.
The possible interference in this assay by 17-hy-
droxypregnenolone sulfate which is normally produc-
ed in very young children was not appreciated.
A simple solvent extraction procedure overcame this
problem once it was detected.
It is often the case that one or other of the
simple puriRcation procedures, such as selective
solvent extraction, use of mini-columns or even TLC
on small plates can greatly enhance speciRcity and
there are many examples of this in the literature.
It is often unnecessary to use expensive HPLC sys-
tems to resolve these steroid analytical problems and
consideration of the physicochemical properties of
the steroid of interest may often suggest a simple
non-HPLC solution. Immunoassay is however one of
the most sensitive methods of steroid quantitation
and, coupled with HPLC, even with a relatively non-
speciRc antibody, can provide a system with both
high speciRcity and sensitivity. It does however
require collection of the eluent } so called ofSine
detection.
4240 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography

Figure 6 HPLC of steroids with fluorescent detection } some examples of pre-column derivatives.
 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography 4241

Figure 7 Use of photodiode array detection in HPLC separation of metabolites of 25-hydroxydihydrotachysterol3. Monitoring the
eluent at 251 nm indicates what appears to be a single homogeneous peak. Examination of the UV spectra at the leading edge, the
apex and the trailing edge clearly demonstrates the presence of a contaminant. From Makin HLJ, Gower DB and Kirk DN (eds) Steroid
Analysis (1995). With permission from Kluwer Academic Publishers.
4242 III / STEROIDS / Liquid Chromatography and Thin-Layer (Planar) Chromatography
Other detection methods can be used, such as UV
absorption with or without derivatization, Suorimetry
after chemical modiRcation or derivatization or elec-
trochemical detection. These detection/ quantitation
methods can usually be carried out online } that is to
say, that the HPLC eluent can be directly and con-
tinuously monitored. When chemical derivatization is
required, this can be carried out prior to HPLC (pre-
column derivatization) but this may reduce resolu-
tion. In such cases it is also possible to carry out the
derivatization after the HPLC separation (post-col-
umn derivatization). Some examples of pre-column
derivatization used for Suorescent detection are illus-
trated in Figure 6.
One particular method of UV monitoring of elu-
ents from HPLC separations is the photodiode array
detector. With this detector the absorption of the
eluent is continuously monitored over a range of
wavelengths and the data stored in a computer.
Reconstructed chromatograms can be obtained at
a later date, as can three-dimensional reconstructions
(showing time versus absorbance versus wavelength).
An example of photodiode array detection is illus-
trated in Figure 7, which demonstrates that what
appears to be a single peak when monitored at
251 nm is in fact composed of two unresolved com-
pounds and this can be demonstrated by comparing
the UV spectra obtained at the leading edge, the apex
and the trailing edge of the peak. This lack of resolu-
tion is also seen in the three-dimensional picture. This
particular separation was obtained when examining
the metabolites of a chemical analogue of 25-hy-
droxyvitamin D.
HPLC can also be linked to mass spectrometry and,
increasingly, techniques are becoming available for
the direct linking of the column to the mass spectrom-
eter. In the past, mass spectrometry of HPLC eluents
had, like immunoassay, required that eluents be col-
lected and prepared for mass spectrometry. The avail-
ability of ionization techniques (such as atmospheric
pressure ionization, electrospray, etc.) now allow the
HPLC eluent to go directly to the mass spectrometer.
Many steroids are susceptible to ionization in such
systems but others require derivatization to achieve
satisfactory ionization. In these systems the elution
solvents should contain water or salts, which implies
the use of reversed-phase separation. This is not
a particular problem with most steroids but for par-
ticular steroid groups (e.g. metabolites of vitamin D)
it may require the development of new solvent sys-
tems. Such ionization procedures are inevitably low
energy and fragmentation is limited. The advantage
of this low energy ionization is that intact steroid
conjugates can be examined. However, low energy
ionization is inefRcient, limits the sensitivity of detec-
tion and does not yield information about structure.
To some degree this structural limitation can be re-
solved by utilizing LC-MS-MS where the Rrst mass
spectrometer allows only the major ion obtained to
pass through to a collision cell. Here the ion is sub-
jected to higher energy ionization or bombardment,
producing further fragments which can then be ana-
lysed by a second mass spectrometer, yielding struc-
tural information. Such systems are expensive but are
immensely valuable. Unfortunately, LC-MS and LC-
MS-MS are not as sensitive as GC-MS, which has
been widely used for the measurement of many ster-
oids present at low concentrations in human body
Suids. High resolution GC-MS can of course increase
sensitivity of detection even further.
Conclusion
The advent of simpler and cheaper mass spectrom-
eters which allow direct coupling of the LC column
has meant that less attention needs to be paid to
resolution and the development of solvent systems
and column packings to achieve improved resolution
is no longer as important. Attention has therefore
shifted towards increasing sensitivity by the use of
microbore columns with the low Sow rates required
for maximum ionization in the MS and the use of
nanospray ESI. Today excellent mass spectra can be
obtained using such systems with femtomole concen-
trations of analyte. There is considerable scope for
further enhancement of sensitivity and selectivity by
improved MS design of both hardware and software.
Acknowledgements
Readers are encouraged to seek further information
from the texts quoted below which are fully refer-
enced and allow entry to the extensive research litera-
ture on this topic. In preparing this review,
I acknowledge the debt I owe to all the researchers in
this area whose results I have used but whose work
I have not been able to acknowledge directly.
See also : II/Chromatography: Liquid: Mechanisms:
Normal Phase; Mechanisms: Reversed Phases. III/Ste-
rols: Thin-Layer (Planar) Chromatography.
Further Reading
Fieser LF and Fieser M (1959) Steroids. New York: Van
Nostrand Reinhold.
Heftmann E (ed.) (1983) Chromatography: Fundamentals
and Applications of Chromatographic and Electro-
phoretic Methods. Part B: Applications. Amsterdam:
Elsevier.
 III / STEROIDS / Supercritical Fluid Chromatography
Hill RA, Kirk DN, Makin HLJ and Murphy GM (eds)
Dictionary of Steroids. London: Chapman & Hall.
Kautsky MP (ed.) (1981) Steroid Analysis by HPLC: Re-
cent Applications. New York: Marcel Dekker.
Makin HLJ (ed.) (1984) Biochemistry of Steroid Hormones,
2nd edn. Oxford: Blackwell ScientiRc Publications.
Supercritical Fluid Chromatography
K. Yaku, K. Aoe, N. Nishimura and
T. Sato, Tanabe Seiyaku, Osaka, Japan
F. Morishita, Kyoto University, Kyoto, Japan
Copyright ^ 2000 Academic Press
Supercritical Suid chromatography (SFC) has been
recognized as a powerful separation technique com-
plementing gas chromatography (GC) and high perfor-
mance liquid chromatography (HPLC). Klesper et al.
published results of the use of supercritical dichloro-
diSuoromethane and monochlorodiSuoromethane to
separate involatile nickel porphyrin in 1961. The
development of the technique was limited by instru-
mental and experimental difRculties due to the high
temperatures and pressures required to maintain the
mobile phase in a supercritical state. Novotny and
Lee, however, developed SFC with a capillary column
(cSFC) in 1981, which led to signiRcant advances in
the technique. In 1982 Gere et al. developed an in-
strument for packed-column SFC (pSFC) based on
modiRcation of an HPLC. They demonstrated, using
polycyclic aromatic hydrocarbons as probe mole-
cules, that the resolution per unit time in pSFC
was 5}10 times better than in HPLC with the same
columns, due to more favourable diffusivity in super-
critical Suids.
The advantages of SFC have been described else-
where in this Encyclopedia. Pure carbon dioxide
Suid is a solvent of inadequate polarity. For the analy-
sis of polar compounds by SFC, alcohol is generally
added to a mobile-phase Suid as a modiRer. Small
amounts of polar modiRers signiRcantly increase the
solvent strength of the mobile phase and make it
possible to elute polar compounds. In particular,
pSFC has been applied to various kinds of polar
compounds such as drugs, and shown to be superior
to HPLC with respect to analysis time, efRciency and
selectivity.
The nonpolar steroids cholesterol and ketosteroids
are easily eluted by either cSFC or pSFC with pure
carbon dioxide. Synthetic corticosteroids, which are
widely used therapeutically for the suppression of
4243
Makin HLJ and Newton R (eds) (1988) High Performance
Liquid Chromatography in Endocrinology. Berlin:
Springer Verlag.
Makin HLJ, Gower DB and Kirk DN (eds) (1995) Steroid
Analysis. London: Blackie.
adrenocortical functions, inSammatory and allergic
diseases, have multiple hydroxyl functional groups in
the structures. In order to modify the efRcacy and
suppress adverse reactions, many corticosteroids have
been synthesized. Thin-layer, normal and reversed-
phase chromatography have been used for the analy-
sis of these compounds. For a number of the synthetic
corticosteroids used in therapy, very little work has
been carried out by pSFC. These polar steroids are
probably difRcult to elute with pure carbon dioxide
due to its poor solvent strength.
In this article, the pSFC retention behaviour of
synthetic corticosteroids, possessing one to four
hydroxyl groups, are focused on. The effect of several
parameters (stationary phase, modiRers, pressure and
temperature) on retention and efRciency are con-
sidered. The chemical structures of corticosteroids
are shown in Figure 1. They were chromato-
graphed using a pSFC instrument modiRed from
a commercial HPLC system. The addition of meth-
anol to carbon dioxide and adoption of an amino-
propyl stationary phase produced both good
resolution and symmetric peak shapes. Both plate
number and resolution indicated that the maxima
were around the critical temperature (40}503C) of
the binary Suid. The selectivity and separation of the
analytes in pSFC are superior to those in existing
normal and reversed-phase HPLC. Seven polar cor-
ticosteroids, possessing one to four hydroxyl groups,
showed baseline separation within 6.5 min with
a modiRer gradient method.
Instrumentation of pSFC
Most studies have been done using commercial pSFC
instruments. However, a pSFC with the same perfor-
mance as a commercial instrument can easily be con-
structed. The HPLC for pSFC operation requires
some simple adaptations to allow use of supercritical
carbon dioxide as a mobile phase. The pSFC system
constructed by the authors by modifying a Shimadzu
HPLC is shown in Figure 2.
4244 III / STEROIDS / Supercritical Fluid Chromatography
Figure 1 Structures and symbols of synthetic corticosteroids.
Effect of Analytical Parameters
Stationary Phase
Peak shapes of polar solutes on a packed column are
often poor when pure carbon dioxide is used as the
mobile phase. The separation of steroids has been
performed using columns with phenyl, nitrophenyl,
diol, aminopropyl, octadecyl and cyanopropyl-modi-
Red silica and pure silica as packing materials. It is
likely that only polar modiRers used with polar sta-
tionary phases produce both good resolution and
symmetrical peaks. As shown in Figure 3, an amino-
propyl column exhibited the best selectivity and peak
shape with a reasonable retention time in comparison
with the others. Octadecyl and phenyl columns
showed poor separation: the former did not retain
any solutes and the latter did not separate under the
operating conditions used. On the silica support, the
solutes showed appropriate retention but poor separ-
ation and peak shape. Although the retention times of
triamcinolone acetonide, Suocinolone acetonide,
hydrocortisone and betamethasone were almost the
same as those on the aminopropyl column, the separ-
ation factor,
, of the two pairs } steroids possessing
two hydroxyl groups, and steroids possessing three
hydroxyl groups } decreased remarkably on silica.
A reversed elution order, however, was observed on
the silica, which showed that it is possible to change
selectivity by selecting the stationary phase.
Modi\er
In pSFC, the addition of a modiRer to a mobile phase
should be considered from the viewpoint of its effect
on either the stationary phase or on the mobile phase.
Berger et al. have studied the effect of column and
mobile-phase polarity using steroids. They concluded
that polar modiRers tended to decrease the intensity
of the solute}silanol interaction, and more polar sta-
tionary phases produced greater retention, requiring
the use of modiRers to obtain reasonable retention
times. Blilie and Greibrokk indicated that the modi-
Rers functioned as deactivation agents by direct inter-
actions with residual silanol groups, and also as
modiRers of the eluting power of the mobile phase.
 III / STEROIDS / Supercritical Fluid Chromatography 4245

Figure 2 Schematic diagram of packed-column supercritical fluid chromatography. 1, Carbon dioxide cylinder; 2, modifier; 3, cooling
bath; 4, LC-6A pump; 5, LC-9A pump; 6, dynamic mixer; 7, injector; 8, oven; 9, packed column; 10 and 11, pressure monitor; 12,
detector; 13, back-pressure regulator; 14, dry thermo unit.

No corticosteroids were eluted from the column
packed with Cosmosil 5NH2 with pure carbon diox-
ide as the mobile phase and a modiRer had to be
added. The effect of modiRers with different polar-
ities on the retention of corticosteroids is shown in
Figure 4. The addition of 11.8% (v/v) methanol to
carbon dioxide gives the best resolution and symmet-
rical peak shapes within 14 min. In comparison, the
addition of the same amount of 99.5% ethanol, and
of 95% ethanol reduced resolution but remarkably
improved the peak shape of the most polar com-
pound, triamcinolone, in the corticosteroids. This
should be attributed to deactivation of the active sites
on the silica support by the water in the 95% ethanol.
Janssen et al. conRrmed that the effect of a few per
cent of modiRer in pSFC is largely due to deactivation
of residual silanol groups on the silica support, and
the accessibility to the active sites depends strongly on
the size and structure of the modiRer molecules. Ac-
cording to Janssen et al., the same volume percentage
of tetrahydrofuran (THF) and methanol was needed
to cover 95% of the surface, but since no corticos-
teroid was eluted under these conditions when meth-
anol was replaced with THF, the effect of the
modiRer on retention of corticosteroids consists in
the enhancement of the solvent strength of the mobile
phase rather than deactivation of the active sites on
the silica support.
The capacity factor of every corticosteroid de-
creased two- to fourfold with a 1.8-fold increase in
methanol concentrations over the range 9.1}16.7%
v/v. All solutes were eluted within 5 min using carbon

Figure 3 Effect of column on retention of corticosteroids. (A) Cosmosil NH2; (B) Ultron VX-SIL; (C) Zorbax phenyl. Operating
conditions: mean pressure 213 kg cm\2, flow rate of CO2 3 mL min\1, flow rate of methanol 0.4 mL min\1, temperature 403C. Peaks:
as in Figure 1.
4246 III / STEROIDS / Supercritical Fluid Chromatography

Pressure
In a study of seven corticosteroids, the capacity factor
of each solute decreased by a factor of two with an
increase in the range of 107}223 kg cm\2. A few
researchers have measured the densities of modiRed
supercritical Suids experimentally. Berger measured
the density of binary Suids using a U tube densito-
meter and drew the constant density lines in plots of
the pressure against the composition for methanol}
carbon dioxide system at three temperatures. The
densities of CO2}methanol (12%, v/v) at different
pressures can be calculated by extrapolating the lines
in the pressure range 105}180 bar. The plots of ln k
against the binary Suid density revealed that there is
a linear relationship between them in SFC, as ex-
pected.
Except for Suocinonide, theoretical plate numbers
(N) reached maximum values at 126 and 144 kg cm\2,
as shown in Figure 5. The maximum N values were c.
4700}9800. Corresponding to the behaviour of the
N values, the resolutions between the adjacent solutes
also showed a maximum at 126}162 kg cm\2. Since
the mass Sow rate was kept constant, the linear velo-
city varied with pressure. The minimum plate height
was obtained in this pressure range. These results re-
veal that pressure is one of the signiRcant parameters
for optimizing the operating conditions.

Temperature
The retention of corticosteroids increases with an
increase in the temperature (decrease in the density).
The N values of each solute also increase with tem-
perature as shown in Figure 6, and reach maximum
values at 39 or 493C, with the exception of hydrocor-
tisone. The maximum N value for triamcinolone is c.
8400 at 393C but only c. 3200 at 583C, correspond-
ing to about a 60% decrease. Although little variation
in the separation factor (
) of any pair of neighbour-
ing solutes was observed over the wide range of
temperature measured, resolution reached maximum
Figure 4 Effect of modifiers on retention of corticosteroids. values at 39}493C, corresponding to the behaviour of
(A) Methanol; (B) ethanol (95%); (C) ethanol (99.5%); (D) 1-
the N values. The critical temperature and pressure
propanol; (E) 2-propanol. Operating conditions: column Cosmosil
5NH2, inlet pressure 224 kg cm\2, outlet pressure 191 kg cm\2,
flow rate of CO2 3 mL min\1, flow rate of modifier 0.4 mL min\1, Table 1 Repeatability (RSD%, n"6)
temperature 403C. Peaks: as in Figure 1.
Corticosteroids tR (min) k Peak area
dioxide modiRed with 16.7% (v/v), and the resolu- Fluocinonide 0.37 1.15 1.01
tions among them were more than 1.6. Dexamethasone acetate 0.35 1.10 0.75
Calculated relative standard deviations (RSD) of Triamcinolone acetonide 0.49 1.39 1.08
0.35}0.70% for tR, 0.82}1.47% for k and 0.50} Fluocinolone acetonide 0.60 1.47 1.34
Hydrocortisone 0.64 1.39 0.67
1.34% for peak area are shown in Table 1, indicate
Betamethasone 0.70 1.39 1.09
that the pSFC modiRed from a commercial HPLC Triamcinolone 0.45 0.82 0.50
system has a good performance and is useful for
a routine analysis. Operating conditions as in Figure 3.
 III / STEROIDS / Supercritical Fluid Chromatography
Figure 5 Relationship between theoretical plate numbers and
pressure. Operating conditions: mean pressure 107, 126, 144,
162, 184, 205 and 223 kg cm\2; other conditions as in Figure 4.
Symbols: squares, hydrocortisone; diamonds, fluocinolone
acetonide; circles, betamethasone; triangles, triamcinolone.
were reported to be 36.853C and 80 bar and 503C
and 95 bar for 2% methanol and 16% methanol
in carbon dioxide, respectively. So, the critical
temperature for 12% methanol in carbon dioxide,
which we used as the mobile phase, can be assumed
to be in the range of 40}503C: the maximum
Figure 6 Relationship between theoretical plate numbers and
temperature. Operating conditions: temperature 22, 29, 39, 49
and 583C, mean pressure 213 kg cm\2, other conditions as in
Figure 4. Symbols: filled squares, fluocinonide; filled circles,
dexamethasone acetate; filled triangles, triamcinolone acetonide;
diamonds, fluocinolone acetonide; open squares, hydrocortisone;
open circles, betamethasone; open triangles, triamcinolone.
4247
N and resolution were obtained around the critical
temperature.
Separation with Modi\er Gradient
A wide range of polar corticosteroids has been separ-
ated in a modiRer gradient elution mode. As shown in
Figure 7, all solutes were eluted within 6.5 min by
increasing the methanol content from 11.8% (v/v) to
17.0% (v/v) at 0.52% (v/v) per min, and keeping the
CO2 Sow rate constant. Good peak shapes, com-
pletely baseline separated, were observed. The stable
baseline without drift and noise is considered to be
due to the good mixing process of the binary Suid. In
pSFC, a modiRer gradient is one of the most effective
techniques for the analysis of polar steroids.
Comparison with HPLC
The retention of a wide range of corticosteroids by
pSFC using an aminopropyl silica column has been
compared with that in normal and reversed-phase
Figure 7 Gradient elution of corticosteroids. Operating condi-
tions: column Cosmosil 5NH2, flow rate of CO2 3 mL min\1, meth-
anol gradient 11.8}17.0% (v/v) at 0.52% (v/v) per min, mean
pressure 206 kg cm\2, temperature 403C. Peaks: as in Figure 1.
4248 III / STEROIDS / Supercritical Fluid Chromatography
Figure 8 Chromatograms of corticosteroids. (A) Packed-col-
umn SFC; operating conditions as in Figure 3; (B) reversed-
phase-HPLC (40% acetonitrile), (C) reversed-phase-HPLC
(55% methanol); (D) normal-phase-HPLC. Peaks: as in Figure 1.
HPLC. The observed chromatograms are shown in
Figure 8. Since the most polar corticosteroid among
the analytes, triamcinolone, was not dissolved in the
mobile phase, it could not be eluted in the normal-phase
mode. The elution order of corticosteroids in pSFC is
the same as that in normal-phase HPLC and is mainly
determined by the number of hydroxyl groups present
in the compound: Rrstly, Suocinonide with a single OH
group; dexamethasone acetate, triamcinolone acetonide
and Suocinolone acetonide with two OH groups;
then hydrocortisone and betamethasone with three
OH groups; and Rnally triamcinolone, with four OH
groups. Corticosteroids are eluted almost in reversed
order in reversed-phase HPLC, but it is noteworthy
that the pairs of triamcinolone acetonide and
Suocinolone acetonide, and hydrocortisone and be-
tamethasone are eluted in the same order as in pSFC.
The elution order of these compounds with the
same number of OH groups seems to be closely re-
lated to their dipole moment. The estimated dipole
moments were 1.19 and 2.04 debye for triamcinolone
acetonide and Suocinolone acetonide, and 0.52 and
2.24 debye for hydrocortisone and betamethasone.
A similar range of N values was obtained in each
chromotographic mode, i.e. c. 3600}8000 in pSFC,
c. 4800}8700 in normal-phase HPLC and c.
2300}11 000 in reversed-phase HPLC. The separ-
ation of triamcinolone acetonide and Suocinolone
acetonide, which show the lowest resolution and the
same elution order in the normal- and reversed-phase
systems, are comparable. The resolution of these sol-
utes is 2.73 by pSFC, 2.04 by normal-phase HPLC,
0.53 by reversed-phase HPLC with a methanol mix-
ture, mobile phase, and 2.20 by reversed-phase HPLC
with an acetonitrile mixture, mobile phase. The elu-
tion time of all solutes in pSFC is about four times
faster than in normal-phase HPLC and about 1.5
times faster than in reversed-phase HPLC.
These results suggest that the pSFC conditions used
give a higher selectivity and better separation efRcien-
cy than normal-phase and reversed-phase HPLC for
the analysis of corticosteroids which possess one to
four hydroxyl groups.
Conclusion
PSFC is useful for the analysis of polar steroids, and
its application as a rapid method for quality control
and routine analysis can be expected in the future.
Acknowledgements
This article was adapted from Yaku K, Aoe K,
Nishimura N, Sato T and Morishita F (1997) Reten-
tion behavior of synthetic corticosteroids in packed-
column supercritical Suid chromatography. Journal
of Chromatography A 773: 277}284. Copyright
1997, with kind permission from Elsevier Science.
Further Reading
Anton K and Berger C (eds) (1998) Supercritical Fluid
Chromatography with Packed Columns } Techniques
 III / STEROLS / Supercritical Fluid Chromatography
and Applications. Chromatographic Science Series,
vol. 75. New York: Marcel Dekker.
Charpentier BA and Sevenants MR (eds) (1988) Supercriti-
cal Fluid Extraction and Chromatography } Techniques
and Applications. ACS Symposium Series 366. Washing-
ton, DC: American Chemical Society.
Chester TL, Pinkston JD and Raynie DE (1994) Supercritical
Suid chromatography and extraction. Analytical Chem-
istry 66: 106R.
STEROLS
Supercritical Fluid
Chromatography
F. J. Sen oraf ns, Universidad AutoH noma de Madrid,
Spain
K. E. Markides, Uppsala University, Uppsala, Sweden
Copyright ^ 2000 Academic Press
Introduction
In supercritical Suid chromatography (SFC) the mo-
bile phase is a Suid subjected to pressures and temper-
atures near or above its critical point. This fact
determines the mobile phase properties (e.g. diffus-
ivity, density, viscosity, etc.) that are intermediate be-
tween those of gases and liquids and can be varied and
controlled by small changes in the pressure or temper-
ature. The most common Suid used in SFC is carbon
dioxide, which has a critical temperature of 313C,
allowing the separation of thermally labile compounds
under mild conditions. SFC can be carried out with
open tubular and packed columns, with differences in
selectivity, detection and need of modiRer addition to
the carbon dioxide. Both types have been employed in
the separation of sterols in a wide variety of samples.
Supercritical carbon dioxide has an adequate solvat-
ing power for sterol separation with both column
types without the need of modiRer addition. It is
therefore possible to separate sterols in complex sam-
ples at lower temperatures than gas chromatography
and in shorter times than liquid chromatography.
Importance of the Analysis of Sterols
The analysis of sterols is of great signiRcance from the
health point of view and for the quality control of
numerous food products.
With respect to quality control of food and nutri-
tion studies, the sterols of vegetable origin (phyto-
4249
Gere DR (1983) Supercritical Suid chromatography.
Science 222: 253.
Smith RM (ed.) (1988) Supercritical Fluid Chromato-
graphy. RSC Chromatography Monographs Series.
London: Royal Society of Chemistry.
Smith RM and Hawthorne SB (eds) (1997) Supercriti-
cal Suids in chromatography and extraction (complete
in one issue). Journal of Chromatography A 785.
sterols) are found in complex mixtures in numerous
plants, with the mixture containing some major
sterols and a great variety of minor compounds
(Figure 1). Thus, the sterol proRle can be indicative of
the origin, or species or variety of food from vascular
plants, as well as from fungi or marine organisms.
Additionally, these compounds are fundamental in
the study of several metabolic pathways.
In the animal world, the variety of sterols is not so
broad, and the main constituents are cholesterol and
derivative esters. For that reason, supercritical carbon
dioxide has been employed for the selective extrac-
tion of cholesterol from meat products and edible
animal fats, to obtain healthier food for human intake.
On the other hand, some sterol oxides (oxysterols)
are known for their toxicity, mutagenicity or carcino-
genic properties, a fact that makes the determination
of their concentrations in natural matrices very neces-
sary, especially in studies of food quality and
physiology.
Characteristics of the Separation
of Sterols Using Supercritical Fluids
The main properties of supercritical Suid chromato-
graphy which affect sterol separation are related to
the high solvating power of supercritical Suids and
a low viscosity, which yields a high resolving power
and rapid throughput. In addition to its other advan-
tages, the ability of SFC to resolve complex mixtures
of low volatility compounds allows the direct injec-
tion of samples that contain sterols with no or little
pretreatment.
Some sterols can be degraded or lost during expo-
sure to light, heat or extreme values of pH. In the SFC
of sterols, all these factors can be avoided, providing
a separation under mild conditions that preserves the
integrity of the sample.
Finally, the relatively good solubility of compounds
with intermediate polarity and volatility such as
the sterols has also frequently been utilized in
4250 III / STEROLS / Supercritical Fluid Chromatography
Figure 1 Structures of the main sterols analysed by SFC.
supercritical Suid extraction (SFE) for sample frac-
tionation of the sterols from complex matrices, cre-
ating new possibilities for the use of supercritical
Suids in multidimensional systems. The solubilities
and chromatographic data of the main sterols in
supercritical carbon dioxide are well known, and can
be found in various books and review articles (see
Further Reading).
Isolation of Sterol Subclasses
and Sample Preparation
The determination of the sterol fraction in complex
matrices such as food provides valuable information
on the quality of the product, as well as its purity,
origin and plant variety. This analysis presents some
difRculties owing to the complexity of the matrix and
the relatively low concentration of the sterols in these
samples. The most widely used method includes a sol-
vent extraction of the lipid material and isolation of
the sterolic fraction, usually after removing the trig-
lycerides through saponiRcation of the lipids and sub-
sequent extraction of the unsaponiRable matter with
an organic solvent.
This unsaponiRable fraction contains the sterols
and other minor components, such as tocopherols,
terpenic alcohols, and hydrocarbons, and therefore
often needs to be fractionated. This is conventionally
performed by thin-layer chromatography or by solid-
phase extraction, prior to high resolution chromato-
graphic analysis.
When the objective is to separate many individual
sterols, there is an additional fractionation of the
unsaponiRable components after the extraction and
sample pretreatment, to isolate the main sterol
subclasses (i.e. 4,4-dimethylsterols, 4-methylsterols
and 4,4-desmethylsterols, and their esters, oxides and
other derivatives), and before the separation of the
individual sterols of the subclass of interest. This last
fractionation is usually carried out by open column
liquid chromatography, thin-layer chromatography,
high performance liquid chromatography or more
recently supercritical Suid chromatography, after
which the sterols can be injected in a gas chromato-
graph, with or without derivatization.
This whole procedure including saponiRcation, ex-
traction and fractionation is not only time consuming
and error prone, and may degrade labile sterols cre-
ating artefacts. Consequently, new approaches have
been developed in the last few years that avoid several
or all these steps by using multidimensional systems
or even direct injection into a SFC column.
 III / STEROLS / Supercritical Fluid Chromatography
SFC in Multidimensional Systems
Multidimensional systems take advantages of two
different chromatographic separations with com-
plementary characteristics, e.g. one extraction stage
and one chromatographic stage. The Rrst step is
aimed at producing a clean and undiluted sample
containing the fraction of interest, and the second
provides a high resolution separation of the target
analytes.
It is often the case in chromatography that the
largest source of error in the quantitative analysis of
sterols and the most time-consuming part of the ana-
lytical method is the sample preparation and extrac-
tion stage. The main advantages of online systems is
that they provide a fast and easily automated sample
preparation procedure which reduces or avoids many
of the errors of manual extraction. Also, less solvent
consumption gives reduced exposure, toxic hazards
and lower disposal cost.
The online methods applied to sterol analysis have
conventionally consisted in the direct coupling of
normal or reversed phase liquid chromatography
with capillary gas chromatography (LC}GC), which
allows the isolation of the sterolic fraction by LC,
followed by online transfer to the gaseous separation
in a fast and effective way. An alternative approach is
to use packed columns to isolate the sterolic fraction
by SFC, as shown recently by Medvedovici
et al. in 1997, who used a conventional packed col-
umn (20 cm;4.6 mm) with aminopropylsilica gel as
stationary phase, for further analysis by capillary gas
chromatography}mass spectrometry. This approach
can replace the classical preparation methods, yield-
ing a much shorter time with good reproducibility,
and it can be automated.
A more interesting approach is the online coupling
of SFE and SFC, that allows the transfer of the solutes
from the solid matrix to the chromatograph, reducing
solvent usage and the need for sample clean-up. This
technique has been applied to the separation in open
tubular columns of cholesterol in food, and may
prove to be of great importance in the future, with the
use of packed column SFC; such columns have a large
sample capacity and shorter analysis time.
The use of SFC in multidimensional chromato-
graphic systems has a number of advantages. The
most common multidimensional system to date,
LC}GC, is limited to the determination of thermally
stable and volatile solutes, but SFC can replace either
the Rrst step (fractionation), or the second step (high
resolution chromatography), or even both. In the case
of SFE}SFC, the transfer is performed without cha-
nges in the mobile phase providing less risk of loss of
analytes.
4251
Direct Introduction in SFC
While the analysis of sterol esters does not present
difRculties by gas chromatography, a sample prepara-
tion step is needed however to remove the fatty ma-
terial, in order to minimize interferences and protect
the GC system. In many cases, the advantage of SFC
is that the untreated sample can be injected directly
onto a packed column allowing estimation of several
lipid fractions at the same time.
Another approach consists in the use of SFC either
for fractionation of the oil or for direct selective
analysis of the sterol composition of the sample, with-
out previous treatment. This direct injection is a par-
ticularly promising technique, and has been applied
to the analysis of oils from marine sources to obtain
Rngerprints of different oil compositions, taking ad-
vantage of the very simple sample preparation re-
quirements of SFC.
Separation of Individual Sterols
For the determination of individual sterols, SFC pro-
vides the same resolution as gas chromatography and
short run times, at temperatures as low as 50}803C in
packed column SFC.
A particular problem is the detection system.
Simultaneous detection of many sterols is difRcult
with ultraviolet detection, as some of the sterols do
not possess high absorptivity. Hence this detection
mode must often be combined with others to provide
a comprehensive detection capability. For this reason,
the most usual detection systems for sterol determina-
tion are Same ionization detector (FID) or mass spec-
trometry, which can easily be used with either gas
chromatography or SFC.
Separations of sterols have been performed by open
tubular column SFC, in samples such as soybean oil
derivatives or commercial antioxidant mixtures. Other
compounds of interest such as tocopherols, squalene, or
even di- and triglycerides can be determined simul-
taneously (see Table 1). (Note that Table 1 is not
intended to be a comprehensive review, but aims to
provide general information on selected applications.)
Although the most common method for the SFC of
free sterols is to employ open tubular columns, separ-
ations of sterol-related compounds have also been
achieved with columns having different packing ma-
terials, ranging from pure silica, to phenyl-, diol-,
amino-, or octadecyl-modiRed silica. For packed
columns, the peak symmetry is improved and the re-
tention times of the sterols are shortened when a modi-
Rer is added to the mobile phase. This moderates the
inSuence of the free silanol groups of the silica and is
analogous to the end-capping of an HPLC silica-based
4252 III / STEROLS / Supercritical Fluid Chromatography

Table 1 Determination of individual sterols by supercritical fluid chromatography

Column type Sample Identified compounds Detector Reference

Open tubular Hamster faeces Free sterols FID Pinkston et al. (1991) Journal of
10 m;50
m i.d. Sterol esters High Resolution Chromatogra-
SB methyl 100 phy 14: 401}406

Open tubular Soybean oil Free sterols FID Synder et al. (1993) Journal
10 m;100
m i.d. deodorizer distillate MS/EI of the American Oil Chemists
SB-octyl-50 Society 70: 349}354

Open tubular Soybean oil Four sterols, FID Galuba and Gogolewski (1997)
10 m;50
m i.d. condensate Three tocopherols Chemia Analityczna
SB phenyl 5, 0.25
m 42: 245}248
film

EI, electron impact; MS, mass spectrometer.

stationary phase. In general, the separation of the
sterols in SFC is affected by parameters such as
the number, location, nature and conformation of the
functional groups present in the molecules.
One special case is the determination of choles-
terol, which is very often performed on samples with
few or no other sterols and where the main objective
is to separate this analyte from other non-sterolic
lipids and minimize the sample preparation, as will be
discussed below.
Determination of Cholesterol
and Cholesteryl Esters
In the last few years, the relationship between plasma
cholesterol levels and the risk of atherosclerosis and

Table 2 Determination of cholesterol by supercritical fluid chromatography

Column type Sample Studied analytes Detector Reference

Packed Human serum Cholesterol UV and FID Nomura et al. (1993) Analytical
column Cholesteryl esters Chemistry 65: 1994}1997
250 mm;4.6 mm
i.d. particle size
5
m
Kaseisorb
ODS-300-5
Open tubular Human serum Cholesterol FID Kim et al. (1994) Journal of Chromato-
10 m;50
m i.d. Cholesteryl esters graphy B 655: 1}8
SB-octyl-50,
0.25
m film
Open tubular Milk fat Cholesterol FID Huber et al. (1995) Journal of
20 m;50
m i.d. Chromatography A 715: 333}336
SB phenyl 5
Open tubular Fish oils Cholesterol FID Staby et al. (1994) Journal of the Ameri-
20 m;100
m i.d. Other lipids can Oil Chemists Society 71: 355}359
DB-5
Open tubular Marine oils Cholesterol FID Staby et al. (1994) Chromatographia
20 m;100
m i.d. Cholesteryl esters 39: 697}705
DB-5, 0.4
m film
Open tubular Fish and shark oils Cholesterol FID Borch-Jensen and Mollerup (1996)
25 m;100
m i.d. Cholesteryl esters Chromatographia 42: 252}258
DB-5, 0.1
m film Other lipids
Open tubular Shark liver oils Cholesterol FID Borch-Jensen et al. (1997) Journal of the
20 m;100
m i.d. Cholesteryl esters American Oil Chemists Society 74:
DB-5, 0.1
m film Other lipids 497}503
 III / STEROLS / Supercritical Fluid Chromatography 4253

coronary heart disease has been conRrmed, resulting
in an increase in concern about dietary and blood
cholesterol levels. As a consequence, the determina-
tion of serum cholesterol is one of the most frequent
clinical diagnostic measurements currently under-
taken. It is usually performed after hydrolysis, quan-
tifying the total cholesterol by routine enzymatic or
colorimetric methods. In many cases, it would be
more useful to separately determine free sterols and
cholesteryl esters in serum to provide more complete
clinical information. This analysis can be carried out
by SFC at low temperature, without saponiRcation or
derivatization (see Table 2). (Again, Table 2 is not
a comprehensive review but aims to provide general
information on selected applications.) The deter-
mination of individual cholesteryl esters cannot be
performed by the enzymatic methods available, while
gas chromatography requires high temperature to
elute the high-molecular-mass unsaturated esters,
causing thermal decomposition. Moreover, the detec-
tion of cholesterol and related compounds is not very
sensitive in HPLC with ultraviolet detection, since
free sterols generally have low absorption coefR-
cients. These problems are avoided with SFC in com-
bination with the highly sensitive FID or with mass
spectrometry.
Cholesterol Analysis in Human Serum
The analysis of cholesterol in human serum has been
performed with both open tubular and packed
column SFC, at temperatures of 65 and 453C, respec-
tively. With open tubular columns, it is possible to
determine quantitatively free, total and individual
esteriRed cholesterol with a simple liquid}liquid ex-
traction without derivatization. The use of FID
gives detection limits of 4}6 pg. The quantitative re-
sults for the analysis of total cholesterol in reference
sera and real samples show good agreement between
the SFC, GC (with derivatization), and enzymatic
methods.
With columns packed with inert octadecylsilica,
there is no need to add any modiRer to the carbon
dioxide, allowing the simultaneous use of ultraviolet
and FID. In addition, the use of carbon dioxide as the
supercritical Suid allows ultraviolet detection at
wavelengths as low as 190 nm, which are usually
below the practical limit with HPLC because of
the general absorption of most solvents at this

Figure 2 Chromatograms of cholesterol and cholesteryl esters from human serum reference material with ultraviolet (UV)
(wavelength 190 nm) and flame ionization detection (FID). Peak identification: (1) cholesterol, (2) cholesteryl laurate (internal
standard), (3) cholesteryl myristate, (4) cholesteryl palmitoleate, (5) cholesteryl linolenate, (6) cholesteryl palmitate#cholesteryl
linoleate#cholesteryl arachidonate, (7) cholesteryl oleate, (8) cholesteryl stearate. Chromatographic conditions: reversed-phase
HPLC column (250 mm;4.6 mm i.d.; particle size, 5
m), column temperature 453C, CO2 pressure 200 atm, CO2 flow rate
750 mL min\1 at room temperature under 760 mmHg. (Reproduced from Nomura et al. (1993) Analytical Chemistry 65: 1994}1997,
with kind permission from the authors and the publisher. Copyright 1993 American Chemical Society.)
4254 III / STEROLS / Supercritical Fluid Chromatography
wavelength. Use of lower wavelengths in ultraviolet
detection results in a greater sensitivity than higher
wavelengths for all cholesterol and cholesteryl esters.
Both ultraviolet and FID showed good agreement in
a study of cholesterol and their esters, with a detec-
tion limit of 20 ng for cholesterol and cholesteryl
palmitate, though better reproducibility was obtained
with UV detection (Figure 2).
Supercritical Suid chromatography can be a useful
tool in studies on cholesterol and cholesteryl ester
metabolism in serum and biological Suids in general,
though some improvements in the selectivity are still
needed to properly resolve some difRcult pairs of
steryl esters.
Cholesterol Analysis in Edible Oils
There are numerous applications of SFC with capillary
columns for the determination of components of oils
of marine animals especially from Mollerups group at
the Technical University of Denmark (see Table 2). In
these samples, numerous lipids have been analysed
including cholesterol and its esters, together with vit-
amin E, squalene, and di- and triglycerides. Open
tubular column SFC is therefore very convenient for
these complex samples owing to its high resolution and
relatively low analysis temperature (150}1703C versus
250}3003C for GC). For example, the analysis of
cholesterol and cholesteryl esters in seafood, shark
liver, seal and other Rsh oils has been accomplished
with open tubular column SFC both with direct injec-
tion of the diluted oil and with previous saponiRca-
tion and fractionation by thin-layer chromatography
(see Table 2). The latter puriRcation avoids the over-
loading of the open tubular columns (100
m internal
diameter) by the squalene that is present in high
concentrations in these samples. SaponiRcation fol-
lowed by fractionation is, however, time consuming
and the recovery of minor components can be difR-
cult. With direct injection in SFC, simultaneous deter-
mination of cholesterols, triglycerides and other lipids
has been achieved, with enhanced separation power
compared with HPLC, while gas chromatography
always requires sample pretreatment of these oils to
ensure chemical stability at high temperatures.
In other solid and more heterogeneous foods,
sample preparation is the most time-consuming step
in the routine determination of, for instance, choles-
terol levels in daily diet and remains the largest source
of error in the quantitative analysis of sterols.
Future Trends
Current developments in new types of columns,
equipment and detectors for SFC show that this tech-
nique is still developing and expanding and will
achieve greater use in the future, particularly with the
advent of new chromatographs for packed capillary
columns and with automation of modiRer addition,
which will be very valuable in the determination of
more polar sterols and their oxides in samples where
high resolution and mild conditions are imperative.
Another important development is the more fre-
quent use of solvents under subcritical conditions
which, in practice, is eliminating the rather artiRcial
boundary between SFC and liquid chromatography.
One of the main advantages of using packed capil-
lary columns over conventional packed columns is
the improved performance when this separation is
coupled to mass spectrometry, thus providing struc-
tural determination of the sterols. This is expected to
be especially important in the coming years owing to
the development of new commercial equipment for
SFC}MS.
Another potential source of improvement is the use
of new detectors with higher sensitivity than ultra-
violet but compatible with the use of modiRers in
packed capillary column SFC. The amperometric de-
tector is a good example.
These anticipated developments in SFC technology
will be important in the Reld of sterol analysis,
though an automated SFE}SFC without tedious
sample preparation and large solvent consumption
would be one of the most valuable future develop-
ments for sterol analysis.
Further Reading
Berg BE, Lund HS and Greibrokk T (1997) Separation and
quantiRcation of components of edible fat utilizing open
tubular column in SFC. Sample introduction by direct
injection and SFE coupled on-line to SFC. Chromato-
graphia 44: 399}404.
Chester TL (1996) Supercritical Suid chromatography for
the analysis of oleochemicals. In: King JW and List GR
(eds) Supercritical Fluid Technology in Oil and Lipid
Chemistry. Champaign, Illinois: AOCS Press.
Heupel RC (1989) Isolation and primary characterization
of sterols. In: Nes WD and Parish EJ (eds) Analysis of
Sterols and Other Biologically SigniTcant Steroids. San
Diego, California: Academic Press.
Hoving EB (1995) Review. Chromatographic methods in
the analysis of cholesterols and related lipids. Journal of
Chromatography B 671: 341}362.
Jinno K (ed.) (1992) Hyphenated Techniques in Supercriti-
cal Fluid Chromatography and Extraction. Amsterdam:
Elsevier.
Lee ML and Markides KE (eds) (1990) Analytical Super-
critical Fluid Chromatography and Extraction. Provo,
Utah: Chromatography Conferences.
Medvedovici A, David F and Sandra P (1997) Analysis of
sterols in vegetable oils using off-line SFC/capillary
GC}MS. Chromatographia 44: 37}42.
 III / STEROLS / Thin-Layer (Planar) Chromatography
Staby A and Mollerup J (1993) Separation of constituents
of Rsh oil using supercritical Suids: a review of experi-
mental solubility, extraction and chromatographic data.
Fluid Phase Equilibria 91: 349}396.
Thin-Layer (Planar) Chromatography
J. Novakovicf , PRO.MED.CS, Prague,
Czech Republic
K. Nesme\ raf k, Charles University, Prague,
Czech Republic
Copyright ^ 2000 Academic Press
Sterols are steroid compounds widely distributed in
various biological materials, e.g. variety plant and ani-
mal lipids, medications, food and dietary supplements.
They are basic metabolites in living organisms and
they are also precursors of a variety of bile acids,
provitamins and steroid hormones. Therefore the anal-
ysis of sterols is important in biochemistry, medicine
and pharmacy. There is considerable interest in the
study of the relationship of plasma cholesterol concen-
trations to the risk of developing coronary artery dis-
ease. Determination of phytosterols and cholesterol is
important for the diagnosis of phytosterolaemia and in
dietary treatment of hypercholesterolaemia.
The collective name for sterols has been adopted
for all naturally occurring crystalline unsaponiRable
steroid alcohols. The basic sterol is 5-cholestane,
and the structure numbering system for sterols are
given in Figure 1. In general, these compounds are
3-monohydroxysteroids, having 27, 28 or 29 carbon
atoms and nearly all have one or more double bonds.
The double bond is most commonly found at position
5, with double bonds at C7 and C22 also being preva-
lent. Sterols are classiRed into Rve groups: cholesterol
and its companions, zoosterols, phytosterols, myco-
sterols and vitamin D. Examples of sterols from each
group are given in Table 1.
Figure 1 5-Cholestane skeleton and numbering system for
sterols.
4255
Xie LQ, Markides KE, Lee ML, Hollenberg NK, Williams
GH and Graves SW (1993) Bioanalytical application of
multidimensional open tubular column supercritical
Suid chromatography. Chromatographia 35: 363}371.
There are four principal methodologies used in
sterol chromatography: gravity Sow column liquid
chromatography (GCC), high performance liquid
chromatography (HPLC), gas chromatography (GC)
and thin-layer chromatography (TLC). For its selec-
tivity, sensitivity and efRciency, TLC is one of the
most frequently employed procedures for the separ-
ation of sterols, both for their characterization and
for their quantitative analysis.
Preparation of Sample
Since sterols are present in different materials,
sample preparation is a very important part of their
analysis.
The Rrst step is an extraction procedure which
is performed either directly on the previously de-
proteinized sample or after cleavage of any conju-
gates present. Diethyl ether, dichloromethane, ethyl
acetate, chloroform and other medium polarity
organic solvents can be used for extraction.
The next step is puriRcation of the extract or, more
exactly, separation of sterols from other lipids, usu-
ally by TLC on a silica gel G plate using an
n-heptane}diethyl ether}acetic acid (85 : 15 : 1) mix-
ture as the mobile phase. Under these conditions the
cholesterol and phytosterols are concentrated in one
band. Separation and quantitative analysis of indi-
vidual sterols are performed after elution from the
plate by GC, TLC or some other technique.
Some sterols (e.g. vitamin D) are sensitive to
atmospheric oxygen, traces of acids and bases,
light and heat. Therefore, all steps should be carried
out in a cool place, protected from exposure to direct
light, and only highly puriRed solvents should be
used.
Stationary Phases and Solvent
Systems
Generally, the chromatographic separation of indi-
vidual sterols is difRcult due to the large number of
epimers and unsaturated isomers. Various forms
of silica gel are most frequently used in the TLC of
4256 III / STEROLS / Thin-Layer (Planar) Chromatography

Table 1 Examples of sterols

Number of carbon atoms Trivial name Systematic name Group of sterolsa

27 Vitamin D3 (cholecalciferol) 5,7-Cholestadiene-3-ol with open B-ring 5

Cholesterol 5-Cholestene-3-ol 1
7-Dehydrocholesterol 5,7-Cholestadiene-3-ol 1
(provitamin D3)
Desmosterol 5,24-Cholestadiene-3-ol 2
Dihydrocholesterol 5
-Cholestane-3-ol 1
28 Brassicasterol 24--Methyl-5,22-cholestadiene-3-ol 3
Campesterol 24-
-Methyl-5-cholestene-3-ol 3
Ergosterol (provitamin D2) 24--Methyl-5,7,22-cholestatriene-3-ol 3
Fungisterol 24--Methyl-7-5
-cholestene-3-ol 4
Vitamin D2 (ergocalciferol) 24--Methyl-5,7,22-cholestatriene-3-ol, with 5
open B-ring
29 Dihydro--sitosterol 24--Ethyl-5
-cholestane-3-ol 3
-Sitosterol 24--Ethyl-5-cholestene-3-ol 3
Stigmasterol 24--Ethyl-5,22-cholestadiene-3-ol 3
30 Lanosterol 4,4,14-Trimethyl-8,24-cholestadiene-3-ol 1

a
Group of sterols: 1, cholesterol and its companions; 2, zoosterols; 3, phytosterols; 4, mycosterols; 5, vitamin D.

sterols; separation can be classiRed, according to the The separation of cholesterol and cholesterol esters
type of stationary phase, into four main groups: silica from other lipid fractions on silica gel pre-coated
gel, silver nitrate silica gel, reversed-phase and plates using the following solvent systems: Rrst,
bromine-system TLC. chloroform}methanol}water (65 : 25 : 4), second,
n-hexane}acetone (100 : 1) and third, n-hexane}
Silica Gel TLC
acetone (100 : 3) has also been reported. The plates are
The separation of individual sterols by adsorption developed with the Rrst solvent system to 8 cm from
chromatography on silica gel pre-coated plates is rela- the origin. After drying, the plates are developed to
tively easy when there are differences in the type, 18 cm above the origin with either of the other sol-
number, position or conRguration of polar groups, vent systems. Cholesterol and four cholesterol ester
but it is difRcult in the absence of such differences. subfractions are separated from other lipid fractions.
The chromatographic properties of sterols on silica Silica gel GF plates have been used to separate
gel G plates using two solvent systems, benzene}di- cholesterol, cholesteryl propionate and low molecu-
ethyl ether (9 : 1) and benzene}diethyl ether (85 : 15) lar weight cholesteryl esters by one-stage one-dimen-
have been studied. The resulting RF values are listed sional TLC. This work employed four solvent
in Table 2. However, the separation of sterols with systems, the best separation among cholesteryl for-
these systems is incomplete. mate, cholesteryl acetate and cholesteryl propionate
was achieved using chloroform}diethyl ether}acetic
Table 2 Relative retentions of some sterols on silica gel G acid}1-propionic acid (92 : 1.5 : 1 : 5 : 0.5).
The solvent systems used for the separation of eight
Sterol RF value 3-sterols of considerable biological interest, which
differ only in ring B and/or in the side chain, on silica
System I System II
gel G pre-coated plates has also been investigated.
Cholesterol 0.20 0.47 The separation was performed using Rrst, cyclo-
7-Dehydrocholesterol 0.19 hexane}ethyl acetate}water (600 : 400 : 1); second,
Desmosterol 0.20 0.46 cyclohexane}heptane (1 : 1), third, cyclohexane}
Brassicasterol 0.20
ethyl acetate}water (1560 : 440 : 1), and fourth,
Ergosterol 0.19 0.43
Vitamin D2 0.19 isooctane}carbon tetrachloride (19 : 1). Differences
-Sitosterol 0.20 in resolving power between polar and nonpolar systems
Stigmasterol 0.20 were observed. Resolution of the pairs with saturated
Lanosterol 0.31 0.62 and unsaturated side chain -sitosterol}stigmasterol
and cholesterol}desmosterol was Rnally effected by
Stationary phase: silica gel G; mobile phase: (I) benzene}diethyl
ether (9 : 1); (II) benzene}diethyl ether (85 : 15); visualization: a mixture of saturated hydrocarbons.
sulfuric acid. Separation of vitamin D from its close structural
Date from Xu et al. (1988). analogues, including provitamin D, irradiation
 III / STEROLS / Thin-Layer (Planar) Chromatography 4257

products of provitamin D and decomposition prod- Table 3 Separation of sterols and sterol acetates on silica gel
ucts, has been carried out on nonimpregnated layers G}silver nitrate layers
of silica gel G with the solvent system cyclohexane}
Sterol RS value
dichloroethane}diethyl ether (5 : 3 : 2). TLC has also
been used for pre-puriRcation of saponiRed samples System I System II
before GC analysis as well as for their in situ quanti- (sterols) (sterol acetates)
tative analysis. Determination of the maximum
Cholesterol ,1.00 ,1.00
permissible limit of concentration of ergosterol in
7-Dehydrocholesterol 0.44 0.43a
ergocalciferol using silica gel G as the stationary phase Dihydrocholesterol 1.14 1.25
with a cyclohexane}peroxide-free ether (1 : 1) mixture Brassicasterol 0.98 0.68
containing 0.1 g L\1 butylhydroxytoluene is an ofRcial Vitamin D2 0.64
method in the European Pharmacopoeia 1997. Dihydro--sitosterol 1.14 1.30
-Sitosterol 1.00 1.00
In general, monosaturated sterols like cholesterol,
Stigmasterol 0.98 0.87
provitamin D (e.g. ergosterol) and vitamins D are Lanosterol 1.70 0.78
separable, but closely related sterols like cholesterol,
stigmasterol and -sitosterol are not resolved on a
After two developments.
silica gel. Stationary phase: silica gel G}silver nitrate. Mobile phase: (I)
chloroform}diethyl ether}acetic acid (97 : 2.3 : 0.5); (II) chloro-
Silver Nitrate TLC (Argentation Chromatography) form}light petroleum (b.p. 60}803C)}acetic acid (25 : 75 : 0.5).
Visualization: dibromofluorescein.
Several methods have been published for separation Data from Copius-Peereboom and Beekes (1965).
of structurally related sterols. A procedure utilizing -
complex formation between Ag(I) ions and the
double bonds occurring in various locations in the The critical pair cholesterol}brassicasterol was not
sterol molecules has been frequently applied. Argen- separated in these RP systems and RP-TLC separation
tation TLC is a method for separating compounds according to the degree of unsaturation using the so-
based on differences in number and position of called bromine system was suggested (see below).
double bonds in the molecule. In this case, silica gel is A good separation of the pairs vitamin D2/D3 and
suspended in an aqueous solution of silver nitrate pre-vitamin D2/D3 on silica gel and Kieselguhr G im-
before spreading on the plate. Silver nitrate can also pregnated with silicone oil eluted with acetone}water
be sprayed on to a pre-prepared layer. mixture has been achieved.
Argentation TLC of sterols has been thoroughly
investigated. The RS values (relative retention relating The Bromine System TLC
to cholesterol) of selected sterols and sterol acetates The separation of unsaturated sterols from their
separated on silica gel G}silver nitrate pre-coated saturated analogues can be substantially improved by
plates using the solvent systems chloroform}diethyl
ether}acetic acid (97 : 2.3 : 0.5) and chloroform}light
Table 4 R S values of some sterols and sterol acetates obtained
petroleum (b.p. 60}803C)}acetic acid (25 : 75 : 0.5)
in RP-TLC
are listed in Table 3.
Sterols that differ in the number and position of Sterol RS value
double bonds are clearly separated by means of silver
nitrate TLC, but separation of cholesterol from the System I System II
phytosterols was not achieved. (sterols) (sterol acetates)

Reversed-phase TLC Cholesterol ,1.00 ,1.00

7-Dehydrocholesterol 1.12 1.26
One of the pioneer works on reversed-phase TLC Dihydrocholesterol 0.90 0.89
(RP-TLC) used silica gel impregnated with parafRn Brassicasterol 1.00 1.00
-Sitosterol 0.86 0.83
oil as the stationary phase and methyl ethyl ketone as
Stigmasterol 0.93 0.91
the mobile phase for the separation of lipids, includ- Lanosterol 0.84 0.97
ing cholesterol esters. Kieselguhr G has been used
following impregnation with undecane as the station- Stationary phase: Kieselguhr G impregnated with undecane; mo-
ary phase with the solvent systems acetic acid}water bile phase: (I) acetic acid}water (90 : 10); (II) acetic acid}water
(92 : 8), visualization: phosphomolybdic acid.
(90 : 10) and acetic acid}water (92 : 8) for the separ-
Data from Copius-Peereboom JW and Beekes HW (1962). The
ation of sterols and sterol acetates, respectively. analysis of mixtures of animal and vegetable fats. III. Separation
RS values of some sterols and sterol acetates obtained of some sterols and sterol acetates by thin-layer chromatography.
by RP-TLC are given in Table 4. Journal of Chromatography 9: 316.
4258 III / STEROLS / Thin-Layer (Planar) Chromatography

Table 5 R S values of some sterols acetates in the bromine ated on Kieselguhr G impregnated with undecane
system TLC using solvent system acetic acid}acetonitrile (1 : 3)#
0.5% bromine. RS values of some sterol acetates are
Sterol acetates RS
given in Table 5.
Cholesterol ,1.00 Bromination of the double bonds before or during
7-Dehydrocholesterol Front chromatography completely changes the mobilities of
Dihydrocholesterol 0.85 the unsaturated compounds, promoting their separ-
Brassicasterol 1.13
ation from the saturated derivatives. In this way the
-Sitosterol 0.82
Stigmasterol 1.06 critical pair cholesterol}brassicasterol can be clearly
Lanosterol Front separated.

Stationary phase: Kieselguhr G impregnated with undecane; mo-

bile phase: acetic acid}acetonitrile (1 : 3)#0.5% of bromine,
visualization: antimony trichloride.
Data from Copius-Peereboom and Beekes (1965).
bromination. After spotting the sterol sample at the
starting zone, some drops of bromine are spotted on
the same place. The plate is then developed with
a benzene}ethyl acetate mixture (2 : 1), by means of
which the spots of cholesterol and dihydrocholesterol
are separated. Thus, sterol acetates have been separ-
Detection and Quantitation
Detection
Sterols that have UV absorbance can be detected at
254 nm (providing that TLC separation is performed
on a layer with a Suorescent indicator). Since a num-
ber of sterols do not have UV absorbance suitable for
detection, most applications still involve visualization
based on chemical reactions. Visualization proced-
ures used to detect and characterize sterols are well

Table 6 Detection procedures based on chemical reactions used in TLC analysis of sterols

Detection reagent Visualization procedure Sterols

Acids
Perchloric acid Spray Vitamin D2
(70%)
Phosphomolybdic acid Spray, heat at 1103C for 10 min Various sterols
(15% ethanolic solution)
Sulfuric acid Spray, heat at 1103C for 15 min, observe C27 sterols, cholestane and lanostane
(conc. or 50%) in day and UV light before and after heating series, ergosterol

Metal salts
Antimony pentachloride Spray, heat at 1203C for 5 min Various sterols
(30% in chloroform)
Antimony trichloride Spray, heat at 1003C for 10 min Brominated sterols
(50% in conc. acetic acid)
Cadmium chloride Spray, heat at 903C for 15 min, observe in Brominated sterols, cholesteryl esters
(50% in 50% ethanol) UV light
Copper sulfate Saturated species
Cupric acetate Spray, heat at 1503C for 30 min Unsaturated species
(3% in 8% phosphoric acid)

Aldehydes
p-Ansialdehyde Spray, heat at 903C for 10 min Sterol and sterol acetates
(1% in acetic acid}sulfuric acid
(98 : 2) mixture)
Salicylaldehyde Spray with pure salicyladehyde, heat at Sterol and sterol acetates
803C for 5 min, spray with 0.5 mol L\1 sulfuric
acid, heat again at 903C for 10 min
Vanillin Spray, heat at 1003C for 5 min Cholestanols and cholestanones
(0.5% in sulfuric acid}ethanol
(4 : 1) mixture)

Ketone reagents
2,4-Dinitrophenylhydrazine Spray and spray again with conc. sulfuric acid Ketonic sterols
(5% in methanol)
 III / STEROLS / Thin-Layer (Planar) Chromatography
established. Detection reagents can be classiRed into
four groups: acids, metal salts, aldehydes and ketone
reagents. The most frequently used detection proced-
ures are listed in Table 6.
TLC coupled with Same ionization detection (TLC-
FID) has been used to detect sterols. The separation is
performed on specially prepared thin quartz rods
coated with adsorbent sintered on to the surface of thin
rods. The adsorbent is usually silica gel and the sol-
vent system is basically the same as in classical TLC.
TLC-FID is a useful technique for the separation of
cholesterol and its esters from other lipid classes.
Separation of individual sterols, especially phyto-
sterols (e.g. -sitosterol, campesterol, brassicasterol,
stigmasterol, etc.) using TLC-FID is not possible.
On the other hand, groups of sterols differing
in the number of methyl groups in position 4
(i.e. 4-demethylsterols, 4-methylsterols and 4,4-
dimethylsterols) can easily be separated by TLC-FID.
The TLC separation of sterols is often used for
preparative purposes. After elution from the plate,
the sterols can be analysed by some other technique
(spectrophotometry, Suorimetry, GC, GC-MS,
HPLC), which is why it is sometimes necessary to
visualize them using nondestructive reagents such as
iodine vapour, water spray or Suorescent reagents
(e.g. Suorescein, Rhodamine B). Fluorescent reagents
can be incorporated in the slurry, during the prepara-
tion of the layer. An example of a nondestructive
detection using radiolabelled [4-14C] cholesterol and
cholesteryl [14C]oleate added as internal standards
has also been reported where desmosterol in monkey
spermatozoa was separated on silica gel G. The free
sterol band containing both cholesterol and desmos-
terol was detected, extracted from the plate and after,
derivatization, analysed by GC.
Quantitative in situ Analysis
Due to progress in plate technology and instrumenta-
tion, modern TLC has become a comparable method
to other chromatographic techniques in terms of
accuracy, precision, reliability and repeatability. In
modern TLC, the main steps are automated, includ-
ing the sample application on the plate } the step
considered the most critical for quantitative evalu-
ation. Several examples of direct quantitation of ster-
ols on TLC plate are discussed below.
Cholesterol ester mapping of human serum by high
performance TLC (HPTLC) has been performed.
Quantitative analysis was carried out at 546 nm after
postchromatographic derivatization with phos-
phomolybdic acid. The densitogram of a standard
mixture containing cholesterol esters is given in
Figure 2.
4259
Figure 2 Densitogram of a standard mixture containing 33 ng
of each cholesterol ester. 1, Cholesterol palmitate; 2, cholesterol
oleate; 3, cholesterol linolate; 4, cholesterol linoleate. Separation
was performed on HPTLC Kieselgel 60 F254 (Merck) in carbon
tetrachloride solvent system. (Reproduced with permission from
KovaH cs et al., 1989. Copyright 1989 American Association for the
Advancement of Science.)
HPTLC silica gel plates and a dual solvent system,
consisting of a run with isopropyl ether}acetic acid
(96 : 4) followed by a run in the same direction with
petroleum ether (b.p. 60}703C)}diethyl ether}acetic
acid (90 : 10 : 1) has been to determine cholesterol in
egg yolk as well as in butter and cream samples.
Cholesterol was detected with cupric acetate reagent,
lipid zones were quantiRed by densitometry.
Quantitative measurement of free cholesterol in
serum on a silica gel/sodium carboxymethylcellulose
plate has also been reported. The solvent system was
petroleum ether}ethyl acetate}glacial acetic acid
(80 : 20 : 1). Spraying with vanillin was used for vis-
ualization. The colour of the cholesterol spot was
stable for c. 20 min. In situ measurement was done by
densitometry at 525 nm with a detection limit of
40 ng per spot. The peak area was linearly related to
the amount of cholesterol over the range 80}700 ng
per spot.
Cholesterol, cholesteryl esters and other neutral
lipids have been analysed in plasma by TLC-FID.
Separation was performed on Chromarods S with
hexane}diethyl ether}formic acid (52 : 8 : 0.1).
Quantitative in situ analysis of vitamin D in
cod liver oil has been demonstrated by measuring
absorbance at 268 nm on silica gel layers after dual
development (Rrst with n-hexane, then with cyc-
lohexane}diethyl ether (1 : 1) mixture). Vitamin
D was also determined in foods and in human milk by
in situ reSectance measurement.
TLC and Characterization of Sterols
The chromatographic behaviour of each compound
depends on the stereochemistry and location of polar
4260 III / STEROLS / Thin-Layer (Planar) Chromatography
Figure 3 Relationship between RS values and the carbon numbers for sterol acetates in RP-TLC; coefficient of linear correlation,
R"0.9793. Stationary phase: Kieselguhr G impregnated with undecane; mobile phase: acetic acid}acetonitrile (1 : 3). Nc"number of
carbon atoms!number of double bonds. (Data from Copius-Peerboom and Beekes, 1965.)
substituents, the solubility, partition coefRcients and
equilibrium constants of the compound in the solvent
system, the size and shape of the molecule and the
degree of hydration. The quantitative structure}
chromatographic retention relationship study be-
tween sterols, TLC mobilities and their structures has
been investigated by several authors.
A separation of steroids according to the degree of
unsaturation has been investigated. In structural anal-
ysis, argentation TLC and the bromine system can
give information about the number and position of
double bonds in a molecule. A linear relationship
between the RS value and carbon numbers
(Nc"number of carbon atoms!number of double
bonds) in the system undecane/acetic acid}acetonit-
rile (1 : 3) for saturated and 5-unsaturated sterols
has been found. This linear relationship is shown in
Figure 3.
Adsorption TLC is not only a method of sample
puriRcation, but the RF value also provides a clue to
the compounds structure. The structural feature that
mostly contributes to the chromatographic behaviour
of sterols in adsorption TLC is the presence of a free
3-OH group. Converting the 3-OH to an acetoxy,
methoxy, keto, or 3
-OH results in a steroid with
a less polar RF value relative to the RF value obtained
for cholesterol.
Separation of individual cholesterol ester subfrac-
tions according to the sum of the carbon atoms and
numbers of double bonds in their fatty acid moieties
has been performed.
The elution order of vitamin D photoisomers can
be correlated with the increasing planarity of the
molecules. RF values of vitamin D photoisomers
on silica gel with solvent system cyclohexane}
dichlorethane}diethyl ether (5 : 3 : 2) were 0.18 (pro-
vitamin D3), 0.23 (tachysterol3), 0.27 (lumisterol3)
and 0.31 (pre-vitamin D3).
A visualization procedure can give additional in-
formation about sterol structure, since different re-
agents produce different colours with individual
sterols. Some reagents are speciRc for individual func-
tional group (e.g. 2,4-dinitrophenylhydrazine for
keto group).
Future Developments
A general tendency in modern TLC is separation on
HPTLC stationary phases, online coupling with other
separation techniques (e.g. HPLC-TLC), as well
as online coupling with spectroscopic methods.
In situ recording of UV-visible spectra is most
commonly used. However, the recording of Fourier
transform infrared, Raman or mass spectra is more
informative. Although these combinations have fre-
quently been reported in the literature, there is still
no example of their application in the Reld of sterol
analysis. Due to the great variety of chemically
closely related sterols, online combination of TLC
and spectroscopic methods can be considered
a powerful tool for their isolation and in situ
characterization.
 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE
See also: II /Chromatography: Thin-Layer (Planar):
Densitometry and Image Analysis; Spray Reagents.
III/Flame Ionization Detection: Thin-Layer (Planar)
Chromatography. Impregnation Techniques: Thin-
Layer (Planar) Chromatography. Silver Ion: Thin-Layer
(Planar) Chromatography. Sterols: Supercritical Fluid
Chromatography.
Further Reading
Bennett RD and Heftmann E (1962) Thin-layer chromatog-
raphy of sterols. Journal of Chromatography 9: 359.
Copius-Peereboom JW and Beekes HW (1965) The
analysis of mixtures of animal and vegetable fats. V.
Separation of sterol acetates by thin-layer chromatogra-
phy in reversed-phase systems and on silica gel G}
silver nitrate layers. Journal of Chromatography
17: 99.
Hung GWC and Harris AZ (1989) Separation of low-
molecular-weight cholesteryl esters by thin-layer
chromatography. Microchemical Journal 40: 208.
STRONTIUM FROM NUCLEAR WASTES:
ION EXCHANGE
P. Sylvester, Texas A & M University, College Station,
TX, USA
Copyright ^ 2000 Academic Press
Introduction
The development of new inorganic ion exchange ma-
terials for the selective removal of strontium and
other radionuclides from nuclear waste has pro-
gressed rapidly in recent years. 90Sr is an important
component of many nuclear wastes and is a high yield
Rssion product of 235U. It is relatively short-lived with
a half-life of 28.8 years and, along with 137Cs, is
the source of a large percentage of the initial radioac-
tivity and heat generation of spent nuclear fuel.
During the reprocessing of nuclear fuel, irradiated
uranium fuel rods are dissolved in nitric acid
and uranium and plutonium are separated from the
Rssion products and other actinides by means of
the Purex process. Tributylphosphate (TBP) dissolved
in an organic phase, such as odourless kerosene,
is contacted with the nitric acid solution, and
plutonium and uranium nitrates are selectively
complexed by the TBP and extracted into the organic
4261
KovaH cs L, Martos ED , Pick J and Pucsok J (1989) Cholesterol
ester mapping of human serum by HPTLC. Journal of
Planar Chromatography 2: 155.
Lisboa BP (1969) Chromatography of sterols and steroids.
In: Marinetti GV (ed.) Lipid Chromatographic Analysis,
vol. 2, pp. 57}147. New York: Marcel Dekker.
Ranny M (1987) Thin-layer Chromatography with Flame
Ionization Detection. Prague: Academia.
Sherma J and Fried B (eds) (1996) Handbook of Thin-layer
Chromatography, 2nd edn. New York: Marcel Dekker.
TvrzickaH E and Votruba M (1994) Thin-layer chromatogra-
phy with Same-ionization detection. In: Shibamoto
T (ed.) Lipid Chromatographic Analysis, pp. 51}73.
New York: Marcel Dekker.
Xu S, Norton RA, Crumley FG and Nes WD (1988) Com-
parison of the chromatographic properties of sterols, se-
lect additional steroids and triterpenoids: gravity-Sow
column liquid chromatography, thin-layer chromatogra-
phy, gas-liquid chromatography and high-performance
liquid chromatography. Journal of Chromatography
452: 377.
phase. The majority of Rssion products, includ-
ing 90Sr, remain in the aqueous acidic phase,
which can then be concentrated by means of evapor-
ation and stored prior to permanent disposal. In
addition to the acidic high level waste stream,
numerous other streams are generated during rep-
rocessing operations as a result of washing, decon-
tamination and scrubbing operations. Details of
some speciRc streams generated by the nuclear indus-
try from which 90Sr needs to be selectively removed
from large excesses of inert ions will be given later in
this article.
A convenient method of selectively removing
contaminant species from higher concentrations of
inert ions is by ion exchange. Organic ion ex-
change resins are used in many industries for the
selective removal of ions from aqueous streams.
These materials consist of a polymeric backbone
(commonly polystyrene) to which has been attached
functional groups such as carboxylic or sulfonic
acids to produce cation exchangers, or tertiary or
quartenary amines to produce anion exchange resins.
However, the use of organicresins in the nuclear
industry is limited for a number of reasons. These
include:
4262 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE
E low resistance to damage by ionizing radiation,
thus limiting operational life;
E low thermal stability;
E limited chemical stability;
E low selectivity in comparison to inorganic ion ex-
change materials;
E incompatibility with grout or cement, making the
Rnal disposal of spent resins a problem.
Inorganic ion exchangers offer a number of advant-
ages over conventional organic resins including
greater selectivity and both chemical and radiolytic
stability. Additionally, they are compatible with cur-
rent waste encapsulation techniques and are stable
enough that they can be used as a Rnal waste form for
long-term storage. The major drawback to the use of
inorganic ion exchangers is that they are typically
synthesized as Rne powders, which are unsuitable for
use in column operations. However, there are now
a number of techniques available to allow these pow-
ders to be produced as pellets or particles suitable for
column operations, while still retaining fast ion ex-
change kinetics and the ion selectivity of the original
material. A number of reviews of available materials
and their ion exchange selectivities have been written,
and as new materials and methods of manufacture
are being developed, the use of inorganic materials
both in the nuclear industry and elsewhere will un-
doubtedly expand. Some of the major classes of ma-
terials that are currently being used (or are under
evaluation) for the selective removal of 90Sr from
nuclear wastes are described in the following sections.
Zeolites
Zeolites are hydrated aluminosilicates with open
framework structures. These consist of building blocks
of +SiO4, and +AlO4, tetrahedra which can be interlin-
ked to give a wide range of different materials with
regular tunnels and cavities. The presence of trivalent
aluminium in the framework results in a net negative
charge that is neutralized by the absorption of ca-
tions. SpeciRc zeolites exhibit high selectivities for
strontium and caesium over other alkali and alkaline
earth cations. This has led to their use in the treat-
ment of some nuclear waste streams. Details of zeolite
synthesis, structures, applications and information on
their ion exchange properties can be found in the
literature and will not be detailed in this article.
Clinoptilolite is a common natural zeolite with the
ideal formula Na6Al6Si30O72 ) 24H2O, though due to
interactions with natural groundwaters, some of the
Na# ions will have ben replaced by K#, Mg2# and
Ca2# ions. It is currently used on a large scale both in
the UK and the USA for the treatment of nuclear
waste solutions such as cooling pond water. This is one
of the largest waste streams in terms of the volume of
liquid, and consists of water used to cool and shield
irradiated uranium fuel rods prior to their disposal or
reprocessing. For example, spent pressurized water
reactor (PWR) fuel is generally stored under water for
up to 5 years to allow short-lived radioisotopes, such
as 131I (T1/2"8.06 days) and 106Ru (T1/2"367 days),
to decay away, and thus make the fuel rods easier to
handle. Storage is often accompanied by the release of
tiny amounts of radioactivity, primarily 137Cs and
90
Sr, from the fuel rods into the cooling water thus
necessitating removal of the radioactivity before the
water can be discharged to the environment. Modern
fuel is typically clad in zircalloy or stainless steel and
the release of radioisotopes is minimal. However,
older fuel types such as the UKs Magnox fuel and
fuel stored at the Hanford site in Washington State,
USA, do release signiRcant radioactivity.
Fuel storage pond waters are relatively pure and
contain minimal dissolved cations that can compete
with the 90Sr and 137Cs for the available ion exchange
sites. Compositions of a fuel pond simulant from the
Hanford site and the composition of an average pond
water from the British Nuclear Fuels plc. (BNFL)
SellaReld, UK site are given below in Table 1.
At the SellaReld plant, BNFL employs two 9.6 m3
beds of clinoptilolite in the site ion exchange efSuent
plant (SIXEP) to decontaminate the pond water used
for storage of Magnox fuel before controlled dis-
charge to the sea. This clinoptilolite originates from
the Mud Hills deposit in the Mojave Desert, Califor-
nia, and has been crushed and sieved to give a particle
size of 0.4}0.8 mm in diameter. A schematic of the
SIXEP plant is given in Figure 1.
Table 1 Composition of two fuel cooling ponds
Component Hanford N-basin (ppm) Sellafield (ppm)
Al 0.78 0.3
B 28.4 nd
Ba 3.1 0
Ca 33.4 1.7
Cs 6.47;10\5 3217 Bq mL\1a
K 2.5 3.6
Mg 0.70 0.3
Na 37.2 48.5
Sr 0.39 287 Bq mL\1a
pH 8.2 11.4
a
The activities of 90Sr and 137Cs correspond to pondwater
concentrations of 5.38;10\5 ppm and 8.69;10\4 ppm, respec-
tively. There is unlikely to be significant nonactive caesium
present; however, the amount of nonactive strontium is likely to be
significantly greater than the 90Sr concentration. nd, not deter-
mined.
 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE
Figure 1 Schematic of BNFLs site ion exchange effluent plant (SIXEP). (Reproduced with the permission of BNFL.)
The feed is pumped through a sand Rlter to remove
any particulates and is then treated with carbon diox-
ide to decrease the pH to approximately 7. The feed
passes through two beds of clinoptilolite in series
before being sampled and discharged to the Irish Sea.
On average, approximately 3000 m3 of efSuent per
day pass through the plant, which corresponds to
a contact time of only 4.6 minutes per ion exchange
bed. The clinoptilolite is very effective and typically
removes 98.7% of the strontium and over 99.7% of
the caesium from the stream prior to its discharge, and
each bed lasts for approximately 6 months online.
Another area in which zeolites can be used for the
removal of radioactive strontium is in groundwater
remediation. Groundwaters have relatively low ionic
strengths similar to pond waters, but differ in that the
predominant inactive ions in solution are magnesium
and calcium rather than sodium. There is also natu-
ral, nonradioactive strontium present, typically in the
order of a few tenths of a part per million, which will
also be removed along with the radioactive 90Sr. Since
Mg2# and Ca2# compete strongly with Sr2# for the
available ion exchange sites on the zeolites, the ob-
served distribution coefRcients (Kds) for Sr tend to be
considerably lower than in pond waters, and the
higher concentration of strontium results in a shorter
ion exchange bed life. However, the low cost of
natural zeolites (less than US $0.5 per lb for clinop-
tilolite) means that they are economically viable.
Sodium Nonatitanate
Sodium nonatitanate (NaTi), Na4Ti9O20 ) xH2O, dis-
plays a very high selectivity for strontium in basic
4263
media. The synthetic procedure is relatively simple
and has been scaled up to allow the titanate to be
produced on an industrial scale. A soluble source of
titanium, such as titanium isopropoxide, is added to
a 50% sodium hydroxide solution, resulting in the
immediate formation of a white precipitate. The mix-
ture is then heated in a hydrothermal bomb for ap-
proximately 21 h at a temperature of 2003C. The
product is Rltered, washed to remove excess NaOH,
and dried to produce a white powder. The Rnal ma-
terial has a low crystallinity and consequently it has
not been possible to determine the crystal structure.
However, the titanate is believed to consist of layers
of TiO6 octahedra separated by exchangeable sodium
ions and water molecules. At room temperature, the
interlayer space is approximately 10 A> , but the dis-
tance can vary considerably depending upon the dry-
ing temperature, and hence the number of water
molecules in the interlayer space. This material is now
available from Allied Signal Inc. (Des Plaines, Illinois,
USA) and similar products can also be obtained from
Selion Inc. (Merden, Connecticut, USA) and Boulder
ScientiRc Company (Mead, Colorado, USA).
Sodium nonatitanate exhibits a very high selectivity
for strontium over alkali and other alkaline earth
metals in basic media. In acidic media, the material has
a high afRnity for protons, so strontium selectivity is
negligible. Consequently, this allows the nonatitanate
to be stripped of absorbed strontium using dilute acid
and reused. Sodium nonatitanate readily hydrolyses in
water, exchanging protons for sodium ions; this results
in a considerable increase in the solution pH. Conse-
quently its use for treating groundwaters contaminated
with 90Sr is limited. However, its stability in highly
4264 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE

alkaline conditions makes it ideally suited for the

treatment of alkaline nuclear wastes. This aspect will
be discussed later.

Titanosilicates
Two separate classes of titanosilicate ion exchange
materials have been developed for the selective ex-
traction of strontium from nuclear wastes. Both
classes are composed of a titanosilicate framework,
but the crystal structures and the Ti : Si ratios are
different and hence so are the ion exchange
properties.
The Rrst class of materials is exempliRed by sodium
titanosilicate (NaTS), with the ideal formula
Na2Ti2O3SiO4 ) 2H2O. This can be synthesized in
a crystalline form which has allowed its structure to
be determined using X-ray powder methods. The
titanosilicate was found to have a tetragonal unit cell
with a"b"7.8082(2) A> and c"11.9735(4) A> . Figure 2 The structure of the caesium-exchanged titanosilicate
Edge-sharing TiO6 clusters reside in all eight corners showing Cs# ion in the centre of the tunnel.
of the unit cell and silicate tetrahedra are located
midway between the clusters and link them together. system. Titanosilicates with the general formula
This arrangement produces tunnels parallel to the M3H(AO)4(BO4)3 ) 4}6H2O (M"H, K, Na, etc.;
c-axis where the exchangeable sodium ions and the A"Ti, Ge; B"Si, Ge) have been prepared using
water molecules reside. The remaining sodium ions hydrothermal techniques. A homogeneous gel of
are located in the framework, bonded by silicate appropriate stoichiometry was hydrothermally
oxygens and are thus not exchangeable. Due to steric treated in an excess of either KOH or CsOH at 2003C
repulsions and space limitations, some of the sodium for 1}3 days. Sodium and proton forms were
ions in the tunnels are replaced by protons, leading to then prepared by exhaustively ion exchanging the
an actual formula of Na1.64H0.36Ti2O3SiO4 ) 1.84H2O. material with either HCl or NaCl. The best studied of
This exchanger is synthesized by hydrothermally these materials is the potassium pharmacosiderite,
heating a titanium silicate gel of appropriate K3H(TiO)4(SiO4)3 ) 4H2O (KTS-Ph), in which a"
stoichiometry in 6 mol L\1 NaOH at 1703C for b"c"7.7644(3) A> . Each unit cell consists of
2 days.
This material has been shown to have a high selec-
tivity for Cs# ions in both acid and alkaline pH and
a high selectivity for strontium in alkaline media.
Caesium ions exchanged onto the titanosilicate are
strongly held in the tunnels, as shown in Figure 2, and
are not readily leached off, thus the material is not
regenerable. Strontium is readily removed by washing
with dilute acid. A related material is currently mar-
keted by UOP as a crystalline silicotitanate (CST)
under the tradename IE-910 for the powder, and
IE-911 for an engineered form suitable for use in
column operations. The CST has shown excellent
selectivity for ppm levels of caesium ions in the pres-
ence of 7 mol L\1 Na# ions and is currently being
considered for use removing 137Cs and 90Sr from alka-
line nuclear wastes in the USA.
The second class of titanosilicate materials has the
crystal structure of the natural mineral pharmaco- Figure 3 The structure of potassium pharmacosiderite
siderite. Pharmacosiderite has the ideal formula HK3(TiO)4(SiO4)3 ) 4H2O with the K# ion located in the centre of
KFe4(AsO4)3(OH)4 and crystallizes in the cubic the tunnels. a"b"c"7.7644(3) A> .
 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE 4265

clusters of four titania octahedra linked to each other ity for strontium in the presence of high sodium
by silicate groups, as shown in Figure 3. This pro- concentrations, and they are also unstable in highly
duces a series of intersecting tunnels parallel to the a, alkaline conditions. However, the titanate and the
b and c axes with the exchangeable ions residing close titanosilicates are synthesized in strongly alkaline me-
to the face-centres of the unit cell. dia and thus exihibit a high stability in these tank
It has proved possible to substitute Ge for both Si wastes. In addition, all of the exchangers exhibit good
and Ti in the pharmacosiderite framework, thus radiation stability, thermal stability and excellent res-
allowing the size of the tunnels to be carefully istance to extreme chemical environments.
tailored. These materials have shown selectivity to- The sodium nonatitanate and the titanosilicate ion
wards both Cs# and Sr2# but are not as effective as exchange materials were evaluated in preliminary
the sodium titanosilicate, NaTS, described pre- batch experiments using 89Sr as a surrogate for 90Sr.
viously. However, the caesium ion can be eluted from Here, 0.05 g of exchanger was contacted with 10 mL
the exchanger, so unlike NaTS, the pharmacosiderites of waste simulant spiked with 89Sr, giving a volume to
are regenerable making them more cost-effective. mass ratio of 200 : 1, for 18 h using a rotary shaker.
The mixtures were then Rltered through a 0.2
m
Removal of Strontium from High Rlter and the activity of the aqueous phase deter-
mined using liquid scintillation counting. Kd values
Ionic Strength Wastes for strontium were then calculated according to
In the USA, there are over 100 million gallons of eqn [1] below:
radioactive mixed waste stored in 332 tanks distri-
buted over a number of Department of Energy (DOE) Kd"(Ai!Af)/Af;v/m [1]
sites. Much of this tank waste is highly alkaline and is
typically over 7 mol L\1 in Na#. The majority of this where Ai is the initial activity of solution (counts per
waste is found at the Hanford site in Washington minute mL\1); Af is the Rnal activity of solution
State, where there is approximately 65 million gallons (counts per minute mL\1); v is the volume of solution
of high-level waste stored in 177 tanks. All of the (mL); and m is the mass of exchanger (g).
Hanford tanks are highly alkaline and were generated
as by-products of the production of 239Pu for nuclear Table 2 The composition of two Hanford tank waste simulants
weapons manufacture. Initially, the waste was in a ni-
Species NCAW (mol L\1) 101-SY (mol L\1)
tric acid matrix, but to minimize corrosion of the steel
tanks, sodium hydroxide was added to neutralize the Al 0.43 0.42
wastes and to precipitate much of the radioactivity. Ca 0 4.20;10\3
The composition of each tank is different, but in Cs 5.00;10\4 4.19;10\5
Fe 0 1.96;10\4
general, the wastes consist of three distinct phases. At
K 0.12 0.034
the bottom of the tank is a metal hydroxide sludge, at Mo 0 4.20;10\4
the top is a salt cake, predominantly made up of Na 4.99 5.1
nitrate salts, and between these layers is an alkaline Ni 0 2.50;10\4
supernate. 90Sr is found in all three layers, but tends to Rb 5.00;10\5 4.20;10\6
Sr 2.70;10\7 4.10;10\6
predominate in the sludge layer. However, in cases
Zn 0 5.00;10\4
where there are signiRcant amounts of complexing Carbonate 0.23 0.038
agents, considerably greater 90Sr activity is found in Fluoride 0.09 0.092
the supernate. The composition of two supernate Hydroxide 3.4 3.78
simulants developed at PaciRc Northwest National Hydroxide (free) 1.68 2.11
Nitrate 1.67 1.29
Laboratory (PNNL) to mimic the tank wastes are
Nitrite 0.43 1.09
given in Table 2. Sulfate 0.15 4.75;10\3
Both waste simulants represent dilution of actual Phosphate 0.025 0.02
tank wastes, which is envisaged to be necessary to Citric acid 0 5.00;10\3
allow ease of handling without excessive precipitation Na4EDTA 0 5.00;10\3
HEDTA 0 3.75;10\3
of salts occurring. NCAW represents tank 241-AZ-
Iminodiacetic acid 0 0.031
102, and 101-SY represents tank 241-SY-101, both Na3 nitrilotriacetate 0 2.50;10\4
diluted to approximately 5 mol L\1 in Na#. This lat- Sodium gluconate 0 0.013
ter tank contains signiRcant amounts of complexants
pH 14.5 14.4
such as ethylenediaminetetraacetic acid (EDTA) and
citric acid. Zeolites are unsuitable for the treatment of Na4EDTA, ethylenediaminetetraacetic acid, tetra sodium salt.
these tank wastes because they lack sufRcient selectiv- HEDTA, N-(2-hydroxyethyl)ethylenediaminetriacetic acid.
4266 III / STRONTIUM FROM NUCLEAR WASTES: ION EXCHANGE
Table 3 The removal of strontium from Hanford tank waste simulants by inorganic ion exchange materials
Ion exchanger NCAW, Kd (mL g\1) % Sr removed
Clinoptilolite 48 19.35
NaTi 235 000 99.93
NaTS 270 000 99.93
KTS-Ph 20 200 99.55
The results obtained for the three ion exchangers
are displayed in Table 3. For comparative purposes,
the strontium Kd for the Mud Hills clinoptilolite was
also included, although this number should be viewed
with caution because clinoptilolite is not stable in
highly alkaline media and will have undergone sub-
stantial decomposition.
It can be seen that all of the synthetic ion exchange
materials exhibited a very high selectivity for stron-
tium from NCAW with Kd values in the tens or
hundreds of thousands. Clinoptilolite performed very
poorly, with a Kd of only 48 mL g\1 compared with
the best material, the sodium titanosilicate, which
had a Kd of 270 000 mL g\1. By contrast, the
Kd values from the 101-SY simulant were very low for
all of the materials, indicating that the presence of
relatively high concentrations of EDTA, citric acid
and other complexants has resulted in the strontium
being strongly chelated and thus not readily extract-
able by ion exchange. However, recent studies have
indicated that this problem can be overcome by the
addition of Ca2# or other ions to the waste in sufR-
cient quantity to saturate all of the EDTA and other
complexants present, and thus release the strontium
into solution where it can be removed by strontium-
selective ion exchangers. Alternatively, the com-
plexants can be destroyed using an appropriate chem-
ical oxidation technique and the strontium removed
by ion exchange.
Column Experiments
Column experiments using 89Sr-spiked NCAW were
performed to further evaluate the efRciency of the
sodium nonatitanate, the sodium titanosilicate and
the potassium pharmacosiderite at removing Sr under
dynamic conditions. This necessitated pelletizing the
ion exchange materials using approximately 15% by
weight of a hydrous titania binder. Hydrous titania
also shows some afRnity for strontium in alkaline
media, but tests proved that the Kd values were insig-
niRcant in comparison to the ion exchange materials.
Approximately 1 mL of material was slurried into
a column and NCAW, spiked with 89Sr to give a total
strontium concentration of 2.7;10\7 mol L\1, was
then passed through at a Sow rate of approximately
20 bed volumes per hour (BV h\1). The 89Sr activity
101-SY, Kd (mL g\1) % Sr removed
Not tested }
295 61.1
231 54.7
31 13.2
in the solution exiting the column was then analysed
using liquid scintillation counting. Percentage break-
through was calculated according to eqn [2]:
%Breakthrough"(Af/Ai);100 [2]
where Af is the 89Sr activity exiting the column and
Ai is the 89Sr activity entering the column.
The breakthrough curves obtained for the mate-
rials are given in Figure 4. Also included is the break-
through curve for the commercially available IE-911
determined under the same operating conditions. The
nature and percentage binder present in the IE-911 is
unknown, but in simple batch equilibrium experi-
ments Sr Kd values in excess of 25 000 mL g\1 were
obtained for 89Sr in NCAW. Figure 4 shows that
breakthrough of 89Sr from the IE-911 bed is almost
instantaneous, indicating very poor kinetics of ex-
change. By contrast, all three of the other exchangers
show (5% 89Sr breakthrough for over 1500 bed
volumes. Breakthrough was Rrst obtained for the
potassium pharmacosiderite and was followed by the
sodium titanate and had reached approximately 25%
and 17% respectively after 3000 bed volumes had been
passed. By contrast, the breakthrough for the titanosili-
cate was still only around 5% when the experiment was
terminated after the passage of 3500 bed volumes. This
indicates rapid kinetics for all of the materials except
the IE-911 and also an appreciable capacity for
Figure 4 89Sr breakthrough curves for IE-911, NaTi, NaTS and
KTS-Ph for NCAW at a flow rate of 20 BV h\1.
 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
strontium. Thus, all of the materials, particularly the
sodium titanosilicate, have good potential for the
decontamination of high-salt, alkaline nuclear wastes.
Conclusions
Inorganic ion exchangers have a wide number of
applications within the nuclear industry and are pre-
ferred over conventional organic resins. Zeolites are
ideal for the treatment of dilute wastes, provided that
the pH is not too extreme, and their relatively low
costs make their use highly economical. For more
extreme wastes like those encountered in the Hanford
storage tanks, new titanium-based materials have
been developed that are able to withstand the high
alkalinity and have sufRciently high selectivity to re-
move trace levels of strontium in the presence of
molar quantities of other ions. Although these
synthetic exchangers cost hundreds of US dollars per
kilogram, their extreme selectivity and ability to be
regenerated makes them viable options for the treat-
ment of these extremely complex wastes.
Acknowledgements
I would particularly like to acknowledge Professor
Abraham ClearReld, Dr. Elizabeth Bluhm and Gina
Graziano at Texas A&M University, who worked
with me on the titanate and titanosilicate ion ex-
change materials.
See also: I/Ion Exchange. II/Ion Exchange: Catalysis:
Organic Ion Exchangers; Historical Development;
SUB-CRITICAL WATER: EXTRACTION
See III / SUPERCRITICAL FLUID EXTRACTION-SUPERCRITICAL FLUID CHROMATOGRAPHY
SUGAR DERIVATIVES:
CHROMATOGRAPHY
S. C. Churms, University of Cape Town, South Africa
Copyright ^ 2000 Academic Press
Because sugar derivatives are generally present as
complex mixtures, chromatographic techniques are
crucial in their analysis. The spectrophotometric
methods, and other methods mentioned in this art-
icle, serve primarily as chromatographic detection
4267
Inorganic Ion Exchangers; Novel Layered Materials:
Non-Phosphates; Novel Layered Materials: Phosphates;
Organic Ion Exchangers; Surface Complexation Theory:
Multispecies Ion Exchange Equilibria; Theory of Ion
Exchange.
Further Reading
Amphlett CB (1964) Inorganic Ion Exchangers. Amsterdam:
Elsevier.
Barrer RM (1982) Hydrothermal Synthesis of Zeolites.
London: Academic Press.
Breck DW (1984) Malabar, FL: Robert E. Krieger.
ClearReld A (ed.) (1982) Inorganic Ion Exchange Mate-
rials. Boca Raton, FL: CRC Press.
ClearReld A (1988) The role of ion exchange in solid state
chemistry. Chemical Reviews 88: 125}148.
ClearReld A (1995) Inorganic ion exchangers: a technology
ripe for development. Industrial Engineering and Chem-
istry Research 34(8): 2865}2872.
Dyer A (1988) An Introduction to Zeolite Molecular Sieves.
Chichester: J. Wiley & Sons.
Dyer A, Hudson MJ and Williams PA (eds) (1993) Ion
Exchange Processes: Advances and Applications.
Cambridge: The Royal Society of Chemistry.
Helfferich F (1962) Ion Exchange. New York: McGraw
Hill.
Lombardo NJ and Schulz WW (eds) (1998) Science and
Technology for the Disposal of Radioactive Tank
Wastes. New York: Plenum.
Streat M (ed.) (1988) Ion Exchange for Industry. Chiche-
ster: SCI/Ellis Horwood Ltd.
Willliams PA and Hudson MJ (eds) (1990) Recent Develop-
ments in Ion Exchange 2. Barking: Elsevier.
systems, and spectroscopic methods are frequently
used in conjunction with chromatography.
Detection Reagents for Planar
Chromatography and for Qualitative
and Spot Tests
Detection reagents that are speciRc for particular
derivatives, or can distinguish certain classes from
4268 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
others, are listed in Table 1, which also shows the
colours produced in each case and, where known, the
detection limit. These reagents are used in spot tests
and as detection reagents in planar chromatography,
those not containing corrosive acids being applicable
to both paper and thin-layer chromatography.
Gas Chromatography
Some sugar derivatives are sufRciently volatile to be
analysed by gas chromatography (GC) without fur-
ther derivatization; this particularly applies to the
partially methylated methyl glycosides and methyl
glycoside methyl esters produced by methanolysis of
methylated polysaccharides. Multiple peaks, corre-
sponding to
and  anomers of pyranoside and
furanoside forms, are given by each glycoside, a
factor that complicates analysis but can aid the identi-
Rcation of the individual components of simple, well-
resolved mixtures. Unsubstituted methyl glycosides
require derivatization for analysis by GC; they are
successfully analysed as either trimethylsilyl (TMS)
ethers or triSuoroacetyl (TFA) esters. Here again the
characteristic patterns of multiple peaks produced
can facilitate identiRcation. However, for analysis of
complex mixtures it is desirable to simplify the
chromatogram by elimination of the anomeric centre.
The mixtures of partially methylated sugars ob-
tained in methylation analysis of polysaccharides are
usually submitted to GC as the acetylated alditol
derivatives, for which a large body of mass spectro-
metry (MS) data is available. However, for some
carbohydrates, notably amino- and acetamidodeoxy
sugars, the GC retention times of the derived alditol
acetates are excessively long. For aminodeoxyhexoses
this problem can be overcome by nitrous acid
deamination of the amino sugars before reduction
and acetylation, or by N-methylation of the
aminodeoxyalditols prior to acetylation. The most
satisfactory procedure in the analysis of the mixtures
of sugars obtained on hydrolysis of bacterial cell wall
polysaccharides or glycoconjugates is derivatization
to O-methyloximes, followed by acetylation or
trimethylsilylation. No more than two peaks are pro-
duced by each component of the mixture and simulta-
neous analysis of neutral and amino sugars, as well
as N-acetylneuraminic acid, muramic acid and
its N-acetyl derivative and 3-deoxy-D-manno-2-
octulosonic acid (KDO), within 40 min is possible
by capillary GC as the acetylated O-methyloximes.
GC analysis of uronic acids also requires derivatiz-
ation by speciRc methods if the multiple peaks given
by methyl glycoside methyl esters or TMS ethers are
to be avoided. Conversion to the oxime is an option
in this case too, or the acids may be reduced to
aldonic acids (by sodium borohydride reduction of
the alduronates) and analysed as the TMS derivatives
of the aldonolactones or the acetylated derivatives of
the N-alkylaldonamides produced on reaction of the
aldonolactones with a L-alkylamine in pyridine. Both
methods of derivatization proceeding via the aldonic
acids result in the production of a single GC peak for
each uronic acid present. The latter method has the
advantage that simultaneous analysis of aldoses, as
the alditol acetates, is possible } the alditol acetates
have much shorter retention times than the N-alkylal-
donamide acetates. Complete analysis of neutral and
acidic sugars within 20 min is possible by capillary
GC of these derivatives.
Oligosaccharide-alditols, up to tetrasaccharide
level, can be analysed by GC as their permethylated
derivatives. The volatility of those containing
acetamidodeoxyhexose residues can be increased by
N-triSuoroacetylation of these residues (through
transamidation by triSuoroacetolysis under carefully
controlled conditions) prior to methylation. This
procedure permits GC analysis of oligosaccharide-
alditols containing up to seven sugar residues and also
allows the use of the electron capture detector, with
a hundredfold increase in sensitivity.
Recommended conditions for GC analysis of vari-
ous sugar derivatives are listed in Table 2. Compre-
hensive retention data are available in the literature.
Liquid Chromatography
Carbohydrate derivatives can be analysed by liquid
chromatography (LC) in various modes, depending
on the polarity of the molecule and whether acidic or
basic groups are present. Nonpolar compounds, or
those rendered nonpolar by derivatization to increase
the sensitivity of analysis, are amenable to reversed-
phase LC or adsorption chromatography on silica.
For hydroxylic compounds such as alditols, several
options are available, including normal-phase LC on
bonded amino phases or amine-modiRed silica (the
column packing being modiRed in situ by addition of
a polyfunctional amine to the mobile phase); LC on
a cation exchange resin in the Ca2# form (ion-moder-
ated partitioning) or, as borate complexes, on an
anion exchange resin; and ion chromatography, with
pulsed amperometric detection. Recently, cyclodex-
trin-bonded silica has also proved effective.
The oligosaccharide-alditols obtained in degrada-
tive structural studies of glycoproteins can also be
analysed by LC in various ways; normal-phase LC,
ion exchange, ion chromatography and size exclusion
chromatography. Amino- and acetamidodeoxy-
hexoses and the hexitols derived from them can
be analysed by normal-phase LC; ion-moderated
Table 1 Reagents used in qualitative analysis for detection of sugar derivatives

Derivatives Reagent mixture Procedure Colour produced

Alditols and cyclitols Fleurys reagent: Spray with (a); heat at 90}1003C for Specific for polyols; cyclitols give orange spots
(a) HgO (5%, m/v) in HNO3 (5%, m/v), diluted 1 : 1 before use 10 min; spray with (b); heat at 1003C (detection limit 5}10
g inositol); alditols and
(b) Barium acetate (10%, m/v) mixed 1 : 10 with glacial acetic for 10}30 min other polyols grey to black
acid
Periodate-p-anisidine: Spray with (a); heat at 1053C for Polyols and aldonic acids give white spots on
(a) p-Anisidine (1%, m/v) in ethanol (70%, v/v) 5}10 min; dip in (b) brown background; distinguished from
(b) NaIO4 (0.1 mol L\1 aqueous solution) mixed 1 : 10 with aminodeoxy sugars, uronic acids and neutral
acetone sugars (see below)
Vanillin}perchloric acid: Spray; heat at 853C for 5}10 min Polyols give pale blue to lilac spots (detection
(a) Vanillin (1%, m/v) in ethanol limit 20}30
g); cyclitols and most aldoses do
(b) HClO4 (3% aqueous solutioin). Mixed 1 : 1 with (a) before use not react, ketoses give grey-green spots,
rhamnose brick-red

Aldonic acids and Periodate-p-anisidine: see above See above Aldonic acids give white spots on brown
aldonolactones background (see above)
Hydroxylamine}iron (III) chloride Spray with 1 : 1 mixture of (a) and (b); Aldonolactones revealed as purple spots;
(a) Hydroxylamine hydrochloride (1 mol L\1) in methanol dry at room temperature for 10 min; other lactones and esters also react
(b) KOH (1.1 mol L\1) in methanol spray with (c)
(c) FeCl3 (2%, m/v) in HCl (1% aqueous solution)

Amino- and Elson}Morgan reagent Dip through mixture of (a) and (b) Specific for amino- and
acetamidodeoxy sugars (a) KOH (25%, m/v) in aqueous ethanol (80%, v/v) (1 : 10); heat at 1103C for 5 min; dip acetamidodeoxyhexoses. Transient purple
(b) Pentane-2,4-dione (acetylacetone), redistilled (1%, v/v) in through mixture of (c) and (d) (1 : 1); spots at room temperature; heating to 803C
95% ethanol, fresh solution dry in stream of cold air gives permanent red spots for free
(c) N,N-Dimethyl-p-aminobenzaldehyde (10%, m/v) in conc. HCl aminodeoxy sugars, purple-violet for
(d) Ethanol (95%) acetamidodeoxy sugars
Fluorescamine Spray with (a); dry in air; spray with Specific for amino- and acetamidodeoxy
(a) Triethylamine (10%, v/v) in dichloromethane (b); dry in air; spray again with (a) sugars. Fluorimetric scanning (excitation
(b) Fluorescamine (0.05%, m/v) in acetone 390 nm, emission 475 nm) detects amino
sugars (detection limit 100 pmol)
Ninhydrin Spray, heat at 105}1103C for 10 min Specific reagent; purple spots produced
0.1% (m/v) solution in 1-butanol
Periodate-p-anisidine See above Yellow spots on brown background
See above

Esters and lactones Hydroxylamine-iron (III) chloride See above Purple spots
See above

Ketals 2,4-Dinitrophenylhydrazine Spray; heat at 1053C for 5 min Specific for keto group; ketoses and ketals
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

0.4% (m/v) solution in 2 mol L\1 HCl (e.g. pyruvate) give orange spots on light
yellow background
4269
Table 1 Continued
4270

Derivatives Reagent mixture Procedure Colour produced

Methyl ethers and methyl p-Anisidine hydrochloride Spray; heat at 1103C for 10 min Methyl ethers give highly characteristic pink,
esters 3% (m/v) solution in 1-butanol red or brown spots, some fluorescent under
UV. Methyl esters of methylated uronic acids
give bright pink colour

Neuraminic acids Resorcinol}HCl}Cu(II) Spray; heat at 110}1203C for about Neuraminic acids give blue spots; ketoses also
(a) Resorcinol (0.2%, m/v) in 4 mol L\1 HCl 20 min react
(b) Aqueous solution of CuSO4 ) 5H2O (0.1 mol L\1). Mix (a)
and (b) (40 : 1) at least 4 h before use
Periodate}thiobarbiturate Spray with (a); leave at room Only neuraminic acids give red spots
(a) NaIO4 (0.5 mol L\1) in 0.025 mol L\1 H2SO4 temperature for 15 min; dry (detection limit about 3
g)
(b) Ethylene glycol}aetone}conc. H2SO4 (50 : 50 : 0.3 v/v/v) thoroughly in stream of air; spray with
(c) Sodium 2-thiobarbiturate, 6% (m/v) in H2O (b); dry similarly. If odour of
formaldehyde persists spray again
with (b). Finally, spray with (c); heat at
1003C for 10 min

Phosphates Ammonium molybdate Spray Phosphorylated derivatives give immediate

10% (m/v) aqueous solution (20 mL) added to conc. HCl (3 mL) yellow colour of ammonium
with shaking; NH4Cl (5 g) added phosphomolybdate
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

Uronic acids p-Anisidine hydrochloride See above Red spots; do not fluoresce under UV
See above
Periodate-p-anisidine See above Red spots on brown background; do not
See above fluoresce (distinction from spots given by
neutral sugars)
Mixed indicators Spray
Thymol blue (0.025%, m/v) and bromothymol blue (0.06%, m/v) Uronic acids and oligomers (e.g.
in 95% ethanol; 1 mol L\1 NaOH added until blue-green colour oligogalacturonic acids) give red spots on
reached green background
Table 2 Conditions recommended for GC analysis of sugar derivatives

Compounds Derivatives for GC Column type Phase Temperature (3C) Gas; flow rate
(mL min\1)

Acetals, isopropylidene Run as such Packed (1) OV-225, 3% on Supelcoport 90P1903C at 43C min\1 N2; 20
(100}120 mesh)
(2) ECNSS-M, 3% on Gas-Chrom
Q (100}120 mesh); 1 and
2 mixed 7 : 4

Alditols Acetates Packed OV-225, 3% on Chromosorb 210 He; 40

W-HP (80}100 mesh)
Capillary OV-225 100P2503C at 43C min\1 He; 1

Aminodeoxyhexoses Alditol acetates, deaminated Capillary Silar 10C 1903C (4 min); H2; 9
190P2303C at 43C min\1;
2303C (8 min)
Trifluoroacetates Packed OV-101, 5% on Chromosorb W 120 N2; 50
AW DMCS (60}80 mesh)
Capillary OV-225 703C (2 min); 70P1803C at N2; 1.5
53C min\1; 1803C (15 min)

Aldonic and aldaric acids TMS ethers Packed OV-1 or OV-17, 0.5% on 160 N2; 30
Chromosorb G (100}120 mesh)
Capillary OV-101 1003C (2 min); 100P2003C at H2; 2
203 min\1; 2003C (5 min)

Aldonolactones TMS ethers Packed OV-1 or OV-17, 0.5% on 160 N2; 30

Chromosorb G (100}120 mesh)
Acetylated N-alkylaldonamides (1-propyl or Packed SP-2340, 3% on Supelcoport 190P2603C at 53C min\1 He; 20
1-hexyl usual alkyl substitutents) (100}120 mesh)
Capillary SP-2330 2003C (2 min); 200P2353C at He; 10
33C min\1; 2353C (5 min)
Capillary DB-1701 (bonded phase) 220P2703C at 13C min\1 He; 12
(fused silica)

Aminodeoxyhexoses Alditol acetates, N-methylated Packed EGSS-X, 2% on Chromosorb 195 N2; 45

W AW DMCS (60}80 mesh)
Alditol acetates, deaminated Packed SP-2340, 3% on Supelcoport 150P2203C at 23C min\1 N2; 40
(100}120 mesh)
Capillary Silar 10C 1903C (4 min); 190P2303C at H2; 9
43C min\1; 2303C (8 min)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

Alditols, trifluoroacetylated Packed OV-101, 5% on Chromosorb 120 N2; 50

WAW DMCS (60}80 mesh)
4271
Table 2 Continued
4272

Compounds Derivatives for GC Column type Phase Temperature (3C) Gas; flow rate
(mL min\1)

Amino- and Methyl glycosides, trifluoroacetylated Capillary OV-210 120P2103C at 13C min\1 Ar; 1
acetamidodeoxy-hexoses
SE-30 903C (4 min); 90P2503C at He; 1.5
83C min\1
Methyl glycosides, trimethylsilylated Packed SE-30, 3% on Chromosorb W-HP 80P2503C at 23C min\1 N2; 25
(100}120 mesh)
Capillary CP-Sil 5 1403C (2 min); 140P2603C at He; 1
(fused silica) 83C min\1
O-Methyloximes, acetylated Capillary OV-1 1753C (4 min); 175P2603C at He; 0.5
43C min\1; 2603C (5 min)
O-Methyloximes, trimethylsilylated Capillary SP-2100 180 He; 1
(fused silica)

Anhydroalditols TMS ethers Capillary CP-Sil 5 130P2203C at 23C min\1 N2; 1.5
(fused silica)

Cyclitols Trifluoroacetates Packed Dexsil 410, 3% on Chromosorb 1003C (1.5 min); 100P3103C N2; 20
W-HP (80}100 mesh) at 3.53C min\1 (3.5 min),
63C min\1 (5 min), 153C min\1
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

(5 min); 253C min\1 (4 min);

3103C (6 min)
TMS ethers Packed SE-30, 3% on Gas-Chrom Q 130P1903C at 23C min\1 N2; 25
(80}100 mesh)

Methyl ethers Alditol acetates Packed OV-225, 3% on Chromosorb 175 He; 40

W-HP (80}100 mesh)
Capillary DB}225 (bonded phase) 195 He; 1
(fused silica)
Capillary OV-275 165P2153C at 23C min\1; He; 1.5
2153C (15 min)
Capillary OV-275 (bonded phase) 150P2503C at 43C min\1; He; 0.8
(fused silica) 2503C (10 min)

Methyl glycosides Trifluoroacetates Capillary SE-30 903C (4 min); 90P2503C at He; 1.5
83C min\1
TMS ethers Packed SE-30, 3% on Chromosorb W-HP 80P2503C at 23C min\1 N2; 25
(100}120 mesh)
Capillary CP-Sil 5 1403C (2 min). 140P1603C at He; 1
(fused silica) 83C min\1
Capillary DB-5 (bonded phase) 150P2203C at 23C min\1 N2; 1
(fused silica)
Table 2 Continued

Compounds Derivatives for GC Column type Phase Temperature (3C) Gas; flow rate
(mL min\1)

Methyl glycosides, Run as such Packed Ethylene glycol succinate, 155 He; 40
methylated polyester, 14% on Chromosorb
W (80}100 mesh)

Muramic acid, KDO and O-Methyloximes, acetylated Capillary OV-1 1753C (4 min); 175P2603C at He; 0.5
neuraminic acid derivatives 43 min\1; 2603C (5 min)

Oligosaccharide-alditols Permethylated Packed Dexsil 300, 1% on Supelcoport 150P3203C at 43C min\1 He; 40
(100}120 mesh)
Permethylated, N-trifluoracetylated Capillary OV-101 2003C (2 min); 200P3503C at He; 0.8
43C min\1

Uronic acids Methyl glycoside methyl esters, Capillary CP-Sil 5 1403C (2 min); 140P2603C at He; 1
trimethylsilylated (fused silica) 83C min\1
Capillary DB-5 (bonded phase) 150P2203C at 23C min\1 N2; 1
(fused silica)
O-Methyloximes, trimethylsilylated Capillary SP-2100 180 He; 1
(fused silica)
Aldonolactones, trimethylsilylated Packed OV-1 or OV-17, 0.5% 160 N2; 30
on Chromosorb G (100}120 mesh)
N-Alkylaldonamides, acetylated Packed SP-2340, 3% on Supelcoport 190P2603C at 53C min\1 He; 20
(100}120 mesh)
Capillary SP-2330 2003C (2 min); 200P2353C at He; 10
33C min\1; 2353C (5 min)

Capilary DB-1701 (bonded phase) 220P2703C at 13C min\1 He; 12

(fused silica)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY
4273
4274 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
partitioning on a cation exchange resin with an aque-
ous}organic solvent system as eluent; cation
exchange chromatography; or ion chromatography.
Uronic acids, on the other hand, are best analysed by
anion exchange chromatography or ion-moderated
partitioning on a cation exchange resin in the
H# form. The same applies to aldonic acids and
aldonolactones. Oligogalacturonic acids are similarly
analysed, but ion chromatography and ion pair
chromatography (with the tetrabutylammonium ca-
tion in the mobile phase) are further options in this
case. The ion pair method has been applied to both
normal oligogalacturonic acids and the unsaturated
products (with 4,5-unsaturated residues at their non-
reducing termini) given on digestion of pectic acid
with endo-polygalacturonic acid lyase. The un-
saturated acids obtained on lyase digestion of
glycosaminoglycuronans can also be analysed by this
method, as well as by anion exchange chromatogra-
phy and ion-moderated partitioning on a cation ex-
change resin with an aqueous}organic solvent system.
The various LC systems applicable to analysis of
carbohydrate derivatives are listed in Table 3. Reten-
tion data have been published elsewhere.
Electrochemical Methods Linked to LC
The pulsed amperometric detector, in which analytes
are oxidized at the surface of a gold electrode, a se-
lected potential being applied between this and a sil-
ver/silver chloride reference electrode, with a glassy
carbon counterelectrode maintaining the potential
without excessive drain on the reference electrode,
has proved highly successful when applied in ion
chromatography of carbohydrates at high pH
(512). Not only neutral sugars but also alditols,
amino- and acetamidodeoxyhexoses, neuraminic acid
derivatives and uronic acids can be analysed in this
way. If the concentration of NaOH in the eluent is
too low for optimal response of the detector, post-
column addition of NaOH at higher concentration is
required; an example of this is the analysis of amino-
and acetamidodeoxyhexoses, which are best resolved
with eluents containing 10}15 mmol L\1 sodium hy-
droxide, but are only detected satisfactorily after ad-
dition of 0.3 mol L\1 sodium hydroxide to the
column efSuent. The method is applicable to
oligosaccharides, including the complex series, neu-
tral, sialylated or phosphorylated, derived from
glycoconjugates, and is now extensively used in ana-
lysis of such oligosaccharides.
It is only readily oxidizable compounds that can be
analysed by oxidation at the surface of a glassy car-
bon electrode, and this permits the determination of
L-ascorbic acid in the presence of other carbohydrates
that are not electroactive with this electrode. Exam-
ples include the analysis of algal extracts for L-ascor-
bic acid and its C5 diastereoisomer, D-erythorbic
acid, at nanogram levels, after LC on a microparticu-
late cation exchange resin (H# form), eluted with
0.1 mol L\1 formic acid; co-eluting reducing sugars
and lactones do not interfere when the carbon elec-
trode is used as a detector. The use of this electro-
chemical detector has also proved invaluable in the
determination of L-ascorbic acid in beers, to which it
is added as an antioxidant; in a recommended pro-
cedure the glassy carbon electrode is used as a de-
tector in LC of the beer samples on C18-silica, eluted
with a citrate buffer (pH 4.4) containing N-
methyldodecylamine (1 mmol L\1) as an ion-pairing
reagent. The detection limit for ascorbic acid is
about 1 ng.
Conductivity detectors can be used in the analysis
of charged molecules. An example is afforded by the
simultaneous determination of inositol phosphates,
sugar phosphates and aliphatic organic anions such as
pyruvate, lactate and citrate in physiological samples
(rat brain and liver) by ion chromatography with
conductivity detection. A post-column micromem-
brane suppressor, continually regenerated with dilute
sulfuric acid, replaces the sodium ions in the eluent
(NaHCO3}Na2CO3; see Table 3) with hydrogen ions,
thus removing the eluent anions by conversion to
carbon dioxide and water. This method permits de-
tection of phosphates in the range 20}100 pmol.
Supercritical Fluid Chromatography
Carbon dioxide, widely regarded as the most useful
mobile phase for supercritical Suid chromatography
(SFC) is a poor solvent for polar solutes and those
having high molecular mass. For this reason such
solutes require derivatization to nonpolar products
before analysis by SFC is possible. In the carbo-
hydrate Reld the main successes of the method have
been its application to series of homologous oligosac-
charides, such as the maltodextrins, as their per-
methylated or trimethylsilylated derivatives, and to
permethylated glycoconjugates. Coupled to chemical
ionization mass spectrometry (CI-MS), SFC affords
a sensitive analytical method (with detection limits at
the picomole level) in such applications as monitoring
of degradation of polysaccharides (e.g. starch) and
proRling of glycoconjugates. With ammonia as
the reactant gas for CI-MS, selected-ion monitoring
of the [M#NH4]# ions as the analytes emerge from
the SFC column permits sensitive detection of
derivatized glucooligosaccharides to a degree of pol-
ymerization (DP) of 15 and above; for the glycocon-
jugate derivatives the molecular mass limit is not in
Table 3 LC systems applicable to analysis of carbohydrate derivatives

Compounds Column packing Mobile phase Temperature (3C) Detection system

Acetates Silica n-Hexane}acetone (10 : 1) or n-hexane}ethyl acetate RT RI or UV

(1 : 1)

Acetylated C18-silica Acetonitrile}water, linear gradient, 10P70% 65 UV

oligosaccharides acetonitrile (80 min)
(to DP 30}35)
Alditols NH2-silica Acetonitrile}water (4 : 1) RT RI
Amine-modified silica (impregnated with Acetonitrile}water (3 : 1), containing RT RI
tetraethylenepentamine) tetraethylenepentamine (0.02%)
Cyclodextrin-bonded silica Acetonitrile}water (4 : 1) RT RI
Cation exchange resin (Ca2# form) Water 65 RI
Cation exchange resin (Pb2# form) Ethanol}water (1 : 4) 80 RI
Anion exchange resin (quaternary 0.50 mol L\1 borate buffer, pH 7.1 (35 min) 65 Photometric or fluorimetric automated
ammonium type) 0.35 mol L\1 borate buffer, pH 8.1 (30 min) periodate}petane-2,4-dione method
0.50 mol L\1 borate buffer, pH 10.5 (25 min)
Pellicular anion exchanger used in ion 0.15 mol L\1 NaOH RT Pulsed amperometric
chromatography

Aldonic and Cation exchange resin (H# form) 4.5 mmol L\1 H2SO4 25 UV
aldaric acids Anion exchange resin (quarternary 0.2 mol L\1 ammonium formate, pH 3.2 45 UV
ammonium type) For aldaric acids:
0.16 mol L\1 NaCl containing MgCl2 (20 mmol L\1) 85 UV

Amino and NH2-silica Acetonitrile}0.15 mmol L\1 phosphate buffer, pH 5.2 RT UV

acetamidodeoxy- (4 : 1)
hexoses and
-hexitols
Cation exchange resin (H# form) Acetonitrile}water (23 : 2) 30 UV, automated 2-cyanoacetamide method
Cation exchange resin (Na# form) used 0.1 mol L\1 sodium citrate, pH 7.2 40 (15 min); Photometric, automated ninhydrin method
in amino acid analyser 63 (45 min)
20 mmol L\1 Na2B4O7, pH 8.0 60 Fluorimetric, automated o-phthalaldehyde
method
Pellicular anion exchanger used in ion 10 mmol L\1 NaOH; post-column addition of RT Pulsed amperometric detector
chromatography 0.3 mol L\1 NaOH to raise pH to optimum for detection
method

As benzoylated Silica n-Hexane}dioxane}dichloromethane, linear gradient, RT UV

hexitols 22 : 2 : 1P4 : 2 : 1 (80 min)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

As pyridylamino C18-silica 0.25 mol L\1 sodium citrate buffer, pH 4.0, containing RT Fluorimetric
derivatives acetonitrile (1.0%)
4275
Table 3 Continued
4276

Compounds Column packing Mobile phase Temperature (3C) Detection system

Ascorbic acid Cation exchange resin (H# form) 4.5 mmol L\1 H2SO4 or 0.1 mol L\1 HCOOH 25 UV

30 Amperometric (glassy carbon electrode)

C18-silica 25 mmol L\1 sodium citrate buffer, pH 4.4, containing RT Amperometric, as above
N-methyldodecylamine (1 mmol L\1)

Cyclitols Amine-modified silica (impregnated with Acetonitrile}water (3 : 1), containing RT RI

tetraethylenepentamine) tetraethylenepentamine (0.02%)
Cation exchange resin (Ca2# form) Water 85 RI

Gangliosides NH2-silica (A) Acetonitrile}5 mmol L\1 phosphate buffer, pH 5.6 RT UV

(83 : 17)
(B) Acetonitrile}20 mmol L\1 phosphate buffer, pH 5.6
(1 : 1)
Gradient elution:
Solution A (7 min); APA#B (33 : 17) in 53 min;
PA#B (9 : 16) in 20 min

Benzoylated Silica n-Hexane}dioxane, linear gradient, 7P23% dioxane RT UV

(18 min)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

Glycolipids, Silica n-Hexane}dioxane, linear gradient, 2.5P25% dioxane RT UV

benzoylated (13 min); isocratic (5 min)

Glycosides, Silica Acetonitrile}water (9 : 1) RT RI

methyl C18-silica Water RT RI
Cation exchange resin (Ca2# form) Water 1.5 RI

Benzoylated C18-silica Acetonitrile}water, linear gradient, 35P90% RT UV

acetonitrile (65 min)

Gycosides, other
Acetylated Silica Benzene}ethyl acetate (9 : 1) RT UV
For aryl glycosides:
Chloroform}carbon tetrachloride (3 : 2)
C18-silica Methanol}water (13 : 7) RT UV

Benzoylated Silica Benzene}ethyl acetate (99 : 1) RT UV

Lactones Cation exchange resin (H# form) 4.5 mmol L\1 H2SO4 25 UV

Methyl ethers, as C18-silica Water}acetonitrile (99 : 1) RT RI

alditols
Table 3 Continued

Compounds Column packing Mobile phase Temperature (3C) Detection system

Neuraminic acid Anion exchange resin (quaternary 0.75 mmol L\1 Na2SO4 RT UV
derivatives, KDO ammonium type) For KDO disaccharides:
and derivatives 10 mmol L\1 Na2SO4
Neuraminic acids, C18-silica Water}methanol}acetonitrile (77 : 15 : 8) RT Fluorimetric
DDB derivatives
Oligogalacturonic Cation exchange resin (H# form) 5 mmol L\1 H2SO4 85 RI
acids, to DP10
DP 5-20 Pellicular anion exchanger used in ion 0.15 mol L\1 NaOH, with sodium acetate, gradient: RT Pulsed amperometric
chromatography 0.40 mol L\1 (2 min); 0.40P0.90 mol L\1 (43 min)
Oligosaccharides, NH2-silica Acetonitrile}water (7 : 3) 25 RI
chitin, to DP 5
Oligosaccharides, NH2-silica Acetonitrile}15 mmol L\1 phosphate buffer, pH 5.2 RT UV
from (4 : 1): isocratic (30 min); buffer 20P45% (50 min)
glycoproteins
For higher oligosaccharides : (8}12 sugar residues):
Linear gradient: buffer 20P44% (80 min)
Pellicular anion exchanger used in ion For neutral oligosaccharides: (2}11 sugar residues): RT Pulsed amperometric
chromatography (A) 0.10 mol L\1 NaOH
(B) 0.10 mol L\1 NaOH containing sodium acetate
(0.15 mol L\1)
Gradient elution: A (10 min) APA#B (1 : 4) in 60 min
Post-column addition of 0.3 mol L\1 NaOH to raise pH
for detection method
For sialylated oligosaccharides
(3}8 sugar residues):
50 mmol L\1 NaOH containing 100 mmol L\1 sodium
acetate
Oligosaccharides NH2-silica 0.1 mol L\1 KH2PO4, pH 4.75 RT UV
(2}8 residues),
from hyaluronic
acid
Oligosaccharide-
alditols, from
glycoproteins,
2}7 residues NH2-silica Acetonitrile}1 mmol L\1 KH2PO4, pH 5.4 (3 : 2) RT UV
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

2}20 residues Size exclusion chromatography packing Water 55 RI or scintillation counting after labelling
with 3H
4277
Table 3 Continued
4278

Compounds Column packing Mobile phase Temperature (3C) Detection system

Oligosaccharide- C18-silica Acetonitrile}water, various proportions from 1 : 1 to RT RI, MS

alditols, ethylated 3 : 2, or linear gradient, 50P65% acetonitrile (45 min)
and methylated
(2}6 residues)
Oligosaccharides, Two-dimensional mapping: For column 1: 55 Fluorimetric
pyridylamino (1) C18-silica (A) 10 mmol L\1 phosphate buffer, pH 3.8;
derivatives (2}20 (B) A containing 1-butanol (0.5%)
residues) Linear gradient, 20P50%B (60 min)
(2) NH2-silica For column 2: 40 Fluorimetric
(C) Acetonitrile}3% acetic acid in water containing
triethylamine, pH 7.3 (13 : 7)
(D) As C, but proportions 1 : 1
Linear gradient, CPD (50 min)

Phosphates Pellicular anion exchanger used in ion 2.4 mmol L\1 NaHCO3}1.92 mmol L\1 Na2CO3 RT Conductivity anion micromembrane
chromatography suppressor
C18-silica For monophosphates: 20 mmol L\1 HCOOH, 38 UV photometry of adduct with Eu complex
containing tetrabutylammonium hydroxide (8 mmol L\1)
as ion-pairing reagent and Eu complex (0.02 mmol L\1)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

as detection reagent
For diphosphates:
20 mmol L\1 HCOOH}20 mmol L\1 HCl}40 mmol L\1
NaCl; concentration of tetrabutylammonium hydroxide
increased to 30 mmol L\1, that of Eu complex
unchanged

Phosphorylated Pellicular anion exchanger used in ion 0.1 mol L\1 NaOH (5 min); linear acetate gradient, RT Pulsed amperometric
oligosaccharides chromatography 0P0.6 mol L\1 sodium acetate in 0.1 mol L\1 NaOH
(2}5 sugar (67 min); isocratic (5 min)
residues)

Unsaturated
oligosaccharides
(2}7 residues),
from lyase
digestion of:

Alginate C18-silica Acetonitrile}0.1 mol L\1 phosphate buffer, pH 6.5 RT UV

(1 : 9), containing tetrabutylammonium hydroxide
(10 mmol L\1)
Table 3 Continued

Compounds Column packing Mobile phase Temperature (3C) Detection system

Pectic acid C18-silica Methanol}0.05 mol L\1 phosphate buffer, pH 7.0 (3 : 7), 40 UV
containing tetrabutylammonium bromide (25 mmol L\1)
Anion exchanger (quaternary ammonium) 0.4 mol L\1 sodium acetate buffer, pH 5.4 40 UV
bonded to silica

Hyaluronic acid C18-silica (A) Acetonitrile}8 mmol L\1 H3PO4 (1 : 4), pH of RT UV

mixture 7.6; contains tetrabutylammonium
hydroxide (10 mmol L\1)
(B) Acetonitrile}6 mmol L\1 H3PO4 (3 : 2), pH of RT UV
mixture 7.5; concentration of ion-pairing reagent as
in A
Linear gradient, APA#B (19 : 1) in 18 min

Unsaturated Anion exchanger (quarternary For oligosaccharides to hexasaccharide: Linear RT UV

sulfated ammonium) bonded to silica gradient, 0.2P0.8 mol L\1 NaCl, pH 3.5 (50 min)
oligosaccharides,
from lyase
digestion of
glycosamino-
glyronans
NH2-silica Disaccharides only: 10 mmol L\1 Na2SO4}1 mmol L\1 50 UV
CH3COOH
Cation exchange resin (Na# form) Disaccharides only: Acetonitrile}methanol}0.8 mol L\1 70 UV
ammonium formate buffer, pH 4.5 (13 : 3 : 4)

Uronic acids Cation exchange resin (H# form) 4.5 mmol L\1 H2SO4 25 UV
Anion exchanger (quaternary ammonium) 5 mmol L\1 KH2PO4 (pH 4.6) containing methanol (5%) RT UV
bonded to silica
0.7 mol L\1 CH3COOH 40 RI

RT, Room temperature; RI, refractive index.

III / SUGAR DERIVATIVES: CHROMATOGRAPHY
4279
Table 4 Solvent systems useful in TLC and paper chromatography of sugar derivatives
4280

Derivatives Stationary phase Solvent systema

Acetals and ketals Silica gel Benzene}ethanol (2 : 1)

Benzene}acetic acid}ethanol (2 : 2 : 1)
For pyruvate:
Ethyl acetate}acetic acid}formic acid}water (12 : 3 : 1 : 4) (threefold
development)
HPTLC plates with bonded aminopropyl phase, impregnated Acetonitrile}water (9 : 1)
with NaH2PO4 (0.2 mol L\1)

Acetates Silica gel Benzene}ethyl acetate (7 : 3)

Alditols Paper
Paper impregnated with tungstate (0.15 mol L\1)
CMC paper (La3#, Ca2# or Ba2# forms)
Cellulose plates impregnated with tungstate (0.15 mol L\1)
Silica gel impregnated with NaH2PO4 (0.5 mol L\1)
Aldonic acids and aldonolactones Cellulose plates
Silica gel, HPTLC
Benzene}methanol (9 : 1)
1-Butanol}ethanol}water (4 : 1 : 5), upper layer
2-Butanone}acetic acid}saturated aqueous solution of H3BO3 (9 : 1 : 1)
Acetone}1-butanol}water (5 : 3 : 2)
1-Butanol}ethanol}water (10 : 1 : 2)
Acetone}1-butanol}water (5 : 3 : 2)
2-Propanol}acetone}0.2 mol L\1 lactic acid (6 : 3 : 1)
1-Butanol}acetic acid}water (6 : 1 : 2)
Ethyl acetate}pyridine}acetic acid}tetrahydrofuran}water
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

(50 : 22 : 4 : 15 : 15)
Amino- and acetamidodeoxyhexoses Paper Ethyl acetate}pyridine}acetic acid}water (5 : 5 : 1 : 3)
1-Butanol}pyridine}benzene}water (5 : 3 : 1 : 3)
Cellulose plates 1-Butanol}pyridine}0.1 mol L\1 HCl (5 : 3 : 2)
Two-dimensional development:
(1) 2-Propanol}90% HCOOH}water (20 : 1 : 5);
(2) Lutidine}water (13 : 7)
Silica gel 1-Propanol}water (7 : 1)
Dansyl derivatives Silica gel Cyclohexane}ethyl acetate}ethanol (6 : 4 : 3)
1
Anhydroalditols Silica gel, HPTLC, impregnated with borate (0.1 mol L\ ) 1-Butanol}acetone}water (5 : 4 : 1)
Anhydro sugars HPTLC plates with bonded aminopropyl phase, impregnated Acetonitrile}water (9 : 1)
with NaH2PO4 (0.2 mol L\1)
Branched-chain sugars (apiose, hamamelose Paper 1-Butanol}pyridine}acetic acid}water (60 : 40 : 3 : 30)
and derivatives) 1-Butanol}ethyl acetate}acetic acid}water (8 : 6 : 5 : 8)
Cyclitols Paper Acetone}water (4 : 1)
Gangliosides Silica gel, HPTLC Methyl acetate}2-propanol}33 mmol L\1 KCl (9 : 6 : 4)
Acetonitrile}2-propanol}50 mmol L\1 KCl (10 : 67 : 23)
Acetonitrile}2-propanol}2.5 mol L\1 aqueous ammonia (2 : 13 : 5)
Glycosides, aryl, acetylated Silica gel 2-Butanone}light petroleum (1 : 3)
Table 4 Continued

Derivatives Stationary phase Solvent systema

Glycosides, methyl Paper t-Pentanol}1-propanol}water (8 : 2 : 3)

Cellulose plates Ethyl acetate}pyridine}acetic acid}water (5 : 5 : 1 : 3)
1-Butanol}acetic acid}water (3 : 1 : 1)
Methylated Silica gel Benzene}ethanol}water (170 : 47 : 15), upper layer

Methyl ethers Paper 1-Butanol}ethanol}water (4 : 1 : 5), upper layer

Muramic acid (separated from
aminodeoxyhexoses)
Neuraminic acids, N-acetyl and N-glycolyl
Oligogalacturonic acids (to DP 9)
Oligosaccharides, chitin (to DP 6)
Oligosaccharides, from hyaluronic acid (2}8
residues)
Phosphates
Unsaturated disaccharides, from lyase
digestion of glycosaminoglycuronans, run as
dansylhydrazones
Uronic acids and alduronolactones
2-Butanone}water azeotrope (85 : 7)
Cellulose plates 2-Butanone}saturated with water
Silica gel 2-Butanone}water azeotrope
Benzene}ethanol}water}aqueous ammonia (200 : 47 : 15 : 1), upper
layer
Silica gel impregnated with H3BO3 (0.1 mol L\1) 1-Butanol}acetone}water (4 : 5 : 1)
Cellulose plates Two-dimensional development :
(1) 2-Propanol}90% HCOOH}water (20 : 1 : 5),
(2) Lutidine}water (13 : 7)
Silica gel Acetonitrile}ethanol}acetic acid}water (13 : 2 : 1 : 4)
Silica gel Methanol}water (5 : 2)
Cellulose plates Ethyl acetate}acetic acid}water (2 : 1 : 2), twofold development
Silica gel, HPTLC Ethanol}25 mmol L\1 CH3COOH (21 : 29), 353C
HPTLC plates with bonded aminopropyl phase, impregnated Acetonitrile}water (18 : 7)
with NaH2PO4 (0.2 mol L\1)
Silica gel 2-Propanol}water (33 : 17), containing NaCl (50 mmol L\1)
Paper Methanol}90% HCOOH}water (16 : 3 : 1), containing tetrasodium salt of
EDTA (0.05%, m/v)
Silica gel 1-Propanol}2-propanol}1-butanol}water (6 : 9 : 1 : 4), containing NaCl
(40 mmol L\1) and ammonia (10 mmol L\1)
Paper 1-Butanol}acetic acid}water (2 : 1 : 1)
Ethyl acetate}acetic acid}formic acid}water (18 : 3 : 1 : 4)
Ethyl acetate}acetic acid}pyridine}water (10 : 3 : 3 : 2)
DEAE-cellulose paper Ethyl acetate}acetic acid}water (3 : 1 : 1)
Cellulose plates 1-Butanol}acetic acid}water (6 : 1 : 2)
Silica gel impregnated with NaH2PO4 (0.3 mol L\1) 1-Butanol}ethanol}0.1 mol L\1 H3PO4 (1 : 10 : 5)
III / SUGAR DERIVATIVES: CHROMATOGRAPHY

a
All proportions are by volume.
4281
4282 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
Table 5 Lectins used in affinity chromatography of oligosaccharides
Lectin
Concanavalin A (Con A)
Datura stramonium agglutinin (DSA)
Griffonia simplicifolia
Helix pomatia (HP)
Lens culinaris (lentil)
Phytohaemagglutinin, erythroagglutinating (E4-PHA)
Phytohaemagglutinin, leukoagglutinating (L4-PHA)
Pisum sativum (pea)
Ricinus communis agglutinin (RCA-I)
Sambucus nigra
Wisteria floribunda
excess of 5000. For these the addition of methanol to
the carbon dioxide mobile phase has proved advant-
ageous. Fused-silica microbore capillary columns,
with a bonded methylpolysiloxane stationary phase
(DB-1 and, especially, DB-5 are very effective), are
used at temperatures ranging from 90 to 1203C and
with pressure programming over the range
10}40 MPa (100}400 bar) at about 0.5 MPa min\1
(5 bar min\1). Under these conditions there is resolu-
tion of
and  anomers (more pronounced with the
TMS derivatives) and Rne structure is discernible in
glycoconjugates.
Thin-Layer Chromatography (TLC) and
Paper Chromatography
While nonpolar derivatives can be separated by thin-
layer chromatography (TLC) on unmodiRed silica
plates, resolution of polar molecules is generally poor
unless the silica gel layer is impregnated beforehand
with an inorganic salt capable of interacting with
carbohydrates. Borate or phosphate buffers are most
Specificity

-D-Man, terminal or substituted only at O2; terminal -D-
GlcNAc at O2 promotes binding
[-D-Gal (1P4) -D-GlcNAc (1P3)]n, i.e. poly (N-acetyllac-
tosamine); binds tri- and tetraantennary oligosaccharides
lacking this sequence if outer Man residue is substituted at O2
and O6 by N-acetyllactosamine
Terminal
-D-Gal
Terminal
-D-GalNAc
Terminal
-D-Man: outer Man residue substituted at O2 and
O6 by GlcNAc
Bisecting GlcNAc at O4 of inner Man residue and sequence
-D-Gal (1P4) -D-GlcNAc (1P2)
-D-Man in outer chains
Tri- and tetraantennary oligosaccharides with outer Man resi-
due substituted at O2 and O6 by N-acetylactosamine
Terminal
-D-Man;
-L-Fuc at O6 of 4-linked GlcNAc in inner
core of N-linked oligosaccharide
Terminal -D-Gal

-NeuAc (2P6) Gal 

-NeuAc (2P3) Gal
-D-GalNAc (1P4) Gal'

-D-GalNAc (1P3) Gal and
-D-GalNAc (1P3) GalNAc: sub-
stitution of 4-linked Gal by NeuAc at O3, or of 3-linked Gal by

-L-Fuc at O2 weakens binding
often used for this purpose; tungstate can also prove
effective, especially in TLC of alditols. The same
applies to TLC on high performance (HP) TLC
plates, particularly those carrying a bonded amino-
propyl phase, which is liable to react covalently
with sugars and derivatives containing hydroxyl
groups. Some separations, particularly those of
aminodeoxy sugars, that are not well resolved on
impregnated silica gel plates, are better on unmodi-
Red silica plates. Cellulose plates also give satisfac-
tory resolution of these derivatives, and of neutral
sugars and uronic acids, but two-dimensional devel-
opment is often required. Impregnation of these
plates with tungstate greatly improves their resolving
power for alditols.
Although paper chromatography has largely been
superseded by TLC, there are groups of sugar deriva-
tives that are far better resolved on paper than by TLC
methods. The mixtures of partially methylated sugars
obtained in methylation analysis of polysac-
charides afford a prime example: resolution on
cellulose plates is better than that on silica plates but
 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
paper chromatography remains the most effective
method. As in the case of TLC, separation of alditols
on paper is improved by impregnation of the paper
with tungstate. Papers having ion exchange properties
can also be used to good effect in separations of some
sugar derivatives: uronic acids and aldobiouronic
acids are well resolved on DEAE}cellulose paper (ani-
on exchanger), while carboxymethylcellulose paper,
converted beforehand to the La3#, Ca2# or Ba2#
forms, gives excellent resolution of alditols.
Some solvent systems that have proved effective in
TLC and paper chromatography of sugar derivatives
are listed in Table 4.
Af\nity and Enzyme Methods
Af\nity Chromatography
Lectin afRnity chromatography is a valuable tech-
nique in analyses of glycoconjugates, as the isolation
and identiRcation of glycopeptides and the various
oligosaccharides obtained on removal of the carbohy-
drate side chains from the protein or lipid moieties
are greatly facilitated by chromatography on a series
of short columns, each containing a different lectin
covalently coupled to agarose gel. The lectins are
selected according to their speciRcity towards carbo-
hydrates having certain of the main structural fea-
tures found in the oligosaccharides, and in this way
the complex mixture of oligosaccharides can be frac-
tionated according to structure. Some of the lectins
that have proved useful in such studies are listed in
Table 5, together with their carbohydrate-binding
speciRcities.
The oligosaccharides are usually applied to
the lectin columns in phosphate-buffered saline
(PBS), pH 7.2, Tris-buffered saline (TBS), pH 8.0, or
10 mmol L\1 Tris}HCl buffer, pH 7.5; sodium azide
(0.02%, m/v) is added as a preservative and small
amounts of calcium, magnesium and manganese
chlorides (1 mmol L\1) are essential to the binding
action of some lectins, notably concanavalin A.
Oligosaccharides that are not bound or are only re-
tarded on the lectin column are eluted with these
buffers, but those that are strongly bound require the
addition of a competing hapten to the eluent. Hap-
tens applicable to the lectins listed above include
methyl
-D-mannopyranoside, lactose, GalNAc,
GlcNAc and N,N-diacetylchitobiose.
A recent development in afRnity chromatography is
the use of monoclonal antibodies as ligands; these are
highly speciRc but less strongly reactive than lectins,
and the dissociation constants of the complexes for-
med with bound solutes are sufRciently low to permit
rapid fractionation, the oligosaccharides reacting
4283
with the ligand being merely retarded on the column,
not totally immobilized. An example of the use of this
technique is afforded by the complete separation of
two of the oligosaccharides of human milk,
-
NeuAc(2P3)-D-Gal(1P3)-D-GlcNAc(1P3)-D-
Gal(1P4)Glc (lactosialyltetrasaccharide, LSTa) and
that designated sialyl Lea, which carries
-L-Fuc at O4
of GlcNAc. On a short column containing mon-
oclonal antibody 19.9 coupled to agarose gel, with
10 mmol L\1 Tris}HCl buffer, pH 7.5, as eluent, the
two oligosaccharides are rapidly separated, the
fucosylated sialyl Lea being the more retarded.
The active oligosaccharides of blood group A are
similarly fractionated according to chain lengths and
degree of fucosylation by chromatography on immo-
bilized IgM antibody, with TBS as eluent. Use of
columns in which such antibodies are coupled to
microparticulate silica makes possible very rapid sep-
arations of oligosaccharides (in 20 min or less). This
new technique of high performance liquid afRnity
chromatography (HPLAC) has great potential in ap-
plications such as clinical analysis, for which methods
that are highly speciRc but also efRcient are required.
Enzyme Methods
Enzyme methods are particularly useful in analyses of
glycoconjugates, for the release of mono- or oligosac-
charides that are not easily liberated by acid hydroly-
sis or are acid-labile, and in the determination of
some constituents. The determination of neuraminic
acid derivatives in glycoproteins or glycolipids is
a striking example of this use of enzymes. The sample
(&200
g), dissolved in 60 mmol L\1 phosphate buf-
fer (pH 7.0, 800
L), is incubated at 373C for 1 h
with Clostridium perfringens neuraminidase (EC
3.2.1.18) and N-acylneuraminate pyruvate-lyase (EC
4.1.3.3). The former (0.5 U) liberates the neuraminic
acid derivatives from glycosidic linkages and the lat-
ter (0.3 U) cleaves the molecules to produce N-acyl-
mannosamines and pyruvate. The mannosamine
derivatives are well separated from GlcNAc, GalNAc
and neutral sugar components of glycoconjugates by
LC (H# form cation exchange resin, 92% acetonit-
rile in water; see Table 3) and may be determined in
this way. Alternatively (or in addition), the propor-
tion of neuraminic acids may be found by determin-
ing the pyruvate released, using the deRnitive lactate
dehydrogenase method.
For release of N-linked oligosaccharides from
glycoproteins, digestion with N-oligosaccharide glyco-
peptidase (EC 3.5.1.52) offers a milder alternative to
the standard hydrazinolysis procedure. After pepsin
digestion of the protein moiety, the product, dis-
solved in 0.1 mmol L\1 citrate}phosphate buffer,
4284 III / SUGAR DERIVATIVES: CHROMATOGRAPHY
is digested with the glycopeptidase (1 mU per
1000 nmol of oligosaccharides) at 373C for 15 h.
For sequencing purposes, smaller oligosaccharides
may be obtained by subsequent digestion with vari-
ous exoglycosidases, such as
-L-fucosidase (EC
3.2.1.51), -D-galactosidase (EC 3.2.1.23) and
-N-acetylglucosaminidase (EC 3.2.1.30). The mix-
tures of oligosaccharides are separated by LC (see
Table 3).
Enzyme methods are also important in the analysis
of glycosaminoglycuronans, which are very resistant
to acid hydrolysis. Hyaluronidase (EC 3.2.1.35) ran-
domly cleaves the (1P4) bonds linking the acet-
amidodeoxyhexose residues to glucuronic acid in
both hyaluronic acid and the chondroitin sulfates, to
yield the disaccharide repeating unit and oligomers. An
exception to this is leech hyaluronidase (EC 3.2.1.36),
which speciRcally cleaves the -D-GlcA (1P3)-D-
GlcNAc linkages in hyaluronic acid, yielding a differ-
ent series of oligomers. All of these, including some
with odd numbers of sugar residues, obtained by
removal of the nonreducing GlcA end-groups with
-glucuronidase (EC 3.2.1.31) or of nonreducing
GlcNAc end-groups with -N-acetylglucosaminidase,
are well separated by LC (see Table 3).
A sensitive analytical method for glycosamino-
glycuronans is afforded by LC of the unsaturated
oligosaccharides produced on digestion with enzymes
having lyase activity, which give disaccharides or, in
the case of hyaluronic acid, tetra- and hexasacchar-
ides with 4,5-unsaturated residues (4-deoxy-L-threo-
hex-4-enopyranosyluronic acid from D-glucuronic
acid or L-iduronic acid) at their nonreducing ends.
Chondroitinase ABC (EC 4.2.2.4) digests chondroitin
4- and 6-sulfate and dermatan sulfate, whereas chon-
droitinase AC (EC 4.2.2.5) does not act upon derma-
tan sulfate, and LC analysis (Table 3) of the mixtures
of unsaturated, sulfated disaccharides produced by
each enzyme permits quantiRcation of the respective
parent glycosaminoglycuronans. Typically, the pro-
teoglycan (1}1000
g) is digested at 373C for 16 h
with the enzyme (0.05 U) in Tris buffer (pH 6.0).
Hyaluronic acid can be determined speciRcally by LC
analysis of the unsaturated tetra- and hexasaccharide
produced on digestion with the lyase from Streptomy-
ces hyalurolyticus (H-1136), which cleaves this poly-
mer selectively. Recently the LC proRles of the
products of digestion of heparin with heparin lyase
(EC 4.2.2.7), from Flavobacterium heparinum, have
been suggested as a means of characterizing this poly-
disperse glycosaminoglycuronan: di-, tetra- and hexa-
saccharides, differing in degree of sulfation and
proportion of iduronic acid, are produced, their pro-
portions in the mixture varying with the source of the
heparin.
Structural analysis of alginates, which contain
blocks of mannuronic acid and guluronic acid resi-
dues, all (1P4) linked, is facilitated by the use of
enzymes acting exclusively on one of these acids,
leaving intact blocks of the other. These enzymes are
lyases, producing unsaturated oligosaccharides from
the portions of the polymer that they attack. For
example, a -D-mannuronase has been isolated from
actively growing tissues of the seaweed Sargassum
Uuitans and an
-L-guluronase from the bacterium
Klebsiella aerogenes type 27. Digestion may be
monitored by LC analysis of the unsaturated
oligosaccharides (Table 3). This applies also to diges-
tion of pectic acid with endo-polygalacturonic acid
lyase (EC 4.2.2.10). The saturated oligogalacturonic
acids produced on digestion of this polymer with
endo-polygalacturonase (pectinase: EC 3.2.1.15) are
also analysed by LC.
Speci\c Problems: Analysis of Acidic
Derivatives
Whereas the enzyme methods discussed above are
used in the degradation of glycoproteins and
glycolipids, which contain sugar derivatives } such as
the neuraminic acid derivatives } that are unstable
when heated in acid, and of glycosaminoglycuronans
and polyuronans, which are strongly resistant to acid
hydrolysis, it is the latter technique that is most wide-
ly used to liberate the constituent monosaccharides
from other heteropolysaccharides. For those contain-
ing aldobiouronic acid linkages, which are far less
readily hydrolysed than are glycosidic linkages be-
tween neutral sugars, slow release and low yields of
both hexuronic acids and the contiguous (interior)
sugar residues make quantiRcation difRcult. The use
of vigorous conditions or prolonged exposure to acid
in attempting to improve the yields of these constitu-
ents is liable to cause both decarboxylation of the
acid and decomposition of some of the neutral sugars
already liberated (pentoses being especially vulner-
able). For quantitative GC analysis, the difRculty can
be obviated by prior reduction of the carboxyl groups
in the uronic acid compounds. This is best effected
by treatment with a carbodiimide at pH 4.75,
followed by reduction with sodium borohydride or
borodeuteride at pH 7.0; if the latter is used, the
hexoses produced from the hexuronic acids are label-
led with deuterium and thus readily identiRable by
GC-MS of the derived alditol acetates.
In methylation analysis, the problems posed by
resistance to acid hydrolysis of linkages involving
methylated uronic acid residues (present as methyl
esters) are similar. In this case the recommended
procedure is reduction of the carboxylate ester groups
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
with lithium aluminium deuteride in dry oxolane
(tetrahydrofuran) at 703C for 16 h. The ester residue
is then converted to a 6,6-dideuteriohexose residue,
the O-methyl ethers of which are easily distinguish-
able by GC-MS of the derived alditol acetates.
An alternative to acid hydrolysis that is applicable
to most polysaccharides and glycoconjugates, includ-
ing those containing acid-labile residues or glycosidic
linkages resistant to hydrolysis, is afforded by meth-
anolysis, in which the sample is heated in methanolic
HCl, the conditions employed depending upon the
nature of the sugar residues present. After suitable
derivatization, all components of methanolysates,
now present as methyl glycosides or, in the case of
hexuronic acids, methyl glycoside methyl esters, can
be analysed simultaneously, either by GC (Table 2)
or by LC (Table 3). The procedure is also applicable
to methylation analysis, the methylated methyl glyco-
sides and methyl glycoside methyl esters being amen-
able to GC without further derivatization (Table 2).
See Colour Plate 118.
See also: II/Chromatography: Paper Chromatography.
Chromatography: Gas: Derivatization; Detectors: Mass
Spectrometry. Chromatography: Liquid: Derivatization.
Chromatography: Thin-Layer (Planar): Spray Reagents.
III/Impregnation Techniques: Thin-Layer (Planar)
Chromatography. Polysaccharides: Centrifugation;
Liquid Chromatography.
SULFUR COMPOUNDS: GAS
CHROMATOGRAPHY
W. Wardencki, Technical University of GdanH sk,
GdanH sk, Poland
Copyright ^ 2000 Academic Press
Introduction
Sulfur compounds, of both biogenic and anthropo-
genic origin, constitute a large group of compounds,
ranging from simple gases up to complex polycyclic
aromatics. These compounds can be present in vari-
ous, usually complex matrices, such as air (gaseous),
water systems (aqueous), various petroleum fractions
(gaseous, liquid and solid), in beverages and food-
stuffs and in pharmaceutical formulations.
Environmentalists believe that these compounds
are responsible for the damage of our environment
through acid deposition, rapid acidiRcation of lakes,
4285
Further Reading
Churms SC (1982) CRC Handbook of Chromatography:
Carbohydrates, vol. I. Boca Raton, FL: CRC Press.
Churms SC (1991) CRC Handbook of Chromatography:
Carbohydrates, vol. II. Boca Raton. FL: CRC Press.
Churms SC (1996) Recent progress in carbohydrate separ-
ation by high-performance liquid chromatography based
on hydrophilic interaction. Journal of Chromatography A
720: 75}91.
Dey PM (ed.) (1990) Methods in Plant Biochemistry, vol. 2,
Carbohydrates. London: Academic Press.
El Rassi Z (ed.) (1995) Carbohydrate Analysis. High Per-
formance Liquid Chromatography and Capillary Elec-
trophoresis. Amsterdam: Elsevier Science.
Ginsburg V (ed.) (1987) Lectin afRnity chromatography of
glycopeptides. Methods in Enzymology 138: 232}259,
AfRnity puriRcation of oligosaccharides using monoclonal
antibodies. Methods in Enzymology 138: 307}313.
Ginsburg V (ed.) (1989) Analysis of complex oligosacchar-
ides from glycoconjugates by afRnity chromatography
and high-performance anion-exchange chromato-
graphy. Methods in Enzymology 179: 30}82.
Reinhold VN, Sheeley DM, Kuei J and Her GR (1988)
Analysis of high molecular weight samples on a double-
focusing magnetic sector instrument by supercritical
Suid chromatography/mass spectrometry. Analytical
Chemistry 60: 2719}2722.
Whistler RL, Wolfrom ML, BeMiller JN and ShaRzadeh
F (eds) (1962) Methods in Carbohydrate Chemistry, vol.
I. New York: Academic Press.
Whistler RL and BeMiller JN (eds) (1976) Methods in
Carbohydrate Chemistry, vol. VII. New York: Aca-
demic Press.
the loss of forests, the corrosion of metal structures
and historical monuments. The interest in bio-
geochemistry results from the role some sulfur com-
pounds play in global chemical cycles. Dimethyl
sulRde (DMS) in sea water, produced in the oceans, is
believed to play a critical role in the global sulfur
cycle and the radiation balance of the Earth. Also,
other sulfur compounds may contribute signiRcantly
to the sulfur Sux in the atmosphere. In foods, bever-
ages and in water, trace levels of sulfur-containing
compounds are responsible for taste and odour prob-
lems. They are also the source of malodorous con-
ditions in municipal sewage systems. ReRners
worldwide give particular attention to these com-
pounds because in petrochemical and chemical ap-
plications even trace levels of sulfur impurities may
cause concern. They can poison the catalysts, impart
4286 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
undesirable properties to Rnal products or produce
general air pollution when fuel is burnt.
For these reasons sulfur-containing compounds are
of constant concern in many Relds. Gas chromatogra-
phy, due to combination of separation capability and
sensitive detection is still a prime technique for the
analysis of these compounds in various matrices.
General Problems of the
Determination of Sulfur Compounds
by Gas Chromatography
The analysis of sulfur compounds in different envir-
onmental matrices is still a big challenge for the
analytical chemist. The main difRculties in their de-
terminations are related to the two main obstacles.
The Rrst is common with general problems encoun-
tered in trace analysis. Most of these compounds are
present at low concentrations, frequently at the low
parts per trillion (ppt) level. They may be encountered
in very complex matrices and in a broad range of
concentrations (often several orders of magnitude).
Complex mixtures can cause interference problems
between major and minor constituents.
The second difRculty is due to the highly reactive
nature of sulfur compounds. It is well known that
these compounds have absorptive, adsorptive, photo-
oxidative and metal catalytic oxidative features. This
can lead to irreversible adsorption, reaction with each
other, catalytic reactions, rearrangements catalysed
by different materials and reactions with substances
they come into contact with. Because of these reasons
special precautions should be undertaken during all
steps of their analysis, e.g., during sample treatment
(sampling, storage, pre-concentration and isolation)
as well as during the gas chromatographic analysis.
When sulfur content is relatively high (up to per-
centage level) and the matrix is very complex, like
crude oil, direct GC analysis can be frequently done,
reducing the analysis time and eliminating the possi-
bility of analyte losses. Such samples, due to possible
interference problems, require very effective separ-
ation systems and very selective (speciRc) detectors.
The choice of a detector with high selectivity for
sulfur over hydrocarbon is crucial.
Due to the diversity of the matrices in which sulfur
compounds can be present it is convenient to discuss
each type of sample separately.
Atmospheric Sulfur Gases
Sulfur gases are released into the atmosphere from
various natural and anthropogenic sources. The most
abundant atmospheric sulfur compounds are: hydro-
gen sulRde (H2S), carbonyl sulRde (COS), dimethyl
sulRde (DMS), dimethyldisulRde (DMDS), carbon
disulRde (CS2) and methanethiol (methyl mercaptan
} MeSH). These compounds have received a great
deal of attention because of the suggestion that the
emission of natural compounds may be substantial
even compared to anthropogenic sources of sulfur
dioxide (SO2). Frequently, all these compounds are
called reduced sulfur compounds, S(-II), abbreviated
to RSCs. These compounds, together with other sul-
fur species with boiling points up to ca. 2003C, are
usually termed volatile sulfur compounds (VSCs).
Considering VSCs the emphasis is especially put on
DMS, which is the predominant form of volatile
sulfur compounds in the oceans.
Sampling
Sampling vessels (glass bottles, bulbs, canisters and
polymeric bags) for sulfur gases should be as inert as
possible in order to minimize adsorption losses and to
avoid possible reactions during sampling. For these
reasons, all materials in sampling vessels, tubing and
unions in contact with the sample should be carefully
chosen. The conditioning or covering of surfaces with
inert materials or application of surface deactivation
procedures such as silanization is usually necessary.
Glass sampling bottles or bulbs are commonly used
for collecting and transporting gas samples or to
blend calibration gas mixtures. Stainless steel
canisters and TeSon bottles are very convenient.
Frequently, the canisters are conditioned by heating
under vacuum before use. Sampling bags made of
Tedlar Rlm which is a polyvinyl Suoride (PVF) are
chosen because of their inertness. To prevent losses of
sulfur compounds, sampling vessels and connections
may be covered with aluminium foil to avoid photo-
chemical reactions.
Preconcentration
Due to the low concentration of sulfur species in
air (ppb or ppt level) different preconcentration
techniques have been applied before the gas
chromatographic analysis proper. The most fre-
quently used methods for these purposes are sorption
on certain metals, sorption on solid sorbents and
cryogenic trapping.
Sorption on metals This pre-concentration method
is based on the ability of certain metals (mainly gold,
palladium and platinum) to chemisorb sulfur gases.
Glass or quartz tubes Rlled with gold wool, gold-
coated glass beads, gold-plated sand or metal foils are
used for this purpose. The sample may be passed
through a TeSon tube containing a thin metal foil of
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
palladium (Pd), platinum (Pt) or gold (Au). Custom-
fabricated Pd on Pt has the advantage of the analyti-
cal collection efRciency of Pd and an increased dura-
bility and lifetime. Rapid desorption of the sulfur
compounds is achieved by passing a current through
the foil. Such a technique (metal foil collection/Sash
desorption and Same photometric detection) has
demonstrated a detection limit for total sulfur con-
centration of around 10 pptv (10\11).
Sorption on solid sorbents Adsorption on solid
sorbents is one of the simplest and most efRcient
methods of concentration of volatile compounds. Ad-
sorbent trapping is very popular, especially when
traps are kept at low temperatures. Ambient temper-
ature trapping may frequently give poor recoveries
due to poor collection efRciency.
Many sorbents, such as activated charcoal, silica
gel, aluminium oxide, graphitized carbon black, mo-
lecular sieves and porous polymers have been applied
to collect volatile sulfur species. The use of porous
polymers is the most widespread since the collected
substances can be desorbed from porous polymers
more easily, compared to desorption from charcoal.
Furthermore, collection efRciency on porous poly-
mers is less sensitive to water vapour in the sampling
atmosphere. The trapped compounds are usually re-
leased by thermal desorption and injected into a GC
column. Before this operation, they may be subjected
to cryothermal focussing in a capillary in order to
obtain a narrow injection band.
Among the porous sorbents, Tenax has the highest
popularity. Tenax has a low afRnity for water, and
breakthrough volume is relatively independent of hu-
midity. It is well suited for thermal desorption tech-
niques as it exhibits high thermal stability (3753C)
and can be subjected to repeated temperature cycling
without deterioration. The determination of several
sulfur gases can be easily conducted, even though
Tenax has a relatively low speciRc surface area (ca.
19 m2/g) which consequently limits the sampling vol-
ume. In practice, Tenax GC or TA is used together
with Chromosorb 106 and Spherocarb as backup
adsorbents. In order to retain the low boiling organic
sulfur compounds that are present in many samples,
cooling the trap with liquid nitrogen may be neces-
sary but this creates a problem when excessive
amounts of methane are present. Cooling with solid
carbon dioxide is suitable for trapping of VOS com-
pounds under these circumstances. Carbosieve ad-
sorption tubes can be used for collecting CS2 after
purging it from seawater samples.
For moist air samples (96% relative humidity)
acceptable recoveries have been observed for the
following sorbents: silica gel (recovery for MeSH
4287
'95%), molecular sieve (recovery for MeSH 73.9%
for COS 75%) and Carbosieve III S (recovery for
COS 71.7%) used with calcium chloride as a drying
agent. For methanethiol, recovery values showed no
signiRcant changes during 36 h storage or using dif-
ferent Sow rates in the range of 10}80 mL min\1.
Cryogenic trapping Cryogenic trapping is the tech-
nique of choice for collecting VSCs from air samples
but is not always practical due to transportation and
storage difRculties at remote locations.
Cryogenic trapping is very popular after purging
VSCs from various water samples and therefore is
also discussed in the next section.
Analysis of VSCs in air is complicated by the oxy-
gen, SO2 and NO2 which can cause variable and often
severe sampling losses by oxidation of these com-
pounds. Scrubbers for oxidant removal include Tef-
lon and Tygon shavings, and various substrates (glass
Rbre Rlters, Chromosorb, Anakrom, and glass beads)
coated with Na2CO3 or manganous oxide, MnO.
Sulfur Compounds in Aqueous
Matrices
Sampling
Aqueous samples for the analysis of VSCs are usually
collected in glass or polymer bottles. Glass vessels are
frequently silanized in order to minimize losses due to
adsorption on the walls. Brown glass is used to stop
biological and chemical processes which can occur
under the inSuence of light. TeSon and polyethylene
are frequently used. During sampling the vessels
should be Rlled to the top to exclude air and minimize
head space losses.
Isolation and/or Preconcentration
Because direct analysis of sulfur compounds in water
matrices is often impossible, various preconcentra-
tion or isolation procedures are applied before the
analysis proper. Solvent extraction and static and
dynamic headspace techniques are most popular.
Liquid extraction Solvent extraction is not as fre-
quently applied as formerly because this technique
has several disadvantages, i.e., handling toxic
solvents, the trapped substances become diluted,
automation is difRcult and the procedures are time
consuming. The most popular solvents for the VSCs
are diethyl ether, hexane or mixtures of these
solvents.
Static gas extraction methods Headspace-gas chro-
matography (HS-GC) analysis can be applied
4288 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
successfully for the analysis of VSCs in different
liquid matrices. It can be also applied in physical
chemistry studies of these compounds, being a valu-
able tool for acquiring data on gas}solid and
gas}liquid systems. For example, it was used for the
determination of distribution coefRcients, K, of se-
lected organosulfur compounds in air}water systems
as well as their temperature, ionic strength and con-
centration dependencies.
Generally, the detection limit of the static head-
space technique is 10 to 100 times poorer than that of
the dynamic technique, i.e., purge and trap (PT).
Dynamic gas extraction methods Purge and trap
assemblies can be used for isolation and preconcen-
tration of volatile sulfur species in water samples.
The extraction efRciency varies with the gas
considered and the extraction facilities employed
such as the dimensions of the purge vessel, bubble size
distribution, sample volume and temperature, purge
gas Sow rate and sparge time. All these parameters
should be carefully considered before applying the
technique for a particular purpose.
Due to the low detection limits which can be ob-
tained with the PT technique, it is extensively used to
determine VSCs in water. Several PT procedures have
been developed especially for the most important
natural sulfur compound } dimethyl sulRde (DMS)
} a climatically active trace gas.
Recently, there has been an interest in dimethyl
sulfoxide (DMSO) determination. DMSO is also an
environmentally signiRcant compound because of its
potential role in the biogeochemical cycle of DMS.
Direct injection and separation of aqueous DMSO
offers a simple and fast application, but exhibits lim-
ited sensitivity due to limitation on injection volumes.
More frequently, DMSO reduction and subsequent
analysis of the evolved DMS by purge-and-trap pre-
concentration has been used.
The P&T technique can also be applied for the
determination of sulfur species in sediments.
Sulfur Compounds of Fossil
Fuel-Origin
Trace level sulfur speciation and detection in crude oil
and in different petroleum products is traditionally
difRcult due to the complex hydrocarbon matrix.
Additionally, the fact that sulfur compounds are po-
lar and the hydrocarbons matrices are non-polar fa-
vours the loss of sulfur compounds to active sites in
analytical instruments and sample vessels. The de-
velopment of sulfur-speciRc detectors for gas
chromatography has added impetus to use of this
technique for the analysis of petroleum fractions. For
example, selectivity of the sulfur chemiluminescence
detector (SCD) allows the determination of sub-ppm
level of sulfur compounds in the presence of percent
levels of co-eluting hydrocarbons.
The usual approach for characterization of the very
complex nature of different individual sulfur com-
pounds in a crude oil is to fractionate the oil into
narrow boiling range cuts (prefractionation) and to
analyse each fraction, which simpliRes the analysis.
Sample preparation/cleanup is needed, especially for
analysis of high boiling fractions (coal-derived
liquids, shale oil), before GC, Solid phase extraction
(SPE), using various cartridges with different solvent
mixtures, followed by normal-phase liquid
chromatography has been applied for separation of
polycyclic aromatic sulfur heterocyclic compounds
(PASHs) from polycyclic aromatic hydrocarbons
(PAHs). PASHs can be found not only in fossil fuels,
but also in sediments, mussels, Rsh and airborne par-
ticulate matter. The separation and determination of
individual alkyl-substituted PASHs isomers in envir-
onmental matrices is difRcult because of the isomeric
structures of these species due to asymmetry imposed
by the sulfur atom. Relatively good resolution of
many PASHs isomers has been obtained on a smectic
liquid crystal column (Figure 1).
In research and in everyday practice, one frequently
encounters situations where not only the concentra-
tions of both sulfur and non-sulfur compounds but
also both percentage levels and low concentrations
(ppm and ppb levels) of sulfur species have to be
determined. In such cases two parallel detectors can
be used. For example, coupling of sulfur chemilumin-
escence (SCD) and thermal conductivity detectors
(TCD) enables the determination of concentration of
both sulfur-containing (from percentage to ppb
levels) and other gaseous compounds through simul-
taneous sampling, separation, and detection. Also
simultaneous SCD and FID detection can be useful in
many cases.
The sulfur-selective detectors, mainly SCD and
atomic emission detector (AED), can be interfaced to
simulated distillation (SimDis) systems to measure
the boiling point range distribution of heteroatoms
(S and N) in various petroleum fractions. Such
an approach is applied for process control, quality
assurance and product speciRcation purposes. Very
good sulfur SimDis chromatograms have been ob-
tained considering the fact that typical sulfur levels in
reRnery streams are several orders of magnitude
lower than the hydrocarbon levels.
The sulfur-selective detectors have been used for oil
spill identiRcation by Rnger-printing of various
crude oils. The speciRc identiRcation of different
dibenzothiophenes by GC}high resolution MS has
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY 4289

Figure 1 GC}MS separation of 3 benzo[b]naphthothiophene (BNT) (m/z 234) isomers and 30 methylbenzo[b]naphthothiophene
(MeBNT) isomers (m/z 248) on different stationary phases: DB-5MS, DB-17 and SB-Smectic. BN12T"benzo[b]naphtho[1,2-
d ]thiophene, BN21T"benzo[b]naphtho[2,1-d ]thiophene, and BN23T"benzo[b]naphtho[2,3-d]thiophene. Numbers identify the
specific methylbenzo[b]naphthothiophene isomers, e.g., 1}12"1-methylbenzo[b]naphtho[1,2-d]thiophene, 8}21"8-methylbenzo
[b]naphtho[2,1-d ]thiophene, 11}23"11-methylbenzo[b]naphtho[2,3-d]thiophene, etc.

permitted differentiation of very similar crude oils,
even from the same Reld.
The ability to speciate the sulfur compounds is an
advantage of GC method over elemental analysis, but
total sulfur can also be determined by GC by summa-
tion of all the sulfur-containing peaks.
Sulfur Compounds in Beverages
and Foodstuffs
Volatile sulfur compounds have been detected in
wine, beer (Figure 2), dairy products, coffee,
Rsh, garlic and tobacco smoke. In food chemistry
these compounds contribute signiRcantly to odour
and Savour because they often possess characteristic
smells and sensory thresholds (ca. 1
g kg\1
for DMS).
For isolation of sulfur compounds from different
food matrices, headspace sampling (HS) is the best
method. For example, HS-GC has been used for the
determination of VSCs in water}alcohol solutions nd
brandies. It was found that headspace concentrations
of sulfur, H2S, MeSH, EtSH, DMS, CS2, DES,
thiophene, DMDS and DEDS increased with increas-
ing ratio between the gas and liquid phase volumes
and was proportional to the temperature. However, it
4290 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY

Figure 2 GC}SCD chromatograms of (A) a beer with a sulfury character and (B) a non-sulfury beer. GC conditions: column: DB-5,
30 m, 0.53 mm I.D., 1.5
m film thickness: injector temperature: 1503C; column temperature programme: 20}503C at 53C min\1,
50}1803C at 83C min\1, 10 : 1 split injection. Peaks: (1) methanethiol; (2) dimethyl sulfide; (3) ethylene sulfide; (4) diethyl disulfide; (5)
dimethyl disulfide; (6) isopropyl sulfide (internal standard) and (7) dimethyl trisulfide.

diminished with increasing ethanol content and was
insensitive to the liquid phase salt concentration. HS
sampling has also provided qualitative and quantitat-
ive data of sulfur species in dairy products.
Storage Stability of Samples
In order to avoid losses or possible transformations of
sulfur compounds, samples should be analysed as
soon as possible. Keeping the samples at sub-ambient
temperature can improve the stability of sulfur com-
pounds. Sulfur concentrations in an air sample
collected cryogenically and stored in a freezer were
found not to change over a 2 week period. A Tenax
trap containing VSCs collected from air was stored
for at least 1 week at 1963C in liquid nitrogen with-
out any loss of sulfur compounds.
For sulfur gases the most convenient method of
storage seems to be in Tedlar bags. The concentra-
tions of the Rve sulfur gases (COS, CS2, MeSH and
EtSH) in such bags were stable for two weeks even at
the ppb concentration. Tedlar bags are not suitable
for SO2 and H2S. In these cases, SO2 concentration
decreased from 22 ppb to less than 1 ppb in 2 h and
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
H2S lost half of its original concentration of 70 ppb in
about 10 days. The stability of sulfur gases in glass
sampling bulbs is inSuenced by the gas matrix (nitro-
gen and air) and moisture. Reduced sulfur gases col-
lected in glass bulbs can remain in the bulbs for
approximately 24 h without major changes in gas
concentrations if the sample is dry and does not
contain oxygen (concentration decreased less than
5%) but dried air samples should be analysed within
3 h. Glass bulbs are not useful for collecting sulfur
gases if the sample in the bulbs contains moisture
(signiRcant decrease in H2S and MeSH concentra-
tions was observed).
The stability of freshwater samples is strongly
affected by the temperature at which it is stored. For
example, it was reported that the stability of DMS in
freshwater is shorter than the 48 h found in seawater
samples.
Because the presence of reduced sulfur compounds
in seawater is closely related to biological activity, the
stability of samples may depend on the depth of
sampling. When a sample was taken from the Baltic
Sea at 4 m depth and was stored at 53C in the dark,
the concentration of DMS Rrst rose dramatically after
4 days (nearly 10 times) and later decreased. Concen-
tration of sample taken from 50 m depth did not
change over a 2 week period. Samples can most
probably be stored longer if the cold trap is main-
tained in liquid nitrogen.
Figure 3 Trace light sulfur gases and C1}C3 mercaptans. (From Supelco Bulletin 722, reprinted with permission of Supelco,
Bellefonte, PA 16823, USA.)
4291
To suppress microbial activity, compounds such as
phenols, mercuric chloride, sodium azide and HCl
can be added to water samples.
When immediate analysis is not possible, refriger-
ation of sample for analysis of sulfur compounds in
aqueous solutions is recommended as the best way to
maintain sample integrity at least for periods up to
48 h.
Separation Systems
Column packing for chromatographic determination
should be chosen not only with respect to the com-
plete separation of a given mixture but should also be
selected with respect to minimize losses due to ad-
sorption and catalytic reactions and rearrangements.
These are particularly important when packed metal
and glass column are used.
The most common material used for packed col-
umns in the analysis of VSCs is TeSon. Supelpack
S (specially treated Porapack QS), different Chromo-
sorbs, Porapack Q, N or QS, Triton X 305, Chromo-
sil 310 or 330 (specially treated silica gel), Carbopack
B or BHT 100 and 3% polyphenyl ether and 1%
phosphoric acid on Chromosorb T have all been
reported for VSCs analysis.
Good separation of many gaseous sulfur com-
pounds can be obtained on Chromosil 310 and 330
and Supelpack (Figure 3). The latter can resolve the
4292 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
large peak of CO2 (evolved from acidiRed seawater)
from the much smaller and neighbouring H2S and
COS peaks.
Development of fused silica capillary columns has
provided more inert surfaces for trace sulfur analysis.
As with most analysis, no single capillary column can
assure the combination of sample capacity, good res-
olution and reasonable analysis time for the wide
range of sulfur species in different sample matrices.
The analysis of VSCs has been achieved with methyl
silicone phases like BD1 or Rtx1 with thick Rlms
(4}5
m). Generally, columns with thicker Rlms pro-
vide increased separation of volatile sulfur com-
pounds and are better suited for analysis of low level
volatile sulfur compounds in gases (Figure 4). On
such non-polar columns retention times are governed
primarily by boiling points and the retention se-
quence can be predicted from boiling-point data.
Figure 4 Effect of stationary film thickness on the separation of sulfur-containing compounds. Sample: gas phase standard of 10
component mixture of sulfur compounds: 1 mL split injection (split ratio 10 : 1), SCD, temperature programme 1 min at 353C, 35}2003C
at 10 min. Reproduced with permission from Hutte (1990).
Thick Rlms separate most VSCs in programmed tem-
perature analysis with an initial temperature of
40}503C. For the separation of H2S, COS and SO2
sub-ambient column temperatures must be used
(Figure 5).
Recently, porous layer open tubular (PLOT) col-
umns have become commercially available. The use-
fulness of such columns has been demonstrated for
analysis of sulfur compounds such as COS, H2S and
DMS.
Smectic liquid crystalline columns may offer
unique selectivity for isomeric polyaromatic sulfur
compound (PASHs) mixtures that are not possible
with other columns. Unfortunately, extensive use of
the SB-smectic column at the upper temperature limit
(2503C isothermal, 2703C during temperature pro-
grammed) can reduce the useful lifetime and column
selectivity often changes dramatically with use.
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY 4293

Figure 5 Sub-ambient column temperature separation of sulfur gas standard: GC conditions: 1 mL split injection (split ratio 10 : 1),
injection port at 2503C, FID temperature 3003C. Column temperature programme: 2 min at !303C, !30}2003C at 203 min\1.
Detector conditions: SCD integration time 0.03 s. Reproduced with permission from Hutte (1990).

Detection Systems ent in liquid fuels have a quenching effect on the

The value of GC for sulfur compounds analysis is
found in the availability of selective and sensitive
detectors. These detectors are especially useful be-
cause matrices requiring sulfur analysis are often very
complex. Such detectors can reduce the analysis time
by eliminating laborious and time-consuming proced-
ures of sample preparation, which can also often
cause contamination or loss of analytes. Selective
detectors have found extensive application in the de-
termination of sulfur compounds in various matrices
because of these reasons.
Table 1 lists the basic characteristics of the most
frequently used sulfur-selective detectors.
The Same photometric detector (FPD) is still the
most widely used sulfur selective detector. The FPD
exhibits a non-linear (exponential) response to
sulfur compounds and response factors may be com-
pound-dependent but it is relatively inexpensive, ro-
bust and adequate for many applications. The major
advantage of the FPD is its application to gases and
fuels. However, major co-eluting hydrocarbons pres-
sulfur response. Also the injection of aqueous samples
directly into a GC-FPD system is not recommended
because the injected water can extinguish the detector
Same and non-volatile material contained in the
sample can contaminate the injection port and col-
umn. An increase of the detector temperature pre-
vents the Same from being extinguished but working
at temperatures higher than 2503C may produce
a poor baseline. An improved FPD called a pulsed
Same photometer detector (PFPD) employs a pulsed
Same and time-resolved emission detection with
gated electronics. The improvements include one to
two orders of magnitude sensitivity enhancement,
about an order of magnitude of increased selectivity
and reduced quenching effects.
A more recent alternative to the FPD is the sulfur
chemiluminescence detector (SCD). Recent applica-
tions of this detector have shown that it gives good
performance in terms of detectability, selectivity, lin-
earity and a nearly equimolar response to sulfur. It
does not suffer signiRcantly from quenching or inter-
ferences. The combination of fused silica capillary
4294 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY

Table 1 Basic characteristics of gas chromatographic sulfur-sensitive detectors

Detector Detection Selectivity Linear Ease of General characteristics

limit [S/C] concentration operation
[gS/s] range (decades)

FPD 10\11 10\3}105 3a Moderate Exponential and compound-dependent re-

sponse, susceptible to flame-out and
quenching effects
PFPDH 10\13 106 3a Moderate Less quenching than for the regular FPD
ECD Variable Variable 4 Simple Strongly compounds-dependent response,
up to 10\15 very high sensitivity to SF6 itself or post-
column converted sulfur compounds
SCDHH (flame 10\12 '106 4}5 Moderate Linear and nearly equimolar response, non-
version) susceptible to quenching or interferences,
very convenient for petroleum applications
SCD (non-flame 10\13 '107 4}5 Simple
version)
AEDHHH 10\12 104 3}4 Difficult Small susceptibility to quenching or inter-
ferences, possibility of elemental composi-
tion confirmation
HECD# 10\11 104}106 3}5 Complicated Possible interferences of other organic com-
pounds
PID## 10\12 Poor 6 Moderate Many factors influence the detecter re-
sponse
MS 10\11 Specific 5 Complicated Convenient for identification of complex mix-
ture, new membrane techniques assure
lower detection limit
FT-IR 10\12 Specific 4 Complicated Applied only for highly complex mixture.
Strongly compound-dependent

a
After linearization.
HPulsed flame photometric detector.
HHSulfur chemiluminescence detector.
HHHAtomic emission detector.
#
Hall electrolytic conductivity detector.
##
Photoionization detector.

columns and the SCD provides a powerful tool for the
measurements of trace levels of sulfur containing
compounds in complex matrices (Figure 6). The per-
formance of the SCD can be improved by changing
the means of sulfur-chemiluminescent-species pro-
duction from a hydrogen Same (commonly referred
to as a Same SCD) to a closed hydrogen/air burner (a
Sameless SCD). The Sameless SCD is typically an
order of magnitude more sensitive than the Same
version. An extremely low detection limit of 25 fgS/s
has been reported but most authors have observed
a limit between 0.1 and 1 pgS/s.
The atomic emission detector (AED) has a good
combination of speciRcity and sensitivity for the
analysis of volatile sulfur-containing compounds. The
AED is better than the FPD because it does not
exhibit as many problems with interferences, quench-
ing, and compound-dependent responses. The AED
can be used to conRrm the elemental composition of
a compound by its ability to monitor several atomic
lines simultaneously. The response of the AED to
sulfur at 180.7 nm is reported to have linear range of
2;104, and sensitivity of 1.7 pgS/s and a selectivity
over carbon of 1.5;103.
The electrolytic conductivity detector (HECD or
Hall detector) has found limited applications in
analysis of VSCs probably because it requires regular
attention. The electrolyte must be kept extremely
clean and sulfur speciRcity is limited by high concen-
trations of co-trapped carbon dioxide. Despite
these problems, the HECD performs well in the sulfur
detection mode. The detector response is linear up
to 50 ng sulfur, selectivity of sulfur to carbon is
typically better than 104, and the limit of detection
is 1 pgS/s.
The electron-capture detector (ECD) can also be
used for determination of a variety of sulfur contain-
ing compounds, e.g., SO2, H2S, thiols and organic
 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY 4295

Figure 6 Comparison of column resolution for the analysis of sulfur components of raw naphtha feedstock (SCD). Capillary columns:
A: 15 m;0.53 mm i.d. DB-1 (1.5
m film thickness); B: 30 m;0.32 mm i.d. SPB-1 (1
m film thickness); C: 100 m;0.25 mm i.d.
SPB-1 (0.5
m film thickness). GC conditions: 1
L direct injection for column A, 1
L split injection for column B (split ratio 10 : 1) and
column C (split ratio 100 : 1). Injection port temperature 2503C. Column temperature programme: column A: 1 min at 353C 35}2003C at
83 min\1; column B: 1 min at 353C, 35}2003C at 103 min\1; column C: 13 min at 353C, 35}453C at 103C min\1, 15 min at 453C, 45}603C
at 13C min\1, 60}3003C at 23C min\1. Reproduced with permission from Hutte (1990).

mono- and di- and trisulRdes. Although the ECD is
only moderately sensitive to SO2 and H2S, both are
detected in the concentration range 0.1}1
g g\1 in
a 1 mL air sample. For COS, CS2, MeSH, H2S and
DMDS the detector displays a sensitivity comparable
to the FPD but is much less sensitive towards DMS
and thiophene. SF6 can be detected at extremely low
levels with minimum detectable peaks lower than
0.2 fmol. The inertness and extremely low detectabil-
ity of SF6 has led to the development of a method in
which the original sulfur compounds are Suorinated
and then detected as SF6.
The application of GC-MS systems is becoming
more popular in the analysis of various sulfur com-
pounds. Determination of sulfur compounds in
underground reservoirs of natural gas and town gas
(RSH, RSR and RSSR type compounds) by GC-MS
has been carried out using the ion CH2"S#H
 4296

Table 2 Representative examples of sulfur gas determination by gas chromatography

Matrix Analysed compounds Sample preparation Column Temperature conditions Detector Detection
limit
Material Packing Dimension

Atmospheric COS, MeSH, Chemisorption on gold Fused silica BP-1 30 m;0.32 mm Isothermally, 353C AED 0.1 nmol m\3
sulfur gases CS2 , DMS, SF6 wool#cryotrapping or ;1
m
only direct cryotrapping
Air from a H2S, C1}C7 thiols, DMS Trapping on Tenax TA, Fused silica GS-Q 30 m;0.53 mm 1 min at 1003C, FPD pgS
petroleum placed in PTV injector, 100}2403C at 83C
plant cryotrapping on column min\1, 30 min at 2403C
Gases from H2S, COS, CS2, Direct injection Fused silica CP-Sil 5CB 50 m;0.32 mm 3 min at 753C; 75}1203C SCD ppb level
sweetening thiophene, MeSH, ;0.5
m at 203C min\1 TCD
process (ppm level) in CH4, Stainless Porapak Q 4 m;3.2 mm o.d. 10 min at 1203C
CO2, C2H4 (% level) steel
Gases from DMS, DES Trapping on Tenax TA, Fused silica DB-1 15 m;0.52 mm 2 min at 603C; FPD#
g m!3
workplace cryofocusing (!1503C) ;1.5
m 60}1003C at 103C min\1, FID
environment 1 min at 1003C;
100}2503C at 303C min\1
Atmospheric DMS, DMDS in presence Cryogenic trapping on Fused silica Paraplot Q 10 m;0.32 mm 7 min at 553C; 55}2163C MS ppt level
gas of volatile organic Tenax (Peltier effect) at 123C min!1
III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY

compounds
Tropospheric COS, H2S, CS2, DMS Trapping in Teflon trap FPD ppt level
airborne (liquid argon) MS ECD
gases After separation
fluorination
to SF6
Atmospheric H2S, COS, SO2, DMS Cryogenic trapping Fused silica DB-1/DB-Wax 30 m;0.53 mm 1.2 min at 303C; FPD 50 pgS
sulfur gases MeSH, CS2, DMDS in Teflon trap or in FSOT ;5
m/ 30}1403C at 303C min\1,
EtSH, iPr SH retention gap 3 m;0.53 mm
;1
m
DB-Wax 60 m;0.53 mm;1
m 2 min at 303C; MS 5 pgS
30}1403C at 153C min\1
Antarctic SF6 Sorption on Porapak Stainless Molecular 30 cm pre- Isothermally, 653C ECD ppt level
atmosphere Q (!773C) steel sieve 5A column#3 m (back flushed when
SF6 reaches main
column)
Table 3 Representative examples of sulfur compound determinations in aqueous matrices by gas chromatography

Matrix Analysed compounds Sample preparation Column Temperature conditions Detector Detection
limit
Material Packing Dimension

Antarctic H2S, COS, SO2, Purging, preconcentration Teflon Carbopack 3 m;0.25 mm 40}1403C}temperature FPD 6}160 ng L\1
marine waters MeSH, DMS on Tenax TA (ambient B/1.5% XE programming
temperature), 60/1% H3PO4
thermal desorption
Atlantic DMS, DMDS, CS2, Purging, trapping in cold trap Fused silica SE-54 50 m;0.32 mm 3 min at 703C, AED or 0.5}0.8 ppm
surface water dimethyl selenide, ;5
m 70}1803C at 33C min\1 ECD
dimethyl iodide
Aqueous H2S, MeSH, Filtration with glass fibre Fused silica DB Wax 30 m;0.53 mm 5 min at 403C; 40}1603C FPD 0.3 mg L!1
matrices DMS, DMDS filters, direct injection ;1
m at 303C min\1, 3 min at
(distilled water, split mode (10 : 1) 1603C, 160}2003C
tap water, kraft at 403C min\1
mill condensate)
Heavily polluted H2S, COS, MeSH, Purging, cryogenic Teflon Carbopack 1.4 m;3 cm 1 min at 53C; 5}503C FPD ng S
water DMS, CS2, DMDS trapping (!803C) BHT-100 at 303C min!1, 2 min
at 503C; 50}1003C
at 303C min\1; 7 min at
1003C
Water samples, H2S, COS, MeSH, Purging, cryocondensation Teflon Carbopack 1.4 m;3.2 cm !5}1003C at 303C min\1, FPD pg S
sediments DMS, CS2, DMDS secondary cryofocusing BHT-100 8 min at 1003C
Surface and 21 organosulfur Purging, trapping on Tenax, Fused silica DB-5 25 m;0.25 mm 4 min at !503C, MS ppb level
potable water, compounds (thiophenes, extraction with hexane ;0.33
m !50}2003C at
industrial sulfides, sulfones, or methylene chloride 163C min!1, 8 min
effluents, benzothiazole) at 2003C; 200}2203C
sediments at 163C min\1
Anoxic-lake H2S, COS, MeSH, Purging, cryotrapping on Glass UCON 50 25 m and 50 m 03C}453C at 33C min\1 FPD, MS ng L\1
water CS2, EtSH, DMS Tenax TA (solid CO2), HB 5100
heating with hot air (2003C)
III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
4297
 4298

Table 4 Representative examples of sulfur compounds determination in beverages and food stuff by a gas chromatrography

Matrix Analysed compounds Sample preparation Column Temperature conditions Detector Detection
limit
Material Packing Dimension

Garlic samples 15 sulfides, disulfides Head space (5 and 45 min) Fused silica HP-5 25 m;0.31 mm 2 min at 403C, AED
(whole non- and trisulfides ;0.53
m 40}703C at 33C min!1, and MS
peeled, peeled 70}2053C at 7.53C min!1;
cloves, crushed 205}2503C
cloves, juices) at 253C min!1
Beer, coffee 13 volatile sulfur 200
L of beer#10 mL of Fused silica HP-1 25 m;0.32 mm 1 min at !203C, MIP-AED from
compounds, DMS water}silicone mixture, ;0.17
m !20 to !103C 0.4 ng L\1
DMDS, CS2, other purging (853C), cryogenic at 103C min\1, !10 to for EtSH to
sulfides, C1}C4 trapping (!1703C) !403C at 303C min\1, 0.9 ng L\1
thiols 0.5 g coffee powder#10 mL 40}1503C at 703C min\1 for DMS
hot water, purging (853C),
(!1703C)
Grape juice COS, CS2, DMS Head space Glass Porapak QS 5 min at 803C; 80}1203C FPD
III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY

at 123C min!1
Alcoholic H2S, MeSH, EtSH, Head space Fused siliica SPB-1-Sulfur 30 m;0.32 mm 1 min at 353C; SCD 10
g L!1
beverages DMS, CS2, ethyl-methyl ;4
m 35}553C at 103C min!1,
sulfide, DES, 55}2503C at
DMDS, DEDS 253C min!1
Packaged SO2 in presence of Head space Teflon Chromosorb 108 1.2 m;2 mm 903C and 1203C HECD 0.81 ng L!1
food (wine, CO2 and H2O
orange juice)
Table 5 Representative examples of sulfur compounds determination in fossil fuel related samples by gas chromatrography

Matrix Analysed compounds Sample preparation Column Temperature conditions Detector Detection
limit
Material Packing Dimension

Gases from H2S, COS, SO2, Vacuum extraction from Teflon Chromosorb 105 80 cm;2 mm 1 min at 603C, 60}1803C FID and 10\10 gS/s
plasma reactors CS2, thiols, sulfides reaction tube or direct at 203C min\1, 2 min FPD 10\12 gHC/s
(also atmosphere in the presence of C1}C5 injection from pressurized at 1803C
from microelec- hydrocarbons (HC) cylinders and syringes
tronic processes
and polluted air)
Natural gas H2S, SO2, CS2, Direct injection, Fused silica SPB-1 Sulfur 30 m;0.32 mm 3 min at 103C, SCD and
C1}C4 thiols, DMS, split mode, 10 : 1 ;4
m 10}3003C at 103C min\1 FID
DMDS, DEDS,
methylethylsulfide,
2-ethylthiophene
Light petroleum Thiophene, Direct injection, split Fused silica DB-1 60 m;0.25 mm 35}1003C SCD and 0.01% (w/w)
fractions (cracked benzothiophene, mode, 10 : 1 ;0.25
m at 103C min\1, FID of total S
gasolines, diesels) dibenzothiophene and 100}2253C at
their alkyl substituted 23C min\1, 20 min at 2253C
homologues
Hydrotreated H2S, thiophene, Direct injection, Fused silica SPB-1 30 m;0.32 mm 35}1253C at AED H2S-single
naphtha alkylthiophenes, split mode, 1 : 100 ;4
m 303C min\1, 125}2603C ppm thiols-
thiols at 53C min\1 tens ppm
total
S-hundreds
ppm
Commercial Thiophene, Direct injection after Fused silica HP-5 25 m;0.32 mm 50}1103C at FPD
diesel fuels benzothiophene, dilution with n-hexane ;0.52
m 303C min\1, 110}2103C
dibenzothiophene, (1 : 100), on-column mode at 2.53C min\1, 210}2803C
hiolur sulfides at 23C min\1, 5 min at 2803C
(up to C16) higher
thiols (up to C8)
Fossil fuels 80 polycyclic Prefractionation Fused silica DB-5MS 60 m;0.25 mm 1 min at 603C, 60}1503C MS Only for
(coal tar, aromatic sulfur hetero- solid-phase extraction, ;0.25
m at 453C min\1, 2 min retention
crude oil) cyclic compounds LC clean-up at 1503C; 150}3003C indexes
(3}5 rings) at 23C min\1, 25 min at determina-
3003C tion
DB-17 1 min at 603C, 60}903C at
353C min\1, 1 min at 1903C,
190}3203C at 13C min\1
Liquid crystalline 1 min at 603C, 60}1903C at
III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY

polysiloxane 353C min\1, 1 min at 1903C,

(SB-Smectic) 190}2663C at 13C min\1
4299
4300 III / SULFUR COMPOUNDS: GAS CHROMATOGRAPHY
with m/z 47. The ion with m/z 45 is more intense in
some sulfur compounds, but is often found in oxygen
compounds (C# 2 H4OH) as well. Twenty-one or-
ganosulfur compounds (DMS and DMDS among
others) were detected in the approximate concentra-
tion range 0.1 to 2000 ppb in water, industrial efSu-
ent, sediment and Rsh samples using an automated
GC-MS system. A GC-MS method for DMS and SO2
determination in air in real time at the sub parts per
trillion level with a high pressure selected-ion chem-
ical ionization Sow reactor has been developed. The
use of isotope dilution GC-MS with perdeutered
DMS for DMS determination in sea water gives better
than 2% precision. By using the ratio of the MS
response at m/z 62 and m/z 68, compensation can be
made for instrumental drift as well any losses in
sampling ambient air. Another signiRcant advantage
is the ability to determine DMS concentration by
stripping only a small fraction of DMS from solution,
resulting in artefact-free DMS concentration. In addi-
tion, larger volumes of water can be sampled by
eliminating the need for long sampling periods re-
quired to remove DMS quantitatively from solution.
Highly sensitive and speciRc continuous measure-
ment of DMS in air, using triple quadrupole mass
spectrometry with atmospheric pressure chemical
ionization, has been demonstrated. Detection limits
in continuous direct monitoring were determined for
DMS (24 pptv), H2S (1 ppbv), for MeSH, COS and
CS2 (about 10 ppbv).
Calibration
The preparation of reliable standard mixtures is an
important step in analysis. The simplest way to cali-
brate a GC system for gas analysis is by injecting
a suitable volume of a standard gas. Low concentra-
tion standards, usually needed in trace analysis, can
be obtained by applying the exponential dilution Sask
technique or with permeation tubes. To minimize
non-linear response problems (as for Same photomet-
ric detector) the calibration curves should cover the
anticipated concentration range. For calibration, the
gases from diffusion tubes are diluted with an inert
gas and frequently led through a glass loop injected
onto the column with appropriate valves. A new
concept for the generation of standard mixture of
thiols, based on thermal decomposition of a sub-
stance chemically bonded to the surface of silica gel,
has been developed. The method enables preparation
of a standard mixture containing volatile, malodor-
ous, unstable and toxic compounds. For example,
standard mixture of MeSH and PrSH have been gen-
erated by heating silica gel with anchored dithiocar-
bamate groups.
In analysis of liquids, primary standards are usually
prepared in an appropriate solvent which should not
interfere with the determined compounds. All stan-
dard solutions should be stored in vials with head-
space volume as small as possible and kept at low
temperature (443C).
Examples of Applications
Due to the diversity of sulfur compounds and various
matrices in which they can be present, the chromatog-
rapher is faced with a difRcult task when separation,
identiRcation and quantiRcation of speciRc sulfur spe-
cies are desired. The approach needed for the analysis
of these compounds depends on several factors. In
choosing the appropriate procedure, the analyst
should consider the form of sulfur compounds to be
determined, their levels of concentrations, the phys-
ical state and the complexity of the matrix.
Tables 2}5 present representative examples of sulfur
compounds determination in various matrices using
gas chromatography. The examples include sample
preparation techniques, columns with chromato-
graphic conditions, and detectors.
Conclusions
Gas chromatography, especially high resolution gas
chromatography, perhaps more than other methods,
fulRls the requirements needed for the analysis and
structure elucidation of multicomponent environ-
mental mixtures in which different sulfur species can
be present in nanogram or picogram amounts. It
should be noted that, besides selective high resolution
columns and sensitive sulfur-speciRc detectors, most
qualitative and quantitative determination of sulfur-
containing compounds requires efRcient sample en-
richment techniques and quantitative desorption
from traps. The applied procedure should also min-
imize adsorption losses of the sulfur compounds in
the whole analytical system and reduce possible re-
arrangements of sulfur analytes. SigniRcant progress
is still being made in all steps of sulfur analysis in
various environmental matrices but such procedures
are not still routinely applied in many laboratories.
Future developments should be focused on proced-
ures that can be used during long research expedi-
tions, directly aboard ships, or in situ for real-time
measurements.
See Colour Plate 119.
See also: II/Chromatography: Gas: Detectors: General
(Flame Ionization Detectors and Thermal Conductivity De-
tectors); Detectors: Mass Spectrometry; Detectors: Selec-
tive; Gas-Solid Gas Chromatography; Headspace Gas
Chromatography; Multidimensional Gas Chromatography;
 III / SUPERCRITICAL FLUID CRYSTALLIZATION
Sampling Systems. III/Flavours: Gas Chromatography.
Appendix 2: Essential Guides to Method Development
in Gas Chromatography.
Further Reading
Hutte RS (1995) The sulfur chemiluminescence detector. In
Adlard ER (ed.) Chromatography in Petroleum In-
dustry, Amsterdam: Elsevier.
Hutte RS, Johansen NG and Legier MF (1990) Column
selection and optimization for sulfur compounds analy-
sis by gas chromatography. Journal of High Resolution
Chromatography 13: 421}426.
Karchmer JH (1970) The Analytical Chemistry of Sulphur
and its Compounds, Part I. New York: John Wiley
& Sons Inc.
Karchmer JH (1972) The Analytical Chemistry of Sulphur
and its Compounds, Part II. New York: John Wiley
& Sons Inc.
MoK ssner SG, Lopez de Alda MJ, Sander LC, Lee ML and
Wise SA (1999) Gas chromatographic retention
behaviour of polycyclic aromatic sulfur heterocyclic
compounds (dibenzothiophenes, naphtho[b]thiophenes,
benzo[b]naphthiophenes and alkyl-substituted deriva-
SUPERCRITICAL FLUID
CRYSTALLIZATION
A. S. Teja and T. Furuya, Georgia Institute
of Technology, Atlanta, GA, USA
Copyright ^ 2000 Academic Press
Introduction
Supercritical crystallization processes use the special
properties of supercritical Suids that make these
Suids particularly suitable as solvents or antisolvents.
In both cases, an expansion of a solution is used to
create supersaturation, which is the driving force for
nucleation and growth of the solute.
A supercritical Suid (SCF) is a Suid above its criti-
cal temperature and pressure. It is characterized by
physical properties (such as viscosity and diffusivity)
that can be continuously varied between those of
liquids and gases. The liquid-like density of a SCF is
associated with its ability to dissolve solutes, and
hence its solvent power. Since this density can be
changed signiRcantly by changing the pressure and
temperature in the critical region, the solvent proper-
ties of a supercritical Suid can be tailored for speciRc
applications. Figure 1 shows the relationship be-
tween pressure and density of carbon dioxide. The
region above the critical pressure and temperature
4301
tives) on different derivatives of different selectivity.
Journal of Chromatography 841: 207}228.
Saltzman ES and Cooper WJ (1989) Biogenic Sulphur
in the Environment. Washington: American Chemical
Society.
Simo R (1998) Trace chromatographic analysis of dimethyl
sulphoxide and related methylated sulfur compounds in
natural waters. Journal of Chromatography A 807:
151}164.
Thompson M and Stanisavujevic M (1980) Gas chromato-
graphy and gas chromatographydmass spectrometry of
organosulphur compounds and other labile compounds.
Talanta 27: 477}493.
Tibbets PJC and Large R (1988) Improvements in oil Rnger-
printing: GC/HR MS of sulfur heterocycles. Petroana-
lysis 87: Dev. Anal. Chem. Pet. Ind., pp. 45}57. Chi-
chester: John Wiley and Sons.
Wardencki W (1998) Problems with the determination of
environmental sulphur compounds by gas chromatogra-
phy. Journal of Chromatography A 793: 1}19.
Wardencki W and Zygmunt B (1991) Gas chromatographic
sulphur-sensitive detectors in environmental analysis.
Analytica Chimica Acta 225, 1}13.
(7.38 MPa, 302.3 K) is commonly referred to as the
supercritical region of carbon dioxide. It is important
to note that the largest changes in the Suid density
with changes in temperature and/or pressure in
the single-phase region occur near the critical
point. Therefore, large changes of solvent power
can be achieved with small changes in pressure or
temperature in the critical region. It should be
added here that a supercritical crystallization process
involves mixtures of solute and solvent; however,
these mixtures are generally dilute so that their
critical points are close to the critical point of
the solvent. The behaviour depicted in Figure 1
may therefore be considered to be representative
of the behaviour of dilute mixtures of constant
composition.
If a supercritical Suid loaded with solute is ex-
panded, then the resulting change in density may lead
to precipitation of the solute. If these changes in
density are made to occur rapidly, then the process is
known as the rapid expansion of supercritical solu-
tions, or RESS. Very high supersaturations may be
achieved in RESS processes over a very short period
of time. This generally favours the deposition of small
crystals and narrow size distributions. Also, the crys-
4302 III / SUPERCRITICAL FLUID CRYSTALLIZATION
Figure 1 Pressure}density behaviour of CO2 . 2, 330 K;
***, 310 K; } } } }, phase boundary.
tals are generally free of solvent inclusions because
the solvent is likely to be in the gaseous state at the
end of the expansion.
Introduction of a supercritical Suid into an organic
solvent can also result in expansion of the liquid
phase, and hence, in large changes in density. If
a solution containing a desired solute is expanded
sufRciently by the supercritical Suid, then the liquid
phase will no longer be a good solvent for the solute
and nucleation will occur. In this case, the supercriti-
cal Suid acts as an antisolvent, and the crystallization
process is known as the supercritical antisolvent
(SAS) process, or by a variety of other names that are
discussed below. Changes in the pressure, temper-
ature, or rate of supercritical Suid addition provide
an opportunity for tailoring the SAS crystallization
process for speciRc applications.
Crystallization by the Rapid Expansion
of a Supercritical Solution (RESS)
The rapid expansion of a supercritical solution
(RESS) by decompression can lead to very large cha-
nges in density and, hence, in the solubility of a solute
in the supercritical solvent. This can result in very
high supersaturation when supercritical solutions are
depressurized, leading to the formation of a large
number of nuclei. A typical RESS apparatus is shown
in Figure 2. Solvent is pressurized in a pump until
a pressure above its critical pressure is attained. The
supercritical state is achieved by passing the pressur-
ized solvent through a heat exchanger maintained at
a temperature above the critical temperature of the
solvent in a constant-temperature bath. The super-
critical Suid is then passed through a bed of solute
where it becomes saturated with the solute. The
loaded solution is then heated to a designed pre-
expansion temperature, and Rnally expanded quickly
through an expansion device, such as a nozzle or
a capillary, into a collection vessel. The expansion
device is generally heated to prevent resublimation or
solvent condensation. The collection vessel is main-
tained at a constant temperature and pressure or
vacuum, and the products are collected on a suitable
substrate placed in the path of the expansion jet. The
pressure in the collection vessel is ambient, but may
sometimes be higher in order to control the particle
size; or it may be below atmospheric to prevent con-
densation of any solvent that is a liquid at ambient
conditions. Variations of this equipment are possible,
particularly if the solvent is to be recycled. Also,
a dual RESS or DURESS process has been proposed
whereby two RESS expansions are carried out in
a concentric expansion device and yield, for example,
a solid solute coated with a polymer.
The RESS process is applicable to any material that
can be dissolved in a supercritical solvent and is
particularly useful for materials of low volatility.
A few examples of crystallized materials using the
RESS process are shown in Table 1. Scanning elec-
tron microscopy (SEM) micrographs of crystals ob-
tained by RESS processes are shown in Figure 3.
RESS expansions result in essentially homogeneous
nucleation of the solute. The morphology of the prod-
uct is determined by a number of factors, including
the solute and its concentration, the device used for
the expansion, the pre-expansion temperature, the
Sow rate, and the pressure drop on expansion. High
concentrations of solute tend to produce powders,
whereas low concentrations generally produce thin
layers or Rlms. The particle size has been found to
increase with solute concentration prior to expan-
sion. Also, processing conditions may be chosen such
that the solvent is a gas at exit conditions and can be
easily separated from the deposited solute. If con-
ditions are chosen so that a two-phase mixture is
Figure 2 Experimental apparatus for a RESS process.
 III / SUPERCRITICAL FLUID CRYSTALLIZATION 4303

Table 1 Substances processed using RESS

Materials Supercritical fluid Morphologies, particle size (
m)

Inorganics
-Al2O3 Water Films, &0.08 thickness
SiO2 Water Spheres, 0.1}0.5
Films,'1.0 thickness
GeO2 Water Spheres, 0.5}1.3
ZrO(NO3)2 Ethanol Particles,&0.1
Polycarbosilane Pentane Fibres, 1 diameter
Particles,(0.1

Organics
Benzoic acid CO2 Particles, 3}8
-Carotene Ethylene Particles, 1}2
Cellulose acetate Pentane Fibres, 0.8 diameter
Naphthalene CO2 Particles, 2}50
n-Octacosane CO2 Particles, 2}6
Phenacetin CO2 Particles,&10
Phenanthrene Trifluoromethane Particles, &3
CO2 Particles,1.6}6.6
Stigmasterol CO2 Fibres,&0.2 diameter

Polymers
Poly-1-butene CO2 Spheres, (5
Poly(carbosilane) Pentane Particles, (0.1
Fibres, 1 diameter
Polycaprolactone CO2 Spheres, (5
CDFMa Spheres, 1}5
Fibres, 2}7
Polyethylene succinate CO2 Spheres, (5
Polyethylene methacrylate CDFM Particle/fibre blend
Poly(glycolic acid) CO2 Particles, 10}20
Poly(L#)-lactic acid CO2 Particles, 10}120
CDFM Particles, 0.2}0.6
Poly(DL)-lactic acid CO2 Particles, 10}20
Polymethyl methacrylate Propane Particles, 0.5}1.0
Fibres, 1 diameter
CDFM Particles/fibres
Polyphenylsulfone Propane Spheres,&0.5
Polypropylene Propylene Fibres,&2.5 diameter
Pentane Fibres, 1 diameter
Polystyrene Pentane Spheres, 20
Fibres, 0.8}2.5 diameter

a
CDFM, chlorodifluormethane.

formed during the expansion, solid may condense to
yield a thin solid Rlm.
There is a possibility of Rbre formation from super-
critical solutions when an organic polymer is the
solute. The polymer may form either a liquid or solid
after decompression, depending on the polymer
melting temperature relative to the post-expansion
temperature. Fibres are generally formed when the
expansion is carried out in a capillary nozzle and the
post-expansion temperature is close to the melting
temperature of the polymer so that the polymer de-
posits as a liquid on the nozzle walls. RESS expansion
of polymers yields powders when the temperature is
not close to the melting temperature of the polymer.
The extremely short times of product formation in
the expansion of supercritical solutions also makes it
possible to produce multicomponent mixtures of
powders with uniform distribution of the compo-
nents. Such powders have tremendous commercial
potential in the ceramic industry.
The pressure, temperature, and supersaturation
proRles in and outside the expansion device
determine the size of the crystals produced in the
RESS process and the crystal size distribution. The
pressure and temperature proRles in the expansion
device can be modelled by solving the mass, energy,
and momentum conservation equations for the
adiabatic expansion of the supercritical Suid.
Typical proRles for a capillary nozzle are shown in
Figure 4. The free-jet expansion after the Suid exits
4304 III / SUPERCRITICAL FLUID CRYSTALLIZATION
Figure 3 SEM micrographs of n-octacosane crystals obtained
in RESS expansion of a CO2 solution.
the device can also be modelled and is shown sche-
matically in Figure 5. Calculations have shown that
a Mach disc is formed a few nozzle diameters down-
stream from the nozzle exit and that the pressure and
temperature are very low in the region between the
exit and the Mach disc. High supersaturations may
therefore be obtained before, in, or after the Suid
exits the nozzle and the exact proRle must be known
Figure 4 Density and velocity profiles in a RESS expansion of
CO2 through a capillary nozzle at 443 K and 17.39 MPa. 2,
Velocity; } } } }, density.
Figure 5 Free jet expansion of a supercritical fluid solution from
a capillary.
if control of the crystal size and crystal size distribu-
tion is desired.
Crystallization by the Addition of
a Supercritical Antisolvent (SAS)
In the supercritical antisolvent (SAS) process, a press-
urized Suid is used as an antisolvent for precipitating
a solid that is dissolved in a liquid solvent. The super-
saturation of the solid is created by the volumetric
expansion of the liquid solution. After crystallization
of the solute, it is possible to remove the antisolvent
completely by pressure reduction. Control of the par-
ticle size distribution is also possible by manipulation
of the process variables.
Many organic solvents show at least partial misci-
bility with gases and supercritical Suids at moderate
to high pressures. Introduction of the SCF antisolvent
into such organic solvents will result in dissolution of
the antisolvent and an expansion of the liquid phase.
This expansion can be quite signiRcant, as shown for
ethyl acetate}carbon dioxide mixtures in Figure 6. In
Figure 6 Volumetric expansion of a ethyl acetate with CO2.
***, 253C; } } } }, 303C; 2, 403C.
 III / SUPERCRITICAL FLUID CRYSTALLIZATION 4305

CO2 or light hydrocarbons. However, many cobalt,

Figure 7 Experimental apparatus for a batch SAS process.
this Rgure, V(%) is deRned as follows:
V(%)"100V(p, T)!V0/V0
where V(p, T) is the volume of the liquid phase when
loaded with antisolvent, and V0 is the volume of the
pure liquid phase at atmospheric conditions. This
expansion is large near the critical temperature of the
antisolvent.
The following requirements must be satisRed for
a successful SAS process: the solute must be soluble in
the organic solvent at ambient temperatures and in-
soluble (or sparingly soluble) in the SCF antisolvent.
The organic solvent must be at least partially miscible
with the SCF antisolvent as described above. Many
organic solids satisfy these requirements, although
this is not true of inorganic compounds. Inorganic
compounds are generally soluble in water or acids
such as sulfuric acid, but these solvents do not expand
appreciably when contacted with simple gases such as
nickel iron and chromium salts are soluble in acetone,
cyclohexane or N-methylpyrrolidone, and these sol-
vents have been used to develop SAS recrystallization
processes.
The SAS process may involve antisolvent injection
into a liquid phase (gas injection) or liquid solution
injection into a SCF antisolvent (liquid injection) op-
eration. Both these processes can be operated con-
tinuously or in batch mode.
A typical experimental apparatus for batch opera-
tions is shown in Figure 7. In the case of gas injection,
a vessel is loaded with a known quantity of liquid
solution containing the dissolved solute, and then the
SCF antisolvent is added to the solution from the top
or bottom of the vessel. This causes the liquid phase
to expand and the solute to precipitate. The rate of
[1] antisolvent addition is an important parameter for
the control of morphology and size of the solid par-
ticles obtained in this process. Rapid addition of the
antisolvent generally leads to smaller and more uni-
form particles. Slower addition of the SCF can result
in a range of particle sizes. The morphology of the
particles can also be controlled by the rate of anti-
solvent addition, and by the organic solvent used to
dissolve the solute. Examples of particles precipitated
in gas injection operations are summarized in
Table 2.
In the case of liquid injection, the precipitation
vessel is pressurized by the addition of the SCF and
then the liquid solution is injected into the vessel. The
injected liquid solution is expanded by the dissolving
SCF causing the solids to precipitate. In one variation
of this type of operation, the liquid solution and the
SCF antisolvent are continuously delivered to a

Table 2 Substances processed using SAS with gas injection

Compounds Solvent a Antisolvent Morphologies, particle size (
m)

Explosives and propellants

Nitroguanidine DMF, cyclohexane CO2 Crystals: spheres, snow-balls,
starbursts, 1}100
Cyclonite, homocyclonite Acetone, -butyrolactone CO2 Crystals,'200
Cyclonite Acetone, cyclohexanone CO2 Crystals,(5
Homocyclonite Acetone CO2 Crystals, 2}5
Polymers
Aramids DMF CO2 Crystalline spherulites,
1}10 long fibres
Polyhyaluronic acid methyl ester DMSO CO2 Spheres, 0.3

Pharmaceuticals
Abecarnil Isopropyl acetate CO2 Crystals, 10}50

Inorganics
Cobalt chloride Acetone CO2 Crystals

a
DMF, Dimethylformamide; DMSO, Dimethyl sulfoxide.
4306 III / SUPERCRITICAL FLUID CRYSTALLIZATION

hanced dispersion by supercritical (SEDS) Suid pro-

Figure 8 Experimental apparatus for a continuous liquid
injection SAS process.
precipitation vessel in an apparatus similar to that
shown in Figure 8. In this operation, solids precipi-
tate continuously in the vessel, as the gas phase (SCF)
leaves through a pressure-control valve. The valve
also maintains the pressure inside the vessel constant.
The ratio of the two Sow rates (Sow rate of the
liquid solution and that of the SCF antisolvent), and
the type of contact (co-current or countercurrent) can
be important in the evolution of the precipitation
process. Continuous precipitation using liquid injec-
tion has been given various acronyms such as precipi-
tation by compressed antisolvent (PCA), aerosol
solvent extraction system (ASES) and solution en-
cess. These processes have been carried out using
slightly different precipitation procedures and in
slightly different apparatus. At the end of the precipi-
tation procedure, the vessel is washed with antiso-
lvent to remove the liquid. This washing procedure is
necessary because any liquid solvent remaining after
depressurization could redissolve the solute.
Examples of solutes precipitated using liquid injec-
tion are summarized in Table 3. These examples in-
clude polymer microspheres, where the temperature
of the precipitation vessel and the concentration of
polymer in the solution play an important role in
determining the morphology. There is a tendency for
the polymer particles to agglomerate when the tem-
perature is higher than the glass transition tem-
perature of the polymer. Also, a high polymer
concentration in solution produces Rbres. On the
other hand, micron-sized particles with a narrow size
distribution can be obtained by adjusting the condi-
tions of co-solvent and injection devices.
The liquid solution injection device plays a key role
in SAS operations. The injector is designed to produce
very small liquid droplets that expand in the precipi-
tation vessel. Various geometries have been proposed
to achieve this, including nozzles, capillaries, vibra-
ting oriRces and co-axial capillaries. The precipita-
tion vessel must be designed to mix two phases and to

Table 3 Substances processed using SAS with liquid injection

Compounds Solvent a Antisolvent Morphologies, particle

size (
m)

Polymers and biopolymers

Poly (L-lactide) CH2Cl2 CO2 Spheres, 1}10
Polystyrene Toluene CO2 Spheres, 0.1}20
Microballoons
Polyacrylonitrile DMF CO2 Microfibrils, hollow fibres

Pharmaceuticals
Insulin, catalase, trypsin, lysozyme DMSO, DMF CO2 Spheres, 1}5
Methylprednisolone acetate THF Ethane Crystals, 2.5}8.5
Hydrocortisone acetate DMF CO2 Crystals, 2.5}8.5
Salmeterol xinafoate Acetone CO2 Crystalline modification,
1}10
Sodium cromoglycate Methanol CO2 Spheres, 0.1}20
Tetracycline NMP CO2 Spheres, 0.15}0.6
Salbutamol DMSO CO2 Long rods, 1}3 length

Catalysts, inorganics
Red lake C, pigment yellow 1, Acetone CO2 Spheres,'0.6
pigment Blue 15
Barium acetate, copper acetate DMSO CO2 Spheres, 0.1}0.4
Yttrium acetate DMSO CO2 Spheres, 0.1}0.3
Samarium acetate, neodymium acetate DMSO CO2 Spheres, 0.1}0.3
Zinc acetate DMSO CO2 Spheres, 0.05}0.02

a
DMF, dimethylformamide; DMSO, dimethyl sulfoxide; THF, tetrahydrofuran; NMP, N-methyl-2-pyrrolidone.
 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
provide heating/cooling. Filtration of the particles at
high pressures also requires special equipment.
In summary, both RESS crystallization and SAS
crystallization appear to be promising methods for
generating supersaturation and therefore represent
alternatives to conventional crystallization. Such al-
ternatives may prove attractive in applications such
as polymer and pharmaceutical processing, or in par-
ticle design for drug delivery. It is possible to obtain
a variety of morphologies and particle sizes in these
processes by proper choice of conditions and expan-
sion devices. However, a priori design of supercritical
crystallization processes is not yet possible because
the interaction between phase equilibria, expansion
paths, and crystallization kinetics in these processes is
not yet well understood.
See also: II/Crystallization: Control of Crystallizers and
Dynamic Behaviour; Polymorphism.
Further Reading
Berends EM, Bruinsma OSL and van Rosmalen GM (1993)
Nucleation and growth of Rne crystals from supercritical
carbon dioxide. Journal of Crystal Growth 128: 50}56.
SUPERCRITICAL FLUID
EXTRACTION+SUPERCRITICAL
FLUID CHROMATOGRAPHY
H. J. Vandenburg, Express Separations Ltd.,
Roecliffe, N. Yorkshire, UK
Copyright ^ 2000 Academic Press
Introduction
The transfer of extracted analytes to a chromato-
graphy column can be either ofSine or online. In
ofSine analysis, the extracted analytes are collected
and then an aliquot is manually transferred to the
chromatography system. Online analysis is where the
extracted analytes are automatically transferred to
the analytical column. The intrinsic problems with
ofSine collection are that sample loss and contamina-
tion are possible, the process is difRcult to automate
and the sample must be diluted with solvent to allow
transfer, resulting in higher detection limits. Coupling
extraction and chromatography minimizes many of
these problems. Supercritical Suid extraction (SFE)
and supercritical Suid chromatography (SFC) are
ideally suited for coupling together as the most
4307
Dixon DJ, Johnston KP and Bodmeir RA (1993) Polymeric
materials formed by precipitation with a compressed
Suid antisolvent. AIChE Journal 39: 127}139.
Gallagher PM, Coffey MP, Krukonis VJ and Klasutis
N (1989) Gas anti-solvent recrystallization: new process
to recrystallize compounds insoluble in supercritical
Suids. In: Johnston KP and Penninger JML (eds) Super-
critical Fluid Science and Technology, ACS Symposium
Series 406, pp. 334}354. Washington DC: American
Chemical Society.
Griscik GJ, Rousseau RW and Teja AS (1995) Crystalliza-
tion of n-octacosane by the rapid expansion of super-
critical solutions. Journal of Crystal Growth 155:
112}119.
McHugh MA and Krukonis VJ (1994) Supercritical Fluid
Extraction: Principles and Practice, 2nd edn. Boston:
Butterworth-Heinemann.
Palakodaty S, York P and Pritchard J (1998) Supercritical
Suid processing of materials from aqueous solution:
the application of SEDS to lactose as a model substance.
Pharmaceutical Research 15: 1835}1843.
Reverchon E (1999) Supercritical antisolvent precipitation
of micro- and nano-particles. Journal of Supercritical
Fluids 15: 1}21.
Tom JW and Debenedetti PB (1991) Particle formation
with supercritical Suids } a review. Journal of Aerosol
Science 22(5): 555}584.
frequently used solvent, carbon dioxide (CO2), is the
same for both techniques. In the case where pure CO2
is used, the extracted analytes can be deposited at the
start of the analytical column simply by reducing the
pressure, and chromatography started by increasing
the pressure again. Capillary SFC (cSFC) beneRts par-
ticularly from online methods. The columns are small
and easily overloaded, particularly with injection sol-
vent. For example, a 1-
L injection occupies 0.5 m of
a 50-
m i.d. column. Larger injections can easily
cause band broadening and peak splitting. The limita-
tion of injection size increases the detection limit.
A logical method of solving the intrinsic problems of
ofSine collection and cSFC is to link them online.
Samples for which SFE+SFC is
Applicable
The main alternatives to SFC are GC and HPLC.
Online coupling of SFE and HPLC is difRcult, as the
presence of gaseous CO2 is incompatible with HPLC
4308 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
analysis. If the analytes are thermally stable and vol-
atile GC is the best separation technique to use. Many
Savour and fragrance compounds in complex food
samples should therefore be analysed by SFE}GC.
The same is true of polychlorinated biphenyls (PCBs),
pesticides and polyaromatic hydrocarbons (PAHs) in
environmental samples.
When the sample contains thermally labile or react-
ive compounds, SFE}SFC is recommended. The pro-
cedure is excellent for thermally unstable polymer
additives in commercial plastics or for fatty acids and
triglycerides in food, etc. which cannot be analysed
by GC very easily without derivatization. Natural
products such as those containing terpene com-
pounds or hops which contain highly reactive bitter
compounds such as humulone and lupulone must also
be analysed by SFC or HPLC as rearrangement can
easily occur at elevated temperatures. Speciation
studies on organotins, an important environmental
pollutant, are difRcult using GC or HPLC as derivat-
izations are required to increase volatility or provide
a chromophore. Other application areas speciRc to
SFC include the analysis of explosives and certain
steroids, vitamins and other drug residues in biolo-
gical samples. SFE}SFC Rnds important applications
in environmental science. The analysis of pollutants
in matrices such as soil and sediments, and extraction
of sorbents on which pollutants in air or water have
been selectively adsorbed have been analysed with
this technique.
Capillary and Packed Column SFC
There are two broad categories of SFC, capillary and
packed column. Capillary SFC was developed from
capillary GC, and packed column SFC is more akin to
HPLC. There are advantages to each. cSFC uses open
tubular capillaries with bonded stationary phases.
Compounds with differing solubilities in CO2 are
eluted using pressure programming, where the pres-
sure, and hence density and solvent strength of the
mobile phase is increased during the separation. This
is the equivalent of temperature programming in GC.
Use of modiRers is rare, partly due to difRculties of
mixing at very low Sow rates and partly because the
universal FID cannot be used with modiRers present.
Open tubular capillaries offer little resistance to the
Sow of the Suid and columns can be long. A major
problem with capillary SFC is the low sample capa-
city. The capillary columns are easily overloaded and
very small injections are required, reducing sensitiv-
ity. Packed column SFC uses columns packed with
HPLC packing materials. Small particles offer a high
resistance to the Suid Sow, and hence there is a pres-
sure drop across the column. This results in a reverse
density gradient along the column, in which the Suid
has the lowest solvent strength at the elution end of
the column. This gradient is working against any
pressure gradient appl ied, and can lead to precipita-
tion of solutes. Elution in packed column SFC is now
often controlled by addition of a modiRer such as
methanol rather than pressure programming. Use of
modiRers means that the FID cannot be used, and
detection for packed column SFC is more usually by
UV absorbance detectors. However, modiRers allow
more polar stationary phases to be used, which have
much greater interaction with polar molecules. When
CO2 alone is used, the stationary phase must also be
nonpolar, otherwise the solvent strength is not sufR-
cient to elute polar compounds. The analyte interacts
only poorly with both stationary and mobile phases,
resulting in poor peak shape. The poor results with
polar compounds on packed SFC columns has also
been attributed to polar active sites (residual silanols)
present on the silica. These are thought to be better
shielded in coated capillaries. The solvent strength of
modiRed CO2 can be varied from similar to pentane
for pure CO2 to similar to acetonitrile with addition
of 40% methanol.
The different natures of capillary and packed col-
umn SFC also lead to differences in instrumentation.
The Sow rates in cSFC are very low, and pressure is
usually controlled by restrictors. These can be linear
capillaries whose diameter and length can be adjusted
to provide the required pressure. Adjustable, heated
needle valves have also been used. The problem with
whichever system is used is that the restrictor is
a passive device, limiting mass Sow at the pressure set
by the pumps. Blockages can occur, and the Sow rate
is not well controlled. Flow rates in packed column
SFC are much higher, which allows the use of manual
or automatic back pressure regulators, which control
the pressure independently of Sow rate. Pressure, Sow
rate and solvent composition can, therefore, be much
better controlled in packed column SFC. In reality,
packed column and capillary SFC are very different
techniques, with different areas of application.
SFE+SFC Interface
The analytes extracted during the SFE step can be
introduced onto the analytical column in two main
ways. The SFE extract can be passed through
a sample loop and an aliquot directed to the SFC
column, or the analytes can be trapped after the SFE
and introduced onto the column in one go.
Aliquot Sampling
The simplest of interface for SFE}SFC is by aliquot
sampling. A part of the extract is sampled by passing
 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
it through an injection loop of the SFC system.
A closed- or an open-loop system may be used.
Closed-loop static SFE}cSFC involves the sample be-
ing sealed in an extraction cell for a period of static
extraction. The extraction cell is connected to the
sample loop of an injection valve. The analytes dif-
fuse to the loop, and after equilibrium is reached the
valve is actuated and an aliquot is injected into the
SFC column.
The major advantage of this procedure is that small
aliquots of the extract can be taken for consecutive
analysis with virtually no difference in the extraction
proRle. However, a major disadvantage is that the
solute containing extraction Suid has to reach equi-
librium and diffuse out of the cell and into the injec-
tion valve before sampling is made. This can take
many hours before complete equilibrium is attained.
Recirculating pumps could be used to reach equilib-
rium in a shorter time, but these can easily become
contaminated.
The system can be sampled more rapidly by allow-
ing a portion of the extraction solution to pass
through the loop to atmosphere, to Sush the loop
with fresh solution. A low-Sow restrictor is connected
to a valve inline after the injector, as shown in
Figure 1. Static extraction can be carried out with the
high-pressure valve closed. Opening this valve to the
restrictor allows dynamic extract and Rlling of
the sample loop. Actuation of the rotary valve passes
the contents of the loop to the analytical column, and
either static or dynamic extraction can be continued.
This is known as open-loop SFE, and with this conRg-
uration one also has the opportunity of passing the
sample through a detector (UV or FID). At periodic
Figure 1 Schematic of open-loop aliquot sampling system (A) Filling loop, dynamic extraction mode. (B) Injecting to column.
4309
intervals aliquots of the extract can be injected into
the SFC column for analysis.
Aliquot sampling diverts only a small portion of
the extract to the SFC column, and is therefore not
suitable for quantitative SFC analysis. SFE}SFC with
aliquot sampling is a good technique for basic quali-
tative investigation and for measuring fundamental
parameters such as partition coefRcients of solutes in
supercritical Suids. However, it is limited in that it is
not usually suited to quantitative or trace analysis
where analytes in the whole extract must be accumu-
lated prior to chromatographic analysis.
Trapping of Analytes
In contrast to static extraction with aliquot sampling,
dynamic SFE}SFC operates principally by continu-
ously exposing the analytes to a fresh stream of super-
critical Suid. Extracted components are accumulated
from this stream in a trap of some kind. Only after
extraction is complete are the trapped analytes trans-
ferred to the SFC column for analysis. The major
advantages of dynamic SFE}SFC are that it is much
more rapid than static SFE}SFC and that trace analy-
sis can be performed. The whole of the extracted
material is passed to the SFC column, therefore the
sensitivity is much greater than for ofSine analysis.
Figure 2 shows a schematic of a simple online
SFE}SFC system. A high-pressure syringe pump sup-
plies the extraction cell with Suid. The outlet of the
cell is connected to a capillary Sow restrictor which is
connected to an accumulating trapping system. Dur-
ing extraction the depressurized gas from the restric-
tor passes through the trap and is then vented to the
atmosphere through valve 1. The extracted analytes
4310 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
Figure 2 Schematic of SFE}SFC system.
are concentrated within the trap. After extraction is
completed, valve 1 closes and valve 2 opens, switch-
ing the CO2 onto the SFC column. The rotary valve
switches the Sow to the trap to avoid the cell and
associated restrictor. This raises the pressure within
the trap and the CO2 becomes a supercritical Suid
and capable of dissolving the trapped analytes and
carrying them to the column.
If uncoated fused silica tubing is used to connect
the trap to the analytical column (the retention gap),
the analytes will, in theory, be unretained during the
transfer. The pressure of CO2 needed to effect the
transfer need only be enough to provide some solubil-
ity of the analytes. Once they reach the stationary
phase Rlm of the SFC column they become concen-
trated as a narrow band, as the relatively low
density solvent is not strong enough to elute the
compounds from the stationary phase. After trapping
is complete, the chromatography can be initiated us-
ing a pressure programme. If such phase ratio focus-
ing occurs successfully, then good chromatographic
efRciency is observed during the separation. If this
process works well, the length and internal diameter
of the retention gap do not signiRcantly affect the
resolution.
Other more complicated systems have been re-
ported using on}off and multiport switching valves to
allow continuous extraction or to permit the extrac-
tion cell to be vented during simultaneous chromato-
graphic analysis. The plumbing of such a system can
be constructed to any speciRc requirement.
Since analytical SFE is most often performed with
Suids that decompress to gases at ambient conditions
(such as CO2, 1 mL min\1 of which produces a gas
Sow of approximately 500 mL min\1), the success of
trapping depends on the success of recovering the
analytes from the expanded gas. Fast Sow rates tend
to elute volatile analytes from the trap, thus, for
quantitative results, recovery of extracted compo-
nents should be performed at lower Sow rates. The
problem of loss of volatile analytes is often not severe
in SFE}SFC, as these are likely to be analysed by
SFE}GC. Therefore SFE}SFC traps generally need be
more concerned with trapping less volatile materials.
Trapping procedures
There are several methods of trapping extracted com-
ponents from dynamic SFE in preparation for online
SFC analysis. The requirements are to efRciently trap
all the material from the gas or low-pressure stream
from the extractor, and then to release all the compo-
nents when the Sow is switched to the analytical
column. Two methods are used for this:
E cryogenic trapping; and
E trapping on an adsorbent stationary phase; the
stationary phase can be either on particles packed
into the trap, or coated onto a fused silica capillary.
Cryogenic trapping
Trapping on uncoated fused silica retention gaps A
length of uncoated fused silica capillary can be cooled
by expanding CO2. Solutes passing through the capil-
lary in the depressurized gas stream from the SFE will
be trapped in the cooled section. The cooling can then
be switched off, and the section pressurized with CO2
to redissolve the analytes. Figure 3 shows an arrange-
ment for a cryogenically cooled fused silica trap. In
this arrangement the expanded mobile phase from the
extraction cell is released from a different outlet than
the incoming CO2 for the SFC. This minimizes con-
tamination of the system from previous analyses. The
extracted analytes are in contact only with deac-
tivated fused silica after leaving the extraction cell,
which reduces loss of polar analytes by adsorption on
metal surfaces.
The Sow rate of the expanding extraction Suid and
the temperature at which analytes are trapped mark-
edly affect the recoveries obtained when uncoated
fused silica tubing is used. In many systems, linear
extraction restrictors are used, since they provide the
correct Sow rate range for online coupling with capil-
lary SFC. They also tend not to plug as quickly as
other restrictors when used for SFE. The length and
internal diameter of the capillary restrictor tubing
Figure 3 Cryogenically cooled trap.
 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
and the pressure at which the extraction is performed
should therefore be considered to obtain suitable Sow
rates.
Restrictors with internal diameters greater than
30
m result in higher extraction efRciencies, but
lower recoveries and signiRcant band broadening of
more volatile components. However, restrictors with
internal diameters less than 15
m do not allow sufR-
cient Sow for efRcient extractions over a short period
of time, but yield good chromtographic peak shapes.
As a rough guide, the gaseous Sow rates from 15-cm
lengths of 15-, 20-, 25- and 30-
m restrictors at
a pump pressure of 300 atm are, approximately, 80,
150, 240 and 300 mL min\1, respectively. A good
compromise therefore is to use a restrictor with a Sow
rate of 100}200 mL min\1. Lengths of capillary tub-
ing of 20 or 25
m i.d. are suitable for most needs.
The trapping efRciency is also strongly dependent
on the trapping temperature. The higher the temper-
ature, the more volatile components will be lost from
the trap. Cooling in the region of !403C to !603C
will allow trapping of C10 hydrocarbons with reason-
able efRciency. The trap should only be cooled to
a sufRcient temperature to trap the analytes of inter-
est, as too low a cryofocusing temperature may result
in restrictor plugging, or components, such as water,
freezing in the restrictor. This reduces the rate of
extraction and makes it difRcult to reproduce ana-
lyses. An alternative arrangement for trapping vol-
atile substances is to keep the restrictor hot and
deposit the analytes in the transfer line held in
a cryogenically cooled oven as shown in Figure 4.
The use of micropacked columns has also been
reported. In this case the restrictor can be vented onto
the head of the analytical column. The cooling of the
expanding gas cools the column and the analytes are
deposited on the packing at the start of the column.
Trapping on coated fused silica retaining pre-columns
An alternative to the cryotrapping method is the use
Figure 4 Arrangement for keeping restrictor hot and trapping in
cryogenically cooled oven.
4311
of a coated fused-silica retaining pre-column for
concentrating extracted solutes. Compared to un-
coated fused silica, coated columns such as GC
columns are much more effective at trapping. The
key is to trap effectively, but allow the mobile phase
to elute the trapped materials during the pressure
programme. It is likely that a column coated with
a similar material to the analytical column will be
effective. The phase thickness on the column is also
important, thicker phases having a greater trapping
power. This method allows the trapping at room
temperature using widely available bonded-phase GC
columns.
Trapping on sorbent traps Sorbents may also be
used as an effective method of trapping. This entails
the use of short traps (usually 2 cm in length) packed
with organic sorbents such as Tenax-GC, Carbotrap
or with HPLC packing materials. Bonded silica and
polymeric stationary phases designed for solid-phase
extraction (SPE) are available with a wide variety of
functionality, and would make ideal packing material
for this application. These materials will effectively
trap the analytes from the low-pressure gas stream,
and can then be desorbed by high-pressure supercriti-
cal CO2. The considerations are similar to those when
using coated silica columns. It is important when
using such a system that breakthrough of the analytes
from the sorbent does not occur and also that the
desorption behaviour is suitable for online chromato-
graphic analysis. Desorption is performed by increas-
ing the trap temperature or by using the supercritical
Suid to desorb the sample. The process is effectively
the same as SPE, with supercritical CO2 as the desor-
bing solvent. The stationary phase should be selected
to have a strong enough afRnity to trap the analytes
from the gas stream, but to be desorbed by supercriti-
cal CO2. Supercritical CO2 is essentially non-polar,
and it is unlikely that polar compounds could be
eluted from polar stationary phases. It is not always
possible to elute all the trapped analytes with CO2,
and supercritical nitrous oxide has been found to be
more effective than supercritical carbon dioxide in
removing solutes from adsorbents. However, the
oxidizing nature of this material has resulted in ex-
plosions, and is not recommended. It is therefore
more important to select the most appropriate sta-
tionary phase which will trap the analytes, and then
be desorbed by the mobile phase.
Use of Modi\ers and Solvent Venting
Although CO2 is a versatile extraction solvent, some-
times modiRers are needed to solvate particular
analytes or overcome analyte}matrix interactions.
4312 III / SUPERCRITICAL FLUID EXTRACTION^SUPERCRITICAL FLUID CHROMATOGRAPHY
This presents a problem in SFE}SFC. With cryogeni-
cally cooled traps, the modiRer will be trapped and
block the restrictor, or Sood the column when the
Sow is switched to the analytical column. If the modi-
Rer becomes liquid after depressurization, it will dis-
solve the analytes and elute them from coated traps.
Coated capillaries can be used to trap the analytes,
provided the modiRer is present at a sufRciently low
concentration to remain as a vapour in the CO2 gas
stream. Therefore the upper limit for the modiRer
addition is that at which CO2 is saturated at atmo-
spheric pressure and the trapping temperature. For
methanol the maximum addition at 253C is 14%. It is
important that the pressure in the trap does not rise,
as this may cause the modiRer to liquefy. Wide-bore
coated capillaries may be needed for the trap to re-
duce back pressure, and a second, narrow-bore col-
umn will catch any breakthrough from the wide-bore
trap. A short gas purge will remove any residual
modiRer, and the analytes can then be transferred to
the analytical column dissolved by supercritical CO2.
It may be necessary to introduce a refocusing trap,
which will focus the analytes from the supercritical
CO2, as the trap volumes may be quite large, which
would otherwise lead to band broadening.
Apart from use of modiRers, other situations occur
when large amounts of solvent are trapped with the
analyte. Co-extraction of low-molecular-weight sol-
vents or reactants along with the desired analytes is
one example. Provided the co-extractant is sufR-
ciently volatile and the analyte involatile, then the
unwanted material can be removed from the inter-
mediate trap by gas purge. The analytes can then be
transferred to the analytical column with supercriti-
cal CO2.
SFE as a Sample-Introduction
Technique
As stated previously, one of the problems of cSFC is
sample introduction without Sooding the column
with solvent. Aqueous samples are a problem for
capillary and packed-column SFC, as water is only
slightly soluble in CO2. SFE can be used as a solvent-
less sample introduction technique to avoid this prob-
lem. One method to achieve this is to inject the
sample onto a pre-column Rtted with a restrictor. The
solvent will Sood the column for some distance. The
solvent can be removed by gas purging, leaving the
less volatile analytes behind. The entire pre-column is
then pressurized with supercritical CO2 to dissolve
the analytes and carry them to the analytical column.
In effect, the pre-column is acting as an SFE cell.
Samples dissolved in aqueous media can be concen-
trated and transferred to packed or capillary columns
while maintaining high efRciency. The use of solid
sorbents has proved very useful in sample introduc-
tion to SFC. The dissolved analytes are injected onto
a sorbent, the solvent can then be removed by evapor-
ation and the analytes transferred to the analytical
column using SFE}SFC. The whole process has been
called SPE}SFE}SFC. This method is particularly ap-
plicable to biological samples where the analyte has
no chromophore. These are often thermally labile,
and therefore analysis by GC or HPLC is problemati-
cal. Direct sample introduction to SFC is also a prob-
lem due to the aqueous nature of the samples. Use of
an intermediate trap and solvent purging to remove
the water and introduce the analytes to the SFC
column allows much larger samples to be introduced,
improving sensitivity by a factor of 100 or more. In
environmental analysis, samples of several hundred
millilitres can be passed through a solid-phase extrac-
tion cartridge to concentrate impurities. The car-
tridge can then be eluted with CO2 to the analytical
column. This system could also be used as an
HPLC}SFC interface.
Optimization of Conditions for
SFE+cSFC
A number of parameters must be optimized for suc-
cessful analysis by coupled SFE}cSFC. Principal
among these are the conditions for quantitative ex-
traction. This should begin with a determination of
the supercritical Suid extractability of the analyte(s)
from the non-sorptive matrices (Rlter paper, etc.) to
assess the appropriate solvent, density and temper-
ature conditions. Trial runs on spiked samples then
allow investigations of matrix}solute interactions; if
necessary these may be overcome by a period of static
extraction. The kinetics of extraction must then be
determined in order to deRne the required extraction
time.
Factors affecting the efRciency of intermediate
trapping must then be addressed. The nature of the
analyte is crucial, while the possible presence of co-
extracted, interfering compounds demands either sel-
ectivity during extraction, or the trapping on an
adsorbent from which selective desorption into the
SFC column is possible. The sample size must be
carefully chosen so that the capacity of the SFC
column is not exceeded, and the extracting supercriti-
cal solvent must be of sufRcient purity to avoid
introduction of extraneous material into the column.
Finally, the conditions for efRcient SFC analysis
must be optimized, preferably ofSine. Correct
choice of column, temperature and pressure/density
programme are vital. Compromises may be inevitable
if the extracted analytes have a range of polarities.
 III / SUPERHEATED WATER MOBILE PHASES: LIQUID CHROMATOGRAPHY
Conclusion
Coupled SFE}SFC has shown itself to be a very useful
technique for those samples for which it is applicable.
The ability to transfer all the extract to the analytical
column without manipulation increases sensitivity,
reduces contamination and sample handling. The
overloading of capillary columns is avoided. Now
that methods for using modiRers in the extraction
solvent and SFE sample injection methods have been
developed, there is every likelihood that SFE}SFC
will become a more widely used technique.
See also: II/Chromatography: Supercritical Fluid:
Fourier Transform Infrared Spectrometry Detection; His-
torical Development; Instrumentation; Large-Scale Super-
critical Fluid Chromatography; Theory of Supercritical
Fluid Chromatography.
SUPERHEATED WATER MOBILE PHASES:
LIQUID CHROMATOGRAPHY
R. M. Smith, Loughborough University,
Loughborough, Leics, UK
Copyright ^ 2000 Academic Press
At room temperature, water on its own is an unat-
tractive solvent in liquid chromatography. In rever-
sed-phase chromatography, water is a weak eluent
and is often regarded as an inert component of the
mobile phase. It is mainly used to dilute a stronger
organic component and thus control the overall elu-
ent strength. In contrast, in normal-phase chromato-
graphy, water is a powerful eluent and interacts
strongly with the stationary phase, often deactivating
it. Even trace amounts in a nonpolar eluent (or even
in a sample) will markedly alter the retention proper-
ties of a silica surface. In separation methods aqueous
eluents are used primarily for ion exchange chromato-
graphy or for the size exclusion separation of biolo-
gical molecules.
However, this represents the situation at room tem-
perature and atmospheric pressure. When liquid
water is heated under pressure, its dielectric constant,
viscosity and surface tension all decrease. These
changes in the properties of water are well known but
have largely remained the province of the physical
chemist and chemical engineer. They have been
widely studied because of the importance of water as
a heat transfer agent and they play their part in the
design and construction of steam power generation
4313
Further Reading
Anton K and Berger C (eds) (1998) Supercritical Fluid
Chromatography with Packed Columns. New York:
Marcel Dekker.
Berger TA (1995) Packed column SFC. RSC Chromato-
graphy Monographs. Cambridge: The Royal Society of
Chemistry.
Clifford T (1999) Fundamentals of Supercritical Fluids.
Oxford: Oxford University Press.
Ramsey ED (ed.) (1998) Analytical Supercritical Fluid
Extraction Techniques. Dordrecht: Kluwer Academic
Publishers.
Wenclawiak B (ed.) (1992) Analysis with Supercritical
Fluids: Extraction and Chromatography. Berlin:
Springer-Verlag.
Westwood SA (1993) Supercritical Fluid Extraction and its
Use in Chromatographic Sample Preparation. London:
Chapman and Hall.
plant and in related areas. Above 3743C under a pres-
sure of 221 bar, a single supercritical phase is ob-
tained. Although these conditions seem extreme for
the laboratory, they occur in nature in the ocean
depths at the spreading points in the earths crust
where water issues from fumeroles at 350}4003C and
250 bar.
In recent years organic chemists have been attrac-
ted by the possibility of using superheated or super-
critical water to achieve clean solvent-free conditions
and to generate novel reaction conditions which are
not available at room temperature. It has also been
employed as a solvent for the high temperature oxida-
tion for waste remediation or for the destruction of
hazardous materials such as nerve gases and explos-
ives as an alternative to high temperature inciner-
ation. In inorganic chemistry, supercritical water has
been used as a solvent to enable high temperature
reactions to be carried out without the inconvenience
of using molten salts.
However, the analytical chemist has made little use
of water under pressure, although the potential of
supercritical water as a Suid solvent for chromato-
graphy was recognized by Lovelock in 1958. Some
work has exploited steam as a mobile phase in gas
chromatography, but the condensed phase has largely
been ignored. Although liquid chromatographers
have used elevated temperatures to improve separ-
ations or efRciencies, in almost every case the
composition of the organic}aqueous eluent was kept
4314 III / SUPERHEATED WATER MOBILE PHASES: LIQUID CHROMATOGRAPHY
Figure 1 Change in dielectric constant of water with temper-
ature.
constant and only the effect of the temperature was
studied, typically up to 70}803C.
Recently, the changes in the properties of liquid
water above 1003C have attracted the interest of the
analytical chemists, who have recognized that these
low polarities might provide an environmentally
clean solvent for extraction. Subsequently, the use of
superheated water has also been applied to liquid
chromatography. Some of the published papers have
referred to this region of the phase diagram as sub-
critical water but this could imply any temperature
less than 3743C. This review will instead employ the
expression superheated water, which is deRned as
water held as a liquid under pressure between 1003C
and the critical point.
Properties of Superheated Water
As the temperature of water is raised under a sufR-
ciently high pressure for it to remain in the liquid state
Figure 2 Change in vapour pressure of water with temperature.
there is a steady decrease in its dielectric constant
(Figure 1) from about 80 at room temperature to less
than 25 at 3003C. These changes represent a marked
change in the solvent polarity of the water. By
2503C the dielectric constant of water is about 30,
which is less than that of methanol at room temper-
ature, so that even under conditions well below its
critical point, water will resemble the polarity of com-
mon organic solvents used as eluents in reversed-phase
liquid chromatography. There is also a decrease in the
viscosity of water from 1.0 cp at room temperature to
0.28 cp at 1003C, as well as in the surface tension.
The pressure conditions usually needed to carry out
supercritical Suid chromatography (SFC) with carbon
dioxide can be as high as 300}500 bar in order to
achieve a sufRciently high density to provide a reason-
able solvent strength. However, the vapour pressure
of water (Figure 2) is modest and even by 2003C only
reaches 15 bar, so that only moderate pressures are
required to maintain a liquid state. In addition, the
density of hot liquid water changes by only a small
amount with changes in the applied pressure. The
solvation properties are thus effectively independent
of pressure. This is in marked contrast to supercritical
carbon dioxide where pressure control is critical be-
cause of its marked effect on density.
Application in Analytical Extractions
The Rrst serious analytical chemistry interest in the
potential of water under pressure above 1003C came in
1994, when Hawthorne and colleagues investigated
the extraction of organic pollutants from environ-
mental solids with supercritical and subcritical water.
They had been prompted by reports that the solubility
of benzo[a]pyrene increases from 4 ng mL\1 under
ambient conditions to approximately 10% by weight
 III / SUPERHEATED WATER MOBILE PHASES: LIQUID CHROMATOGRAPHY
at 3503C and 100 bar. This represents an increase in
solubility of about 25 million. They found consider-
able solubility for polar analytes, such as chlorinated
phenols, and even a signiRcant solubility for nonpolar
analytes, such as naphthalene at 503C. Raising the
temperature resulted in an increased solubility for
polynuclear aromatic hydrocarbons (PAHs) and by
2503C all their test compounds except the n-alkanes
had been completely extracted. The alkanes required
supercritical conditions and were completely extracted
at 4003C. These results conRrmed that superheated
water had a sufRciently high solubilizing power to be
used as a solvent for even nonpolar analytes. They
then demonstrated that subcritical conditions of
2503C and 50 bar would also efRciently extract PAHs
from soil samples. In addition, good recoveries were
obtained from air particulates. These results also
showed that pressure was not an important factor, in
marked contrast to the pressure dependence of super-
critical Suid extraction with carbon dioxide.
Subsequently they studied the extractions further
and showed that polychlorinated biphenyls (PCBs)
could be extracted from soils and sediments with
subcritical water. At 3003C and 50 bar, complete
extraction could be achieved in pure water. Other
workers have found that water can be used for the
extraction of PCBs from a range of matrices in good
yield. In a similar study, the pesticides Dacthal and
acid metabolites have been extracted from soil with
superheated water.
Chromatography using Superheated
Water as the Mobile Phase
It was realized that if superheated water could
extract PAHs from soils and dissolve PCBs, then it
Figure 3 Superheated water chromatograph. Components: 1, pump; 2, injection valve; 3, preheating coil; 4, column; 5, detector; 6,
back-pressure regulator.
4315
was behaving as a less polar solvent than the meth-
anol} water and acetonitrile}water mixtures conven-
tionally used as mobile phases in reversed-phase
liquid chromatography. It should therefore be pos-
sible to use superheated water as a mobile phase and
achieve typical reversed-phase liquid chromato-
graphic separations. Many studies have examined the
effect of increasing the temperature on separations
and have shown that there is a consistent drop in
retention with an inverse relationship to the absolute
temperature (k varies as 1/T K). However, most of
this work has either looked at temperature effects
using constant eluent composition or has been limited
to 80}903C by the volatility of the organic compo-
nents of the mobile phase.
In 1996, Smith and Burgess demonstrated that un-
der a modest pressure it was possible to carry out
the reversed-phase separation of phenols using super-
heated water at 1603C on a polystyrene-divinylben-
zene (PS-DVB) column. The equipment was a
combination of high performance liquid chromato-
graphy (HPLC) and gas chromatography (GC) systems
(Figure 3) with some components from a packed-col-
umn SFC system. The water mobile phase was pum-
ped using a single reciprocating pump but, unlike
SFC, there was no need to cool the pump heads to
condense the mobile phase. As the mobile-phase po-
larity can be controlled by temperature, no modiRer
pump was needed. The column was heated in a GC
column oven which enabled the temperature to be
controlled up to 3503C. To maintain the pressure
a SFC back-pressure regulator was used. A detector
Rtted with a high pressure Sow cell was originally
employed but, because the back-pressures required
are relatively low, in later studies standard HPLC
spectroscopic Sow cells were used for Suorescence
4316 III / SUPERHEATED WATER MOBILE PHASES: LIQUID CHROMATOGRAPHY
and ultraviolet-visible spectroscopy. It is also possible
to replace the back-pressure regulator with a length
of narrow-bore PEEK tubing.
A PS-DVB packed column was used in the Rrst
studies because this material can be used without
problems in size exclusion chromatography at 1603C
and is thermally stable at this temperature. Phenols
were examined with a methanol}water eluent and
then the methanol content was reduced while raising
the temperature in steps. With each increase in tem-
perature the retention times decreased. By 1803C in
the absence of methanol, the samples had similar
retention times to those in 20 : 80 acetonitrile}water
at room temperature.
The relative retention of a wide range of analytes
from phenols, amides, esters to simple aromatic com-
pounds demonstrated that the retention followed
a similar pattern to conventional reversed-phase
liquid chromatography. The separation followed the
hydrophobicities of the analytes and the homologous
parabens eluted in order of increasing chain length.
They were also stable, showing neither hydrolysis nor
oxidation. The separations, as expected, were insensi-
tive to the back-pressure applied to the column.
It was also realized that programming the temper-
ature of the column during the separation would
systematically reduce the eluent polarity. This pro-
duced an effect similar to gradient elution, and would
speed up and focus later peaks. Inorganic buffers
could also be added to control the pH without caus-
ing any problems.
Stationary Phases for Superheated
Water Chromatography
Most of the work that has been reported has em-
ployed PS-DVB columns which have shown reason-
able temperature stability. They can be used up to
about 2203C before softening appears to reduce their
lifetime. As in ordinary HPLC, these columns show
a marked difference in the retention of nonpolar and
hydrogen-bonding analytes. The latter, including al-
cohols and phenols, have markedly lower retentions
than nonpolar analytes such as alkylbenzenes and
nitrobenzene. These latter compounds cannot be eas-
ily eluted even at 2303C.
Porous graphite carbon (Hypercarb) has been
examined as an alternative thermally stable station-
ary phase. No instability was observed and the separ-
ations of mixtures of phenols, anilines and aryl
amides were similar to those obtained at room tem-
perature with conventional eluents (Figure 4).
There is particular interest in octadecylsilica
(ODS)-bonded silica phases because of their wide-
spread use in conventional liquid chromatography.
Figure 4 Separation of amides on porous graphitic carbon col-
umn at 1903C. Analytes: 1, benzesulfonamide; 2, benzamide; 3,
m-toluamide.
However, although a number of different silica-based
bonded-phase materials have been examined, they all
show quite rapid degradation at temperatures greater
than 1003C. Even though they have a lower retention
capacity than the PS-DVB columns and do not re-
quire such high temperatures to obtain elution, it
appears that this matrix is insufRciently stable for
routine use.
A second problem was that, when highly end-
capped ODS-bonded materials were examined with
100% water as the eluent, the octadecyl chains col-
lapsed on to the silica surface on cooling the column
to room temperature. As a result, the retention capa-
city of the column dropped markedly and this was not
restored by heating. Instead, the column had to be
treated with methanol}water mixtures. Similar prob-
lems have been reported in reversed-phase liquid
chromatography at room temperature when the mo-
bile phase contains less than 2% methanol.
A promising alternative stationary phase is poly-
butadiene-coated zirconia, a relatively new material,
which is reported to be stable in water at 2003C. It
gives good separations and peak shapes. The order of
elution is similar to that on ODS phases in conven-
tional reversed-phase liquid chromatography.
The elevated temperatures would be expected to
reduce mass transfer effects in the mobile phase be-
cause of higher diffusion rates and this would result in
improved separation efRciencies. Van Deemter curves
of column plate heights against mobile-phase Sow
rate of water at elevated temperature determined for
the PS-DVB and Hypercarb columns have been com-
pared with acetonitrile}water separations at room
temperature. In both cases, at the optimum Sow rates
the height equivalent to one theoretical plate (HETP)
values are similar, but in superheated water the efR-
ciency decreases rapidly at lower Sow rates. This
 III / SUPERHEATED WATER MOBILE PHASES: LIQUID CHROMATOGRAPHY
effect can be ascribed to a higher diffusion rate in the
mobile phase at the higher temperature.
Detection in Superheated Water
Chromatography
One of the advantages of superheated water chromato-
graphy is that it increases the possible number of
detection methods that can be employed compared to
liquid chromatographic methods using an organic
solvent. However, with some detectors the mobile
phase had to be cooled to room temperature to avoid
baseline instability due to refractive index effects. So
far no problems have been experienced due to the
analytes coming out of solution between the column
and detector, probably because the concentrations
are generally low and the transfer time to the detector
is brief.
Because usually only low back-pressures (less than
50 bar) are required to maintain the liquid state in the
column, standard liquid chromatography Sow cells
can frequently be employed for ultraviolet-visible and
Suorescence spectroscopic detection. Alternatively,
high pressure ultraviolet-visible Sow cells designed
for SFC application can be used. In both methods of
detection, one advantage of water as an eluent is that
is it is transparent down to 190 nm. This enables low
wavelength detection of unconjugated double-bond
chromophores without solvent interference. How-
ever, some Suorescence detection is reduced com-
pared to organic solvents because of quenching by the
polar aqueous solvent.
The absence of an organic modiRer raised the pos-
sibility that the eluent could be passed to a Same
ionization detector. This could provide a simple
method of universal detection for liquid chromato-
graphy, which previously has only been obtainable
using mass spectrometry. This possibility was realized
by Miller and Hawthorne, who demonstrated the use
of the FID in 1997 to detect alkanols, phenols and
amino acids, and conRrmed by others.
Another detector that has problems in conven-
tional liquid chromatography, because of mobile
phase interference, is on-line nuclear magnetic reson-
ance spectroscopy (LC-NMR). Many of the problems
can be overcome by employing superheated heavy
water (deuterium oxide) as the mobile phase. Com-
pared to deuterated organic modiRers, deuterium
oxide is relatively cheap and unlike supercritical Suid
chromatography the Sow cell can be at room temper-
ature and pressure. This makes stop-Sow detection
easier and characteristic proton-NMR spectra have
been obtained for a range of compounds, including
barbiturates, sulfonamides and a number of pharma-
ceuticals and natural products.
4317
It has also been demonstrated that superheated
water chromatography can be linked to mass
spectrometry using a standard LC-interface to give a
superheated water LC-NMR-MS system. These sep-
arations using superheated deuterium oxide have also
provided some interesting exchange reactions which
are more selective and speciRc than those reported
with supercritical deuterium oxide.
Application of Superheated Water
Chromatography
A wide range of analytes (Table 1) has been examined
by superheated water chromatography. They have gen-
erally been moderately polar and could be characterized
as analytes where conventional liquid chromatography
would employ a mobile phase with 60% or less organic
modiRer. Less polar analytes, such as alkylbenzenes, can
currently cause problems because they require a mo-
bile-phase temperature above the limit of the poly-
meric stationary phases primarily used so far.
The principal groups of compounds examined so
far have been phenols (Figure 5), alcohols, amino
acids, esters, pharmaceuticals, water-soluble vitamins
and lactone natural products. The method is still
relatively new and further applications are constantly
being developed. Although there was concern that the
separation conditions might cause sample oxidation,
hydrolysis or degradation, so far few compounds
have caused problems. Not surprisingly, aspirin is
hydrolysed but this occurs readily even at room tem-
perature. Nitrobenzene appears to degrade and there
is some suggestion that other nitro-compounds are
also thermally unstable in hot water. In contrast,
compounds such as the parabens (4-hydroxybenzoate
Table 1 Typical compounds which have been separated by
superheated water chromatography
Aryl aldehydes
Amino acids
Aryl alkyl ketones
Aryl amides
Aryl amines
Arylsulfonamides
Parabens
Pharmaceuticals, including:
Barbiturates
Caffeine
Paracetamol
Phenacetin
Sulfonamides
Phenols, including:
Cresols
Guaiacol
Methoxyphenols
Phenol
1,2,3-Trihydroxybenzene
4318 III / SURFACTANTS / Chromatography
Figure 5 Functional group selectivity of PS-DVB column. Conditions: column, PLRP-S; temperature, 2003C. Solutes: 1, hydro-
quinone; 2, p-cyanophenol; 3, phenol; 4, p-methoxyphenol; 5, p-cresol; 6, p-bromophenol; 7, 3,5-xylenol; 8, 2,4-xylenol.
esters) which might be thought to be susceptible
both to oxidation and to hydrolysis, have been separ-
ated without difRculty. One reason may be that, as
the temperature is raised and the water becomes less
polar, it also becomes a weaker hydrolysis agent.
See also: II/Chromatography: Liquid: Mechanisms:
Reversed Phases; Nuclear Magnetic Resonance De-
tectors. Extraction: Supercritical Fluid Extraction. III/En-
vironmental Applications: Pressurized Fluid Extraction.
Porous Polymers: Liquid Chromatography.
Further Reading
Chienthavorn O and Smith RM (1999) Buffered superheated
water as an eluent for reversed-phase high performance
liquid chromatography. Chromatographia 50: 485}489.
Hawthorne SB, Yang Y and Miller DJ (1994) Extraction of
organic pollutants from environmental solids with sub-
and supercritical water. Analytical Chemistry 66: 2912.
SURFACTANTS
Chromatography
J. G. Lawrence, Unilever Research
Bebington, Merseyside, UK
Copyright ^ 2000 Academic Press
Introduction
Surfactant is a contraction of surface active
agent. It has come to be used interchangeably with
Kuhlmann B, Arnett EM and Siskin M (1994) Classical
organic reaction in pure superheated water. Journal of
Organic Chemistry 59: 3098.
Miller DJ and Hawthorne SB (1997) Subcritical water
chromatography with Same ionisation detection. Ana-
lytical Chemistry 69: 623.
Smith RM and Burgess RJ (1996) Superheated water } a
clean eluent for reversed-phase high-performance liquid
chromatography. Analytical Communications 33: 327.
Smith RM and Burgess RJ (1997) Superheated water as
an eluent for reversed-phase high-performance
liquid chromatography. Journal of Chromatography
785: 49.
Smith RM, Chienthavorn O, Wilson ID, Wright B and
Taylor SD (1999) Superheated heavy water as the eluent
for HPLC-NMR and HPLC-NMR-MS. Analytical
Chemistry 71: 4493}4497.
Yang Y, B+wdt S, Hawthorne SB and Miller DJ (1995)
Subcritical water extraction of polychlorinated biphenyls
from soil and sediment. Analytical Chemistry 67: 4571.
detergent, particularly when applied to cleaning
products such as fabric washing powders, soaps,
hard-surface cleaners and the many other products
used for cleaning in and around the home. Solutions of
surfactants exhibit one or more of the properties of
detergency, foaming, wetting, emulsifying, solubilizing
and dispersing.
This article will deal with the main classes of sur-
factants used as commercial detergents, which are
not single compounds but mixtures of compounds
of the same general structure but having a range
of alkyl chain lengths. Surfactants have the follow-

ing general properties:
E They are molecules composed of groups of oppo-
sing solubilities, typically an oil-soluble hydrocar-
bon chain and a water-soluble ionic group.
E They are soluble in at least one phase of a liquid
system.
E They form oriented monolayers at phase inter-
faces.
E They form micelles (aggregates of molecules or
ions) above a limiting concentration in solution.
Chromatographic separations are important in sur-
factant analysis for a number of reasons. The most
important of these is based on their ability to separate
both molecules of different, though similar, structures
and molecules from within a structural family on the
basis of carbon chain length, chain branching or posi-
tional isomer distribution. Many procedures for sur-
factant analysis give average values for the property
determined. Two quite different examples are tit-
rimetric determination of the active level of a surfac-
tant which requires a value for the mean molecular
weight of the surfactant to calculate the weight per-
cent of active material and determination of the de-
gree of ethoxylation of an alcohol or alkylphenol
ethoxylate by proton nuclear magnetic resonance,
which gives a value for the average degree of
ethoxylation but no information on the ethoxamer
distribution. Chromatographic techniques can supply
the mean molecular weight to use with the titration
and both the detailed ethoxamer and alkyl chain
length distribution of the ethoxylate. Such detailed
knowledge of molecular composition is required for
full understanding of surfactants and their properties.
Other areas of surfactant analysis in which
chromatographic techniques are used are determina-
tion of levels and composition of surfactants in prod-
ucts and in the environment, particularly in studying
decomposition, and in studying the detailed composi-
tion of the surfactant itself including low levels of
impurities or contaminants which may result from
the manufacturing process. Legislative and consumer
pressures for cleaner, safer, more environmentally
friendly raw materials and products are resulting in
such analyses becoming increasingly common.
In the following sections, the most common surfac-
tants in the detergents and related industries and the
chromatographic techniques (gas}liquid chromato-
graphy, high performance liquid chromatography,
supercritical Suid chromatography, thin-layer
chromatography, capillary electrophoresis) used in
their analysis will be described. The examples have
been chosen to illustrate the variety of sample prep-
aration, separation and detection procedures which
can be used.
III / SURFACTANTS / Chromatography 4319
Anionic Surfactants
Alkylbenzenesulfonates
Alkylbenzenesulfonates are the most common of the
commercial anionic surfactants. Their general struc-
ture is p-alkylbenzenesulfonic acid, sodium salt
(Figure 1) where the alkyl chain may range from C9
to C14 with the benzenesulfonate moiety attached in
different proportions at each carbon atom from C2 to
the central carbon of the chain. The carbon chain is
essentially linear to permit biodegradation, though
there is a minor usage of branched alkylbenzenesul-
fonates for speciRc applications.
Gas chromatography Gas chromatography (GC) of
alkylbenzenesulfonates requires some pretreatment of
the molecule to enable it to be volatilized. The most
common pretreatments are desulfonation and derivat-
ization. The following examples demonstrate a num-
ber of sample preparation and detection options:
E Desulfonation of alkylbenzenesulfonate with
phosphoric acid to the corresponding linear
alkylbenzene (LAB) followed by separation on
a fused silica capillary column (Figure 2). This
procedure enables both the alkyl chain lengths
and the attachment points of the benzene ring
along the chain to be determined.
E Desulfonated linear alkylbenzenesulfonate (LAS)
or LAB can be prefractionated from complex
detergent and environmental samples using ar-
gentation thin-layer chromatography (TLC), in
which the TLC plate is coated with silver nitrate
to modify the separation process. The separated
spots are recovered from the plate and analysed
using similar conditions to those in Figure 1.
Electron impact mass spectrometry (EIMS) de-
tection is used to give component identiRcation.
E LAS and dialkyltetralinsulfonates are converted
to their sulfonyl chlorides by reaction with phos-
phorus pentachloride and then to their trif-
luoroethyl derivatives by reaction with tri-
Suoroethanol. These derivatives are separated
using similar conditions to those described in
Figure 2.
Figure 1 C12 2-phenylalkylbenzenesulfonate, sodium salt.
Courtesy of RSC.
4320 III / SURFACTANTS / Chromatography
Figure 2 C10}C13 linear alkylbenzene. Column 25 m;0.2 mm i.d. fused silica capillary of HP1 (0.33
m film) programmed from
1203C to 2403C at 33C per minute, final temperature held for 20 min, injector and detector at 2753C, FID, carrier helium at 1 mL min\1,
splitless injection. Courtesy of RSC.
E Aqueous solutions of LAS are shaken with tetra-
butylammonium hydrogensulfate to form ion
pairs. Injection of the ion-pair solution into a GC
injection port at 3003C forms the butylsulfonate
derivatives of the LAS which are separated on
a HP-5 column (20 m;0.2 mm i.d., 0.33
m
Rlm) programmed from 1103C to 2203C at 103C
per minute then 3003C at 63C per minute with
a Rnal 3-minute hold. Detection is by EIMS to
give full spectral information.
E LAS is rapidly and efRciently converted to its
methylsulfonate ester by derivatization with
trimethoxyorthoacetate at room temperature
and separated as in Figure 3.
High performance liquid chromatography The ad-
vantage of high performance liquid chromatography
(HPLC) for the separation of LAS is that, in contrast
to GC, sample preparation speciRcally to make the
LAS compatible with HPLC is unnecessary. LAS can
be determined by HPLC by a number of procedures
of which examples to demonstrate different separ-
ation and detection conditions are given below.
The chain-length distribution of LAS can be deter-
mined on a column (300;3.9 mm) of
-Bondapak
C18 (10
m) using a linear gradient from 70 : 30
acetonitrile}0.15 mol L\1 sodium perchlorate solu-
tion to 90 : 30 of the two solutions. UV photometric
detection is at 230 nm. Peaks are eluted in order of
increasing alkyl chain length.
The positional isomer distribution is determined
using a column (250;4 mm i.d.) of Spherisorb ODS
II (3
m) with an isopropanol}water}acetonitrile
gradient with 0.02 mol L\1 sodium perchlorate
added and UV detection at 225 nm.
Both chain length and positional isomer distribution
can be obtained using a Zorbax ODS column (250;
4.6 mm i.d.) with a gradient from acetonitrile}water

Figure 3 Chromatogram of methyl esters of C10}C13 linear alkylbenzene. Column 25 m;0.2 mm i.d. fused silica capillary of HP1
(0.33
m film) programmed from 1203C to 2403C at 33C per minute, final temperature held 20 min, injector and detector at 2753C, FID,
carrier helium at 1 mL min\1, splitless injection. Courtesy of RSC.
(40 : 60)#100 mmol L\1 sodium chloride to
acetonitrile}water (60 : 40) with UV photometric
detection at 225 nm.
Other separation techniques The analysis of LAS
using a silica gel G layer (a standard TLC silica) impreg-
nated with 10% ammonium sulfate and 2-methyl-
4-pentanone}propyl alcohol}0.1 mol L\1 acetic acid}
acetonitrile (20 : 6 : 1.6 : 1, v/v/v/v) and visualization
by spraying with phosphomolybdic acid in ethanol
followed by charring by heating has been described.
Supercritical Suid chromatography has been used
for analysis of alkylbenzenesulfonates on a fused sil-
ica open tubular column (10 m;0.53 or 0.25 mm
i.d.) coated with a 0.1 or 0.2
m Rlm of SE54 with
carbon dioxide as mobile phase and FID detection.
The LAS is derivatized before analysis.
Capillary electrophoresis can be used for the separ-
ation of LAS by alkyl chain length. Conditions are
a fused silica capillary (60 cm;50
m i.d., 40 cm
to detector) with a buffer of acetonitrile}water
(40 : 60), 3.0 mmol L\1 magnesium ion, pH 6.0,
10 mmol L\1 sodium acetate. Applied voltage is
30 kV and detection is by UV at 220 nm.
Alkyl Sulfates
Alkyl sulfates normally exist as a group of com-
pounds with a range of alkyl chain lengths (Figure 4).
III / SURFACTANTS / Chromatography 4321
They can be readily determined by GC following acid
hydrolysis, recovery of the parent alcohols, and sep-
aration either as the alcohols or after conversion of
the alcohols to their trimethylsilyl ether derivatives.
Typical conditions are a 10 m;0.53 mm i.d. column
of methylsilicone phase programmed from 703C to
2403C at 53C per minute with helium as carrier and
FID. The HPLC separation of alkyl sulfates by carbon
chain length uses a column 25 cm;4.6 mm i.d. ODS
material with gradient elution from 60 to 30% aqueous
acetonitrile containing 0.01 mol L\1 disodium
hydrogenphosphate and 0.01 mol L\1 sodium ni-
trate. Detection is at 242 nm. This is an example of
inverse photometric detection where nitrate in the
mobile phase absorbs a constant level of radiation
apart from when the level of nitrate is reduced by the
presence of the non-absorbing alkyl sulfate anion. It
is the reduced absorbance which is monitored.
Alkyl sulfates can be separated on a synthesized
cross-linked amine-Suorocarbon polymer on silica
column with 0.2 mmol L\1 naphthalenedisulfonate}
35% acetonitrile mobile phase. Both indirect con-
ductivity and indirect photometric detection can be
used.
Figure 4 Sodium decyl sulfate. Courtesy of RSC.
4322 III / SURFACTANTS / Chromatography
Figure 5 C12 alkyl 3-ethoxysulfate. Reproduced with permission from the American Chemical Society.
TLC separation of alkyl sulfates and alkyl ether
sulfates using a silica gel layer with acetone}tetra-
hydrofuran (9 : 1, v/v) and visualization with Pina-
cryptol Yellow has been described.
Alkylethoxy Sulfates
Alkylethoxy sulfates (AES) have the general formula
CH3(CH2)n(OCH2CH2)mSO4Na where n is com-
monly in the range 9 to 17 and mean values of m are
in the range 2 to 20 (Figure 5). They are formed by
reaction of alcohols with ethylene oxide to give a de-
sired molar ratio of ethoxylate though in practice
they are a broad distribution of ethoxylate ratios
peaking at the desired mole ratio. The alcohol
ethoxylate is subsequently sulfated.
The hydrophobe (alkyl chain) distribution of al-
kylethoxylated sulfates can be obtained through GC
by reaction of the AES with 30% HBr in glacial acetic
acid at 903C overnight to give alkyl bromides fol-
lowed by separation on a column (6 ft;14 in o.d.) of
10% OV-17 on Chromosorb W with temperature
programming from 1003C to 2503C at 83C per min-
ute with helium as carrier gas and FID. AES can be
analysed by HPLC on a 2.5 cm;2 mm i.d. column of
C18 reversed-phase material with a water}tetrahyd-
rofuran gradient system. The detector is the evapor-
ative light-scattering detector as the molecules being
separated have no strong chromophore. There are
a number of alternative gradient systems.
To obtain more detailed distributions by either GC
or HPLC, the molecule can be desulfated and ana-
lysed as described below for ethoxylated alcohols.
Sulfonates
As for alkyl sulfates, alkylsulfonates exist in a range
of alkyl chain lengths with the added complication
that they can be primary, secondary or
-oleRnsulfon-
ates and also mono- and disulfonates with hydroxy
and -ene substitution (Figure 6).
Alkylsulfonates are separated by HPLC on a syn-
thesized cross-linked amine-Suorocarbon polymer or
silica column with 0.2 mmol L\1 naphthalenedisul-
fonate}35% acetonitrile mobile phase. Both indirect
conductivity and indirect photometric detection can
be used.
Separation of positional isomers of alkylmonosul-
fonates is obtained using Hypersil ODS I phase with
Figure 6 C13
-olefin sulfonate, sodium salt.
acetonitrile}water gradient and N-methylpyridinium
chloride as visualization reagent for indirect photo-
metric detection. Baseline separations are obtained.
As for LAS, aqueous solutions of secondary al-
kanesulfonates (SAS) can be shaken with tet-
rabutylammonium hydrogensulfate to form ion pairs.
Injection of the ion-pair solution into a GC injection
port at 3003C forms the butylsulfonate derivatives of
the SAS which are separated using on an HP-5 col-
umn (20 m;0.2 mm i.d., 0.33
m Rlm) programmed
from 1103C to 2203C at 103C per minute then 3003C
at 63C per minute with a Rnal 3-minute hold. Detec-
tion is by EIMS to give full spectral information.
The separation of
-oleRnsulfonates (AOS) into
their hydroxy, alkene, and disulfonate isomers to-
gether with chain length information is achieved us-
ing a column (25 cm;4.6 mm) of Zorbax TMS with
a mobile phase of methanol}water (75 : 25, v/v) and
refractive index detection. This separation has also
been demonstrated at an operating temperature of
553C.
An alternative column for separation of AOS by
carbon chain and isomer is a Novapak Phenyl column
(150;2 mm) with a mobile phase of 70 : 20 : 10
10 mmol L\1 ammonium acetate}acetonitrile}tetra-
hydrofuran (THF) at 0.2 mL min\1 with full-scan MS
detection to conRrm peak identiRcation. Supercritical
Suid chromatography has been used for analysis of
alkylsulfonates on a fused silica open tubular column
(10 m;0.53 or 0.25 mm i.d.) coated with 0.1 or
0.2
m SE54 with carbon dioxide as mobile phase
and FID detection. The sulfonates are derivatized as
described above for LAS before analysis. Capillary
electrophoresis can be used to separate C4 to C12
alkanesulfonates by alkyl chain length. Conditions
are a fused silica capillary (60 cm;50
m i.d., 40 cm
to detector) with a buffer of aqueous pH 7.0,
1.0 mmol L\1 magnesium ion, 5.0 mmol L\1 phos-
phate, 5.0 mmol L\1 salicylate solution. Applied
voltage is 30 kV, and indirect photometric detection
is at 230 nm.
Other Anionics
Sodium salts of alkyl (C10}C14) sulfosuccinates are
separated on a 250;4.6 mm i.d. column of 10
m
Nucleosil C8 with aqueous 0.01 mol L\1 tetrabutylam-
monium hydrogensulfate}methanol (23 : 77) at pH
3 as mobile phase with refractive index detection.
The separation of sodium isethionate from its alkyl
isethionate ester on a Vydac 302 IC column with

Figure 7 C10 alkyl 3-ethoxylate.
methanol}20 mol L\1 phthalic acid}water (3 : 5 : 12)
and conductivity detection has been described.
Nonionic Surfactants
The two most common nonionic surfactants are
ethoxylated alcohols and ethoxylated alkylphenols
(Figure 7). Synthesis of alkyl ethoxylates is described
above under AES. Alkylphenyl ethoxylates are syn-
thesized similarly but with the added structural fea-
ture that the phenyl ring is attached at different
carbon atoms along the alkyl chain. The alkyl chain is
commonly nine carbon atoms.
Ethoxylated alcohols and alkylphenols can be separ-
ated by GC on a fused silica capillary column (30 m;
0.25 mm; 0.25
m Rlm) of SE-54 using helium as
carrier gas and EI or CI (chemical ionization) (methane)
MS detection. Temperature programming is from 703C
(1 min) to 3003C (10 min hold) at 33C per minute.
Ethoxylates up to 6 EO units are detected.
Separation of alkylphenyl ethoxylates and alkyl
ethoxylates on an aluminium-clad fused silica column
Figure 8 Separation of C12/C14/C16 3 EO alkyl ethoxylate as trimethylsilyl ethers. Column 10 m;0.53 mm i.d. methylsilicone
(1.0
m film) programmed from 703C to 2403C at 103C minute, injector and detector 2703C, carrier helium at 15 mL min\1, FID.
III / SURFACTANTS / Chromatography 4323
(10 m;0.53 mm i.d.) of OV-1 with helium as carrier
gas, FID, and temperature programming to 3253C
has been demonstrated. The hydrophobe distribution
can be obtained via reaction to alkyl bromides as
described above for AES.
Silylation of alcohol ethoxylates to their trimethylsilyl
ether derivatives, when combined with temperature-
programmed GC using a non-polar methylsilicone
column, gives an extremely complex pattern of peaks
(Figure 8).
HPLC analysis of alkyl ethoxylates is made complic-
ated by the lack of a strong UV chromophore. Derivat-
ization to introduce a chromophore is an option.
Reaction to form phenyl isocyanate derivatives which
can then be detected by UV after separation on a
-
Bondapak C18 column according to alkyl chain length
or on a
-Bondapak amine column according to de-
gree of ethoxylation. The evaporative light-scattering
detector reduces the need for derivatization for HPLC.
Separation by ethoxamer is shown in Figure 9.
Alkylphenol ethoxylates can be readily detected by
UV. Columns and separation conditions are similar
4324 III / SURFACTANTS / Chromatography

Figure 9 HPLC of C12/C14/C16 alkyl 3 EO ethoxylate. Column 250;4 mm i.d. Nucleosil 50 silica. Linear gradient: ethyl
acetate}water (99 : 1, v/v) to acetone}water (90 : 10, v/v) over 60 min. Evaporative light-scattering detector.

to those for alcohol ethoxylates. Ethoxamer distribu-
tion can be determined on a LiChrosorb amine col-
umn (250;4.6 mm i.d.) with a hexane}isopropanol
to aqueous isopropanol gradient system and UV de-
tection at 277 nm.
Separation of ethoxylated fatty acids is obtained
using a column (250;4.6 mm i.d.) of Nucleosil
DIOL with hexane}isopropanol}water}acetic acid
(105 : 95 : 10 : 1, v/v).
TLC can also be sued for the determination of
nonionic surfactants using a silanized silica gel
GF254 layer with aqueous 80% methanol as develop-
ing solvent and a scanning densitometer at 525 nm
for detection.
Alkylphenyl ethoxylates may be analysed using
a Kieselgel F60 layer with chloroform}methanol as
developing solvent. IR detection is feasible with such
systems.
Supercritical Suid chromatography (SFC) has been
extensively applied to analysis of alcohol ethoxylates.
Examples are separation of ethoxylated alcohols on
a 20 m;0.1 mm i.d. column of poly(dimethyl-
siloxane) with density programmed carbon dioxide at
1003C as mobile phase and FID. Response factor
corrections are required for quantitative analysis.
SFC can be used to determine alkyl chain distribu-
tions of ethoxylated alcohols after reaction with 50%
HBr in glacial acetic acid to give their alkyl bromides.
An alternative to FID detection for SFC analysis of
these molecules is evaporative light-scattering detection.
Cationic Surfactants
Cationic surfactants are generally based on a quater-
nary ammonium structure with a number of long
('C10) alkyl chains attached either directly to
the nitrogen atom or through an ester linkage
(Figure 10).
Figure 10 Dialkyldimethylammonium chloride. R1 and R2 are
typically based on hardened tallow.

As cationics are nonvolatile, the contribution
which GC can make to their analysis is in determina-
tion of their alkyl substitution. This is achieved by
a degradation reaction. Alkyltrimethylammonium
and dialkyldimethylammonium cationics are con-
verted by Hoffmann degradation to their alk-1-enes
by heating on a water bath for 30 min with potassium
t-butoxide in benzene}DMSO (4 : 1). After extrac-
tion and clean-up, the alkenes are separated on a glass
column (2 m;3 mm i.d.) of 5% SE-30 on Chromo-
sorb W AW-DMCS (80}100 mesh) temperature pro-
grammed from 1603C to 2703C at 63C per minute
with nitrogen as carrier gas and FID.
Alkyl chain distribution can also be obtained by
thermal decomposition in the chromatograph injec-
tion port followed by separation on a 1 m;2 mm i.d.
column packed with 8% Carbowax 20M (KOH
treated) on acid-washed Chromosorb W (80}
100 mesh) and programmed from 703C to 2103C at
83C per minute with nitrogen as carrier gas and FID.
The chain-length distribution for cationics contain-
ing ester linkages is obtained by alkaline hydrolysis
followed by extraction of the resulting fatty acids and
conversion of the acids to their methyl esters. (See
below for separation conditions.)
Cationic surfactants are more amenable to HPLC
than to GC analysis as HPLC can analyse the intact
molecule. The separation of mono-, di-, and trialkyl
methylammonium quaternaries uses a column
250;4.6 mm of 5
m RSil Polyphenol with guard
column, a mobile phase gradient from 90 hexane}10
THF}methanol to 10 : 90, both solvents containing
5 mmol L\1 triSuoroacetic acid, and evaporative
light-scattering detection.
Cationics of the structure alkylamidopropyl-N-
(2,3-dihydroxy)-N,N-dimethylammonium chloride
have been analysed on a column (150;4 mm i.d.)
of
-Bondapak CN with water}acetonitrile}THF
(57 : 42 : 1, v/v) containing 0.1% triSuoroacetic acid
as mobile phase and differential refractive index detec-
tion. QuantiRcation is with an external standard and
the method has also been applied to cosmetic as well as
detergent products. Cationics of the structure difatty
acid ester of 2,3-dihydroxypropyltrimethylammnoium
chloride can be separated into mono- and di-esters on
a Partisil PAC column 250;4.6 mm i.d. with chloro-
form}methanol}acetic acid (94 : 6 : 0.1, v/v) as mo-
bile phase and refractive index detection.
The separation of imidazoline type cationics on
a column (150;4.6 mm i.d.) of 3
m Develosil ODS-
3 with 0.1 mol L\1 sodium perchlorate in meth-
anol}acetonitrile}deionized water (60 : 60 : 5, v/v)
as mobile phase and UV detection at 240 nm has been
demonstrated. This procedure gives separation of
different chain-length alkyl substitution.
III / SURFACTANTS / Chromatography 4325
A Rnal example of separation of dialkyl-
dimethylammonium quaternaries on a column of
5
m PLRP-S with a mobile phase of 5 mmol L\1
methanesulfonic acid in 70% acetonitrile uses post-
column ion suppression and atmospheric pressure
ionization mass spectrometry for component identiRca-
tion. TLC can be used to separate and compare
cationics of the dialkyldimethylammonium, fatty
acid esters of 2,3-dihydroxypropyltrimethylammonium,
and fatty acid esters based on methyltriethanolamine
quaternaries on a single plate. Separations are by
numbers of substituent groups and the different struc-
tures are also separated. Conditions are Merck
HPTLC Silica HF254 with a developing solvent
chloroform}methanol}acetic acid}water (72 : 20 : 5 :
3, v/v). Visualization can be with iodoplatinate spray
reagent.
Soap
Soap, the sodium salt of fatty acids in the chain length
range C10 to C18 (Figure 11) is analysed by protona-
tion of the salts to their acids and derivatization of the
acids with boron triSuoride}methanol to give fatty
acid methyl esters. A wide range of stationary phases
and conditions have been used for the separation
(Figure 12). The fatty acids obtained from soap can
be analysed by HPLC without derivatization using
a column (150;4 mm i.d.) of 5
m Hitachi Gel 3056
at 503C with methanol}5 mmol L\1 tetrabutyl-
ammonium phosphate (3 : 1, v/v) at pH 7.5 with con-
ductivity detection.
Betaines
Betaines are amphoterics. The two most common
types are alkylbetaines and alkylamidopropyl-
betaines. The alkyl moiety of the latter is generally
based on coconut fatty acids.
Separation by chain length of both types of betaine
is obtained on a cation-exchange column, Nucleosil
100-5 SA, 5
m, 250;4 mm i.d., with a mobile
phase of 70% acetonitrile}30% 0.05 mol L\1 lithium
hydroxide in water adjusted to pH 1.6 with phos-
phoric acid (v/v). A column temperature of 403C is
used together with diode array detection at 210 nm.
Contaminants
Common contaminants in commercial surfactants
are ethylene oxide, 1,4-dioxane, sultones and
dialkyltetralins.
Figure 11 Sodium salt of palmitic acid.
4326 III / SURFACTANTS / Chromatography
Figure 12 C8 to C22 fatty acid methyl esters. Column: Chrompack CPSIL5, 10 m;0.32 mm i.d., film 0.12
m, temperature
programme 503C to 2003C at 153C per minute. Carrier helium at 1.5 mL per minute, FID, cool-on-column injection.
Ethylene oxide is used for ethoxylation of alcohols
and alkylphenols. Levels of unreacted ethylene oxide
are deRned and must not be exceeded in the Rnished
raw material.
Ethylene oxide in alcohol ethoxylates is determined
by equilibrium headspace analysis. An aliquot of the
vapour is analysed on a column (8 ft;0.125 in) of
Chromosorb 102 (80}100 mesh) (a gas}solid phase)
programmed from 1203C (5 min) to 1903C (10 min)
at 83C per minute, with helium as carrier gas and FID.
A detection limit of 1
g g\1 is achieved. An alterna-
tive column, also used for ethylene oxide in AES, is
3 m;1.8 mm i.d. 0.8% THEED/Carbopack C (80}
100 mesh).
For improved quantiRcation, both methods can be
adapted to a method of standard additions.
1,4-Dioxane is a by-product of the sulfation of
alcohol ethoxylates. The industry standard method is
equilibrium headspace GC and involves sample prep-
aration using the method of standard additions for
quantiRcation. There is some Sexibility as to whether
capillary or packed columns are used and the actual
phase required.
An alternative approach describes the use of a to-
tally deuterated 1,4-dioxane analogue with isotope
dilution and MS detection to minimize matrix
effects. The separation is carried out on a 60 m;
0.32 mm i.d. column of Supelcowax 10, temperature
programmed from 503C (2 min) to 1003C at 53C
per minute.
An HPLC approach for 1,4-dioxane in alkylether
sulfates is a column (250;4.6 mm i.d.) of 5
m
LiChrospher C-8 with an aqueous acetonitrile gradi-
ent and UV detection at 200 nm. An external cali-
bration curve is used for quantiRcation.
1,3- and other sultones occur as by-products of the
formation of
-oleRn sulfonates. Certain sultones are
potent sensitizers and must be controlled. 1,3-Sul-
tones are determined in
-oleRn sulfonates by extrac-
tion with diethyl ether, trapping from a silica column,
and GC on a column (1 m;3 mm) of 2% DEGS on
Chromosorb W AW-DMCS (60}80 mesh) at 2203C
with helium as carrier and Same photometric (sulfur
mode) detection.
Alternatively, detection limits down to 0.2 ng g\1
can be obtained with negative chemical ionization
MS with methane as reagent gas. 1,3-Sultones can
also be determined by HPLC following extraction
from
-oleRn sulfonate and separation on a column
(200;4.6 mm i.d.) of 5
m CPS Hypersil with
hexane}ethyl acetate (90 : 10, v/v) as mobile phase
and differential refractive index detection. QuantiR-
cation is by external standard.
Dialkyltetralins occur in linear alkyl benzene, the
precursor of LAS. Again, there are set limits as to
permissible levels. They are determined on a column
(250;4 mm i.d.) of 5
m Lichrosorb Si60 with dry
iso-octane as mobile phase and UV detection at
254 nm. A reference standard is available from ECO-
SOL to calibrate the method.

Future Developments
As has been described above, there are many alterna-
tive approaches to the chromatographic analysis of
commercial surfactants. The approach to be used for
any analysis may be determined by a number of
factors, the information required, the equipment
available, the time available to carry out the analysis.
As we move into the twenty-Rrst century, there will
be a trend towards faster analysis and analysis requir-
ing less sample handling. Automation will increase,
requiring less and less skilful human input. Data
handling will become faster and more intelligent in
order to deal with greater volumes of data generated
in shorter times. Use of new separation techniques
will be investigated, particularly electro-driven tech-
niques and some will replace existing techniques for
certain analyses. Overall, separation systems will be-
come smaller with advantages of lower solvent and
carrier gas use, less use of laboratory space, and with
Liquid Chromatography
T. M. Schmitt, BASF Corporation, Wyandotte, MI, USA
Copyright ^ 2000 Academic Press
Introduction
Liquid chromatography (LC) is very useful for the
characterization of individual surfactants. Most com-
mercial surfactants are mixtures of members of
homologous series, and LC is capable of deRning
these mixtures according to their homologue distribu-
tion, indicating, for example, alkyl chain length or
degree of polymerization. LC is also the preferred
technique for the quantitative determination of many
surfactants, especially ionic surfactants in mixtures.
The utility of LC stems from the properties of surfac-
tants } these compounds have good solubility in the
usual LC mobile phases and possess diverse chemical
functionality, but at the same time they are not vol-
atile enough for ready analysis by alternative tech-
nologies such as gas chromatography (GC) or simple
mass spectrometry (MS). The structures of common
surfactants are given in Table 1.
Surfactants are usually analysed in LC systems con-
taining a substantial percentage of organic solvent so
as to inhibit micelle formation. The presence of
micelles will confound LC analysis.
Formulations and Mixtures of Surfactants
LC is used for quality control of formulations such as
cleaning compounds and pharmaceutical prepara-
III / SURFACTANTS / Liquid Chromatography 4327
increased portability to permit use away from the
laboratory.
See also: II/Chromatography: Gas: Detectors: Selec-
tive. Chromatography: Liquid: Derivatization; Detectors:
Evaporative Light Scattering; Ion Pair Liquid Chromato-
graphy. Chromatography: Thin-Layer (Planar): Den-
sitometry and Image Analysis; Layers. III/Detergent
Formulations: Ion Exchange. Surfactants: Liquid
Chromatography. Thin-Layer Chromatography-Vibra-
tion Spectroscopy.
Further Reading
Cullum DC (ed.) (1994) Introduction to Surfactant Analy-
sis. Glasgow: Blackie A & P.
Kirk-Othmer (1997) Encyclopedia of Chemical Technol-
ogy. Surfactants, 4th edn. New York: John Wiley.
Spitz L (ed.) (1996) Soaps and Detergents. A Theoretical
and Practical Review. AOCS Press.
tions. LC is often the easiest and most speciRc method
for determining surfactant concentration in a well-
understood mixture. On the other hand, LC is not
often useful for analysis of unknown formulations
unless MS detection is available. This is because of
the limited separation range of any single LC system.
Sometimes, especially in quality control where
there are no unknown components, no preliminary
sample work-up is necessary. This is particularly true
of ionic surfactants. More often, especially for non-
ionics, a gross separation of the surfactants from the
matrix is required. This can be accomplished by sol-
vent extraction of the dried solids or by liquid}liquid
extraction or solid-phase extraction (SPE) of an aque-
ous solution.
Alkylarylsulfonates and alkylphenol ethoxylates
can be determined with a minimum of sample work-
up because of the availability of a speciRc detection
method, Suorescence, to distinguish them from other
surfactant and nonsurfactant compounds that may
also be present. For the very common mixtures of
anionic and nonionic surfactants, ion exchange
chromatography systems result in nonionic surfac-
tants eluting prior to anionics, while reversed-phase
systems result in the nonionics being retained longer
than anionics.
Environmental Analysis
LC is widely applied in environmental analysis, but it
is not used for routine monitoring of efSuents, except
by industry for the analysis of speciRc process streams
4328 III / SURFACTANTS / Liquid Chromatography

Table 1 Structures of common surfactants

Surfactant Structure

Cationic surfactants
Quaternary amines RR
RR


N#Cl\
R,R
,R, R


"H or C1}C18 alkyl or C6H5CH2
Anionic surfactants
#
Linear alkylbenzene sulfonates 4-RC6H4SO\3 Na
R"C10H21}C14H29
#
Alkyl sulfates ROSO\ 3 Na
R"C8H17}C18H37
#
Alkanesulfonates RSO\3 Na
R"C8H17}C18H37
# #
Ether sulfates 4-RC6H4O(CH2CH2O)xSO\ 3 Na or R
O(CH2CH2O)xSO\
3 Na
R"C9H19; R
"C12H25}C18H37; x"2}10
-Olefin sulfonates (mixtures of alkenesulfonates RSO\3 Na
#

and hydroxyalkanesulfonates) R"C11H21}C20H39 or C11H22OH}C18H36OH

Ether carboxylates RO(CH2CH2O)xCH2COO\Na#

R"C12H25}C18H37; x"5}25
Sulfosuccinate esters ROOCCHSO3CH2COOR\Na#
or HOOCCHSO3CH2COOR\Na#
R"C8H17
# #
-Sulfofatty acid methyl esters RCH(SO\3 )COOCH3Na and RCH(SO\
3 )COO\2Na
R"C12H25}C16H33

Nonionic surfactants
Alkylphenol ethoxylates 4-RC6H4O(CH2CH2O)xH
R"C8H17, C9H19 or C12H25; x"3}50
Alcohol ethoxylates RO(CH2CH2O)xH
R"C12H25}C18H37; x"5}60
Acid ethoxylates RCOO(CH2CH2O)xH
R"C11H23}C17H35; x"5}20
EO/PO copolymers HO(CH2CH2O)y(CH(CH3)CH2O)x(CH2CH2O)y H
x"16}70; y"1}100
Esters
CH2CHOHCHOORCHCH2OHCH2OOR
R"C16/C18 alkyl

where the composition is uniform and well under-
stood. The standard wastewater methods, for example
those based on colorimetric tests, give a gross value for
total surfactant concentration and are most suitable
for routine environmental monitoring.
LC is used, however, in special investigations of
environmental impact to give information on the con-
centration and degradability of speciRc surfactants in
particular environmental areas, relying on the ability
of LC methods to precisely characterize surfactant
homologues. A preliminary separation or preconcen-
tration is always necessary. The most common pre-
treatment method nowadays is SPE, especially when
low levels of surfactant must be determined. C18
media are most often applied, in the form of SPE
cartridges or extraction discs. Nonpolar resin of the
poly(styrene/divinylbenzene) type is also used. A sec-
ondary separation of the surfactants into nonionic,
cationic and anionic fractions can be performed on
ion exchange media.
LC is the method of choice for determining anionic
surfactants in the environment. It is also preferred for
the determination of the anionic degradation prod-
ucts of these surfactants. LC is the best method for the
determination of nonionics (especially when coupled
with MS), if detail on homologue distribution is
needed. LC is less often applied to environmental
determination of cationics, since interference is not as
serious a problem with the standard methods for
determining cationics in wastewater as it is for other
surfactants.
Analysis of Individual Surfactants
Anionic Surfactants
LC analysis of anionic surfactants is a mature techno-
logy. Detection is a simple matter, either by direct UV
absorption of aromatic surfactants or by inverse
photometric detection of the aliphatic compounds.
 III / SURFACTANTS / Liquid Chromatography 4329

Alkylarylsulfonates The commercial product, linear

alkylbenzenesulfonate (LAS) is a mixture containing
a range of alkyl chain lengths, typically C10}C14. Re-
versed-phase LC with a C4 or C8 column effectively
separates LAS according to the length of the alkyl
chain (Figure 1). A C18 packing gives a more complex
chromatogram because the individual compounds of
discrete alkyl chain length are themselves partially
resolved into isomers; in many cases this resolution is
not needed or desired (Figure 2). In any case, GC
analysis after desulfonation is a better method for
determining isomers. Aqueous mixtures of acetonit-
rile, methanol and tetrahydrofuran (THF) are appro-
priate mobile phases, generally containing a salt such
as 0.1 mol L\1 sodium perchlorate. Detection is by
UV absorbance at 225 nm or Suorescence with exci-
tation at 225 nm and emission at 290 nm. Fluores-
cence detection is advantageous for trace analysis.

Alkyl sulfates Alkyl sulfates with chain lengths in
the surfactant range (C10 and higher) are readily sep-
arated according to increasing alkyl chain length in
a reversed-phase system with methanol/water mobile
phase containing a salt such as sodium perchlorate.
The pH is often adjusted to 2.5 or 3.0. Gradient
programming is impractical if detection is by direct
low wavelength UV, differential refractive index
(DRI) or conductivity. Gradients are successful with
detection by indirect UV (typically, methylpyridinium
chloride is added to the eluent) or evaporative light
scattering (ELS). If anion exchange chromatography
is used, elution is in order of decreasing alkyl chain
length.
Figure 1 LAS isolated from river water and analysed isocrati-
cally using a C4 reversed-phase column. Labels indicate alkyl
chain length. Column: Wakosil 5C4, 4.6;150 mm. Mobile phase:
0.1 mol L\1 sodium perchlorate in 50 : 50 CH3CN/H2O. Detection:
UV, 220 nm. (Reproduced with permission from Yokoyama Y and
Sato H (1991) Reversed-phase HPLC determination of linear
alkylbenzenesulphonates in river water by precolumn concentra-
tion. Journal of Chromatography 555: 155}162. Copyright (1991)
Elsevier Science.)
ditions. A single peak for easy quantiRcation can
sometimes be obtained by using a very short reversed-
phase column and a step gradient.
Normal-phase systems are also used for analysis of
ether sulfates, with stationary phases of bare silica or

Alkanesulfonates These are generally separated

with the same systems used for alkyl sulfates. In a
mixture, the peaks of the alkyl sulfates and alkane-
sulfonates are interspersed, with the alkyl sulfates
more strongly retained on reversed-phase column
packings than alkanesulfonates of the same chain
length. Separation from anionic surfactants of other
types is usually straightforward. Interference from
alkyl sulfates can be eliminated by subjecting the
sample to acid hydrolysis to convert them to the
corresponding alcohols and sulfuric acid; sulfonates
are not affected by this treatment.
Figure 2 Chromatogram of commercial LAS mixture analysed
Ether sulfates Alkylphenol ether sulfates and alco- using a reversed-phase column and gradient elution. Labels indi-
hol ether sulfates can be resolved by reversed-phase cate groups of isomers for various alkyl chain lengths. Column:
chromatography with elution in the order of both Zorbax ODS, 4.6;250 mm. Mobile phase: CH3CN/H2O gradient
increasing alkyl chain length and increasing (or de- with increasing concentration of NaCl. Detection: UV, 225 nm.
(Reproduced with permission from Chen S and Pietrzyk DJ
creasing) ethoxy chain length (Figure 3). An al-
(1994) Reversed-phase LC separation of linear alkylbenzenesul-
kylamine ion-pairing agent may be added to increase phonates. Effect of mobile phase ionic strength. Journal of
retention time. Unsulfated alcohol or alkylphenol Chromatography A 671: 73}82. Copyright (1994) Elsevier
ethoxylate impurities elute Rrst under paired-ion con- Science.)
4330 III / SURFACTANTS / Liquid Chromatography
Figure 3 Chromatogram of a lauryl ether sulfate isolated from
shampoo. Labels indicate alkyl chain length and number of moles
of ethylene oxide. Column: Alltech Surfactant C8, 4.6;250 mm.
Mobile phase: MeOH/H2O, 45 : 55, 0.00023 mol L\1 in NH4OAc.
Detection: conductivity. (Reproduced with permission from Stemp
A, Boriraj VA, Walling P and Neill P (1995) Ion chromatographic
characterization of ethoxylated anionic surfactants. Journal of the
American Oil Chemists Society 72: 17}21.)
amino- or cyanopropyl-silica. In this case, the
nonionic material elutes Rrst, followed by the anionic
material, each in order of increasing ethoxylation. An
ion-pairing agent such as cetyltrimethylammonium
chloride may be added to give more rapid elution of
anionic material. A relatively hydrophilic mobile
phase is often used in conjunction with the ion-pair-
ing agent.

-OleVn sulfonates These compounds give complex
chromatograms that reSect the complexity of their
composition. The commercial product is formed of
approximately equal portions of alkenesulfonates
and hydroxyalkane sulfonates, each carrying the
chain length distribution of the alkene feedstock. Di-
sulfonates may also be present. -OleRn sulfonates are
analysed by reversed-phase LC with methanol/water
and added salt. DRI or low wavelength UV detection
is suitable. Elution is in order of increasing chain
length, with hydroxyalkanesulfonates eluting prior
to the corresponding alkenesulfonates and with all
disulfonates eluting prior to all monosulfonates
(Figure 4). Complete resolution is not obtained if the
starting -oleRn was a mixture of many chain lengths,
as is usually the case with commercial products.
Petroleum sulfonates and alkylnaphthalenesulfo-
nates These are frequently separated on anion ex-
change packings with elution according to increasing
degree of sulfonation. Reversed-phase systems have
also been used. Resolution by alkyl chain length is
sometimes attained, but, more often, the practitioners
are content with a single peak for the active agent.
Ether carboxylates Only reversed-phase systems
have been demonstrated for separation of these prod-
ucts, generally with acetonitrile/water mobile phase.
Elution is always according to increasing chain length
of the alkyl or alkylphenol moeity. Depending on the
system, there may be an overtone of separation ac-
cording to increasing or decreasing ethoxy chain
length. As with ether sulfates, separation from
nonionic impurities is straightforward. Alkyl-
phenolether carboxylates are easily detected in the
UV (225 or 254 nm), while the alkylether car-
boxylates require low wavelength UV, DRI or ELS
detection.
Sulfosuccinate esters These are analysed most read-
ily by reversed-phase LC in the presence of an ion-
pairing agent. If not monodisperse as to alkyl chain
length, they are eluted in order of increasing alkyl
length. Separation from other anionic surfactants can
usually also be accomplished by reversed-phase
chromatography.

-Sulfofatty acid methyl esters These also are easily
separated according to alkyl chain length by reversed-
phase methods. Ion-pairing agents are rarely used.
Soap Fatty acids are separated according to increas-
ing alkyl chain length on a C18 column with
methanol/water mobile phase and refractive index
Figure 4 Chromatogram of -olefin sulfonate after hydrogen-
ation and formation of methyl esters. Peak identification: 4- and
3-hydroxyhexadecylsulfonate, methyl ester; 4- and 3-hydroxy-
octadecylsulfonate, methyl ester; hexadecylsulfonate, methyl
ester; octadecylsulfonate, methyl ester. Column: Inertsil C18,
4.6;250 mm. Mobile phase: MeOH/H2O, 85 : 15. Detection: re-
fractive index. (Reproduced with permission from Matsutani
S and Endo Y (1991) Separation and determination of sulfonate
type anionic surfactants including 2-sulfonatofatty acid methyl
ester by methyl ester derivatization and HPLC analysis.
Yukagaku 40: 566}573.)

detection. Ionic strength and pH must be controlled.
Analysis is either at low pH without an ion-pairing
agent or at high pH with ion-pairing. It should be
noted that most practitioners prefer GC for analysis
of fatty acids (as methyl esters).
Cationic Surfactants
LC is the most generally useful method for deter-
mination of cationic surfactants. All commercial
cationic surfactants are salts of quaternary amines
(quats). Most of these, like the long-chain alkyl-
trimethylammonium salts, have good water solubility
and are readily analysed by reversed-phase LC. Con-
trol of pH and ionic strength is necessary and inverse
spectrophotometric detection gives the best results for
quats without an aryl moiety; toluenesulfonate or
xylenesulfonate are used as counterions for detection.
Retention times are inSuenced by the hydrophobicity
of the counterion. Conductivity detection is also ap-
plicable, especially if ion chromatography instrumen-
tation is used with a nonpolar stationary phase (this
conRguration is sometimes called mobile-phase ion
chromatography). Of course, UV detection can be
used for compounds with pyridyl or benzyl substitu-
ents. Since quats exhibit long retention times on
C18 stationary phases, reversed-phase LC is most of-
ten performed on cyano columns, usually with meth-
anol/water or acetonitrile/water mobile phase.
Quats are sometimes analysed using cation ex-
change packings. While the separating ability of the
ion exchange systems is not as great as that attained
with normal- or reversed-phase systems, ion ex-
change is sometimes preferred for formulation analy-
sis because interference is minimized in that only the
cationic materials are seen.
The quats used as fabric softeners for household
laundry contain two long alkyl chains and have poor
water solubility. For reasons of solubility, these are
most easily analysed by normal-phase chromatogra-
phy. Bare silica stationary phases are never used, but
rather cyanoamino, amino, DIOL or polyphenol
phases. Chloroform/methanol mobile phases work
well, usually with a little added acetic acid. Conduct-
ivity and ELS are suitable for detection.
Common fabric softener quats, including ester quats,
can be characterized by normal-phase chromato-
graphy on a polyphenol column with a hexane/
methanol/THF gradient and added triSuoroacetic
acid. Elution is in order of decreasing alkyl length,
with the quats well resolved from unquaternized
amine impurities.
For trace analysis, detection is sometimes accom-
plished by paired-ion extraction of the HPLC
efSuent with a Suorescent anion, followed by phase
separation and Suorescence detection.
III / SURFACTANTS / Liquid Chromatography 4331
Nonionic Surfactants
Ethoxylated nonionics are most easily characterized
by normal-phase chromatography. This permits the
separation of the compounds according to the length
of the ethoxy chain, with the longer chain homo-
logues eluting later. Amino-bonded stationary phases
are often used along with the usual nonpolar mobile
phases such as hexane.
Separation of homologues can also be accomp-
lished by reversed-phase chromatography. Reversed-
phase LC is usually applied to higher ethoxylates
because normal-phase LC resolution deteriorates
with higher molecular weight. Solvent systems of
methanol/water or acetonitrile/water are usual. Or-
der of elution can be according to increasing or
decreasing order of ethoxylation, depending on the
particular reversed-phase media and solvents used,
the particular nonionic surfactant and whether it has
been derivatized. The order of elution even depends
on the molecular weight: elution is sometimes accord-
ing to reverse order of ethoxylation for lower mem-
bers of a series and according to increasing order of
ethoxylation for higher members of the same series.
Reversed-phase HPLC is effective for separation of
ethoxylated surfactants according to the identity or
chain length of the hydrophobic moiety. Caution is
required, since under various reversed-phase condi-
tions separation according to degree of ethoxylation
will also occur, as mentioned above. Unless MS detec-
tion is available, this two-dimensional separation
makes quantiRcation difRcult, so the system is usually
optimized to minimize the inSuence of hydrophile
homogeneity. If MS detection is used, then the separ-
ation by hydrophobe is sufRcient for complete char-
acterization of the surfactant, with the MS detector
giving the information on ethoxy homologue distri-
bution. The sensitivity of the MS detector is not
identical for all homologues. For precise work, calib-
ration must be performed over the entire range of
composition.
Size exclusion chromatography (SEC) is sometimes
applied to the analysis of nonionic surfactants, parti-
cularly higher molecular weight surfactants like the
ethylene oxide/propylene oxide (EO/PO) copoly-
mers. Nonaqueous systems are most useful for this
analysis for two reasons. First, formation of micelles
is discouraged. Micelle formation is a function of
concentration, so SEC can show different values for
molecular weight depending on sample concentra-
tion. Second, aqueous SEC column packings often
have a silica backbone. Polyethoxy compounds are
strongly adsorbed to silica, resulting in mixed-mode
separation rather than separation only according to
molecular size.
4332 III / SURFACTANTS / Liquid Chromatography

Polyethylene glycol (PEG) impurity is determined

in most ethoxylated surfactants by reversed-phase
separation with 95 : 5 methanol/water and DRI or
ELS detection. PEG elutes as a single peak prior to the
surfactants.
The refractive index of nonionic surfactants is
a function of degree of ethoxylation. Thus, the re-
sponse of a differential refractive index detector va-
ries for homologues, with the variation being most
signiRcant at low degrees of ethoxylation.

Alcohol ethoxylates (AE) Commercial products are

mixtures of homologues containing a distribution of
alkyl chain length and ethoxy chain length. Conven-
tional LC analysis fails to give a single peak for
alcohol ethoxylate. Rather it yields a series of peaks
more-or-less resolved corresponding to the alkyl or
ethoxy distribution. This limitation is only overcome
by using backSush techniques.
Derivatives may be formed to improve detectability
of AE. Derivatization also inSuences retention time,
so that a gradient system optimized for underivatized
AE must be modiRed for chromatography of the de-
rivatives. Typical derivatizing agents are phenyl
isocyanate, naphthyl isocyanate and 3,5-dinitroben-
zoyl chloride. Fluorescence detection is sometimes
used in environmental analysis with Suorescent de-
rivatizing agents such as 1- and 2-naphthoyl chloride
and naphthyl isocyanate.
Normal-phase chromatography, preferably on Figure 5 Chromatogram of C13 alcohol ethoxylate from a bio-
aminopropyl- or cyanopropyl-bonded silica, will give treatment study. Analysis using a CN normal phase column with
gradient programming and evaporative light scattering detection.
the ethoxy distribution (Figure 5). Reversed-phase Labels indicate ethoxy chain length. Column: Rainin Microsorb
chromatography on C18 media serves to separate by CN, 4.6;250 mm, 453C. Mobile phase: gradient, hexane/
alkyl chain length. In either case, solvent program- THF/(90 : 10 2-PrOH/H2O); from 100 : 0 : 0 to 80 : 20 : 0 in 5 min,
ming is usually required for complete resolution of then to 52 : 30 : 18 in 15 min, then to 40 : 40 : 20 in 5 min. Detec-
a commercial product, so either an ELS detector is tion: ELS (Reproduced with permission from Dubey ST, Kravetz
L and Salanitro JP (1995) Analysis of nonionic surfactants in
used or the surfactant is Rrst derivatized to permit use bench-scale biotreater samples. Journal of the American Oil
of a UV detector. If reversed-phase solvents are not Chemists Society 72: 23}30.)
optimized, a separation by ethoxy chain length is
superimposed on the separation by alkyl chain length.
UV detection (225 or 275 nm) is always used since it
Alkylphenol ethoxylates (APE) Almost all commer- gives a uniform molar response to the homologues.
cial products are based upon a monodisperse hydro- For trace analysis, Suorescence detection is applicable
phobe, usually nonylphenol. Therefore, quantiRcation since APEs have native Suorescence.
is usually performed by reversed-phase chromatogra-
phy on C18 media using a isocratic methanol/water or Ethoxylated acids These compounds can be con-
acetonitrile/water eluent and UV detection, resulting in sidered as esters of PEG and fatty acids, with the
a single peak. Octylphenol ethoxylates are easily separ- commercial products also containing diester, residual
ated from nonylphenol ethoxylates by such systems. fatty acid and free PEG. Reversed-phase chromato-
As with alcohol ethoxylates, separation by degree graphy with methanol/water separates the ethoxylated
of ethoxylation is easily performed with any of the acid, PEG diester, free PEG and free fatty acid, and
usual normal-phase stationary phases, with the cyano usually also serves to separate the compounds from
packings most popular. Amino and bare silica media other surfactants. As with other ethoxylates, normal-
are also used (Figure 6). Reversed-phase LC is also phase chromatography gives resolution by ethoxy
applied, especially for higher degrees of ethoxylation. chain length. SEC is often useful to resolve the

Figure 6 Chromatogram of a mixture of two commercial prod-
ucts, nominal 4-mole and 30-mole ethoxylates of nonylphenol,
analysed by normal-phase HPLC. Column: Hewlett-Packard Si-
100, 4.6;200 mm, 303C. Mobile phase: gradient; A"80/20 n-
hexane/ethyl ether; B"40 : 30 : 20 : 10 : 1 : 0.5 dioxane/ethyl
ether/n-hexane/2-PrOH/H2O/HOAc; 5 to 95% B in 45 min. Detec-
tion: UV, 280 nm. (Reproduced with permission from Anghel DF,
Balcan M, Voicu A and Elian M (1994) Journal of Chromatography
A 668: 375}383. Copyright (1994) Elsevier Science.)
mono- and diesters. DRI and ELS detectors are most
applicable. Low wavelength UV detection must be
used with caution because of the disproportionate
response from unsaturated fatty acid moieties.
Esters Esters of fatty acids with glycerol and sugars
are separated according to degree of acyl character by
either normal-phase or reversed-phase chromatogra-
phy. Reversed-phase elution is according to increas-
ing acyl character, while normal-phase elution is in
order of decreasing acyl character (i.e. according to
both the chain length of the acyl groups and their
number). Normal-phase LC is usually performed
with a DIOL stationary phase and propanol/water
mobile phases. The ELS detector is most applicable,
although historically many applications have been
developed using DRI or UV at wavelengths of 220 nm
or less.
EO/PO block copolymers In the absence of interfer-
ing compounds, polymers of the poloxamer type can
sometimes be determined by reversed-phase HPLC
with methanol, but the most common separation
technique is SEC. The copolymers useful for deter-
gents and pharmaceuticals are higher in molecular
weight than most other synthetic surfactants, so SEC
can be used for both qualitative and quantitative
III / SURFACTANTS / Liquid Chromatography 4333
analysis. If conventional LC is used, depending on the
speciRc product, a reversed-phase system with
acetonitrile/water can be optimized to be indifferent
to EO chain length, separating the surfactant accord-
ing to length of the PO chain. DRI detection is gener-
ally used for the block copolymers.
Alkanolamides These compounds, for example the
C10}C18 fatty acid monoethanolamides, are eluted
according to increasing acyl chain length by reversed-
phase chromatography with methanol/water sol-
vents. These systems may also be used for formula-
tion analysis. Detection is a challenge: DRI and low
wavelength UV are most common, and ELS and ni-
trogen-speciRc detectors have been applied in more
recent times. N-Methylglucosamides are analysed in
the same way.
Alkyl polyglycosides Reversed-phase LC with
methanol/water will separate these compounds, with
elution according to increasing chain length of the
acyl constituents. For compounds of the same acyl
chain length, polyglycosides elute prior to mono-
glycosides. ELS detection is typically used.
Amphoteric Surfactants
Amphoteric surfactants are almost always separated
on C18 columns with methanol/water mobile phase.
The pH is often held as low as the column will
tolerate and a salt such as sodium perchlorate is
added. Under such conditions, the amphoteric surfac-
tant behaves much like a cationic surfactant and the
same detection methods are used as discussed above
for cationics.
Betaines These compounds can be separated, with
elution by increasing alkyl chain length by reversed-
phase LC or by decreasing chain length using cation
exchange chromatography. DRI detection is most
often chosen, although low wavelength UV and ELS
are sometimes applied.
Phosphatides These compounds must be discussed
separately from other amphoterics. They are constitu-
ents of the natural surfactant, lecithin, but they also
have great biochemical importance. Normal-phase LC
serves to separate the main constituents of commercial
lecithin: phosphatidylethanolamine, phosphatidylcho-
line, phosphatidylinositol, phosphatidylserine and
phosphatidic acid. (Each of these consists of a num-
ber of individual compounds containing various acyl
groups.) The normal-phase separation is traditionally
performed on a bare silica column with low
wavelength UV detection. Since it is mainly double
4334 III / SYNTHETIC POLYMERS / Gas Chromatography
bonds that give the detector response, and since each
of the individual components contains acyl chains of
varying unsaturation, quantiRcation by UV is only
approximate. The ELS detector is rapidly becoming
standard for this analysis. Since this detector is toler-
ant of solvent gradients, other normal-phase col-
umns, notably the DIOL column, may be used instead
of bare silica. These give better reproducibility but do
not have sufRcient resolving power for lecithin analy-
sis in the absence of gradient programming.
Separation by acyl chain length is accomplished by
reversed-phase LC of fractions separated by the nor-
mal-phase methods. LC}MS is an obvious way to
simplify the characterization of unknowns. Precise
phosphatide analysis is very much an activity of
specialists and the Reld is advancing rapidly.
Conclusions
LC is the only practical method to characterize many
surfactants according to their oligomer or homologue
distribution. It is also the best way to determine
quantitatively many surfactants, particularly ionic
surfactants.
However, in spite of improvements in instrumenta-
tion and in stationary phases, LC is not easy. It
demands more time and training of the operator than
most analytical techniques. Preliminary sample prep-
aration is very often necessary for mixtures and envir-
onmental samples, making an LC analysis an
expensive analysis.
Continued development in the areas of detection
(especially in element-selective detectors, detectors
speciRc for chemical functionality, and LC}MS inter-
faces) will make LC even more useful in the future.
For example, the ELS detector, even though suffering
from problems in linearity in its present incarnation,
has already greatly expanded the utility of LC for
analysis of lecithin and ethoxylates.
SYNTHETIC POLYMERS
Gas Chromatography
J. K. Haken, The University of New South Wales,
Sydney, NSW, Australia
Copyright ^ 2000 Academic Press
See also: II/Chromatography: Liquid: Mechanisms: Ion
Chromatography; Ion Pair Liquid Chromatography; Mech-
anisms: Reversed Phases. Extraction: Solid-Phase
Extraction.
Further Reading
Balazs PE, Schmit PL and Szuhaj BF (1996) HPLC of soy
phospholipids. Journal of the American Oil Chemists
Society 73: 193}197.
Cross J (ed.) (1998) Anionic Surfactants: Analytical Chem-
istry, 2nd edn. New York: Marcel Dekker.
Cross J and Singer EJ (eds) (1994) Cationic Surfactants:
Analytical and Biological Evaluation. New York: Mar-
cel Dekker.
Di Corcia A (1998) Characterization of surfactants and their
biointermediates by liquid chromatography}mass spec-
trometry. Journal of Chromatography A 794: 165}185.
Evans KA, Dubey ST, Kravetz L et al. (1997) Quantitation
of alcohol ethoxylate surfactants in environmental sam-
ples by electrospray mass spectrometry. Journal of the
American Oil Chemists Society 74: 765}773.
Garti N, Kaufman VR and Aserin A (1983) Analysis of
nonionic surfactants by HPLC. Separation and PuriTca-
tion Methods 12: 49}116.
Miszkiewicz W and Szymanowski J (1996) Analysis of
nonionic surfactants with polyoxyethylene chains by
HPLC. Critical Reviews in Analytical Chemistry 25:
203}246.
Rissler K (1996) HPLC and detection of polyethers and
their mono(carboxy)alkyl and arylalkyl substituted de-
rivatives. Journal of Chromatography A 742: 1}54.
Schmitt TM (1992) Analysis of Surfactants. New York:
Marcel Dekker.
Stache HW (ed.) (1996) Anionic Surfactants: Organic
Chemistry. New York: Marcel Dekker.
Thiele B, GuK nther K and Schwuger MJ (1997) Alkylphenol
ethoxylates: trace analysis and environmental behavior.
Chemical Review 97: 3247}3272.
Wilkes AJ, Walraven G and Talbot GM (1992) HPLC
analysis of quaternary ammonium salts with the evapor-
ative light scattering detector. Journal of the American
Oil Chemists Society 69: 609}613.
Introduction
Successful gas chromatography (GC) requires that the
sample be volatile at the operating temperature. The
majority of synthetic polymers are of substantial mo-
lecular weight, i.e. in excess of 20 kDa, and not
amenable to direct chromatographic examination.
 III / SYNTHETIC POLYMERS / Gas Chromatography
Figure 1 shows the molecular weight limitations of
compounds suitable for gas and liquid chromatogra-
phy. Monomers with molecular weights in the range
50}100 Da are particularly suitable for direct GC.
Performance-enhancing or compounding additives
in polymers are also usually amenable to direct exam-
ination. The major difRculty with these materials is
separation or extraction from the polymer matrix,
which generally accounts for over 90% of the prod-
uct. Plasticizers form a major part of many polymer
compounds and monomeric plasticizers may fre-
quently be examined directly or after extraction.
Polymeric plasticizers, and polymeric additives after
separation, require examination by pyrolysis or by
spectrometric tools. The Rnely divided polymer is
subjected to thermal desorption or extraction or
digestion with solvents. Extraction with supercritical
Suids is Rnding greater use and has the advantage that
the removal of extraction solvent is simpliRed.
To increase the volatility of polymers, a reduction
in molecular weight must be achieved. This is most
commonly carried out by pyrolysis or thermal degra-
dation in the absence of oxygen, or to a lesser extent
by chemical degradation, which is applicable to most
condensation polymers. Both techniques are indirect
methods of analysis where the polymer is character-
ized by analysis of the volatile products of the degra-
dation. A recently developed technique, known as
pyrolytic methylation, effectively combines both
methods and is applicable to condensation polymers.
To date this new technique has found its major ap-
plication in forensic science.
Pyrolysis
Pyrolysis techniques possess several advantages. The
sample preparation is negligible, while the time for
Figure 1 Approximate useful ranges of common chromatographic procedures. (Reproduced from Haken JK (1990) Trends in
Analytical Chemistry 8:14 with permission of Elsevier Science Publishers.)
4335
analysis is relatively short in comparison with that
needed for other instrumental techniques. In addition
the sample required for pyrolysis is small.
Pyrolysis was originally carried out separate from
the GC instrument, but in situ pyrolysis in a device
directly attached to the GC was soon universally em-
ployed. The pyrolysers available are of two basic types:
(1) Furnace type. The polymer sample is introduc-
ed into a heated microfurnace attached to the
injection port of the chromatograph and the vol-
atile pyrolysis products are rapidly swept into the
column by the carrier gas.
(2) Pulse mode type. The polymer is attached to the
pyrolysis element, which is rapidly heated to
a predetermined temperature. The volatile pyro-
lysis products are rapidly swept into the
chromatograph as before. The pyrolysis element
may be either a Rlament or ribbon device that is
resistively heated or a Curie point device. With
Curie point heating the polymer is deposited on
a wire of ferromagnetic material. The wire is
rapidly heated to its Curie point using induction.
A range of wires with Curie points from 3583C
for a nickel wire to 9803C for a wire consisting of
50 : 50 iron and cobalt are available. Ribbon and
Rlament pyrolysers normally use materials of
high resistance that are inert in nature, such as
platinum or other noble metals; this reduces the
possibility of reactions occurring with the de-
graded sample.
Both types of pyrolysers are used, each type having
both advantages and disadvantages, but the two pulse
mode instruments are most widely utilized.
The Rrst application of pyrolysis and gas
chromatography was in 1954, when vinyl polymers
4336 III / SYNTHETIC POLYMERS / Gas Chromatography
and copolymers were heated at 6503C in a stream of
nitrogen. The volatile products were condensed and
subjected to gas chromatography. In situ pyrolysis
quickly followed with the use of a Rlament device,
which was little different to that which has sub-
sequently found extensive usage with almost every
conceivable type of polymer.
Early workers used large samples, which resulted in
poor heat transfer, the occurrence of combination
reactions of the volatile fragments and the production
of nonreproducible results. Bibliographies of papers
and compilations of pyrolysis results, both as pyro-
grams and bar charts, date from the 1960s, but the
early literature is of limited value as the reproducibil-
ity is often poor, and the techniques used do not
always represent current practice. Libraries of pyro-
grams have appeared in texts, but the experimental
conditions are frequently omitted and it is difRcult to
reproduce the results.
It has long been recognized that small samples,
typically much less than 1 mg, good heat transfer,
rapid heating of the pyrolysis element and rapid re-
moval of the degradation products from the heated
zone are essential in achieving reproducible results.
Packed chromatographic columns were widely
used in the early work, but capillary columns that
offer enhanced separation are now almost universally
employed. The detection limits of the Same ioniz-
ation detector, the mass spectrometer and the Fourier
transform infrared detectors in current use are orders
of magnitude better than that required by the smallest
possible pyrolysis sample. In the 1980s a sample of
800 ng of acrylic polymer would produce a mass
spectrum of about 30 compounds when pyrolysed.
More recently samples of 1
g have been used for
forensic casework, while sample sizes of 2
g have
been generally used, the major difRculty being in
handling and weighing these small samples. With
such quantities a chromatogram containing dozens of
peaks can be obtained and a mass spectrometer at-
tached to the GC will provide a mass spectrum of
each peak. The spectra may be interpreted ofSine or
in many cases may be identiRed by simultaneous
online matching with an inbuilt computer containing
a library of spectra, which shows the degree of prob-
ability of the match.
Pyrolysis Temperature and Heating Time
The temperatures used for pyrolysis are variable and
depend to some extent on the nature of the polymer.
Polymers of high thermal stability or which are highly
crosslinked obviously require a higher temperature
than a simple thermoplastic. The bond strengths of
the constituent atoms and the association of atoms
inSuence both the ease and type of degradation that
occurs, as rupture of the weakest bonds will predomi-
nate. The formation of stable free radicals generally
occurs. Little pyrolysis of polymers occurs with the
lowest available Curie point wire, i.e. 3583C and
temperatures of 500}6003C might be considered as
lower limits. The optimum temperature is considered
to be near 8003C. An excessively high pyrolysis tem-
perature is to be avoided, as with increasing temper-
ature fragmentation of the pyrolysis products occurs,
with an increased amount of very low molecular
weight gaseous products being formed. These prod-
ucts are not helpful in achieving identiRcation as they
are typical of organic compounds generally rather
than of a particular polymer.
The heating time, while variable, is short and may
range from seconds to a fraction of a second. With
Curie point pyrolysis the heating period is usually less
than with Rlament or ribbon types. In either case the
time to achieve the Rnal temperature must be short.
Pyrolysis under the conditions selected must be essen-
tially complete, a situation that is readily checked by
reheating the element after pyrolysis and determining
if further pyrolysis products are separated. The Sow
rate of the carrier gas that passes through the pyrolyser
should be such that the pyrolysis products are readily
swept from the heated pyrolysis zone into the column,
minimizing the recombination of reaction products.
Such secondary products may be more characteristic of
the apparatus than of the polymer. The unreactive
carrier gases normally used can be conveniently em-
ployed. Gases that react with the reaction products
are used in special circumstances, the most common
being oxygen for use in oxidation studies or hydrogen
with a suitable catalyst for hydrogenation.
An advantage of the resistively heated pyrolyser is
that stepwise pyrolysis, where the same sample is
pyrolysed at increasing temperatures, may be carried
out. This technique has been used with low temper-
atures to remove the majority of additives and mono-
meric plasticizers, and also with higher temperatures
to study the ease of polymer degradation. A disadvan-
tage of resistively heated pyrolysers is that the resist-
ance of the heating element may vary over time owing
to corrosion and thinning of the wire. With a vari-
ation in the resistance of the wire, the nominal tem-
perature is not achieved when the same current is
applied and the pyrolysis results are variable.
Enclosed Curie point pyrolysis (ECP) has been de-
scribed where the sample is deposited on the Curie
point element and sealed in a capillary tube, with
pyrolysis taking place in the tube. The tube is sub-
sequently broken in the carrier gas Sow. The method
has been used for the study of the oxidation of poly-
isobutylene. A comparison with conventional resis-
tive pyrolysis and ECP shows that a method for
 III / SYNTHETIC POLYMERS / Gas Chromatography
distinguishing gas phase versus melt phase secondary
reaction is possible.
Polymer Degradation
Degradation may occur by a variety of mechanisms,
or combination of these. The common mechanisms
are described below.
Most addition polymers incorporate a carbon}car-
bon backbone. With a polyoleRn such as polyethy-
lene, the bonds are equivalent and the rupture is
random. The strength of the C}C bond is approxi-
mately 349 kJ mol\1 (83 kcal mol\1) and that of
the C}H bond is approximately 393 kJ mol\1
(94 kcal mol\1), so that rupture of the former bond
occurs. In these circumstances a large number of
fragments result and the mechanism is termed ran-
dom scission. The hydrocarbons with terminal free
radical ends require to be stabilized. The fragment
with a free radical end may extract a hydrogen atom
from an adjacent fragment and become a saturated
end. In extraction a free radical is created on the
adjacent fragment. This fragment commonly stabil-
izes by
-scission, where the induced free radical site
becomes an unsaturated molecule end. This process
continues and produces a sequence of three hydrocar-
bons, the Rrst saturated, the next with a double bond
at one end, and the third with a double bond at both
ends, a series of n-alkanes, -oleRns and ,-dioleRns
being formed from methane to hydrocarbons of near
40 carbon atoms.
Where the carbon atoms are not equivalent, such as
in polyvinyl chloride where the bond strength of the
C}Cl bond is 305 kJ mol\1 (73 kcal mol\1), random
scission does not occur } rather aromatic compounds
are formed. Hydrogen chloride is eliminated and the
free radicals on the adjacent sites form a sequence of
double bonds to make an unsaturated backbone. This
then fragments to form a series of aromatic compounds.
A third mechanism that occurs with a few polymers
containing -methyl substitution, i.e. polymethyl
methacrylate and polymethyl methacrylamide, is un-
zipping or reformation of monomer. Here fragmenta-
tion of the C}C backbone occurs with the free radical
fragments formed undergoing
-scission with the
elimination of a molecule of monomer and formation
of a new free radical fragment. Repeated
-scission
leads to the formation of more monomer, which is
frequently in excess of 95% of the original monomer
concentration. The yields of monomer vary greatly
with the polymer. The lower polyalkyl methacrylates
yield essentially all monomer, but as the substituent
alkyl chain becomes larger the monomer yield de-
creases. With polylauryl methacrylate the monomer
yield is approximately 70%, as some degradation of
the alkyl chain occurs.
4337
Polystyrene degrades by a combination of mecha-
nisms, and the monomer yield is approximately 40%.
Unzipping reactions are of little importance in the
degradation of polyacrylates, where the monomer
yield is in the order of 5%.
Microstructure
Pyrolysis GC of polymers allows determination of the
microstructure in addition to the chemical composi-
tion. The polymer of simplest chemical composition,
polyethylene, is prepared by polymerization under
different conditions and with a variety of catalysts to
produce products with greatly differing physical prop-
erties, dependent on the microstructure. Short- and
long-chain branching occurs, as does stereoregularity.
First to be analysed were polymers produced by
high pressure processes. These contain a signiRcantly
branched structure with both short-chain branching
(C1}C6) and long-chain branching, as illustrated in
Figure 2A. The introduction of low pressure pro-
cesses using metal alkyl catalysts has allowed the
production of products with structures as shown in
Figure 2B and 2C.
The low density polymer produced by high pres-
sure processes undergoes random scission and
produces a strong sequence of triplet peaks corre-
sponding to ,-dioleRns, -oleRns and n-alkanes of
each carbon number, with weak multiple peaks of
isoalkanes, isoalkenes and isoalkadienes between the
triplets.
A technique that has been used in polyoleRn pyro-
lysis for decades is in situ simultaneous hydrogen-
ation of the pyrolysis products, hydrogen being used
as the carrier gas with a pre-column containing
Figure 2 Structures of various polyethylenes. (A) Low density
polyethylene (LDPE); (B) high density polyethylene (HDPE);(c)
linear low density polyethylene (LLDPE). (Reproduced from
Wampler TP (1995) Applied Pyrolysis Handbook, p. 81 with per-
mission of Marcel Dekker.)
4338 III / SYNTHETIC POLYMERS / Gas Chromatography
a hydrogenation catalyst inserted between the pyro-
lyser and the injection port. With this technique the
triplets become single peaks of the n-alkane and the
intermediate peaks are reduced in number and in-
creased in intensity, with iso compounds of compar-
able structure forming single isoalkane peaks.
While ethylene is symmetric, propylene and other
oleRns are not, and the possibility of head to tail, head
to head and tail to tail combinations exists; such
differences are evident in pyrograms. With copoly-
mers of polyethylene and polypropylene, pyrograms
containing peaks associated with each monomer are
observed; however, with different catalysts the intensity
of the individual peaks vary. This is used as a measure
of the sequence distribution of monomer units along
the chain. Sequence distribution determination is not
restricted to polyoleRns, but has also been reported
with many other important polymer systems.
The tacticity of various polymers, including
polystrene, has been determined by pyrolysis gas
chromatography. With polystyrene, fragments char-
acteristic of the polystyrene chain, ranging from
monomer to pentamer, were observed in the pyro-
gram. The relative intensities of the tetramer and
pentamer peaks reSect the original tacticity.
Minor and subtle differences in the end groups of
a particular polymer system frequently cause signiR-
cant alterations of the properties of the polymer,
particularly concerning thermal stability and trans-
parency. It is well known in gas chromatography that
the polysiloxane materials frequently used as station-
ary phases possess high thermal stability when appro-
priate end termination is employed. Details of end
termination are also of value in determining polym-
erization mechanisms. The identiRcation and deter-
mination of end groups is difRcult owing to their low
concentration. Pyrolysis-GC has been used to charac-
terize many polymer systems. Examples are poly-
methyl methacrylate radically polymerized in toluene
solution with benzoyl peroxide initiation under vary-
ing conditions. The peak intensities of some products
characteristic of the end groups present have been
interpreted in terms of polymerization temperature
and solvent/monomer in the feed.
Methacrylate end groups in polystyrene samples
have been determined by reaction with tetra-
methylammonium hydroxide where methyl meth-
acrylate was split out and determined. The initiator
used was s- or n-butyl lithium and on pyrolysis the
main product was styrene monomer together with
a considerable amount of dimers and trimers. Various
minor fragments clearly showed the presence of n-
butyl end groups. Similar analyses of end groups in
other polymers have been reported, including end
groups in polycarbonates.
Chemical Degradation
Alkyd resin constituents have been determined for
decades, the method used being chemical cleavage
followed by estimation by gravimetry, colorimetric or
spectrophotometric means. The analysis of alkyd
resins was revolutionized in the 1960s by US Govern-
ment workers who determined the reaction products
by GC.
The Zeisel reaction has also been extended by
the use of GC; traditionally all alkoxy groups
were estimated as methoxy groups. With GC, the
alkoxy groups are converted to the corresponding
alkyl halides by reaction with hydriodic acid in phe-
nol prior to chromatographic separation of the indi-
vidual halides.
Gas chromatography was early applied to the es-
timation of the hydrolysis products of polyurethanes,
polyethers, polysiloxanes and polyamides.
In situ Chemical Degradation
The degradative analysis of many polymers with
in situ GC was developed by scientists working on
determination of the volatile degradation products in
the 1970s. They conducted vigorous hydrolytic cleav-
age in a reactor constructed from a furnace pyrolyser
attached to the injection port of a gas chromato-
graph. A 30 mol% excess of a prefused mixture of
potassium hydroxide (85% KOH approximating to
the hemihydrate) and 1}10% sodium acetate as Sux
was heated for 0.5}1.0 h at temperatures within the
range 200}3503C. Volatile reaction products were
examined by GC while the reaction products that
formed alkanoate soaps remained in the reactor.
All the alkali metal hydroxides have been used for
alkali fusion, although potassium hydroxide is prefer-
red as its melting point is suitable and organic
compounds have greater solubility in a potassium
hydroxide melt than in a sodium hydroxide melt.
Table 1 shows the melting points of the common
alkali metal hydroxides in both the anhydrous and
hydrated forms.
Most acids, both organic and inorganic, have been
used to effect hydrolysis and to achieve cleavage of
ethers. They include phosphoric, hydrochloric, sul-
furic, hydrobromic, hydroiodic and p-toluenesulfonic
acids. Mixed anhydrides of p-toluenesulfonic and
acetic acid or triSuoroacetic anhydride or triSuor-
oacetic acids have also been used. Some of the acids,
particularly sulfuric acid and hydrochloric acid, pro-
duce by-products, while phosphoric and hydrob-
romic acids have often been used successfully.
The ether groups in alkylene oxide polymers have
been cleaved using a mixed anhydride of p-toluenesul-
fonic acid and acetic anhydride. The reaction is
 III / SYNTHETIC POLYMERS / Gas Chromatography
usually conducted by heating in a microSask with an
appropriate condenser. The polymer may be con-
verted to compounds suitable for GC or the reaction
products may be worked up as derivatives.
External Chemical Degradation
The work described above has been extended, by
employing external fusion, to allow all of the reaction
products to be identiRed as volatile products or as
derivatives amenable to GC. Hydrolytic reactions us-
ing alkaline or acidic catalysts are achieved, as is
acidic cleavage of ether groups. In several cases poly-
mers have been examined using simultaneous
hydrolysis and alkylation.
The advantages of external fusion are listed below:
1. Fusion is more rapid, efRcient and more readily
controlled than in situ degradation as the water
necessary for the reaction remains in the reaction
environment rather than tending to be swept into
the cold trap.
2. Multiple fusions can be carried out in an external
heater without restricting the use of a gas
chromatograph, or, more importantly, restricting
examination to GC alone.
3. Materials that would ordinarily be retained in the
reactor as soaps or other material of low volatility
can be examined after appropriate chemical reac-
tion and/or derivatization.
4. Hydrolytic degradation and cleavage of ether
groups can be conducted simultaneously or separ-
ately.
5. Other analytical techniques can be used as appro-
priate.
6. All of the components of a polymer can be ana-
lysed rather than simply those sufRciently volatile
for direct GC.
The quantitative nature of both acid and alkaline
fusion reactions has been reported and a number of
polymers have been studied with acceptable results.
Table 1 Melting points of alkali metal hydroxides
Alkali metal hydroxide Melting point
Anhydrous Hydrate
Potassium hydroxide 360 125a
Sodium hydroxide 318 64.3b
Lithium hydroxide 417 !b,c
a
Commercial potassium hydroxide contains 15% water and is
present as the hemihydrate.
b
Present as the monohydrate.
c
Decomposes to form lithium hydroxide and water.
4339
Nitrogenous polymers Nitrogenous polymers have
been widely studied. These include: polyamides such
as the simple nylon materials; the condensation prod-
ucts of dicarboxylic acids and diamines and the
condensation products of ,-aminoalkanoic acids;
the dimer polyamids (using the C36 dimer dicarboxy-
lic acids prepared from vegetable oils); the aramid
Rbres (using an aromatic diamine and a dicarboxylic
aromatic acid); and the polyhydrazides produced us-
ing hydrazine. Polyimides produced by the polymeriz-
ation of benzene tetracarboxylic acids and aromatic
diamines and copolymers of amides and imides have
also been analysed using alkali fusion.
Polyesters External chemical degradation has been
used to analyse polyesters, both containing oils and
oil-free, as well as silicone alkyds and crosslinked
systems of polyesters with various aminoplasts. In
a crosslinked system the aminoplast butylated
ureaformaldehyde is itself cleaved, while with other
aminoplasts only the butylated groups are removed.
Simple Rbreglass-reinforced plastic (FRP) and vinyl
ester laminates are cleaved by this method. With the
laminates and silicone polyesters, the siliceous Rbre-
glass (normally E-glass, a very low alkali borosilicate
glass containing approximately 50% silica and 10%
boron trioxide) or organic silicone is converted into
an organic derivative amenable to GC examination.
A chromatogram showing the trimethylsilyl (TMS)
derivatives of the polyols, dicarboxylic acids and
organic siloxane moiety produced from a silicone
polyester is shown in Figure 3. Other reinforcement
materials that are used in aerospace and other special-
ized applications include polyester or polyamide
(NomexTM) reinforcements, both of which are amen-
able to hydrolytic cleavage.
Polyurethanes Polyurethanes are conventionally the
reaction products of an isocyanate with an ether or
ester and terminated with hydroxyl groups. While
these materials are relatively resistant to hydrolysis,
they can be readily cleaved by vigorous hydrolytic
reactions. The polyurethane ether materials are more
resistant to simple solution hydrolysis than are the
polyurethane esters. Many polyurethane compounds
have been studied using both alkaline and acidic
hydrolysis. The simple condensation products
} chain-extended materials produced using a short-
chain polyol or an amine; polyether polyurethanes
used in medicine; transparent polyurethanes that use
polycaprolactone diols; isocyanate-based copolyam-
ide resins; and a urethane crosslinking agent used in
reversion-sensitive natural rubber } have all been
examined using vigorous hydrolysis reactions. Deter-
mination of the tertiary amino groups allows the
4340 III / SYNTHETIC POLYMERS / Gas Chromatography
Figure 3 Gas chromatogram showing simultaneous separation
polyols, dicarboxylic acids and silicate TMS derivatives of: 1,
solvent peak; 2, neopentyl glycol; 3, silicate; 4, trimethylol ethane;
5, trimethylol propane; 6, adipic acid; 7, isophthalic acid. (Repro-
duced from Haken JK et al. (1985) Journal of Chromatography
441: 207}212 with permission of Elsevier Science Publishers.)
number of secondary amino groups in polyurethanes
to be determined, and an estimate of the degree of
crosslinking to be made.
Epoxy resins Only simple epoxy systems, either as
the ether or crosslinked with amine compounds, can
be cleaved by fusion. The majority of epoxy systems
are complex networks that are sterically hindered and
resistant to cleavage.
Polysulfones The aliphatic polysulfones are readily
cleaved by vigorous hydrolysis. However, the aromatic
polysulfones that Rnd application as high temperature
polymers, either alone or frequently as copolymers
with polyethers, are resistant to hydrolytic cleavage.
Liquid crystal polyesters Liquid crystal polyesters
based on p-hydroxybenzoic acid, p,p-biphenol,
terephthalic acid and 2-hydroxy-6-naphthoic acid,
which are also used as high temperature polymers,
are readily cleaved by vigorous hydrolysis.
Summary
Systems that have been successfully subjected to
chemical cleavage include polyacrylates, polyamides,
polyimides, polyurethanes, polysiloxanes, poly-
urethanes, polyesters (containing vegetable oil and veg-
etable oil-free), liquid crystal polyesters, polyhydraz-
ides and silicone polyesters. Table 2 shows polymers
and additives that have been examined using both in
situ and external pyrolysis. Copolymers that include
more than one functional class have been examined;
these include polycarbonate}polydimethylsiloxane
block copolymers and isocyanate-based copolyamides.
With a completely unknown sample alkali fusion is
recommended, followed if necessary by acid reaction.
This ensures that some reactants, such as polyethers,
are not cleaved into such small fragments that their
initial composition is not apparent. For some pur-
poses acid reaction is more rapid. For routine pur-
poses some of the extraction steps, which simply
serve to separate functional classes, may be elimi-
nated. The reduction or elimination of extraction
steps increases the quantitative nature of the analyses.
It has been shown that the cleavage reactions are
essentially quantitative and that errors are introduced
by the extraction steps.
Pyrolytic Methylation
The term pyrolytic methylation was Rrst used in 1979
to describe the coinjection of tetramethylammonium
hydroxide with free carboxylic acids and phenols into
the injection port of a gas chromatograph, resulting
in the formation of methyl derivatives. The applica-
tion to polymers did not occur until a decade later.
Simultaneous pyrolysis and alkylation was conducted
by the use of tetramethylammonium hydroxide or
tetrabutylammonium hydroxide mixed with the poly-
mer in a pyrolyser. Typically a 5
g sample of poly-
mer and 2
L of the derivatizing reagent is subjected
to Curie point pyrolysis at 7703C. Separation is by
capillary GC with conRrmation using mass spectro-
metric detection.
The reaction mechanism has been discussed and the
evidence suggests that reactions occur by the following
mechanism. When intimately mixed with tet-
ramethylammonium hydroxide and heated to temper-
atures above 4003C, the polymer undergoes hydrolysis
with the strongly basic agent forming salts of the
hydrolysed products. These then undergo thermal
fragmentation to the methyl derivatives. The term
SPM, or simultaneous pyrolysis methylation, has been
acknowledged to be something of a misnomer and the
process has been renamed thermally assisted hydroly-
sis and methylation, with the abbreviation THM.
Applications
Alkyd and polyester resins The use of THM has
largely been directed towards polymers that Rnd
 III / SYNTHETIC POLYMERS / Gas Chromatography 4341

Table 2 Polymers and additives examined using in situ and external pyrolysis

Material Product/s of reaction Unidentified products

In situ pyrolysis
Nylons Diamine Alkali metal soap of acid
Phthalate esters Corresponding alcohol Alkali metal soap of acid
a
Polyacrylamide Ammonia
a
Polyacrylonitrile Ammonia
Polyamides Diamine Alkali metal soap of acid
Poly(amides-imides) Diamine Alkali metal soap of acid
Polycarboranesiloxanes Amine and diamine Nil
Polychloroacrylate esters Corresponding alcohol Alkali metal soap of acid
Polyimides Diamine Alkali metal soap of acid
Polymethacrylate esters Corresponding alcohol Alkali metal soap of acid
a
Polysiloxanes Aliphatic or aromatic hydrocarbon
Polyurethanes Diamine Alkali metal soap of acid

External pyrolysis
Alkyd resins Acids and polyols as derivatives Nil
b
Alkyd resins, crosslinked Acids and polyols as derivatives
Butylated urea formaldehyde Carbon dioxide and n-butyl trifluoroacetate Nil
b
Epoxy resins Diamines and acetate derivatives
Nylons Diamines and diacids as derivatives Nil
Phthalate esters Alcohols and acids as derivatives Nil
a
Polyacronitrile Ammonia
a
Polyacrylamide Ammonia
Polyamides Diamine and acids as derivatives Nil
Polyhydrazides Hydrazine and diacids Nil
Polyamides Diamines and acids as derivatives Nil
Polysiloxanes Hydrocarbons and silica derivatives Nil
Polyurethanes Diamines and ester or ether derivatives Nil
Silicone polyesters Silicone, acid and diol derivatives Nil

a
Reaction with pendant groups.
b
Limited application.

application in surface coatings. Pyrolysis and methyl-
ation of alkyd resins gives methyl esters of the
constituent acids and methyl ethers of the polyols.
A soyabean oil-pentaerythritol-orthophthalic alkyd
resin produced C8}C16 methyl alkylanoates from the
vegetable oil; tri- and tetramethyl ethers of pen-
taerythritol, dimethyl orthophthalate and methyl
benzoate from the benzoic acid used as the chain
regulator; and cyclopentanone from scission of C}O
bonds of adipic acid, dimethyl isophthalate and
methyl benzoate.
The Rbre polyester, polyethylene terephthalate,
gave benzene, a degradation product obtained on
pyrolysis, dimethyl terephthalate and methyl ben-
zoate as a combination product. It was evident in the
analysis that no product attributable to the ethylene
portion of the polymer was reported. This result is
different from that obtained by hydrolytic degrada-
tion and chromatography, where a component peak
attributable to the ethylene or butylene chains was
identiRed.
The partial structure of alkyd resins may be elucid-
ated; in addition to identiRcation of the carboxylic
acids and polyhydric alcohols as the appropriate
esters and ethers, the drying oil type, degree of
cure, the oil length and modiRcation with rosin
and epoxy resins may be determined. The most com-
mon polyhydric alcohols used in alkyd resins are
glycerol and pentaerythritol, resulting in the
formation of the di- and trimethyl esters and the tri-
and tetramethyl esters, respectively. All three of
the isomeric phthalic acids are readily separated
on the low polarity capillary column used. Methyl
benzoate is observed, which is a problem as some
of this compound is due to decarboxylation of some
of the phthalic acid and differentation of the use
of benzoic acid as a chain terminator is not possible.
All types of oils, both vegetable and the more
highly unsaturated marine types, are readily
characterized before autoxidative polymerization.
However, after drying or autoxidative polymeriz-
ation, little unsaturation remains. The relative pro-
portion of unsaturated to saturated fatty acid methyl
esters gives an indication of the cure or age of the
alkyd resin. A simple alkyd resin based on linseed
oil-pentaerythritol-orthophthalic acid was examined
over 5 months. The following conclusions were
made:
4342 III / SYNTHETIC POLYMERS / Gas Chromatography
1. Before autoxidative polymerization the ratio of
linolenic acid (9,12,15-octadecatrienoic acid) to
palmitic acid (hexadecanoic acid) was signiRcant
but the unsaturation rapidly decreased such that
after 2 days all the linolenic acid had been re-
moved by crosslinking.
2. After 2 weeks the ratio of oleic acid (9-oc-
tadecadienoic acid) to stearic acid (octadecanoic
acid) slowly began to reduce with time. After
4 months the concentration of oleic acid had been
reduced by approximately two-thirds of its initial
concentration.
3. Nonanedoic acid began to appear after 3 days
and increased to a maximum in 1 month. The oil
length of an alkyd resin is the percentage of fatty
acid acylglycerols present in the total resin solids.
By considering the ratio of products from the
drying oil to the aromatic compounds from the
phthalic acids, an approximation of the oil length
may be obtained. Some decarboxylation occurs
however, and the value of the estimate is reduced.
Naturally occurring modiRers, i.e. rosin (as methyl
dehydroabietate), have been determined, as have
epoxy resins.
Epoxy resins A simple epoxy ether, i.e. a bisphenol
A epichlorohydrin condensate, on pyrolysis produced
three component peaks } phenol, isopropenyl phenol
and bisphenol A. However, on pyrolytic methylation
a variety of components was produced including
phenol, isopropenyl phenol, the monomethyl ethers
of these compounds and bisphenol A. The diether of
bisphenol A was also formed.
Polyvinyl acetate Pyrolysis butylation has been used
with low molecular weight products such as vinyl
acetate-containing polymers where the vinyl acetate
formed by pyrolysis is advantageously examined as
vinyl butyrate, the disadvantage being that by-prod-
ucts, i.e. n-butanol and tributylamine, are formed.
Polymethyl acrylates Pyrolytic butylation of a meth-
acrylic copolymer produced n-butyl methacrylate. n-
Butyl acetate and n-butyl butyrate were produced
from a cellulosic acetate}butyrate copolymer and n-
butyl cyanoacrylate was produced from a commercial
cyanoacrylate adhesive.
Rosin adducts The rosin-based resins have been ex-
tensively studied. While these are natural polymers,
they are used in many modiRed forms. Abietic acid is
the principal acid and it contains both a conjugated
diene and a carboxylic acid, both of which are readily
reacted on a commercial scale. Wood rosins contain
a high proportion of diterpentine and mixtures of
seven organic acids. Pyrolytic methylation has al-
lowed the identiRcation of (1) fumaric acid, (2) sand-
araco-fumaric acid, (3) pallstric acid, (4) isofumaric
acid, (5) dehydroabietic acid, (6) abietic acid and
(7) neoabietic acid.
Para-substituted alkylphenol resins or modiRca-
tions with rosin or its esters produce characteristic
pyrograms when subjected to pyrolytic methylation.
Tertiary butylphenol and p-nonylphenol are the ton-
nage phenols used. Traces of the free phenols result
from both phenol modiRcations and methyl- and
dimethyl-substituted phenols. Pentaerythritol rosin
esters produce peaks due to the methyl esters of
dehydroabietic acid and abietic acid in addition to the
tri- and tetramethyl ethers of pentaerythritol. ModiR-
cation of rosin with fumaric or maleic acids produce
dimethyl fumarate. Rosin and reaction products with
fumaric acid have been detected as a size on paper at
the 1% level.
Polycarbonates Polycarbonates have been cleaved
using alkaline reaction. Various phenolic compounds
are formed by C}C bond cleavage as well as by
cleavage of carbonate linkages. Almost quantitative
degradation of the main chain occurs through react-
ive pyrolysis at the carbonate linkages to yield the
dimethyl derivatives of the constituents.
Liquid crystal polyesters As in chemical degrada-
tion, reaction occurs with liquid crystalline polyes-
ters, partial reaction occurring with products based
on p-hydroxybenzoic acid and 2-hydroxy, 6-naph-
thoic acid. Quantitative results were achieved by
varying the reaction conditions. Similar liquid cry-
stalline polyesters based on 4-hydroxybenzoic acid,
terephthalic acid and 4,4-biphenol produced almost
quantitative results.
Conclusion
The application of GC to synthetic polymers has
been outlined using three types of methods } pyroly-
sis, chemical degradation and pyrolytic alkylation.
In all cases a considerable reduction in molecular
weight is achieved before GC. Pyrolysis is applicable
to both addition and condensation polymers and oc-
curs by thermal degradation of the constituent
chemical bonds. Chemical degradation and pyrolytic
alkylation are applicable to condensation polymers
with degradation at the location of constituent func-
tional groups. Pyrolysis in some cases produces
quantitative results, while chemical degradation usu-
ally produces quantitative results. Pyrolytic alkyla-
tion has to date been used only for qualitative
analysis.
 III / SYNTHETIC POLYMERS / Liquid Chromatography
See also : II/Chromatography: Gas: Derivatization;
Detectors: Mass Spectrometry; Pyrolysis Gas Chromatog-
raphy. Extraction: Supercritical Fluid Extraction.
Further Reading
Challinor JM (1991) The scope of pyrolytic methylation
reactions. Journal of Analytical and Applied Pyrolysis
20: 15}24.
Crompton TR (1989) Analysis of Polymers, An Introduc-
tion. London: Pergamon.
Cross J (1987) Non Ionic Detergents } Chemical Analysis.
New York: Marcel Dekker.
Haken JK (1993) Fusion reaction chromatography:
a powerful analytical technique for condensation poly-
mers. Advances in Chromatography 33: 177}231.
Liquid Chromatography
C. H. Lochmu] ller, Duke University, Durham,
NC, USA
Copyright ^ 2000 Academic Press
Introduction
The goals in the use of liquid chromatography for the
separation of polymers and polymer oligomers in-
clude the determination of purity, the production of
pure/purer polymer mixtures and for obtaining qual-
ity control data for polymer intermediates. There are
fundamentally two partition mechanism options for
such separations: size exclusion or sorption in the
sense of surface adsorption or dissolution into a sta-
tionary phase. Size exclusion involves the partition of
the molecules of interest from the mobile phase into
the stationary mobile phase contained in the various
pores of the solid support. The extent to which the
stationary liquid is explored by the polymer mol-
ecules is determined by their Stokes radius (dynamic
size) and the volume of mobile phase in pores of
a diameter large enough for penetration to be pos-
sible. Adsorption onto or sorption into a phase coated
or grafted as a thin Rlm on the surface of a solid
support is dominated by solubility in the mobile
phase and the chemical potential for sorption of the
polymer molecules in a given mobile phase in contact
with a given stationary phase. Because stationary
phase supports used in modern liquid chromato-
graphy are themselves porous, mixtures of size exclu-
sion in the presence of sorption and vice versa are
known.
4343
Haken JK (1996) Degradative polymer analysis by
chromatography. Journal of Chromatography A756:
1}20.
Haken JK (1998) Pyrolysis gas chromatography of syn-
thetic polymers. A bibliography. Journal of Chromato-
graphy A 825: 171}187.
Mitchell J Jr (ed.) (1991) Applied Polymer Analysis and Char-
acterization, vols 1 and 2. Munich: C. Hanser-Verlag.
Taguchi VY (1990) Derivatization reactions. In: Clements
RE (ed.) Gas Chromatography, Biochemical, Biomedi-
cal and Chemical Applications, pp. 129}177. New
York: Wiley.
Wampler TP (1995) Applied Pyrolysis Handbook. New
York: Marcel Dekker.
Whitlock LR and Siggia S (1974) Fusion reaction gas
chromatography. Separations and PuriTcation Methods
3: 299}337.
Size Exclusion
Historically, size exclusion has been the method most
often used for polymer separation, puriRcation and
molecular weight determinations. The technique de-
veloped in parallel in the organic polymer area and
the biological polymer area. When used in organic
polymer work, the term gel permeation is used. In
water soluble biopolymer work, it is called gel Rltra-
tion. There is no fundamental difference in the prin-
ciples involved and both are size exclusion based. In
liquid chromatography molecules move in the direc-
tion of development because of mobile phase Sow.
Most gel electrophoresis methods are, in reality, size
exclusion based separations and that includes the gel
methods used for sequencing of nucleotide fragments.
There the driving force is electromigration of
molecules with essentially identical ionic mobility
which reptate through a porous polymer medium at
rates proportional to size.
Sorption
Despite the potential attractiveness of a method
which could introduce chemical selectivity in to the
separation of polymers, sorption methods have seen
little practical application until more recent times.
The sole exception is the ion exchange puriRcation of
polyelectrolytes such as proteins. The history of the
development of polymer high-performance liquid
chromatography (HPLC) is an interesting one and
is detailed in subsequent paragraphs. The reader
should keep the following introduction in mind when
4344 III / SYNTHETIC POLYMERS / Liquid Chromatography
considering various models for retention of polymers
in sorption techniques.
The liquid chromatography of small molecules is
dominated by solubility of the molecule[s] of interest
in the moving or mobile phase. This is in stark con-
trast to gas chromatography where the stationary
phase contribution dominates. One controls retention
and the large fraction of selectivity (differential mi-
gration) in liquid chromatography by mixing various
solvents to obtain differential solubility sufRcient for
the task of separation. That is not to suggest that the
stationary phase is not important. It is, but generally
as a secondary effect.
The rate of change of retention volume (or time at
constant Sow) for small molecules (m.w.(2 kD) is
not steep compared to polymers. Thus the strategy for
small molecule separations is to use combinations of
solvents in which one solvent is a good solvent and
the other is more hostile or poorer in terms of
solubility of the molecules of interest. Polymers, on
the other hand can have very rapid transitions from
soluble to insoluble over narrow ranges of good/poor
solvent mole fraction.
It is also common practice to inject samples of
small molecules in a solvent which is good compared
to the mobile phase. This technique can have awk-
ward effects in polymer chromatography. If the mo-
bile phase is one in which the solubility is already very
small or near zero, the injected plug of good solvent
moves with the leading and trailing edge of the plug
being depleted of solute. The net effect is to coat the
column with polymer with the excess eluting at the
void volume. Columns are well-designed packed beds
and, as such, are very poor mixers. The solution is to
dissolve the polymer in the mobile phase and to mix
the sample solution with the mobile phase before
column contact.
There has been much interest in methods for the
fractionation of macromolecules using reversed-phase,
high-performance liquid chromatography (RPLC).
Reversed-phase methods are those that have a rela-
tively polar mobile phase and relatively non-polar
stationary phase, e.g. water and parafRn oil. If it were
possible to achieve both isocratic and gradient elution
of polymer oligomers and isomers, then these separ-
ation techniques could provide vastly more insight and
control than is currently the case in a majority of the
applications where size exclusion alone dominates.
Certainly debates over the validity of models are an
important scholarly aspect of the current dialogue.
Successful models based on sound physico-chemical
principles could guide the development of practical
applications. The short-term solution is likely to be
a combination of models if the historical case for the
application of HPLC to small molecules applies here.
Although there are many reported successful
examles of polymer separations and several models
have been suggested, the retention mechanism of
polymers in RPLC remains unclear. GloK ckner sug-
gests a precipitation}redissolution model for the
gradient elution of polymers. In this model, polymer
molecules repeatedly precipitate onto the stationary
phase and redissolve into the mobile phase until
Rnally eluting at a mobile phase composition at which
the polymer is totally soluble. Retention depends sol-
ely on the mobile phase with the column playing
a passive role providing only a large surface area as
support for the precipitate. Armstrong, Martire,
Boehm and Bui proposed a model for critical solvent
composition behaviour found by some in the isocratic
elution of polymers. This model is often called BMAB
theory or critical solution theory. This theory was
developed from a statistical treatment of the equilib-
rium distribution of inRnitely dilute polymer molecu-
les between a mobile phase and a stationary phase
based on the Flory}Huggins theory. According to the
model, the range of the mobile phase composition
within which Rnite retention factor (k) values can be
observed under isocratic elution conditions is very
narrow for high molecular weight polymers. Plots of
log k versus the mobile phase composition show
slopes that mean that polymer molecules are either
inRnitely retained or not retained at all. Therefore,
isocratic retention should be impossible. It can be
concluded from this model that the separation is
strictly mobile-phase controlled and has little to do
with the column length. If either of these models are
correct in every detail, isocratic elution is impossible
because, under isocratic elution, the polymer molecu-
les either Sow through the column without any reten-
tion or strongly adhere to the stationary phase
without ever eluting.
In contrast, Snyder and coworkers assert that no
special model is needed for the polymer retention and
the traditional models can be used in interpreting the
retention behaviour of polymers.
After numerous failures in attempts to reproduce
the published work of others, LochmuK ller and
McGranaghan were the Rrst to consider the likely fate
of the injected polymer sample in the mobile phase
prior to its contact with the column. They found that
traditional retention behaviour was obtained only
when the sample was adequately mixed with the
mobile phase using a low dispersion, crocheted capil-
lary tube placed between the sampling device and the
column. They reported that polystyrenes of molecular
weight ranging from 2000}2 800 000 Da could be
separated under isocratic elution conditions with bi-
nary mobile phases of tetrahydrofuran/H2O and di-
chloromethane/acetonitrile. Finite, non-zero k values
 III / SYNTHETIC POLYMERS / Liquid Chromatography
and linear relationships of log k versus the volume
percentage of tetrahydrofuran and dichloromethane
were observed. Alhedai, Boehm and Matire sub-
sequently reported the isocratic elution behaviour of
polystyrene homopolymers.
In polymer RPLC, separations are achieved by us-
ing a mixture of good and poor solvents as the
mobile phase. A good solvent is one that is thermo-
dynamically favourable for polymer solution and
a poor solvent is thermodynamically unfavourable.
Since polymers have low solubility in poor solvents,
polymer samples are often dissolved in a good solvent
for chromatography. The result is that the injected
sample plug is a better solvent than the actual mobile
phase. For hydrophobic polymers, the good solvent
usually is a strong solvent in RPLC. Often sample
preparation is followed by injection without any fur-
ther treatment of the sample. Because chromato-
graphic columns are, by their very design, poor
mixing devices and the equilibration between the
polymer sample and the mobile phase may be slow on
the chromatographic time scale, the polymer sample
can remain well solvated from its interior in the
injection solvent and isolated from the mobile phase
and stationary phase effects. Under the worst condi-
tions, the mobile phase can be so hostile to the poly-
mer that the polymer sample will co-elute with the
injected plug of the good solvent. This solvent effect
could explain why normal chromatographic retention
behaviour was not observed in some of the previous
studies. In addition, polymer molecules can undergo
various conformational changes in the chromato-
graphic process due to the difference between the
injection solvent and the mobile phase. These confor-
mational changes can complicate separations and
make reproducible, meaningful results difRcult to ob-
tain. The use of a pre-column mixer and an initial
binary solvent whose composition is close to the
mobile phase composition affords equilibration be-
tween the polymer and the mobile phase and also
affords a more uniform elution condition.
There are, of course, signiRcant differences in the
properties that distinguish different polymer types
from each other that must be taken into account in
guiding method development. In the special case of
some biological polymers such as proteins and poly-
peptides, a contribution of size exclusion and ion
exchange is possible and careful manipulation of
the ion exchange effect can provide resolution of
genetically different isoenzymes. In other cases where
the polymer has groups with little acid}base differ-
ence, as in the case in gene sequencing, it is possible to
cleave the macromolecule at speciRc sequence events
and to tag these fragments. The resulting fragments
are then separated through size exclusion in a gel
4345
matrix using electrokinetic driving force through the
gel space. Current applications of HPLC are more
limited and most of the literature examples of protein
resolution are carried out with molecules differing by
thousands of Daltons and are limited to analytical
purposes through the use of denaturing conditions
where narrow peak shape is more important than
retention of native structure.
Many, if not most, common organic polymers are
either non-electrolytes or strong polyelectrolytes. In
the latter case, the most common situation is one in
which the ionized or ionizable function is the same
for each repeating unit and/or the groups have nearly
identical pKa values. Ion exchange is, thus of little use
in the resolution of oligomers. In the case of non-
ionizable polymers, such as the polystyrenes, the meth-
acrylates, acrylates, polyesters and ethers, the use of
RPLC offers the possibility of selectivity by both sol-
vent effect and stationary phase interaction. If ordi-
nary chromatography is possible, then it could be
possible to resolve oligomers, to resolve copolymer
variations by chemical selectivity and thus improve the
quantitation as well as the isolation of such materials.
There are two solutions to the steep gradient in
d ln k/d Vol%. The Rrst is to Rnd a second at a
solvent less hostile to the polymer under investiga-
tion. The second is provided by modern instrumenta-
tion and that is that a gradient mixer will mix
mixtures. This second option involves running
a gradient where the A solvent or poor solvent is
already a mixture and the same for B. In principle,
a gradient can be run from 60% A to 58% A in a way
reproducible to 1% by volume A in B. In this manner,
the steep gradient in d ln k/d Vol% is slowly traver-
sed in the gradient mixing process. The results can be
dramatic and it is possible to resolve individual poly-
styrene oligomers at an average molecular weight of
100 kD.
Figure 1 shows a separation of two polymethyl-
methacylate samples, one a synthetic sample contain-
ing lower molecular weight oligomers after gel
permeation separation, and the other a nominally
monodisperse 75 kD standard. Figure 2 is the linear
dependence of retention as a function of tetrahydro-
furan vol% in water for a wide range of polyethylene
glycol monodisperse standards. Note how the
slope steepens with increasing molecular weight.
Figure 3 is a similar plot for poly-L-, poly-D- and
poly-D,L-tryptophans. Note the linearity of response
but also that the lines for all poly-L and the poly-D,L
are not parallel despite the molecular weight being
the same (&5.5 kD).
There are many good reasons to want to use
a single component, isocratic elution method. The
Rrst is the cost per sample run. The second is that
4346 III / SYNTHETIC POLYMERS / Liquid Chromatography

Figure 1 Separation of a mixture of PMMA 33.5-kD (2) and PMMA 75-kD (3) in gradient mode with mobile phase THF/water from
68/32 to 80/20 in 20 min on a Hypersil 20 cm;2 mm C18 column. (Reproduced, in part, from the authors work with permission of the
Journal of Chromatography Science and Preston Publications.)

only one solvent need be removed from the collected
polymer sample. The potential of modern liquid
chromatography is to produce samples which are
truly monodisperse. The fewer separation solvents
the better. A possible route is the use of a single
solvent and the temperature dependence of k. Of
course, the magnitude of the temperature effect on
retention is in direct proportion to the solubility de-
pendence. The temperature gradient can be applied as
it is in gas chromatography, i.e. as a gradient over
time. Another is a gradient over space, the column
length. The Rrst example of a spatial gradient was in
the separation of polyethylene glycols. These poly-
ethers have a decreasing solubility in water with in-
creasing temperature. Thus a column kept at high
temperature will show large retention volumes for
polyethylene glycols and a gradient run from hot
to cold is an option. A column kept relatively hot at
the inlet and cold at the outlet will show a reduced
rate of development for polyethylene glycols at the

Figure 2 Linear plots of log k  (retention factor) vs. volume fraction of THF in water for PEG samples. (Reproduced, in part, from the
authors work with permission of the Journal of Chromatography Science and Preston Publications.)
 III / SYNTHETIC POLYMERS / Liquid Chromatography 4347

Figure 3 Dependence of log k  versus volume fraction of THF poly(L-tryptophan)s and poly (DL-tryptophan)s. (Data taken in part from
work reported in LochmuK ller and Chun Jiang (1994) Journal of Liquid Chromatography 17: 3179}3189.)

Figure 4 Chromatograms of mixture (1) of PEG 26-KD (A), 46-KD (B) and 95-KD (C) (ACN/H2O 42/58; Top, t"233C; the thermal
gradient was !0.053C min\1 started from 283C. Bottom, tinlet"403C, toutlet"233C, the gradient was 0.73C cm\1 along the 10 cm
column. (Reproduced, in part, from the authors work with permission of the Journal of Chromatography Science and Preston
Publications.)
4348 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
inlet and higher at the outlet. The effect is a function
of molecular size within the polymer class. Figure 4 is
an example of both approaches.
See also: II/Chromatography: Liquid: Mechanisms:
Reversed Phases; Mechanisms: Size Exclusion Chro-
matography; III/Gradient Polymer Chromatography:
Liquid Chromatography. Peptides and Proteins: Liquid
Chromatography.
Further Reading
Armstrong DW and Bohem RE (1984) Journal of
Chromatographic Science 22: 378.
Barth HG and Janca J (1991) Polymer analysis and charac-
terization. Journal of Polymer Science 1991.
Thin-Layer (Planar) Chromatography
L. S. Litvinova, Institute of Macromolecular
Compounds of Russian Academy of Sciences,
Petersburg, Russia
Copyright ^ 2000 Academic Press
The use of thin-layer chromatography (TLC) in poly-
mer analysis was Rrst mentioned in 1968. Belenkii
and Inagaki with their co-workers described separations
according to the composition of random styrene}
methyl methacrylate and styrene}methyl acrylate
copolymers with molecular weights varying from 40
to 200 kDa. Since then polymer TLC has been de-
veloped intensively and other researchers have also
begun to work actively in this Reld.
The mechanisms of polymer TLC have been inves-
tigated and the methods for the determination of
molecular weight (MW) and molecular weight distri-
bution (MWD) of homopolymers, such as poly-
styrene (PS), poly(ethylene oxide) (PEO) and
poly(methyl methacrylate) (PMMA), have been de-
veloped. Impressive results were obtained for separ-
ations in accordance with different features of the
polymer architecture. This investigation of structural
heterogeneity of styrene}methyl methacrylate (S-
MMA) copolymers have made it possible to separate
random, block and alternating copolymers as well as
two- and three-block copolymers. The stereoregular
heterogeneity of PMMA and polybutadiene (PB) has
also been determined. Styrene (S) and butadiene (BD)
block copolymers have been studied and deuterated
and hydrogenous PS separated.
In the vast majority of studies silica gel has been
employed as a sorbent with the occasional use of
Barth HG and Mays JW (1991) Modern Methods of Poly-
mer Characterization. New York: Wiley and Sons.
Bohem RE and Matire DE, Armstrong DW and Bui KH
(1984) Macromolecules 17(3): 400.
de Gennes R (1979) Scaling Concepts in Polymer Physics.
Ithaca: Cornell University Press.
GloK ckner G (1987) Polymer Characterization by Liquid
Chromatography. Elsevier.
LochmuK ller CH and Chun Jiang (1994) Journal of Liquid
Chromatography 17: 3179}3189.
LochmuK ller CH and Chun Jiang (1995) Journal of
Chromatographic Science 561}567.
LochmuK ller CH, Qicai Liu and Chun Jiang (1996) Journal
of Chromatographic Science 34: 69}76.
Shalliker RA, Kavanagh PE, Russell IM and Hawthorne
DG (1992) Chromatographia 33: 427}433.
Snyder LR, Stadalius MA and Quarry MA (1983) Analyti-
cal Chemistry 55: 1412A.
alumina. In 1976 Belenkii and co-workers reported
that critical conditions exist on passing from size-
exclusion to adsorption chromatography of poly-
mers. The Rrst reviews on polymer TLC appeared in
1977. In 1980 GloK ckner demonstrated the import-
ance of gradient elution for polymer separations, and
in 1982 Armstrong and co-workers used reversed-
phase plates to separate homopolymers according to
MW. More recent reviews have been by GloK ckner
(1987) with the most comprehensive review present-
ed by Gankina and Belenkii in 1991.
Here the behaviour of macromolecules and small
molecules are compared and the mechanism of
chromatographic processes and the analysis of differ-
ent types of polymer heterogeneity considered.
Behaviour of Macromolecules under
TLC Conditions
TLC is one of the most efRcient methods used for the
fractionation of polymers and the analysis of their
heterogeneity. The chromatographic behaviour of
polymers differs from that of low MW compounds in
many ways that can be revealed even in the analysis
of narrow-dispersity homopolymers. Unlike low MW
compounds, polymers are characterized by physical
heterogeneity, i.e. they are a mixture of macro-
molecules with different degrees of polymerization
(polymer homologues). The concept of molecular
weight is replaced by the expression average MW,
which is a statistical average value. In addition
to physical heterogeneity characterized by molecu-
lar weight distribution (MWD), chromatographic
 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
methods of analysis also enable the determination of
chemical, functional, structural, stereoregular and
topological polymer heterogeneity.
Macromolecules are characterized by low diffusion
coefRcients in solution and by even lower coefRcients
in pores. This adversely affects the kinetics of en-
trance and exit of macromolecules in and out of
sorbent pores. Macromolecules consisting of many
repeat units are, moreover, characterized by multi-
centre adsorption, which makes the sorption}desorp-
tion process slower and more complex. Both these
factors hinder interphase mass transfer and, hence,
chromatographic zones become broader in the longi-
tudinal direction. At the limit, tails and tracks appear
on chromatograms in both adsorption and adsor-
ptionless chromatography (Figures 1 and 2). In TLC
both these factors become apparent as the mobile
phase velocity is increased as a result of the increase
of sorbent particle size. Hence, it must be borne in
mind that in polymer analysis the eluent velocity is
subject to greater limitations than in the analysis of
low MW substances. To avoid the appearance of false
zones and tails in polymer analysis, sorbents with
a particle size of 5}8
m must be used.
SpeciRc properties of polymers are a consequence
of the large size of their molecules. According to
current concepts, a long Sexible chain molecule in
a dilute solution is coiled. The coil size is comparable
to that of the pores. In the absence of adsorption the
molecules enter the pore when the coil size is smaller
than that of pores. When the coil size is too large, the
molecule is excluded from the pore.
Even dilute polymer solutions are characterized by
considerable viscosity, this viscosity increases with
concentration and MW. The chromatographic zone
forms a region with high viscosity, past which the
mobile phase Sows. This leads to an increase in the
Figure 1 TLC of PS with MW (kDa) of (1) 20, (2) 111, (3) 200, (4) 498, (5) 865 and (6) 2610 on plates coated with silica gel KSKG
with dp (
m) of (A) 9.3, (B) 12.8 and (C) 19. Eluent: cyclohexane}benzene}acetone (12 : 1 : 1).
4349
size of chromatographic zones in the longitudinal
direction (in the direction of mobile phase motion).
Moreover, the longitudinal size of the zone increases
both with concentration of the polymer in the zone
and with increase in MW (Figure 2). Therefore, the
shape of the polymer chromatographic zones is usu-
ally elongated in the direction of mobile phase
motion, not only because of heterogeneity in MW
and spreading as a result of slow interphase mass
transfer, but also because of the viscous effect.
Polymers dissolve much more slowly than low MW
compounds and their dissolution is preceded by
swelling. Since, in TLC, dissolution precedes
chromatography, the slower dissolution rate of poly-
mers with high MW in the starting zone is manifested
as tails on a plate at high eluent velocity (particle size
20
m). Moreover, polymer solubility depends on
MW, and the polymers usually become less soluble as
MW increases.
The coil size (hydrodynamic volume) at a Rxed
MW is not constant. It depends on the thermodyn-
amic quality of the solvent (thermodynamic quality
expresses the measure of thermodynamic afRnity of
the solvent for the polymer) and also varies on pas-
sing from the mobile to the stationary phase.
There are several points of view about the mecha-
nism of macromolecule adsorption. One view is that
during adsorption the macromolecule diffuses from
the solution into the sorbent pore, more or less retain-
ing its globular shape. Another hypothesis states that
in adsorption on the pore surface the molecule is
uncoiled and lies Sat on this surface. This becomes
possible as a result of enthalpy gain when the chain
segments interact with active centres on the sorbent
surface. This gain exceeds the increase in the free
energy because of entropy decrease. This process
is accompanied by a considerable decrease in the
4350 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
Figure 2 Adsorptionless TLC of PS with MW (kDa) of (1) 20, (2) 111, (3) 200, (4) 498 and (5) 867 on silica gel plates with sorbent
particle size dp (
m) of (A) 6.5 and (B) 19.0. Mobile phase: toluene.
volume occupied by the macromolecule, i.e. by an
increase in density. As a result, the macromolecules
are adsorbed on the surface of small pores inaccess-
ible to them in a size-exclusion regime. Consequently,
the change in free energy
G, when a macromolecule
enters a pore is the sum of the change in entropy
S
and enthalpy
H of the system (
G"
H!T
S).
The entropy change is caused by a decrease in the
number of conformations of the macromolecule inside
the pore over that in solution. The value of
H is due
to the interaction of chain units with the pore walls.
It has been established experimentally, and con-
Rrmed theoretically, that in polymer chromatography
the adsorption region and the molecular-sieve region
are separated by a critical point at which the change
in the conformational free energy of the macro-
molecule on passing from the mobile into the station-
ary phase is equal to zero. At this point the
macromolecules undergo Rrst-order phase transition.
Under critical conditions homopolymers are not sep-
arated according to molecular weight; the pore struc-
ture of the sorbent and its speciRc area do not
inSuence the behaviour of macromolecules. Hence,
the number of separation mechanisms for polymers is
greater than that for low MW compounds: besides
adsorption chromatography (!
G'0, kd'1) at
least two other chromatographic regimes are possible:
E !
G(0, kd(1 } size-exclusion chromatogra-
phy (SEC);
E !
G"0, kd"1 } adsorption chromatography
under the critical conditions (ACCC), where kd is
the distribution coefRcient.
It is possible to pass from adsorption regime to size
exclusion and vice-versa by varying temperature and
eluent composition or by modifying the adsorbent.
These facts experimentally conRrm a single mecha-
nism of the liquid chromatography of polymers.
Chromatographic Regimes Applied
to the TLC of Polymers
Adsorption+Exclusion Regimes
Adsorption TLC Adsorption TLC (ATLC) is used
more extensively in polymer analysis. Enthalpy
change is the dominant factor in the mechanism of
adsorption separation of macromolecules. By select-
ing appropriate chromatographic conditions, adsorp-
tion activity of macromolecules can increase with
increasing MW or the fraction of adsorption-active
polar groups (for copolymers or homopolymers with
functional groups).
ATLC is used to separate homopolymers according
to their MW, functional groups and stereoregularity.
It can also be used for the separation of copolymers
according to composition and for the analysis of the
MWD of oligomers with complete separation into
oligomer homologues.
The most complex problem in ATLC is mobile
phase selection. It is solved for the separation of
homopolymers according to MW. In this case, the
eluent for corresponding oligomer homologues is se-
lected Rrst (using, for example, the Prizma model).
Subsequently, a small amount of the adsorption-
active component, the displacer, is added.
Exclusion TLC Entropy change is the dominant fac-
tor in the mechanisms of exclusion separation of
macromolecules (ETLC). The chromatographic mo-
 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
bility of macromolecules is determined by the ratio of
the size of macromolecules to pore size and increases
with the increasing hydrodynamic volume of the
macromolecules. The exclusion effect is seen in TLC
when two conditions are obeyed: the adsorption ac-
tivity of the sorbent is suppressed and its pores are
Rlled with the solvent. The interparticle volume
should remain free. Pore Rlling can be accomplished
either by pre-elution with subsequent removal of the
solvent from the interparticle volume: the solvent
passes along the plate before sample spotting; or else
capillary condensation takes place during preliminary
plate saturation with solvent vapour in a saturated
chromatographic chamber. In the latter case the sam-
ples are spotted before saturation (or pore Rlling).
Solvents or their mixtures in which analytes migrate
along a dry sorbent layer with the front are used as
eluents. The pore size distribution of the sorbent is of
great importance and resolution in ETLC is lower
than that in ATLC.
Adsorptionless TLC Adsorptionless (or viscomet-
ric) TLC is a type of exclusion TLC. It is carried out
when the mobile phase moves along the dry sorbent
layer. In this case the polymer is concentrated near
the eluent front because the pores of the sorbent are
accessible to it.
Moreover, a viscous polymer solution in the
chromatographic zone plays the role of a kind of plug
along which the solvent Sows. Hence, the chromato-
graphic zone acquires a droplike elongated shape and
distinct boundaries. The zone length increases with
both polymer concentration and its MW. For a Rxed
polymer quantity, the length of the chromatographic
zone (L) is a function of intrinsic viscosity ([], at
Figure 3 Dependence of zone length (L) on intrinsic viscosity
for (1) PS, (2) PMMA, (3) polycarbonate, (4) polyisoprene (PI),
(5) poly(vinyl chloride) and (6) poly(-methyl styrene) with MW
(kDa) of 20.8, 111, 200, 498, 867, and 2610 (1); 60.6 and 75 (2);
36.4 and 50.2 (3), 32.5 and 109 (4); 75 (5); and 97 and 610 (6) at
a fixed polymer quantity (
g) of (I) 4 and (II) 10. Binder: gypsum;
layer thickness, 500
m.
4351
inRnite dilution). This dependence is obeyed for poly-
mers of different nature, i.e. it is of universal charac-
ter and can be obtained with the aid of any type of
polymer standards (Figure 3). To determine [], it is
necessary to plot the calibration dependence L"f []
by using four or Rve polymer standards in the re-
quired MW range at C"const. Subsequently, the
length of the chromatographic zone is determined
under the same conditions. Viscosity average MW
value can be easily determined according to the
Mark}Kuhn}Houwink equation with the aid of its
coefRcients available from reference books.
Adsorption TLC under critical conditions In ad-
sorption TLC under critical conditions (ATLC), the
changes in enthalpy during polymer interaction with
the sorbent surface are compensated for by increasing
entropy of the macromolecule when it enters the
pore: !
H"T
S. ATLC under critical conditions
is used to analyse heteropolymers: polymers and
oligomers containing functional groups, block
copolymers (AB and ABA types) and graft comb-
like copolymers. Using ATLC it is possible to separate
linear, cyclic and branched structures. In this regime
one of the components of the copolymer undergoes
chromatography under critical conditions and re-
mains invisible } having no effect on separation.
Another component of the copolymer, the character-
istics of which should be determined, takes part in the
chromatographic process according to an adsorption
or a size-exclusion mechanism. Critical conditions
can be implemented either by using an adsorbent, the
pores of which are Rlled with the eluent (analogously
to ETLC) or by eluent migration along a dry sorbent
layer. In the former case critical conditions are in-
dicated by the absence of a separation according to
MW for homopolymers with the same chemical com-
position as that component of the copolymer which
should become invisible. In the latter case the indica-
tion of critical conditions is the absence of separation
(RF values are equal) of homopolymers differing
in MW but containing the same functional groups
(Figure 4).
Precipitation+Extraction Regimes
Precipitation TLC Precipitation TLC (PTLC) was
Rrst suggested and its mechanism investigated by
Kamiyama and Inagaki in 1971. The classical exam-
ples are the separation of PMMA according to MW
using a chloroform}methanol (29 : 71) mixture as the
mobile phase.
In PTLC mixtures of polymer, solvent and adsorp-
tion-active precipitant (a thermodynamically poor
solvent) are used as the mobile phase. Separation
4352 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
Figure 4 ATLC of reaction mixtures, obtained by the reaction of
PSLi with (1}4) methyl acetate, and (5) methyl benzoate; of
tertiary alcohols (6) (C8H17)2}C(OH)}CH3 and (7) (C8H17)2}
C(OH)}C6H5, of (8) polymer ketone PS}CH2}C(C6H5)2}COCH3
and its PS precursor with MW 0.4 kDa and PS standards with MW
(kDa): (9) 0.4 and (10) 40. Reaction mixtures (1}5) contain: PS
precursors with MW (kDa): (1, 5) 0.4, (2) 0.6, (3) 6.4 and (4) 13.5,
polymer ketone PS}CO}R with the same MW as that of PS
precursors, and tertiary polymer alcohol (PS)2}C(OH)}R with
double MW, where R is !CH3 or !C6H5. Conditions are close to
the critical point for PS: adsorbent is silica gel KSKG; mobile
phase is toluene}CCl4 (1 : 4) mixture.
occurs as a result of changes in the dissolution proper-
ties of the mobile phase along the chromatographic
plate, for instance, as a result of partial evaporation
of mobile phase components demixing during
chromatography. The dissolution properties of the
mobile phase can also be changed by delivering it
with a changing composition to the plate. The initial
mobile phase should contain the adsorption-active
component at a concentration completely preventing
polymer adsorption.
During PTLC when the individual polymer species
with different MW pass along the plate they undergo
a continuous series of precipitations and extractions.
The elementary PTLC process is the separation of
a polymer solution into the dilute phase, which is
transported with the solvent Sow, and the concen-
trated gel phase which is precipitated on the surface
of sorbent particles.
PTLC is used to separate homopolymers and ran-
dom copolymers by MW and to separate a block
copolymer from the corresponding homopolymers.
The analysis of oligomers with the aid of PTLC is not
effective. The working range of MW exceeds 10 kDa.
Extraction TLC The action of extraction TLC is
opposite to that of precipitation TLC. Extraction
TLC is based on selective dissolution and desorption
of the polymer in the starting zone region. Desorbed
polymer fractions are displaced to the eluent front.
Eluent migration length is not signiRcant. Polymer
zones move along the plate in size-exclusion mode.
Since the eluent moves along a dry sorbent layer,
separation according to MW does not take place. In
this method a single component mobile phase or
binary solvent mixtures are used most often; gradient
column chromatography is the closest analogue.
Under conditions of extraction TLC the gradient is
replaced by stepwise elution.
Using extraction TLC it is possible to separate
compounds differing essentially in solubility or ad-
sorption activity: isotactic and atactic polystyrene,
isotactic and atactic PMMA, poly-1,4 trans- and 1,2-
butadiene and to separate the S-MMA block
copolymer from PS and PMMA.
Some Problems in Polymer Analysis
Determination of MW and MWD TLC, like other
chromatographic methods, is not an absolute
method. Therefore, in order to determine physical or
chemical heterogeneity, polymer standards with
known characteristics are necessary. Belenkii and
Gankina state that in order to determine the MWD, it
is necessary Rrst to establish the dependence of RF on
MW with the aid of narrow-disperse standards, sec-
ondly to obtain a densitogram of the chromato-
graphic zone, thirdly to establish the dependence of
the recorded signal (I) on polymer mass (P) for differ-
ent RFs and to determine (dP/dI) F, fourthly to deter-
0
mine the distribution of polymer mass in the zone
P(RF), and Rnally to calculate MWD according to the
following equation:


dM
P(M)"P(RF)
dRF RF
If the distribution, P(M) is known, it is also possible
to obtain the expressions for weight-average MW
 2
0 M P(M) dM
MM W"

0 MP(M) dM
number average MW:

0 MP(M) dM
MM n"

0 P(M) dM
and heterogeneity:
Mw
H"
Mn
An important point increasing the precision of
MWD determination is the correction for instrumental
broadening in SEC. This is attained with the aid of
 III / SYNTHETIC POLYMERS / Thin-Layer (Planar) Chromatography
two-dimensional chromatography. In the Rrst direc-
tion the chromatographic zone is separated according
to MW and also undergoes chromatographic spreading.
In the second direction the separation according to MW
may be neglected. The dispersion of chromatographic
spreading is equal to the difference between zone disper-
sion after the second and the Rrst elution in the direction
of the second elution. For correction, the calculated
dispersion of chromatographic spreading is subtracted
from the total dispersion of the polymer zone after the
Rrst elution in the direction of mobile phase migration.
Analysis of Copolymers
As polymer molecules very often contain functional
groups or consist of chains differing in chemical
nature, polymers are characterized by a mixed separ-
ation mechanism. The analysis of copolymers exhibi-
ting chemical and structural heterogeneity requires
more complex elution procedures: for example, step-
wise, two-dimensional, gradient and continuous
TLC. In most cases TLC is combined with column
chromatography or pyrolysis}gas chromatography
(GC), as well as with spectroscopic methods.
In the analysis of block and graft copolymers or
branched homopolymers, the following problems
should be solved:
E the diagnosis of a copolymer or a branched
homopolymer;
E the determination of linear homopolymer present
in the copolymer; and
E the investigation of MW and MWD of copolymers.
The Rrst two problems can be solved by comparing
the chromatographic mobility of the polymer being
analysed with that of the corresponding linear
homopolymers in appropriate solvents using ATLC
or PTLC. An indispensable condition of separation is
the difference in adsorption activity or solubility of
A and B homopolymers. DifRculties can appear if
their properties are similar and also when block
copolymer complexes with one of the homopolymers
are formed.
Additional proofs of copolymer presence can be
obtained by double detection of the chromatographic
spots, for example, by using reagents speciRcally
staining homopolymers of different types and by
spectroscopic methods in situ, or after elution of the
polymer zone from the plate.
To evaluate the MW of block-copolymer compo-
nents, one can use ATLC under critical conditions.
Oligomers
Oligomers are built from the same monomer units
as polymers, but their chain is much shorter
4353
(MW(10 kDa). Polymers in dilute solutions are
characterized by the following types of interactions:
solvent}solvent, solvent}polymer segment, polymer
segment}segment, solvent}surface, and polymer
segment}surface, whereas for oligomer segment}
segment interaction and local entropy effects are
small. Oligomers without end groups are readily sep-
arated into oligomer homologues on polar adsorbents
by adsorption chromatography. It is also easy to
separate according to MW, oligomers with end
groups for which the energies of interaction with
adsorbent for the central (c) and end (e) units are
similar. If the energy of interaction with silica gel is
much lower for the central chain unit than for the end
groups (c(e), separation according to the types of
functionality will take place. Therefore, to separate
according to MW it is necessary to use silica gel
modiRed either chemically (for example, RP18) or
dynamically (for example, the separation of PEO
oligomers in a pyridine}water mixture, 0.1 : 10).
Two-component mobile phases are known to sep-
arate some oligomers according to MW on silica gel:
PS, PI, poly(propylene glycol), PEO and its various
derivatives, oligoacrylates, etc. If the MW of one of
the members of the homologous series on the
chromatogram is known, corresponding standards
are not necessary to calculate MWD.
For oligomers used in the production of synthetic
polymers, the distribution according to the types of
functionality (FTD) is very important. It characterizes
the relative content of macromolecules with different
functionalities in the oligomer. The functionality type
of chemical compounds is determined by the number
and nature of functional group. For macromolecular
compounds the concept functionality just as the
concept molecular weight has a statistical signiR-
cance. In order to characterize distribution width
according to functionality types the values of num-
ber-average ( fM n) and weight-average ( fM w) function-
ality are used:
fM n"ni f i/ni
fM w"nif 2i /nifi
where ni"pi/MWi (is the number of moles of macro-
molecules i with molecular weight MWi, functionality
fi and weight pi.
Oligomers are separated according to the number
and nature of functional groups by adsorption
chromatography under the critical conditions or
under conditions close to critical. An example of this
separation is shown in Figure 4. A scanning den-
sitometer or a videodensitometer can be used for
quantitative FTD determination.
4354 III / TERPENOIDS: LIQUID CHROMATOGRAPHY
Future Developments
TLC of polymers is more complex than that of small
molecules for the following reasons: the size of
sorbent pores and that of macromolecules are similar
and co-operative effects exist which are characteristic
of macromolecules as multicentred formations. For
the same reasons, TLC of polymers is a very useful
technique to investigate chromatographic mecha-
nisms. The open sorbent layer in TLC makes it much
easier to detect and investigate chromatographic arte-
facts than by column chromatography. The versatility
of this method, the absence of restrictions with re-
spect to solvents, and high sensitivity to functional
groups (ATLC) make it possible to analyse effectively
the compositional and structural heterogeneity of
polymers.
Further development of this method involves in-
vestigations of mechanisms of adsorption and
chromatography of macromolecules, more extensive
use of chemically modiRed sorbents, more extensive
application of quantitative methods of analysis and
such promising methods as overpressured-layer
chromatography (OPLC). The investigation of macro-
molecules with complex architecture and of reaction
mixtures is not possible without perfecting complex
methods of investigation. The combination of ATLC
and column chromatography in different regimes and
the combination of TLC with spectroscopic methods,
especially with MALDI-TOF-MS are very promising.
See also: II/Chromatography: Size Exclusion Chro-
matography of Polymers. III/Polymers: Field Flow Frac-
tionation; Supercritical Fluid Extraction.
TERPENOIDS: LIQUID CHROMATOGRAPHY
P. K. Inamdar,* Hoechst Marion Roussel India Limited
Research Centre, Mumbai 400080, India
Sugata Chatterjee, Merck Development Centre
Limited, MIDC, Taloja, India
Copyright ^ 2000 Academic Press
Introduction
Essential oils are secondary metabolites of plant ori-
gin and are complex mixtures of fragrance and
* Retired. Present address: Camlin Limited, Pharmaceutical Div-
ision, Camlin House, J. B. Nagar, Andheri (E), Mumbai 400 059,
India.
Further Reading
Alger MSM (1989) Polymer Science Dictionary. London:
Elsevier Applied Science.
Armstrong DW and Bui KH (1982) Nonaqueous reversed-
phase liquid chromatographic fractionation of polysty-
rene. Analytical Chemistry 54: 706}708.
Bui KH and Armstrong DW (1984) Determination of poly-
mer molecular weight and molecular weight distribution
by reverse phase thin layer chromatography. Journal of
Liquid Chromatography 7: 45}58.
Belenkii BG and Gankina ES (1977) Thin-layer chromato-
graphy of polymers. Journal of Chromatography
(Chromatoraphic Review) 141: 13}90.
Belenkii BG and Vilenchik LZ (1983) Thin-layer chroma-
tography of polymers. Modem Liquid Chromatography
of Macromolecules, pp. 361}411. Amsterdam: Elsevier
Science.
Gankina ES and Belenkii BG (1991) Polymers and
oligomers. In: Sherma J and Fried B (eds) Handbook of
Thin-layer Chromatography. Chromatographic Science
Series, vol. 55, 807}862. New York: Marcel Dekker.
GloK ckner G (1987) Thin layer chromatography. Polymer
Characterization by Liquid Chromatography. Journal of
Chromatography Library, vol. 34, pp. 476}507.
Amsterdam: Elsevier Science.
Inagaki H (1977) Thin layer chromatography. In: Tung LH
(ed.) Fractionation of Synthetic Polymers, pp. 23}29.
New York: Marcel Dekker.
Litvinova LS (1998) Practical aspects of adsorption
chromatography of synthetic polymers. Journal of
Planar Chromatography } Modern TLC 11: 114}118.
Litvinova LS, Belenkii BG and Gankina ES (1991) Quanti-
tative analysis of polymers on the basis of the lengths of
chromatographic zones in adsorptionless TLC. Journal
of Planar Chromatography } Modern TLC 4: 304d308.
Savour substances. They originate in roughly
one-third of known plant families and are isolated
from plant bodies by extraction or distillation
procedures. The composition of the constituents are
dependent on the isolation method used, the part of
the plant body from which they are isolated and
also on their thermal, pH and intrinsic chemical stab-
ility. Most essential oils are currently separated from
nonvolatile materials by steam distillation. As
for other classes of natural products, a fully
satisfactory deRnition of essential oils is difRcult
to put into words. Although they are volatile
plant materials, they do not leave a grease stain
on paper.
 III / TERPENOIDS: LIQUID CHROMATOGRAPHY 4355

Terpenoids in Essential Oils

The majority of the essential oils of commercial inter-
est are mixtures of mono- and sesquiterpenoids, con-
taining only minor amounts of compounds belonging
to other classes. The second largest group of essential
oils consists of aromatic compounds including phen-
olic ones. The terpenoid components occur admixed
with the corresponding terpene hydrocarbons, which
generally contribute less to the aroma component.
Among the popular essential oils are bergamot,
grapefruit, lemon, lime, mandarin, orange, petitgrain
and neroli oils belonging to the citrus class of oils.
These oils contain varying proportions of differ-
ent types of monoterpenoids, e.g. linalyl acetate,
linalool, citrals (geranial and neral), corresponding
acetates, and nootkatone. Lime oil, one of the most
important distilled citrus oils contains 1,4- and 1,8-
cineole, terpinen-4-ol and
-terpineol, along with
aromatic p-cymene, which is probably derived from
limonene by oxidation. Another important class of
Lavandula oils contain linalool and linalyl acetate in
essential oils is cedar oils, which are characterized by
the lavender oil variety whereas lavandin oil contains
the sesquiterpenoids cedrol, cedryl acetate, cedrene,
camphor and 1,8-cineole. Essential oils of the mentha
thujone, thujopsene, trans-
-atlantone and the mono-
variety are characterized by the presence of menthol,
terpenoids pinenes and delta-3-carene along with 2-
menthone and menthyl acetate. Spearmint oil of the
trans-4-cis-decadienyl isovalerate. A major class of
mentha variety contains carvone, dihydrocarveol,
essential oils derived from the plant class Eucalyptus
menthone and limonene. Pulegone is also found in
has about 500 species. Monoterpenoids such as cit-
the mentha variety of oils. Balsam and wood terpen-
ronellal, citronellol, isopulegol, piperitone and -
tine oils contain bornyl acetate along with pinenes,
phellandrene are the important constituents apart
limonene, p-menthadienes, whereas the Sage oils
from geraniol and nerol. See [I]}[VII] for structures of
class of essential oils contain sclareol and
- and
some of these components of essential oils.
-thujones.
Other important essential oils are grass oils, e.g.
There are other varieties of essential oil containing
lemon grass oil containing citral, geraniol, etc. Vetiver
terpenoids, e.g. ambrette seed oil; amyris oil contain-
oil has high sesquiterpene content of
and  vetivone.
ing sesquiterpenoids, e.g. elemol, eudesmols and
-
agarofurans. Buchu leaf oils contain menthone,
isomenthone, pulgeone and bifunctional diosphenols
also known as buchu camphors and some sulfur-
containing terpenoids, e.g. p-menthane-8-thiol-3-one
and its thioacetate. The plethora of essential oils con-
taining mono- and sesquiterpenoids occurring in na-
ture is extremely large and a description of the
principal constituents in them is beyond the scope of
this chapter. Among the oils having importance in the
perfumery industry, rose oil contains damascenone,
cis-rose oxide and verbenone in various species; san-
dalwood oil contains the santalols; rosemary oil con-
tains verbenone; patchouli oil has patchoulol and
norpatchoulenol; orris root oil contains neoiso-
-
irone; jasmin oil contains cis-jasmone and methyl jas-
monate; guaiac wood oil contains guaiol; and costus
root oil contains costol and dehydrocostuslactone.
Among the essential oils used as food Savouring,
ginger oil contains the zingiberenes, ar- curcumene
and -sesquiphellandrol, copaiba (balsam) oil con-
4356 III / TERPENOIDS: LIQUID CHROMATOGRAPHY

tains caryophyllene, celery seed oil contains -

selinene, cardamom oil has 1,8-cineole and
-terpinyl
acetate. Some tea aroma components are presumed to
be derived partly from the ionone series of com-
pounds (probably derived from carotene besides
other terpenoids such as methyl jasmonates, which
give the sweet taste to Oolong tea leaves.
Essential oils also have therapeutic signiRcance,
e.g. in antiseptics and antibacterials, analgesics
(clove, peppermint, lavender, birch oils), anti-fungals
(tea tree, thyme), anti-inSammatories (German
chamomile, lavender), anti-toxics (chamomile oil),
anti-virals (Melissa ofTcinalis, Eucaliptus smithii,
Ravensara aromatica, Niaouli), diuretics (juniper
oil), balancers, deodorants, digestives (anise, cara-
way, peppermint, lavender), spasmolytics (basil,
majoram, German chamomile, cypres, laurel), mu-
colytics and expectorants (Eucalyptus globulus, pine,
anise, thyme). They are also used as insecticides and
insect repellents (citronella, cinnamon, geranium).
See structures [VIII]}[XXIV] for some components of
therapeutic essential oils.
 III / TERPENOIDS: LIQUID CHROMATOGRAPHY 4357

discussed later. Before attempting any liquid

chromatography (LC) or gas chromatography (GC)
analysis, the terpene hydrocarbons are usually re-
moved by distillation because of their considerably
lower odour and aroma values. Furthermore, these
oleRns polymerize and are poorly soluble in lotions.

Analysis of Essential Oils
The economic importance resulting from their fra-
grance and aroma which is used extensively in the
perfume and food Savour industries as well as the
therapeutic applications requires that reliable analyti-
cal methods are available for qualitative and quantit-
ative analysis of the individual ingredients as well as
methods for their preparative separation from the
complex mixtures they exist in. In fact, most varieties
of essential oils are so complicated that resolution
and analysis of any single component of interest is
a formidable task. Although liquid and gas chromato-
graphic separations on analytical and preparative
scales have been used, such techniques are plagued
with the problems of weak UV absorbance, volatility
and thermal instability of many of the ingredients.
Most of the essential oils contain only very minor
amounts of the active principles and large amounts of
undesired components that obscure the separation
and reduce the sensitivity. The important develop-
ments that have taken place in liquid chromato-
graphic analysis of essential oils over the years are
Thin-Layer Chromatography
Thin-layer chromatography (TLC) is an inexpensive
technique and was the earliest LC method used to
provide information on the complexity of essen-
tial oils. TLC studies of thymus essential oil on
SiO2 plates using various developing solvent
systems and detection by spray reagents, e.g.
H2SO4 charring, could identify thymol and carvacrol
as the essential components by comparison with
standard substances. This TLC method has been fur-
ther employed to quantify these components by
scanning densitometry. Additionally, quantiRcation
of geraniol, citral, terpinen-4-ol, cineole and gama-
terpiniol have been carried out on thymus essential
oil. Coscia described various reagents used to quan-
tify mono-, sesqui- and other terpenoids (also poly-
phenols) such as carotenoids, tocopherols, retinoids,
etc. in analysis of essential oils by TLC.
HPLC Analysis of Essential Oils
Almost 90% of the components present in essential
oils that are responsible for fragrance and Savour are
in the boiling range of 150}3003C and possess ideal
vapour pressure to be successfully analysed by gas
chromatography. The molecular weights of these

Figure 1 Typical analytical HPLC separation of citrus essential oil mixture. Analytical HPLC of a citral-type ginger oil. Experimental
conditions: 15;0.46 cm. Column filled with Microsorb 5
m C-18 silica gel, solvent MeCN; H2O"6 : 1 to MeCN: H2O"95 : 5 in
30 min, 1 mL min\1, detection 236 nm. (Courtesy Springer-Verlag GmbH & Co.)
4358 III / TERPENOIDS: LIQUID CHROMATOGRAPHY
Figure 2 General protocol recommended for the identification and quantification of essential oil components. (Courtesy, Springer-
Verlag GmbH and Co.)
compounds are mainly in the range up to 300 amu,
thus making GC-MS an ideal technique for analysis
and identiRcation. However HPLC is being used in-
creasingly to analyse essential oils (Figures 1+4) and
in fact HPLC shows certain signiRcant advantages
over currently used open tubular column GC
methods, such as minimal exposure to air, the avoid-
ance of high temperature degradation, ease of separ-
ation of nonvolatile components and higher sample
recovery. Furthermore, although as mentioned
earlier, a majority of the terpenoids are weakly ab-
sorbing in the ultraviolet (UV) region due to the lack
of a chromophore, the availability of highly sensitive
online ultraviolet-visible (UV/Vis) detectors has ex-
panded the utility of high performance liquid
chromatography (HPLC) applications in essential oil
 III / TERPENOIDS: LIQUID CHROMATOGRAPHY
Figure 3 Isolation procedure of aroma constituents from Oolong tea.
analysis. The problem of UV detection on normal
phase columns using eluents that are strongly UV
absorbing has been largely alleviated by using rever-
sed-phase columns which permit the use of solvents
Figure 4 Isolation procedure of linalyl -vicianoside, bornyl -
primeveroside, and 2-phenylephenylthyl -primeveroside from
Gardenia jasminoides. (Figures 3 and 4 } Reproduced from The
International Congress of Flavours, Fragrances and Essential
Oils.)
4359
like methanol and acetonitrile, and detection by end
absorption, i.e. between 200}220 nm. Although dif-
ferential refractive index (Rl) and polarimetric de-
tectors have been used, they suffer from poor
sensitivity. Polarimetric detection is however claimed
to have better selectivity. Pulsed electrochemical de-
tectors have been used for Savour-active alcohols by
Le Fur. Primary terpenols are detectable at ppb level
with others at ppm level. Primary and secondary
alcohols can be quantiRed with good repeatability
and sensitivity.
Since essential oils are highly complex mixtures;
HPLC is of great assistance for both the separation of
the components into classes and of such a class into
its components. Such pre-fractionation is followed by
subsequent high-resolution GC (HRGC) analysis of
the fractions. Multiple LC separations have been fre-
quently carried out in tandem in order to obtain
adequate pre-fractionation before subjecting each
fraction to GC analysis. Thus the labile sesquiterpene
germacrene B has been isolated by Clark from lime
peel oil as an important Savour impact component.
Various types of HPLC columns have been used, e.g.
normal phase, diol-bonded silica and reversed phase,
e.g. C2, C4, C8 and C18. In one example strawberry
jam was previously gel Rltered through Sephadex
LH-20 to remove fatty acids. The resulting material
was subjected to gradient HPLC using a diol-bonded
silica column and gradient elution using pentane-
diethyl ether resulting in two major fractions, the Rrst
containing hydrocarbons, esters, aldehydes and
ketones and the second fraction consisting of polar
4360 III / TERPENOIDS: LIQUID CHROMATOGRAPHY

compounds such as alcohols, hydroxy esters and lac- derivatives of
-pinene has been achieved using
tones. GC-MS analysis of these fractions identiRed amylose tris(3,5-dimethylphenyl)carbamate as the
150 compounds as compared to only 60 in the ab- chromatographic stationary phase. The major contri-
sence of HPLC pre-fractionation. Computer-control- buting factor to achieve such separation is H-bonding
led online HPLC-HRGC has thus emerged as of the analytes with the carbamated amylose.
a powerful method for essential oil analysis. Munari
used a fully automated HPLC-GC instrument for
HPLC pre-separation of citrus oil into four major
Supercritical Fluid Chromatography
fractions } hydrocarbons, aldehydes, esters and Supercritical Suid chromatography (SFC) has also
alcohols } using gradient elution, and the fractions been used for the separation of terpenoid compo-
were automatically transferred to the GC. HPLC
pre-separation with multiple GC transfer from
a single HPLC injection combined with other identi-
Rcation techniques, e.g. mass spectrometry (MS),
Fourier transform infrared (FT-IR) spectrometry
gives additional advantage of online identiRcation.
The most widely used among such techniques is high
performance liquid chromatography-gas chromatog-
raphy-mass spectrometry (HPLC-GC-MS). This tech-
nique has been used in a fully automated form by
Mondello for the analysis of several essential oils:
bergmot, lemon, clementine, sweet orange, bitter or-
ange, grapefruit and Mexican lime. The information
was more accurate than that obtained by only GC-
MS analysis of the essential oil. Using chiral GC
columns Giovanni determined the enantiomeric dis-
tribution of the monoterpene alcohols in citrus essen-
tial oils. In the same vein it can be envisaged that such
multidimensional techniques as microbore HPLC-
electrospray ionization/MS-MS holds great potential
in essential oil analysis since in addition to improved
column efRciency, the second mass analyser permits
mixture analysis by interpretation of the collision
activated dissociation (CAD) spectrum obtained.

Separation of Enantiomeric
Components
The terpenoids present in essential oils are chiral
molecules occuring in enantiomerically pure form.
HPLC analysis using chiral columns and co-elution
with authentic optically pure compounds furnish in-
formation on the chirality of the constituents. HPLC
separation of enantiomeric mixture of
-pinene, -
pinene, camphene and limonene were carried out by
Moeder on a C4 column with UV detection at
210 nm and a mobile phase containing 5}20 mM of

-cyclodextrin, or 0.01}0.8 mM of -cyclodextrin
(90 : 10 mixture of MeOH and 0.1% phosphoric Figure 5 Enrichment of essential oil fractions by semi-
acid). Equations were derived for the apparent forma- preparative HPLC. HPLC fractionation of an essential oil at a flow
tion constants of the diastereoisomeric complexes of rate of 8 mL min\1. Conditions: column, 24 cm;10 mm i.d.
LiChroprep RP 18 (40 m), mobile phases. (A) methanol}water
the solute with cyclodextrin and their stoichiometry 82.5 : 17.5 (v/v), (B) methanol, flow rate 8 mL min\1, detector, UV
estimated. It was observed that efRcient separation 220 nm. 1"oxygen-containing compounds, 2"monoterpene
was achieved with only
-cyclodextrin for these bi- hydrocarbons, 3"sesquiterpene hydrocarbons. (Courtesy
cyclic terpenes. Chiral discrimination of enantiomeric Kluwer Academic Publishers, The Netherlands.)
 III / TERPENOIDS: LIQUID CHROMATOGRAPHY 4361

nents. The advantage of SFC lies in the fact that it has
features of both GC and LC. Both GC and LC col-
umns can be used and various detectors like Same
ionization detector (FID), UV and MS can be used
making it a nearly universal technique. To illustrate
the application of SFC 2-Z- and 3-E-nerolidols, R/S-
geranyliol,
-bisabolol and 2E,6E-fernesol were sep-
arated in 10 min on a column packed with 5
m
Zorbax Z215 silica equipped with a guard column of
the same material operated at 403C with CO2 con-
taining 0.5% MeOH as the mobile phase and
monitoring the components at 220 nm. In another
example eight terpenoid components of cinnamon oil
were separated on a delta-bond SiO2 column at
1253C using a gradient elution of 100% CO2 to 7%
EtOH in CO2. The detection limits were typically
1.0
g mL\1 for the terpenes with a linear response
over four decades.
Preparative Liquid Chromatography
Isolation of individual essential oils is a formidable
task because of their complexity. Open column
chromatography, droplet countercurrent chromato-
graphy (DCCC), rotation locular countercurrent
chromatography (RLCCC) and gel permeation
chromatography are some of the techniques which
have found considerable use. However each of these
techniques has its weaknesses. Multiple separations
on either normal- or reversed-phase packing ma-
terials (40}70
M) using low to medium pressure
(10}40 bar) accomplishes substantial enrichment of
the components in terms of their functional group
type and/or polarity. Using step or gradient elution
techniques the complex mixtures of terpenoid hydro-
carbons, carbonyl compounds, alcohols, esters, etc.
can be cut into fractions. Online UV detection
and automatic fraction cutting considerably enhances
the utility of such medium pressure liquid chromato-
graphy}low pressure liquid chromatography (MPLC-
LPLC) techniques. The fractions can then be
subjected to preparative HPLC using columns of
inner diameter '10 mm to furnish the semipure
components. Even preparative HPLC may have to
be repeated several times before adequate resolution
is achieved. The pooled fractions are then subjected
to semi-preparative HPLC (column i.d. (10 mm)
followed by peak collection from analytical columns
to yield the pure terpenoids. The level of purity de-
sired depends on the use for which the component
materials are required. Some representative HPLC
traces for the separation of essential oil components
are shown in Figures 5+7. In several cases where the
terpenoids exist as glycosides or some such deriva-

Figure 6 HPLC separation of sesquiterpene hydrocarbons. Conditions: 300;4 mm i.d. Lichroprep Si 60 (7 m) column with 48% water.
Eluent: n-pentane. 1"
-copaene, 2"-elemene, 3"-elemene. 4"-caryophyllene, 5"
-bergamotene. 6"-bisabolene,
7"
-humulene. 8"-cadinene, 9"standard. (Courtesy Kluwer Academic Publishers, The Netherlands.)
4362 III / TERPENOIDS: LIQUID CHROMATOGRAPHY
Figure 7 Semi-preparative HPLC of ginger oil. Experimental conditions: load 4 mL oil dissolved in 4 L EtOH, column 25;1 cm.
Filled with Microsorb 5 u C-18 silica gel, solvent MeCN}H2O"9 : 1. 4 mL min\1, detection UV 215 nm. (Courtesy Springer-Verlag,
GmbH and Co.) E indicates UV absorption at 245 nm.
tives, pre-puriRcation using other forms of
liquid chromatography like hydrophobic interaction
chromatography on HP or XAD types of resins
is done followed by separation of individual
compounds on a reversed-phase HPLC column.
Two such examples are the separation of linalyl
-vicianoside, bornyl -primeveroside and
2-phenylethyl -primeveroside from Gardenia
jasminoides. Separation of cis-linalool 3,7-oxide-6-
O--D-apiofuranosyl--D-glucopyranoside from
oolong tea leaves was achieved using similar
methodology.
Conclusion
In view of the importance of essential oils in the
perfume, food and Savour industries as well as their
therapeutic applications, it is imperative that research
in their analysis and isolation continues. The crux of
the problems associated with such endeavours lies
basically in the fact that the constituents of essential
oils most often occur in very minor amounts. There-
fore the aim of further analytical research has to be
directed to the achievement of the highest possible
sensitivity and resolution. HPLC analysis using micro
columns containing either normal- or reversed-phase
silica as packing material coupled to detectors like the
electrospray ionization mass spectrometer with ion-
trap detectors permitting MSn analysis holds good
potential. Electroanalytical techniques like capillary
electrophoresis can go a long way in both qualitative
and quantitative analysis of essential oil constituents.
Capillary electrophoresis interfaced with MS and/or
a photodiode array detector is likely to solve the
problems of separation complexity of essential oil
components.
See also: II/Chromatography: Liquid Chromatography-
Gas Chromatography. III/Essential Oils: Gas chromatog-
raphy; Thin-Layer (Planar) Chromatography. Medium-
Pressure Liquid Chromatography. Natural Products:
Supercritical Fluid Chromatography.
Further Reading
Bauer K and Garbe D (1993) Common Fragrance and
Flavour Materials. VCH Verlagsgesellschaft, mbh.
Hostettmann M and Hostettmann A (1986) Preparative
Chromatographic Techniques (Applications in Natural
Products Isolation). Berlin: Springer-Verlag.
 III / THERMALLY-COUPLED COLUMNS: DISTILLATION
Linskens HF and Jackson JF (1991) Essential Oils
& Waxes. Berlin: Springer-Verlag.
Recent Trends in Flavour Evaluation of Spices Newer
Trends in Essential Oils and Flavours (1991) New
Delhi, India: Tata MacGraw-Hill Publishing Co.
THERMALLY-COUPLED COLUMNS:
DISTILLATION
R. Smith, Centre for Process Integration, UMIST,
Manchester, UK
Copyright ^ 2000 Academic Press
Introduction
A considerable amount of energy is used in distil-
lation operations. Energy integration has proven
to be successful in reducing energy costs for conven-
tional distillation arrangements. However, the
scope for energy integration of conventional distilla-
tion columns into an overall process is often limited.
Also, practical constraints often prevent integra-
tion of distillation columns with the rest of the
process.
If the column cannot be integrated with the rest of
the process or, if the potential for heat integration is
limited by the heat Sows in the background process,
then we must turn our attention back to the distilla-
tion operation itself and look at unconventional
arrangements.
Figure 1 shows two conventional arrangements for
the separation of a three-component mixture. The
sequence shown in Figure 1A is the so-called direct
sequence, in which the lightest component is taken
overhead in each column. The indirect sequence
shown in Figure 1B takes the heaviest component as
bottom product in each column.
One of the most signiRcant unconventional ar-
rangements involves thermal coupling. Figure 2
shows a number of unconventional arrangements
that use thermal coupling. In thermal coupling part of
the heat transfer necessary for the separation is pro-
vided by direct contact via material Sows. Figure 2A
shows a side-rectiRer arrangement and Figure 2B
a side-stripper arrangement. Arrangements similar to
that in Figure 2B are widely used in petroleum reRn-
ing. The fully thermally coupled arrangement in
Figure 2C (sometimes known as the Petlyuk column)
has been known for over 50 years. Various studies
have shown that thermally coupled arrangements can
4363
Svendsen and Scheffer (1985) Essential Oils and Aromatic
Plants. Dordrecht, The Netherlands: Junk Publishers.
Sweig G and Sherma J (1984) CRC Handbook of Chromato-
graphy, Terpenoids, vol. 1. Boca Raton, Florida: CRC
Press Inc.
save up to 30% of energy costs when compared with
conventional arrangements.
Simple Versus Complex Columns
Consider Rrst the design of distillation systems com-
prising only simple columns. These simple columns
employ:
E one feed split into two products;
E key components which are adjacent in volatility;
E a reboiler and a condenser.
For a three-component mixture in which simple col-
umns are employed, the decision is between the two
sequences illustrated in Figure 1.
Consider the Rrst characteristic of simple columns,
which involves a single feed being split into two
products. As a Rrst option to two simple columns, the
possibilities shown in Figure 3 can be considered, in
which three products are taken from one column. The
designs can be both feasible and cost-effective when
compared with simple arrangements, but only for
certain conditions. If the feed is dominated by the
middle product (typically more than 50% of the feed)
and the heaviest product is present in small quantities
(typically less than 5%) then the arrangement shown
in Figure 3A can be an attractive option. If a pure
middle product is required, then it is usually only
possible if there is a large volatility difference be-
tween components B and C, with the middle product
taken as a vapour to assist the separation. The heavy
product must Rnd its way down the column past the
side-stream. Unless the heavy product has a small
Sow and the middle product a high Sow, a reasonably
pure middle product cannot be achieved.
If the feed is dominated by the middle product
(typically more than 50%) and the lightest product is
present in small quantities (typically less than 5%),
then the arrangement shown in Figure 3B can be an
attractive option. This time the light product must
Rnd its way up the column past the side-stream. If
4364 III / THERMALLY-COUPLED COLUMNS: DISTILLATION

Figure 1 The (A) direct and (B) indirect sequences of simple distillation columns for a three-component separation. (Reproduced with
permission from Triantafyllou and Smith (1992) Transactions of the Institution of Chemical Engineers, Part A 70: 1992.)

a pure middle product is required, then it is usually In summary, single-column side-stream arrange-
only possible if there is a large volatility difference ments can be attractive when the middle product is in
between components A and B, with the middle prod- excess and one of the other components is present in
uct taken as a liquid to assist the separation. only minor quantities. Thus, the side-stream column
 III / THERMALLY-COUPLED COLUMNS: DISTILLATION
Figure 2 Thermally coupled columns. (A) Side-rectifier; (B) side-stripper; (C) fully thermally coupled column. (Reproduced with
permission from Triantafyllou and Smith (1992) Transactions of the Institution of Chemical Engineers, Part A 70, 118.)
Figure 3 Distillation columns with three products. (A) More
than 50% middle component and less than 5% heaviest compo-
nent; (B) more than 50% middle component and less than 5%
lightest component. (From Smith and Linnhoff (1988) Chemical
Engineering Research and Design, 66, 195, reproduced with
permission from the Institution of Chemical Engineers.)
4365
only applies to special feed compositions. More gen-
erally applicable arrangements are possible by relax-
ing the restriction that separations must be between
adjacent key components.
Consider a three-product separation as shown in
Figure 4A in which the lightest and heaviest compo-
nents are chosen to be the key separation in the Rrst
column. In such a case, two further columns are
required to produce pure products. However, note
that the bottoms and overheads of the second and
third columns in Figure 4A are both pure B. Hence
the second and third columns could simply be con-
nected and product B taken as a side-stream, as
shown in Figure 4B. The arrangement in Figure 4B is
known as a prefractionator arrangement. Note that
the Rrst column in Figure 4B, the prefractionator, has
a partial condenser to reduce the overall energy con-
sumption. The prefractionator arrangement in
Figure 4B typically requires 30% less energy than
conventional arrangements for the same separation
duty. The extent of the energy saving depends on the
feed composition and the relative volatility of the
components being separated. The energy saving re-
sults from the fact that the prefractionator arrange-
ment is thermodynamically more efRcient than
a simple arrangement.
To understand why this is the case, consider the
sequence of simple columns shown in Figure 5. In the
direct sequence shown in Figure 5, the composition
of component B in the Rrst column increases below
the feed as the more volatile component A decreases.
However, moving further down the column, the com-
position of B decreases again as the composition of
the less volatile component C increases. Thus,
the composition of B reaches a peak, only to be
remixed.
4366 III / THERMALLY-COUPLED COLUMNS: DISTILLATION

Figure 4 Choosing nonadjacent keys leads to the prefractionator arrangement. (A) Sequence for three product separation using
nonadjacent keys; (B) prefractionator arrangement. (Reproduced with permission from Smith (1995) Chemical Process Design,
McGraw-Hill.)

Similarly, with the Rrst column in the indirect se- the more volatile component A increases. Again,
quence, the composition of B Rrst increases above the composition of B reaches a peak, only to be
the feed and reaches a maximum only to decrease as remixed.

Figure 5 Composition profiles for the middle product in the columns of the direct sequence show remixing effects. (From Triantafyllou
and Smith (1992) Transactions of the Institution of Chemical Engineers, Part A 70, 118, reproduced by permission of the Institution of
Chemical Engineers.)
 III / THERMALLY-COUPLED COLUMNS: DISTILLATION
Figure 6 Composition profiles for the middle product in the prefractionator arrangement show that there are no remixing effects.
(From Triantafyllou and Smith (1992) Transactions of the Institution of Chemical Engineers, Part A 70, 118, reproduced by permission
of the Institution of Chemical Engineers.)
This remixing which occurs in both sequences of
simple distillation columns is a source of inefRciency
in the separation. By contrast, consider the prefrac-
tionator arrangement shown in Figure 6. In the pre-
fractionator, a crude split is performed so that
component B is distributed between the top and bot-
tom of the column. The upper section of the prefrac-
tionator separates AB from C, whilst the lower
section separates BC from A. Thus, both sections
remove only one component from the product of that
column section and this is also true for all sections of
the main column. In this way, the remixing effects
which are a feature of both simple column sequences
are avoided.
In addition, one other feature of the prefrac-
tionator arrangement is important in reducing mixing
effects. Losses occur in distillation operations due to
mismatches between the composition of the column
feed and the composition on the feed tray. Because
the prefractionator distributes B between top and
bottom, this allows greater freedom to match the feed
composition with one of the trays in the column to
reduce mixing losses at the feed tray.
Distillation Using Thermal Coupling
Rather than each column having a reboiler and con-
denser, it is possible to use material Sows to provide
some of the necessary heat transfer by direct-contact
thermal coupling.
4367
First consider thermal coupling of the simple se-
quences from Figure 1. Figure 7A shows a thermally
coupled direct sequence in which the reboiler of the
Rrst column is replaced by thermal coupling. Liquid
from the bottom of the Rrst column is transferred to
the second as before, but now the reboiler of the
second column supplies the vapour required by the
Rrst column. The four column sections marked as 1,
2, 3 and 4 in Figure 7A can be rearranged to form
a side-rectiRer arrangement, as shown in Figure 7B.
Similarly, Figure 8A shows a thermally coupled
indirect sequence in which the condenser of the Rrst
column is replaced by thermal coupling. The four
column sections marked as 1, 2, 3 and 4 in Figure 8A
can again be rearranged, but this time forming a side-
stripper arrangement.
Both the side-rectiRer and side-stripper arrange-
ments have been shown to reduce the energy con-
sumption compared with simple two-column
arrangements. This results from reduced mixing
losses in the Rrst (main) column. As with the Rrst
column of the simple sequence, a peak in composition
occurs with the middle product, but now advantage
of the peak is taken by transferring material to the
side-rectiRer or side-stripper.
Side-stripper arrangements are commonly used in
petroleum reRnery separations. Figure 9A shows
a typical arrangement for distillation of crude oil. The
main column is fed with the pre-heated crude oil feed.
Products are taken from various points from the main
4368 III / THERMALLY-COUPLED COLUMNS: DISTILLATION

Figure 7 Thermal coupling of the direct sequence. (A) Thermally coupled direct sequence; (B) side-rectifier arrangement. (Repro-
duced with permission from Smith (1995) Chemical Process Design, McGraw-Hill.)

column via side-stripper columns. Heat is also re-
moved at various points through the main column via
pumparounds. Pumparounds take liquid from the
column, cool it and return it to the column at a higher
point, effectively acting as intermediate condensers.
Heat to the side-strippers is supplied from either
reboilers or live steam injection. The arrangement
shown in Figure 9A is equivalent to a sequence of
simple columns in the indirect sequence, as shown in
Figure 9B.
Consider now thermal coupling of the prefrac-
tionator arrangement from Figure 10A. Figure 10B
shows the equivalent thermally coupled prefrac-
tionator arrangement, sometimes known as the
Petlyuk column. To make the two arrangements in
Figure 10 equivalent, the thermally coupled prefrac-
tionator requires extra plates to substitute for the
prefractionator condenser and reboiler.
Various studies have shown that the thermally
coupled arrangement in Figure 10B requires typically
30% less energy than a conventional arrangement
using simple columns. The saving depends on the feed
mixture. In most cases the fully thermally coupled
column in Figure 10B requires less energy than the
side-rectiRer and side-stripper arrangements, for the
same separation. The prefractionator arrangement in
Figure 10A and the thermally coupled prefrac-
tionator (Petlyuk column) in Figure 10B are similar in
terms of total heating and cooling duties, but there
are differences in the temperatures at which the heat
is supplied and rejected.
Figure 11 shows the evolution from the prefrac-
tionator in Figure 11A to the thermally coupled pre-
fractionator in Figure 11B. Finally, in Figure 11C,
the thermally coupled prefractionator uses a single
shell with a vertical bafSe dividing the central section

Figure 8 Thermal coupling of the indirect sequence. (A) Thermally coupled indirect sequence; (B) side-stripper arrangement.
(Reproduced with permission from Smith (1995) Chemical Process Design, McGraw-Hill.)
 III / THERMALLY-COUPLED COLUMNS: DISTILLATION 4369

Figure 9 The typical crude oil distillation column decomposes to a sequence of simple columns in the indirect sequence.

of the shell into two parts, known as the dividing wall
column or partition column. The arrangements in
Figure 11 require almost the same energy consump-
tion, which typically is 30% less than a conventional
arrangement. However, in the case of the prefrac-
tionator in Figure 11A, the heat load is supplied at
two points and rejected from two points. In addition,
the dividing wall column in Figure 11C requires
typically 30% less capital cost than a two-column
arrangement of simple columns.
Dividing Wall Columns
Dividing wall columns, as shown in Figure 11C, have
been known for over 50 years and yet it is only
recently that they have been applied in practice. It is

Figure 10 Thermal coupling of the prefractionator arrangement. (A) Prefractionator; (B) thermally coupled prefractionator. (Repro-
duced with permission from Smith (1995) Chemical Process Design, McGraw-Hill.)
4370 III / THERMALLY-COUPLED COLUMNS: DISTILLATION

Figure 11 The thermally coupled prefractionator can be arranged in a single shell. (A) Prefractionator arrangement; (B) thermally
coupled prefractionator (Petlyuk column); (C) dividing wall column. (Reproduced with permission from Smith (1995) Chemical Process
Design, McGraw-Hill.)

true that the basic design is more problematic than
a single conventional column, because there are more
degrees of freedom in the design. However, methods
have been developed to initialize the degrees of free-
dom prior to detailed simulation. Detailed simulation
of the dividing wall column is carried out by model-
ling it as a Petlyuk arrangement, as shown in
Figure 10B. Control of the column has also been
a concern. However, such concern is misplaced, as
the control is straightforward, being effectively the
same as control of a side-stream column. Standard
temperature and composition control conRguration
schemes can be employed. The hardware and column
internals for the dividing wall column are also stan-
dard, despite the presence of the dividing wall. How-
ever, it should be noted that the column performance

Figure 12 Thermal coupling reduces the quantity of energy required but makes temperatures more extreme.
 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
suffers if the dividing wall is not insulated in some
way. This can be done in practice by using two plates
separated by a layer of insulation.
Temperature of Heat Supply
and Rejection
So far the beneRts of thermal coupling have been
discussed in terms of the reduced energy required. Let
us now consider the temperature at which the heat
needs to be supplied and rejected if thermal coupling
is used. It is always preferable to add the heat to the
reboiler at the lowest temperature possible and to
reject heat from the condenser at the highest temper-
ature possible. In the Rrst instance, this allows
cheaper hot and cold utilities. In addition, if heat
integration of the reboiler and condenser is to be
considered, heat integration will also always beneRt
from lower reboiler temperatures and higher conden-
ser temperatures.
Figure 12 compares a conventional and a thermally
coupled arrangement in terms of temperature and
enthalpy. In the conventional arrangement there is
freedom to choose the pressures of the two columns
independently, and thus the temperatures of the two
condensers or the two reboilers can be varied inde-
pendently. In the case of the thermally coupled ar-
rangement no such freedom exists. Although the
thermally coupled arrangement requires a smaller
heat load than the conventional arrangement, more
of the duties are at extreme levels. The smaller duties
work to the beneRt of utility costs and heat integra-
tion but the more extreme levels work against them.
THIN-LAYER CHROMATOGRAPHY ^
VIBRATION SPECTROSCOPY
E. Koglin, Research Center Juelich,
Juelich, Germany
Copyright ^ 2000 Academic Press
Introduction
The utility of vibrational spectroscopy in chemical
structure elucidation of separated thin layer
chromatography (TLC) spots has been recognized
for many years. Although the traditional method
has been infrared spectroscopy (Fourier transform
infrared spectrometry (FTIR)) a number of com-
4371
Summary
Thermally coupled distillation columns offer con-
siderable beneRts in terms of operating costs. Side-
stripper, side-rectiRer and fully thermally coupled ar-
rangements such as the Petlyuk column can save
typically 30% of the energy consumption compared
with sequences of simple columns. The magnitude of
the saving depends on the feed composition and
relative volatility of the components being separated.
The dividing wall column also offers large potential
savings in capital cost. Apart from the use of
side-stripper arrangements in the petroleum reRnery
industry there has been reluctance on the part of
process designers to exploit the full potential of
thermal coupling. Control of thermally coupled ar-
rangements does not present any particularly difRcult
problems.
See also: II / Distillation: Energy Management; Model-
ling and Simulation; Theory of Distillation.
Further Reading
Biegler LT, Grossmann IE and Westerberg AW (1997)
Systematic Methods of Chemical Process Design. New
Jersey: Prentice Hall.
Douglas JM (1988) Conceptual Design of Chemical Pro-
cesses. New York: McGraw Hill.
King CJ (1980) Separation Processes. New York: McGraw
Hill.
Smith R (1995) Chemical Process Design. New York:
McGraw Hill.
peting techniques now exist including normal
Raman scattering (RS), Raman microspectroscopy
(Micro-Raman), Fourier transform Raman (FT-
Raman) and surface-enhanced Raman scattering
(SERS). Since each type of spectra provide essential
vibrational proRle of analytes from the TLC plate, the
different disciplines are natural partners in a general
spectroscopic analysis. All methods involve the vibra-
tional energy of the molecule and thus provide
molecular and structural information about the
separated sample. However, since infrared (IR)
absorption, Raman scattering and SERS have differ-
ent selection rules } what is frequently strong in
4372 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
a Raman spectrum is weak in an IR spectrum and
vice versa. For this reason, a combined IR and
Raman system offers the Sexibility of working
with almost any sample, as well as complete
vibrational information from numerous com-
pounds.
Up to now the combination of TLC separation
and IR spectroscopy has been approached in roughly
two ways. The classical approach for recording
IR spectra is to elute the separated TLC zones
from the layer onto an IR-transparent pellet or
powder (sample transfer TLC-FTIR). Numerous
workers have attempted to record diffuse reSect-
ance Fourier transform infrared (DRIFT) spectra
directly on TLC plates (in situ TLC-FTIR). A
small number of research groups have studied the
applicability of near-infrared spectroscopy (NIR)
in the reSectance mode as an in situ detection tool
in TLC.
Although IR spectroscopy still yields the largest
number of publications in the Reld of TLC vibrational
analysis many serious attempts are now being made
to explore the analytic potential of Raman spectro-
scopy in new and challenging areas of TLC spot
identiRcation. The use of Raman scattering eliminates
moisture and background absorption problems
which may be present in the infrared-based tech-
niques. Therefore, a number of important advantages
to this Raman technique exist: (1) most common
TLC matrices can be used with little interference,
(2) spectra can be taken in situ from wet or dry
plates, (3) the well-known advantages of NIR
excitation (
ex"1064 nm) in FT-Raman (avoidance
of Suorescence and photo-induced sample damage),
(4) the use of Raman microspectroscopy which
allowed unambiguous placement of the laser focus
on the TLC plate with spatial resolution of the order
of 1
m.
One of the signiRcant limitations of the applicat-
ion of Raman spectroscopy in the Reld of TLC
chromatogram spot characterization is the lack
of sensitivity. An important step forward was
made by SERS. Raman scattering intensities from
adsorbed substances on nano-metal particles (Ag, Au,
Cu) are increased by a factor of l05}106 compared
to those of the nonadsorbed compounds at equal
concentration. For TLC this SERS effect is accomp-
lished by spraying chromatograms with colloidal sil-
ver solution or silver coating in a vacuum chamber
(post-chromatographic SERS activation in TLC). By
using SERS microprobe techniques (laser spots down
to 1
m in size) and high performance thinlayer
chromatography (HPTLC) plates it is possible to
achieve low picogram detection limits for HPTLC
spots.
Fourier Transform Infrared
Spectroscopy (FTIR)
Sample Transfer TLC-FTIR
The classic approach and most frequently used
method for recording infrared spectra of substances
separated by TLC or HPTLC involves removal of the
sample zone from the plate, usually followed by ex-
tracting the analytes via a solvent to an IR-transpar-
ent pellet or powder. This technique makes it possible
to measure full IR spectra at a reasonable sensitivity
using conventional FTIR transmission or DRIFT de-
tection. Commercial accessories are available for
a simple and convenient procedure for the transfer of
the analyte spots to cups containing IR-transparent
substrates. The use of this method has been illustrated
by the analysis of dyes, quinones, coal extracts and
biochemical substances. Reasonable IR spectra with
full spectral features can be obtained from 1 to
0.01
g of sample per spot. The main reason for using
transfer in TLC-FTIR is to avoid the strong absorp-
tion bands in the mid-IR (400}4000 cm\1) from the
TLC stationary phase. Over the years, numerous
other transfer methods for the combination of TLC or
HPTLC and FTIR detection have been described in
the literature. The normal method involves removal
of zones (scraping off ), elution of the spot analyte,
deposition on the material transparent to IR and the
measurement by FTIR. However, the removal of
zones from the plate increases the risk of contamina-
tion and can result in further reactions. The
`wick-sticka technique consists of pressing KBr micro
pyramids onto the TLC plate at locations correspond-
ing to the analyte spot. With a development perpen-
dicular to the Rrst development, the analytes are
eluted into the pyramids, which are then dried and
pressed into pellets. The Eulochrom system involves
the elution of analyte zones from silica-gel TLC plates
by means of small amounts of methanol, then solvent
evaporation and pellet preparation. A simple and
convenient procedure in conjunction with TLC sheets
with a liquid-permeable support such as the Empore
TLC sheet is the transfer of separated spots to a pow-
der layer of potassium bromide. After ordinary TLC
development, the sheet is put in a sheet holder cham-
ber and a thin layer of KBr powder is applied on the
upper side of the Empore TLC sheet. After this KBr
coating, zones are eluted from the sheet by means of
a wetted fritted glass unit and thus moved into the
powder layer.
Thermal desorption FTIR analysis can be used to
avoid the interference from TLC stationary phases
with analyte because strong interaction between the
analyte and the stationary phase causes signiRcant
 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
band shift. Therefore, vapour-phase spectrum libra-
ries can be used directly for sample identiRcation.
This method can be applied to thermally stable sub-
stances which can be desorbed from thermally stable
TLC stationary phases. The separated TLC spot is
scraped off and loaded onto the sample pan of a ther-
mogravimetric analyser (TGA/FTIR). Spectra are eas-
ily identiRed for samples present at a level of 10
g
per TLC zone. A detection limit of 0.8
g is found for
analysis of methyl benzoate.
Over the years, research and practical applications
have shown that sample transfer TLC-FTIR tech-
nique (ofSine coupling for TLC and FTIR) can be
effective and useful as a reliable TLC spot identiRca-
tion method.
In situ TLC-FTIR
IR spectral identiRcation of a TLC spot by means of
sample transfer TLC-FTIR methods is usually time
consuming and problematic. Alternatively, IR spec-
troscopic information about TLC-separated mater-
ials can be obtained in situ by means of a variety of
FTIR techniques. Monitoring the TLC zones by direct
DRIFT measurements, transmission spectroscopy,
infrared microspectroscopic detection and photo-
acoustic FTIR (PA-FTIR) enables both qualitative
and quantitative characterization of separated spots.
To acquire useful IR spectra the in situ TLC spot
method requires the background measurement of the
adsorbent free of any sample. The Rnal IR spectrum is
obtained by either ratioing sample and background
measurements or subtracting the background
spectrum from the sample spectrum. This part of
the in situ TLC-FTIR detection method is very
important for the quality of the resulting analyte
IR spectrum. Depending on the nature of the isolated
TLC spots and the goal of the analysis, the choice
of IR measurement can be either DRIFT or transmis-
sion.
Today online coupling of TLC(HPTLC) and
DRIFT spectroscopy can be carried out using com-
mercially available equipment. By combination
a computer-controlled x-y stage with a specially con-
structed DRIFT unit and an FTIR spectrometer it is
possible to obtain direct IR chromatograms of TLC
spots. The chromatograms can be generated both
frequency-dependent as spectral windows of a certain
range and frequency-independent as a Gram-Schmidt
trace. Depending on the infrared absorptivity of the
TLC analyte and the distance run in the chromato-
gram, the limits of identiRcation, the validated
detection limits, and the limits of quantiRcation lie
between 10 ng and 2.5
g.
In general, use of automated multiple development
(AMD) for TLC separation can improve the online
4373
DRIFT identiRcation limit by one-third compared to
conventional TLC separation techniques.
The in situ FTIR detection of spots on a plate by
means of IR transmission measurements requires an
IR transparent support, e.g. silica gel coated on AgCl
plates. A thin adsorbent coating and IR transparency
of the silver chloride support permits enough energy
throughput to acquire analyte species at 0.1}10
g of
material. The detection limit of this technique in
conjunction with programmed multiple development
and special postcoatings can be improved to the
nanogram level. This technique is limited to noncom-
mercially available AgCl speciality plates.
The development of microchannel TLC spot identi-
Rcation with diffuse reSectance infrared microspec-
troscopic detection is also a method for speciRc
practical applications. In this case a zirconia station-
ary phase is used instead of silica or alumina. Zirco-
nia shows signiRcantly higher reSectivity than silica
or alumina resulting in only moderate background
interferences. The detection limit for this specially
prepared plate is about 1}10 ng.
Photoacoustic FTIR has been suggested as the tech-
nique of choice over DRIFT analysis of high IR-ab-
sorbing sample matrices. Photoacoustic spectrometry
(PAS) is based on the phenomenon that light imping-
ing on the solid TLC plate, can produce an acoustic
signal. PAS involves therefore the measurement of
oscillating pressure variations of a conRned inert gas
situated above the TLC plate. In combination with
a PAS cell an FTIR spectrometer can yield photo-
acoustic IR spectra, which can be applied for TLC
zone identiRcation purposes. PA-FTIR does not re-
quire sample preparation and avoids the effects of
light scattering and reSection.
TLC-NIR
Both mid-infrared (mid-IR) and NIR spectroscopy
are important techniques for TLC or HPTLC spot
analysis because of their sensitivity and versatility.
The main difference between mid-IR and NIR is that
bands in the mid-IR are primarily due to molecular
fundamental vibrations, and absorption in the NIR
region, 800}2500 nm, is primarily due to overtone
and combination bands of, O}H, N}H, S}H and
C}H functionalities. In the NIR region absorption is
rather weak and TLC adsorbents such as silica gel
have no strong absorption bands in these NIR re-
gions, so background interferences are very small.
Also, nearly all analytes of interest absorb in the NIR
region.
Direct, in situ diffuse transmission FT-NIR micro-
spectroscopic detection of separated HPTLC spots of
different kinds of phospholipids give typical results
4374 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
from which the usefulness of the TLC-NIR method
can be evaluated. The limit of detection with a nar-
row NIR beam of light (0.4 mm2) is under 1
g,
and the correlation coefRcient of the calibra-
tion curve is about 0.98 for phospholipid amounts
from 1.25 to 10
g. Detection limits of less than 1
g
of selected sugar samples on developed TLC plates
have been demonstrated with the use of NIR detec-
tion with a diffuse-transmittance geometry.
Raman Spectroscopy
TLC/Normal RS
In contrast to FTIR spectroscopy, where we have
been concerned with the absorption of infrared light,
RS depends on the frequency of the laser light scat-
tered by molecules as they undergo rotation and
vibration and in this respect it is similar to infrared
spectroscopy. Since the selection rules are different,
the information obtained from the laser Raman spec-
trum often complements that obtained from FTIR
studies and provides valuable structural information.
The intensity of Raman scattering is directly propor-
tional to the laser excitation intensity and to the
concentration of the TLC sample. This is important
in quantitative studies. Therefore, laser RS can be
considered as a tool for in situ analysis of TLC or
HPTLC spots, since materials such as silica gel ma-
trices give weak Raman spectra and minimal interfer-
ence with the spectra of the adsorbed species on the
TLC plate.
Using an argon-ion laser for visible excitation (488
or 514 nm) and dispersive Raman units, detection
limits for different separated nonresonant substances
are in the
g region. In the past in situ Raman micro-
spectroscopic investigations on different TLC plates
have shown that the detection limit by means of this
Raman method could be improved up to the ng re-
gion. As an example, nonresonant Raman spectro-
scopy of representative explosive samples separated
on silica gel plates have shown that visible conven-
tional macro-sampling RS can be used nondestruc-
tively to detect and identify this substance class down
to a few micrograms. The same investigations by
means of visible Raman microprobe spectroscopy
(scanning laser diameter of 8
m) gave improved de-
tection of these explosive substances (the detection
limit of TNT is 0.5
g).
TLC/Resonance RS
As the laser exciting frequency in the Raman experi-
ment approaches an allowed electronic transition in
the molecule being investigated, those normal modes
that are vibronically active in the electronic transition
exhibit a pronounced enhancement in their Raman
intensities. Most examples of resonance-enhanced RS
involve the enhancement of totally symmetric modes.
The resonant Raman effect can enhance Raman in-
tensities by factors of the order of 105. This means
that corresponding lower concentrations of scattering
molecules on the TLC spot can be used, therefore
improving the detection limit. An example of the
enhanced intensity from the resonance Raman
effect is the investigation of silica gel TLC zones of
metalloporphyrins. Using 514.5 nm excitation and an
optical multichannel analyser, nanogram levels
of the strong absorbing analytes (Ni-uroporphyrin,
Ni-protoporphyrin) in the visible spectral range
have been obtained.
TLC/FT Raman
The past few years have seen pronounced growth in
the use of dispersive and interferometric RS in the
Reld of TLC spot identiRcation, largely attributable to
increased awareness of the techniques potential as
well as the new methodology and hardware. These
developments include, in particular, near infrared
Fourier transform Raman spectroscopy (NIR FT-
Raman). Since the development of NIR lasers, both
Suorescence and photodecomposition problems have
been reduced or avoided in most cases. In particular,
the Nd:YAG lasers emitting at 1064 nm have proven
advantageous due to their long wavelength
, high
power, and stable intensity. Absorption of the TLC
plates from the silica coating and the glass substrate
are avoided and the full Raman spectral range may be
collected without interference. Commercially avail-
able X-Y-Z stages for TLC plates can be placed dir-
ectly in FT-Raman instruments and the measurement
of the TLC spot taken in situ. One further advantage
arises from the use of newly developed HPTLC plates
for Raman spectroscopy (Merck, HPTLC aluminium
sheet Si 60 F254s Raman). The potential and beneRt
of this in situ hyphenation of HPTLC and Raman
spectroscopy in its broad spectral range, no interfer-
ence by the silica matrix, identical spectra with stan-
dard Raman spectra, high sensitivity of detection and
about 10-fold intensity of signal/noise compared to
conventional HPTLC plates.
The feasibility of measuring the Raman spectra of
chlorinated hydrocarbon pesticides on TLC adsor-
bents was initially studied using high concentrations
(about 100
g cm\2) of separated pesticides on nor-
mal silica gel TLC plates. In the case of N-hetero-
cyclic pesticides the detection limit on normal
HPTLC plates could be reduced to low
-gram re-
gion. As an example, Figure 1 shows the FT-Raman
investigations of the MPP(1-methy-4-phenylpyridi-
 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
Figure 1 FT-Raman spectra of the pesticide 1-methy-4-phenyl-
pyridiniumiodine (MPP). (A) 500 ng of MPP spotted on a HPTLC
silica gel 60 plate and (B) as the pure crystalline powder. Bruker:
NIR FT-Raman spectrometer RFS 100;
ex"1064 nm, laser
power 630 mW, 4 cm\1 resolution.
niumiodine) pesticide on a silica gel HPTLC plate and
the corresponding pure material. In comparing the
two spectra, no signiRcant band shifts were observed
but all bands in the spectrum of the HPTLC spot were
broader than in the spectra of the polycrystalline pure
substance.
In some basic experiments, the feasibility was dem-
onstrated of obtaining artifact- and Suorescence-free
spectra by in situ FT-Raman spectroscopy of para-
cetamol, Suorene and rhodamine B on silica gel TLC
plates. For the strong Raman scatterer Suorene, the
detection limit was found to be 500 ng for a 3 mm
diameter TLC spot.
Several pharmaceutical test compounds have also
been investigated and together with the use of
FT-Raman spectral libraries for identiRcation of
TLC spectra and different search algorithms have
been compared.
At present further work is in progress to investigate
factors which could improve the in situ spot analysis
of TLC, HPTLC and Raman-HPTLC plates by means
of FT-Raman spectroscopy.
4375
SERS Spectroscopy
The discovery that Raman vibrational signals from
molecules adsorbed on nanometer scale metal par-
ticle structures are enhanced by 106 to l09 has caused
extraordinary interest and excitement. This Raman
technique, known as SERS offers new possibilities as
a spectroscopic probe in the Reld of TLC separation
science. It was shown that excellent Raman spectra
could be obtained for low nanogram to picogram
amounts of nonresonant and Suorescent substances
on Rlter paper, paper chromatographic supports, and
TLC or HPTLC plates using SERS spectroscopy.
The Raman scattering cross-section of an adsorbed
molecule on nanometal structures can be further in-
creased by utilizing a laser excitation frequency
which is in resonance with an electronic transition
in that molecule. This molecular resonance Raman
scattering and the SERS effect can combine to give
surface-enhanced resonance Raman scattering
(SERRS) so that the limit of detection is further in-
creased. Therefore, by detecting resonant molecules
on TLC (HPTLC) plates in conjunction with the
SERRS effect, very low concentrations have been
achieved in many Relds of research. Another striking
feature of SERRS spectroscopy is that the Suores-
cence of the analyte on TLC plates can be completely
quenched by the presence of a nanometal surface.
Therefore, the SERRS quenching effect generates
a high-quality surface Raman spectrum.
For TLC (HPTLC) this SERS or SERRS effect
is accomplished by spraying chromatograms with
colloidal silver solutions (reduction of AgNO3 with
NaBH4 or citrate). In order to further improve the
sensitivity of the TLC-SERS method experiments
have been carried out in the Reld of well-deRned
vacuum-deposited silver Rlms onto the separated and
developed TLC plates. The TLC plates were mounted
on a holder inside a vacuum chamber where silver is
thermally evaporated onto the plate. The evaporation
rate and the silver thickness are controlled in order to
Rnd out the most intense SERS signals. Such SERS-
activated TLC plates are stable for many weeks and
can therefore be considered as an analytical diskette
for Raman spectroscopy. In addition to these post-
activation methods, two other possibilities for SERS
activation can be identiRed: (1) the simultaneous
activation of the plate with spotting of the sample,
e.g. by dissolving the sample in Ag colloidal solution,
(2) pre-activation of TLC plates via in situ decompo-
sition of silver carboxylates.
In order to reveal the optimal conditions in TLC-
SERS spectroscopy, atomic force microscopy (AFM)
was applied to investigate the surface morphology
of the Ag labelled TLC substance spots (colloid
4376 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
Figure 2 (See Colour Plate 120) AFM photograph of a silver-coated HPTLC silica gel KG60 plate (post-overlayer SERS activation).
NanoScope III; tapping mode.
TLC-SERS and overlayer TLC-SERS). Figure 2
shows an example of the AFM picture of a post-
activated silver-coated HPTLC silica gel plate in the

m scale (post-overlayer SERS activation).
TLC/VIS-SERS
The Rrst report on the combination of planar
chromatography and SERS with colourless (non-
resonance Raman scatterer) analyte spots was the
direct analysis of HPTLC spots of nucleic purine
derivatives in 1987. After separation and drying,
HPTLC plates were sprayed to wetness with colloidal
silver solution by a spray atomizer. The chromato-
gram zones were analysed at room temperature by
a computer-controlled double beam monochromator
and the excitation wavelength was the 514.5 nm line
of an argon ion laser. Limits of detection were esti-
mated to be less than 5 ng/spot. Comparison of these
HPTLC/VIS-SERS spectra with normal Raman
spectra of molecules in aqueous solution reveals
signiRcant differences: (1) the relative band intensities
are changed due to vibration-dependent surface-en-
hanced scattering mechanism, (2) band shifts can oc-
cur (e.g. shift of ring breathing mode of adenine from
724 to 736 cm\1), (3) band broadenings were ob-
served in the HPTLC/VIS-SERS spectra.
In clinical chemistry, immunoassay methods in
conjunction with TLC play an important role in the
routine determination of active substances in body
Suids. For instance it is possible to separate theophyl-
line without difRculty from its positional isomers
and from other xanthine derivatives. In situ
identiRcation of these compounds using online
HPTLC/VIS-SERS can be performed in the ng region.
Examples of TLC/VIS-SERS measurements from
many research groups have been selected to illustrate
the high sensitivity, molecular speciRcity, accuracy,
easy SERS sample preparation, and the signiRcant
manifold application.
The result of all these investigations is that the
TLC/VIS-SERS detection appears to depend on dif-
ferent factors which must be considered before any
normal application. Close approach to, or direct con-
tact with, silver colloids of investigated analytes are
a prerequisite of Raman scattering enhancement. Fur-
thermore the size, shape, dimension and electrical
charge density of the silver colloids are very impor-
tant. Control and optimization of all these para-
meters would increase the feasibility of TLC/
VIS-SERS in situ detection to a maximum number of
chemical compounds separated on TLC(HPTLC)
plates.
TLC/SERRS
The surface-enhanced resonance Raman effect in the
Reld of chromatogram spot identiRcation, Rrst re-
ported in 1984 by Tran (see Further Reading), has led
to the study of a variety of separated dye mol-
ecules with different chromatographic techniques.
Three structurally similar dyes, crystal violet, mala-
chite green and basic fuchsin were chosen in order to
show the potential of SERRS for the direct identiRca-
 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
tion of chromatogram spots separated by means of
paper chromatography. The detailed vibration
spectra allowed identiRcation and the limit of detec-
tion was 2 ng cm\2.
After this work, different types of dyes, TLC
(HPTLC) plates, the role of the supporting matrix
and sol preparation protocols were investigated and
examined for their inSuence on the SERRS signal.
The investigations of all these effects and the con-
struction of a remote sensing Raman spectrometer to
investigate the model compound pararosaniline re-
sulted in the maximum SERRS intensity yielding a de-
tection limit of about 108 femtomol (33 pg) of this
pararosaniline dye.
Another very interesting application is the incor-
poration of SERRS as a detector for a liquid
chromatograhpy (LC)-coupled TLC system. In this
SERRS/LC/TLC system, efSuent from the LC system
was deposited onto the TLC plate and an activated
Ag sol was added to the plate. Optical Rbres carried
the 514.5 laser light to the TLC plate and the scatter-
ing radiation to the Raman spectrometer. The dye
molecule, pararosaniline acetate, was found to give
a linear SERRS signal over the concentration of
1;10\5 to 1;10\7 M range and the limit of detec-
tion was 750 fmol.
The measurements of other highly Suorescent mol-
ecules such as acridine orange, rhodamine, Glu-P2 as
well as N-containing PAHs separated on HPTLC
plates have shown that the sensitivity is so high that
in situ vibrational investigations are possible with low
picogram amounts of material.
Today, it is clear that TLC/SERRS spectroscopy can
be widely applied to obtaining structural information
about very small amounts of coloured substances
(dyes, pigments) separated by TLC or HPTLC.
TLC/Micro-SERS
Following the rule that an optimized Raman scatter-
ing sample is a micro sample Raman microspec-
trometers have been designed for investigation of
small particles in the
m range. Although there is
earlier work in the coupling of a microscope and
a Raman spectrometer, by far the largest amount of
work has been carried out in the past 10 years. The
new generation of Raman microprobe spectrometers
comprise modern monochromators with notch Rlter
systems, charged coupled device (CCD) detectors and
confocal optics. This type of laser confocal Raman
microspectrometer permits the acquisition of Raman
spectra from TLC(HPTLC) spots down to 1
m in
size and allows the unambiguous placement of the
laser focus at the chromatogram with a spatial resolu-
tion of 1
m.
4377
In the Rrst study on HPTLC/Micro-SERS, a Raman
microspectrometer consisting of an argon ion laser,
a microscope, a triple monochromator and a multi-
channel detector was used. With this system it was
possible to acquire VIS-SERS spectra of silica gel
HPTLC spots of DNA bases and dibenzofuran at the
low picogram level. Considering the fact that the laser
focus is about 1
m, the irradiated spot mass is only
a few fermtograms. Combining this microsurface-en-
hanced Raman scattering and HPTLC has also en-
abled in situ analysis of chromatogram spots of
cationic surfactants in amounts down to subnano-
gram levels.
Micro-Raman equipment and the SERRS effect
have been used for selective detection of
structurally similar aminotriphenylmethane dyes
separated by HPTLC. In situ SERRS spectra were
recorded after the application of aqueous Lee-Meisel
hydrosol solution (reduction of silver nitrate with
sodium citrate) to the analyte spot. The limits of
identiRcation of the dyes are of the order of 5 ng
(applied amount).
Recently the optical detection and spectroscopy
of single molecules and single nanoparticles
was achieved with the use of SERRS spectroscopy
for rhodamine 6G adsorbed on selected silver
nanoparticles. This was the Rrst application
of Raman spectroscopy in the Reld of probing
single molecules by means of Raman vibration spec-
troscopy.
As a result of the dramatic improvement of the
sensitivity in the coupling of TLC and SERRS micro-
spectroscopy, it is possible to probe low numbers
of molecules on TLC plates. Figure 3 shows the
high sensitivity of TLC/Micro-SERRS spectro-
scopy by demonstrating the identiRcation of rho-
damine 6G on a HPTLC plate at a concentration as
low as 1
L of 10\10 M solution applied on the plate.
Considering the fact that the 488 nm laser spot is only
1
m in diameter, the number of R6G molecules
in the analysed area are substantial between 400 and
600.
Figure 4 shows the topography of a selected area
from the silver colloidal SERS activated silica gel
(KG60) HPTLC spot of R6G. This AFM picture
clearly shows the Ag nanoparticle adsorbed on the
silica gel particles of the HPTLC plate. Depending on
the colloid aggregation on the substance/silica zone,
the diameter of the Ag particles is in the region be-
tween 10 and 60 nm.
According to the short range sensitivity of SERRS
(about 1 nm from the surface) the spectra obtained
from the HPTLC spot are attributed to molecules
which have a direct contact to the surface of the
colloids.
4378 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
Figure 3 HPTLC/SERRS microprobe analysis of rhodamine
6G. 1
L of 10\10 M R6G applied on the silica gel (KG60) HPTLC
plate. 1
m selected with the focused laser beam. Laser excitation
line, 488 nm; laser power at the spot, 8 mW; integration time, 3 s;
number of reads, 80. Number of molecules in the scattering
volume 400I600 R6G molecules.
Figure 4 (See Colour Plate 121) AFM (tapping mode) surface plot of a silver colloidal SERRS activated silica gel plate (KG60,
Merck). A selected X"Y"500 nm spot area of a 1
L of 10\10 M rhodamine 6G spotted on the plate. The silver hydrosol marker has
a diameter of between 10 and 60 nm.
TLC/FT-SERS
Some years ago, NIR FT-SERS spectroscopy was sug-
gested as an alternative experimental method to
conventional dispersive Raman spectroscopy in TLC/
SERS. Because of the NIR excitation at 1064 nm
and the Suorescence-quenching effect of the SERS
effect, TLC plates containing indicator could be used
without any problem. In the Rrst TLC/FT-SERS ex-
periments, good-quality FT-SERS spectra of a spot
of 40 ng rhodamine B was recorded after spraying
with a silver colloid solution. These preliminary
results have shown that the SERS effect also works on
TLC plate spots in the NIR spectral range. In general,
it was also found that the enhancement factor at
1064 nm excitation was increased by one or two
orders of magnitude as compared with the 514.5 laser
excitation.
Direct analysis of sub-femtogram quantities of
carotenoids on different types of TLC plates has
been made possible by associating FT-Raman
micro spectroscopy with the SERS effect. In this
HPTLC/micro-FT-SERS experiment, detection of
10\5 M crocetin corresponded to a deposited mass
of 1.65;10\8 g, thus in the 8
m laser spot an
analysed mass of 0.02 fg. These postactivation
SERS investigations have also shown that SERS inten-
sity is not dramatically inSuenced from the TLC
layer thickness, particle size distribution, mean
particle size and presence or absence of a Suorescence
indicator.
 III / THIN-LAYER CHROMATOGRAPHY^VIBRATION SPECTROSCOPY
Figure 5 HPTLC/FT-SERS analysis of metamitron on a KG60
plate with a silver-coated layer of 20 nm thickness. 1
L of 10\5 M
metamitron spotted on the KG60 plate. Bruker: NIR FT-Raman
spectrometer RFS 100 S;
ex"1064 nm, laser power 400 mW,
4 cm\1 resolution.
The possibility of TLC/FT-SERS pre-activation
of TLC plates has been demonstrated for heterocyclic
and aromatic species (pyrene, anthracene Suoran-
thene, biphenyle and benzoquinoline). The pyro-
lysis of pre-deposited silver oxalate forms FT-SERS
system based on TLC plate. This thermal decomposi-
tion of silver oxalate represents a new, simple
method of pre-activation in TLC/FT-SERS spectro-
scopy. The disadvantage is that RF values on
these pre-activated TLC plates are strongly inSuenced
by the Ag clusters on the plate. Therefore, in
this case one has new separation conditions on the
TLC plate.
The most useful and effective method for TLC/
FT-SERS spectroscopy is the postchromatographic
silver coating of the TLC plate inside a vacuum evap-
orator (overlayer TLC/FT-SERS). As an example to
demonstrate this new technique, a HPTLC/FT-SERS
spectrum of metamitron (1
L of 10\5 M metamitron
spotted on HPTLC-KG60 plate) which was coated
with a 20 nm thick layer of silver, is shown in
Figure 5. Therefore this overlayer TLC/FT-SERS
spectroscopy is very suitable for the investigation of
separated pesticide spots on TLC plates.
4379
As a consequence of their advantages the combina-
tion of TLC/FT-SERS techniques with silver coating
of the TLC plate inside a vacuum evaporator will
extend the use of surface-enhanced Raman spectro-
scopy to a much greater range of samples than exam-
ined so far.
Conclusion
The examples of the coupling of TLC and vibration
spectroscopy in the separation technique reviewed in
this article have been selected to illustrate the sensitiv-
ity, molecular speciRcity of separated substance
zones, accuracy, ease of sample preparation, and the
many signiRcant applications of FT-IR, Raman analy-
sis and SERS spectroscopy for substances in the ad-
sorbed state on TLC (HPTLC) plates. The sensitivity
and rich spectral information that FT-IR, Raman and
SERS provide have spurred the rapid development of
technology for using vibration spectroscopy as an
analytical detection method of separated TLC spots.
With the extremely high enhancement of the Raman
scattering signals on SERS-activated TLC plates it is
possible to identify in situ separated TLC spots at the
pico- and femtogram level. Further work is in pro-
gress to investigate factors which could improve the
utility of vibrational spectroscopy in chemical struc-
ture elucidation of separated TLC zones.
See Colour Plates 120, 121.
See also: II/Chromatography: Thin-Layer (Planar):
Modes of Development: Conventional. III/Dyes: Thin-
Layer (Planar) Chromatography. Pesticides: Thin-Layer
(Planar) Chromatography. Pigments: Thin-Layer (Planar)
Chromatography.
Further Reading
Garrel LG (1989) Surface-enhanced Raman scattering.
Analytical Chemistry 61: 401A}411A.
Gocan S and Cimpan G (1997) Compound identiRcation
in thin layer chromatography using spectrometric
methods. Reviews of Analytical Chemistry 16: 1.
Koglin E, Kreisig SM and Copitzky T (1998) Adsorption of
organic molecules on metal nanostructures: State of the
art in SERS spectroscopy. Prog Colloid Polym Science
109: 232.
Mustillo DM and Ciurczak EW (1992) The development
and role of near-infrared detection in thin-layer
chromatography. Applied Spectroscopy Reviews 27:
125.
Petty C and Cahoon N (1993) The analysis of thin-layer
chromatography plates by near-infrared FT-Raman.
Spectrochimica Acta 49A: 645}655.
Somsen GW, Morden W and Wilson ID (1995) Planar
chromatography coupled with spectroscopic techniques.
Journal of Chromatography A 703: 613}665.
4380 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY
Somson GW, Riet P, Gooijer C, Velthorst NH and Brink-
man, UAT (1997) Characterization of aminotriphenyl-
methane dyes by TLC coupled with surface-enhanced
resonance Raman spectroscopy. Journal of Planar
Chromatography 10: 10}17.
Soper SA, Ratzlaff KL and Kuwana T (1990) Surface-
enhanced resonance Raman spectroscopy of liquid
chromatographic analytes on thin-layer chromato-
graphic plates. Analytical Chemistry 62: 1438}1444.
TOBACCO VOLATILES:
GAS CHROMATOGRAPHY
W. M. Coleman, R. J. Reynolds Tobacco Company,
Winston-Salem, NC, USA
Copyright ^ 2000 Academic Press
Introduction
The leaves of tobacco plants, Nicotiana tabacum,
have been shown by Tso and Stedman to consist of
a wide array of organic and inorganic compounds.
For general classiRcation, the components of the leaf
can be described as volatile, semivolatile and non-
volatile. The nature of the compounds making up the
components of the leaf range from nonpolar constitu-
ents such as hydrocarbons to very polar compounds
such as carboxylic acids and amines. The structural
identity of these components has been determined in
many laboratories, including those of DeMole and
Lloyd, through the use of classical wet chemical sep-
aration techniques coupled with end determinations
involving, in some cases, further separations by gas
chromatography (GC). In some cases, headspace vol-
atiles have been trapped on Tenax or charcoal then
either thermally or solvent desorbed. Supercritical
carbon dioxide has also been used for extraction. In
some of the earlier work, the separations were per-
formed on packed glass GC columns and somewhat
later on coated glass columns.
With the advent of multidimensional GC (MDGC),
the labour-intensive sample preparation techniques
associated with wet chemical separations of selected
fractions were for the most part eliminated. With the
MDGC approach, a concentrated solution of a to-
bacco essential oil in a volatile organic solvent can be
directly injected on to a fused silica capillary pre-
column, followed by heart-cutting on to another col-
umn. The phase of the pre-column of the MDGC
system essentially serves as a replacement for the wet
chemical separation approach. In this way, as many
Stahlmann S and Kovar KA (1998) Analysis of impurities
by high-performance thin-layer chromatography with
FTIR and UV absorbance detection in situ measure-
ments: chlorodiazepoxide in bulk powder and in tablets.
Journal of Chromatography A 813: 145}152.
Tran CD (1984) Subnanogram Detection of Dyes on Filter
Paper by Surface-Enhanced Raman Scattering Spectro-
scopy. Analytical Chemistry 56: 824}826.
as 80 new compounds were tentatively identiRed in
a Sue-cured tobacco essential oil. Detection and iden-
tiRcation of the volatile components of the essential
oils were possible through the use of mass selective
detectors and matrix isolation Fourier transform infra-
red spectroscopy.
Another approach for the lower molecular weight
volatiles employed automated purge-and-trap (P & T)-
GC with Same ionization (FID) and mass selective
detection (MSD). A relatively nonpolar DB-5 ((5%
phenyl)methylpolysiloxane) column was employed
for the separation of the volatiles. In further
advances on the P & T method, the inSuences of
ionic strength and pH on the yield of volatile mater-
ials have been reported. By employing a cryofocus-
ing approach coupled with a DB-1701 ((14%
cyanopropyl-phenyl)methylpolysiloxane) fused silica
column of intermediate polarity, and an internal
standard, semiquantitative analytical data were
provided on the volatiles from a variety of natural
products.
Through the use of multiple fused silica GC col-
umns of different bonded phases, a universal com-
monality describing the nature of the volatile
materials from a wide range of natural products has
been developed. The types and proRles of volatile low
molecular weight compounds found in the headspace
above tobacco leaves has been shown to bear remark-
able similarities to the proRles from a diverse array of
heat-treated natural products such as coffee, peanuts
and soybeans. The common thread linking all of
the products has been ascribed to the presence of
speciRc reactions between amino acids and other ni-
trogen sources with sugar molecules to yield volatile
components such as pyrazines and aldehydes. For
details of the work outlined in this section, see the
papers by Coleman et al. in the Further Reading
section.
 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY 4381

This article describes the current GC approaches to
the analysis of volatiles from tobacco. Results are
described on approaches involving dynamic head-
space-GC-MSD, as well as automated solid-phase
microextraction (SPME)-GC-MSD.
Experimental
Dynamic Headspace P & T-GC-FID-MSD
Solid samples Representative samples of dry cured
tobacco leaves were Rnely ground in a coffee grinder
prior to analysis. After grinding, the moisture content
of the tobacco leaves was adjusted to 20% by weight
and the samples were allowed to age overnight at
room temperature (&233C). Finely ground mois-
ture-adjusted tobacco leaves (3}5 g) were placed in
a 25 mL sampling tube and afRxed to the sampling
port of a Tekmar LSC 2016/LSC 2000 P & T unit. Six
replicates w ere run to obtain adequate statistical
analyses. Average % relative standard deviation
values for the total FID area counts for the solid
samples were approximately $10%. After reaching
the desired temperature, the contents of the 25 mL
sampling tube were swept with helium, at a known
Sow rate for a speciRed length of time. The volatiles
in the gas stream were efRciently trapped on a Tenax
column (12 in ;1/8 in i.d. SS tube) held within the
Tekmar LSC 2000 unit. After sweeping the sample,
the contents of the Tenax trap were thermally desor-
bed and transferred via a heated, aluminium-
clad, deactivated, 0.53 mm i.d. fused silica transfer
line to the injection port of a Hewlett Packard (HP)
5880 GC set at 2503C. Liquid nitrogen was employed
to cool the GC oven to 03C during the transfer of the
volatiles. Upon completion of the transfer, the GC
oven was temperature-programmed up to 2503C. The
efSuent from the DB-1701 column was split via
an SGE low dead-volume splitter to the FID and MSD
detectors at a ratio of 20 : 80. An HP 5970 Mass
Selective Detector operation at 70 eV in the electron
impact mode served as the MSD. The analyses were
automated by means of basic programming of the HP
5880 GC terminal. The parameters employed in the
analysis of the volatiles from solid tobacco leaves are
listed in Table 1.
The volatile constituents were identiRed with the
aid of GC retention indices as well as results of library
mass spectral searches of Wiley, NBS and internal
mass spectral databases.
Aqueous liquids and suspensions To a 5 mL sparge
tube was added 1 mL of the tobacco liquid extract or
suspension. The sparge tube was afRxed to the
Tekmar LSC 2016. Analysis then proceeded as
described in the preceding section.
Table 1 Instrumental parameters for P & T-GC-MSD-FID
analyses
Sample volatilization temperature 703C
Sample volatilization time 20 min
Trapping material Tenax
Trap desorption temperature 1753C
Trap desorption time 10 min
GC oven initial temperature 03C
GC oven first programme rate 23C min\1
GC oven final value 473C
GC oven second programme rate 103C min\1
GC oven final value 2503C
GC oven final time 10 min
GC column DB-1701
GC column length 30 m
GC column i.d. 0.32 mm
GC column film thickness 1
m
Automated SPME-GC-MSD
Aqueous liquids and suspensions A Varian 8200
CX AutoSampler with SPME II Sample Agitation was
mounted on top of an HP 5890 Series II Plus GC
equipped with a HP 5972 MSD operating either in
the electron impact mode at 70 eV or in the selected
ion monitoring (SIM) mode. This GC was Rtted with
a relatively polar DB-Wax (polyethylene glycol
(PEG)) fused silica column (30 m;0.25 mm i.d.,
0.5
m Rlm thickness, J & W ScientiRc). The MSD
interface and GC injection port temperatures were
2503C. The GC oven was temperature-programmed
from 40 to 1403C at 53C min\1, then to 2203C at
103C min\1 and held there for 4 min. Splitless injec-
tions were made and the split was opened after 1 min.
The Rbre was automatically submerged in the solu-
tion, vibrated for 0.75 min, removed, injected, and
held in the injection port for 30 min, employing para-
meters set via operating software.
SPME Rbres for automated injections were
obtained from Supelco and employed strictly follow-
ing the manufacturers instructions for use and
activation.
Prior to analysis by automated SPME, the aqueous
heat-treated tobacco suspensions were manually
Rltered through a Whatman Autovial equipped with
a 0.45
m polyvinylideneSuoride Rlter and designed
for use with aqueous solutions. Then, to a 1.8 mL vial
equipped with TeSon-lined septum, was added, via
a Rainin EDP Plus Motorized Microliter Pipet, 1.7 mL
of the Rltered solution. Strict attention to consistency
in the addition of 1.7 mL was necessary to obtain
reproducible results. The charged vials were loaded on
the sample carousel and automatically sampled
employing the instrumentation software provided by
Varian and HP. In some cases it was necessary to dilute
the heat-treated suspensions to obtain reproducible
4382 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY
Rbre performance. Fresh samples were used for every
injection.
SIM was used for the quantitative determination of
selected pyrazines in the heat-treated suspensions.
The selected compounds and accompanying selected
ions (m/z) are listed as follows: methylpyrazine, 94,
95, 96; C2 pyrazines, 107, 108, 109, 110; C3 pyra-
zine, 121, 122, 123, 124 and C4 pyrazines, 135, 136,
137 and 150. The C2, C3 and C4 notations are used
to denote a class of pyrazines. For example, C2 pyra-
zines would include all of the dimethylpyrazines as
well as ethylpyrazine. Identical arguments are used
for the C3 and C4 terms.
Discussion of Results
The distinct aromas associated with such materials as
cured tobacco, roasted peanuts, roasted coffee and
baked bread are familiar to most people. The identity
of a large majority of these volatile aromatic compo-
nents has been published recently. Within the tobacco
industry, skilled individuals can easily distinguish be-
tween the types of tobacco employed in American
blended cigarettes. This is not surprising in light of
the unique patterns of the headspace volatiles
found for these tobaccos (Figure 1). In a number of
cases, some of the volatiles found in the headspace
above the tobaccos are common to all of them and the
main difference lies in the relative amounts of these
common volatiles. For example, 2-butanone was
common to three tobacco samples; however, the rela-
tive amounts were signiRcantly different. In 1996
Coleman et al. demonstrated that this quantitative
difference in the amount of speciRc headspace
volatiles is common to a diverse array of natural
products.
Use of the DB-1701 column with intermediate
polarity served very well in providing sufRcient
resolution to assist in identifying the vast majority
of compounds. Some of these compounds are listed
in Table 2 with numbers which correspond to
peak identities throughout the entire manuscript.
Each of the components listed in Table 2 has
been previously identiRed by Tso and Stedman in
tobaccos. Thus, P & T-GC-MSD is a viable approach
for the identiRcation and segregation of tobacco
types.
Roasting of natural products has often been asso-
ciated with the production of pleasant aromas and
tastes. In some cases these pleasant sensory responses
have been attributed to the presence of low concen-
trations of volatile pyrazines and aldehydes. The pres-
ence of these compounds has been linked to the
reaction of nitrogen sources such as amino acids with
carbon sources such as sugars. Toasting of burley
tobacco leaves at approximately 1503C for 5 min
produces an aroma associated with toasted natural
products. Analysis of the headspace above the toasted
burley tobacco reveals the presence of low molecular
weight compounds also found in other heated natural
products (Figure 2).
Increased levels of the Strecker aldehydes, 2-methyl-
propanal, 3-methylbutanal and 2-methylbutanal,
relative to the starting tobacco are leading indicators
of the sugar}nitrogen reactions. The reaction to form
Strecker aldehydes is one of a series of complex reac-
tions, collectively referred to as the Maillard reaction,
occurring simultaneously between nitrogen sources
such as amino acids and sugars during the heat treat-
ment of natural products. When valine, leucine,
isoleucine, phenylglycine and phenylalanine react
with, for example, fructose at elevated temperatures,
the following Strecker aldehydes are produced: 2-
methylpropanal; 3-methylbutanal; 2-methylbutanal;
benzaldehyde; and benzeneacetaldehyde. Thus, the
presence of these compounds in the headspace above
heat-treated natural products serves as an excellent
indicator of amino acid}sugar reactions. In addition,
an increased amount of volatile pyrazines in the head-
space above heated natural products serves as a wit-
ness to the reaction between nitrogen sources and
sugars. Further changes can be noted in the proRle of
the toasted burley relative to the starting material.
Substantial increases in the relative amounts of
solanone, neophytadiene and damascenone, all
known burley constituents, can be found in the
toasted material (Figure 3).
As before, the intermediate polarity of the DB-1701
column served well in providing sufRcient chroma-
tographic resolution of the volatile components.
Cooking natural products to produce edible mate-
rials with positive sensory attributes in some instan-
ces occurs in what can be viewed as an aqueous
suspension. Boiling rice, beans or cooking a roast
under pressure are simple examples. It has been
shown that heat treatment of an aqueous suspension
of burley tobacco, at 1703C for 30 min, produces
some similar volatile materials (Figure 4) as are found
in the headspace, for example, above a heat-treated
peanut suspension. The headspace of both samples is
dominated by the presence of Strecker al-
dehydes, sugar thermal degradation products and
pyrazines. On a relative basis, the volatile materials
detected in the heat-treated suspension were not
detectable in the starting tobacco. The volatile
components in the heat-treated aqueous burley
tobacco suspension were also indicative of the sugar}
nitrogen chemistries known to produce Strecker
aldehydes and volatile pyrazines when heating
tobacco alone.
 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY 4383

Figure 1 Headspace volatiles from dynamic headspace analysis of selected tobaccos. (A) Turkish; (B) flue-cured; (C) burley.

Figure 2 Headspace volatiles from dynamic headspace analysis of (A) toasted burley and (B) burley tobacco.
4384 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY

Table 2 Compounds identified and numbered in selected

figures

Compound number Identification

1 Acetone
2 2-Methylpropanal
3 2-Butanone
4 2,3-Butanedione
5 3-Methylbutanal
6 2-Methylbutanal
7 2,3-Pentanedione
8 Acetic acid
9 Hexanal
10 Methylpyrazine
11 Furfural
12 C2 Pyrazinesa
13 Acetylfuran
14 6-Methyl-5-hepten-2-one
15 C3 Pyrazinesb
16 Benzaldehyde
17 C14 Hydrocarbon
Figure 3 Structures of some compounds present in heat- 18 C15 Hydrocarbon
treated tobaccos. 19 Nicotine
20 Solanone
21
-Damascenone
22 Neophytadiene
Certain amino acids have been postulated in model
systems by Waller and Feather to be directly involved a
Pyrazines such as dimethylpyrazine and ethylpyrazine.
in the production of pyrazines and selected aldehydes b
Pyrazines such as methylethyl pyrazines.
by means of the Strecker degradation mechanism.
That is, upon reaction with selected sugars, certain

Figure 4 Headspace volatiles from dynamic headspace analysis of (A) burley tobacco and (B) a heat-treated burley tobacco
suspension.
 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY
amino acids will yield volatile aldehydes correspond-
ing in large part to the structure of the parent
amino acid backbone. SpeciRcally, for example,
phenylalanine upon reaction with sugars yields as
a volatile aldehyde benzaldehyde, while valine will
yield 2-methylpropanal. Dynamic headspace analysis
of aqueous suspensions of burley tobacco with
mmole quantities of added phenylalanine and valine
show elevated levels of benzaldehyde and 2-methyl-
propanal (Figures 5 and 6) respectively. These results
provide the Rrst evidence implicating the Maillard
and Strecker reactions within a tobacco matrix.
Again, the DB-1701 GC column provides the neces-
sary resolution.
Recently, manual SPME-GC-MSD has been shown
to be an excellent analytical approach for the quanti-
tative determination of volatile components associ-
ated with heat-treated natural products. SpeciRcally,
the quantitative analysis of pyrazines, furfurals,
thiazoles and pyridines has been demonstrated in
aqueous media. With the introduction of an auto-
mated injection system analytical precision has
improved. Application of automated SPME (Auto-
SPME)-GC-MSD to the analysis of a heat-treated
tobacco suspension revealed some of the Rrst insights
Figure 5 Headspace volatiles from dynamic headspace analysis of two tobacco dusts: (A) tobacco dust plus phenylalanine;
(B) tobacco dust.
4385
into a variety of reaction mechanisms occurring dur-
ing the heating process (Figure 7). The origin of py-
ridine, myosmine,
-nicotyrine and pyrrole-2-
carboxaldehyde can probably be directly linked to the
decomposition of the alkaloid, nicotine. The origin of
the pyrazines is no doubt related to sugar}nitrogen
reactions (see above), and the furans can be attributed
to the thermal decomposition of sugars. Searches for
the appropriate SPME Rbre resulted in the selection
of a carboxenpolydimethylsiloxane Rbre. This par-
ticular Rbre is suited to the analysis of relatively polar
compounds in aqueous solutions. The best column
for the analysis was found to be a relatively polar
DB-Wax, which provided the most information con-
cerning the reaction mechanisms.
Further application of AutoSPME-GC-MSD in
the SIM mode has provided additional insights into
the formation of volatile pyrazines during the heat
treatment of aqueous tobacco suspensions. Model
studies indicate the formation of pyrazines through
the coupling of two molecules (Figure 8). Thus,
should the source of the N in the two molecules be
15
N, then the possibility of changes in the m/z values
for pyrazines could be either one or two. For example,
the m/z for methylpyrazine could be changes from
4386 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY

Figure 6 Headspace volatiles from dynamic headspace analysis of tobacco dust suspensions. (A) Tobacco dust plus valine;
(B) tobacco dust.

Figure 7 Total ion chromatogram from AutoSPME of aqueous heat-treated tobacco suspension using a carboxen polydimethyl-
siloxane SPME fibre.
 III / TOBACCO VOLATILES: GAS CHROMATOGRAPHY
Figure 8 Formation of methylpyrazine from two molecules.
a molecular ion of 94 to molecular ions of either 95 or
96. In other words, the distribution of the ions m/z"
94, 95, 96, would substantially change if 15N were
incorporated into the methylpyrazine molecule. Model
studies have also indicated that a possible source of
nitrogen for the formation of pyrazines is the am-
monium ion. Labelling the ammonium ion with
15
N would seem to be a viable approach for determin-
ing one of the possible mechanisms associated with
pyrazine formation in a heat-treated tobacco matrix.
Thus, to an aqueous tobacco dust suspension was
added a mmole quantity of 15NH4OAc. A control
reaction was also performed with no additive. After
heat treatment at 1703C for 30 min in a sealed reac-
tor, AutoSPME-GC-MSD-SIM was performed on the
Figure 9 Selected ion-monitoring mass spectra of methylpyrazine produced in heat-treated tobacco suspensions (A) without
additive; (B) with added 15NH4OAc.
4387
Rltrate (Figure 9), with the focus on methylpyrazine.
Alteration of the m/z pattern observed for the methyl-
pyrazine produced in the control experiment was deR-
nitely evident. The shift in abundance toward the ions
m/z"95 and 96 directly implicated the inclusion of
both one and two 15N atoms into the methylpyrazine
molecule, obviously from the added 15NH4OAc. Thus,
for the Rrst time, by employing AutoSPME-GC-MSD-
SIM, a direct link between the presence of naturally
occurring reagents such as ammonium ions and the
production of volatile aromatic compounds such as
pyrazines has been clearly demonstrated.
Conclusion
GC in combination with such sample preparation
techniques as automated dynamic headspace and
SPME has been demonstrated to be an effective ana-
lytical tool for the qualitative and quantitative exam-
ination of tobacco volatiles. Through the use of an
MS detector in the SIM mode, understanding has
been acquired of the formation of the volatiles on
a molecular basis.
4388 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY
See Colour Plate 122.
See also: II/Chromatography: Gas: Column Techno-
logy; Detectors: Mass Spectrometry; Headspace Gas
Chromatography; Historical Development; Theory of Gas
Chromatography. Extraction: Solid-Phase Extraction;
Solid-Phase Microextraction.
Further Reading
Coleman WM III (1992a) The volatile and semivolatile
components of supercritical Suid and methylene chlor-
ide extracts of selected tobaccos. Journal of Essential Oil
Research 4: 113.
Coleman WM III (1992b) Automated purge-and-trap-
gas chromatography analysis of headspace volatiles
from natural products. Journal of Chromatographic
Science 30: 159.
Coleman WM III (1996) A chromatographic study of the
inSuence of ion concentrations and pH on the yield of
volatile materials from heat-treated natural product ex-
tracts. Journal of Chromatographic Science 34: 1.
Coleman WM III (1997) A study of the behavior of polar
and nonpolar solid-phase microextraction Rbers for the
extraction of Maillard reaction products. Journal of
Chromatographic Science 35: 245.
Coleman WM III and Gordon BM (1994) Advances in
Chromatography, vol. 34, Ch. 2, p. 57. New York:
Marcel Dekker.
Coleman WM III, White JL and Perfetti TA (1994) A hy-
phenated GC-based quantitative analysis of volatile
TOXICOLOGICAL ANALYSIS:
LIQUID CHROMATOGRAPHY
A. P. De Leenheer, W. Lambert and J. Van Bocxlaer,
University of Gent, Gent, Belgium
Copyright ^ 2000 Academic Press
Introduction
With the introduction of high performance liquid
chromatography (HPLC) coupled to detection sys-
tems providing spectral information (e.g. photodiode
array detection and mass spectrometric detection,
HPLC-DAD and LC-MS) the development of HPLC
methods for broad-spectrum drug screening has at-
tracted great interest in forensic laboratories. The
information obtained by these methods is two-fold,
i.e. retention data and spectral information, creating
a powerful identiRcation system. The growing litera-
ture describing HPLC as a broad-spectrum technique
demonstrates its unique and essential position in
toxicological investigations.
materials from natural products. Journal of Chromato-
graphic Science 32: 323.
Coleman WM III, White JL and Perfetti TA (1996) Invest-
igation of a unique commonality from a wide range of
natural materials as viewed from the Maillard reaction
perspective. Journal of Science of Food Agriculture 70:
404.
DeMole E and Berthet D (1972) A chemical study of burley
tobacco Savour (Nicotiana tabacum L.) I. Volatile to
medium-volatile constituents (b.p. 4843/0.001 Torr).
Helvetica Chimica Acta 55: 1866.
Gordon BM, Uhrig MS and Borgerding MF et al. (1988)
Analysis of Sue-cured tobacco essential oil by hyphen-
ated analytical techniques. Journal of Chromatographic
Science 26: 174.
Lloyd RA, Miller CW and Roberts DL et al. (1976) Flue-
cured tobacco Savor I. Essence and essential oil compo-
nents. Tobacco Science 20: 43.
Sakaki K, Niino K, Sakuma H and Sugawara S (1984)
Analysis of the headspace volatiles of tobacco using
an ether trap. Agriculture Biological Chemistry 48:
3121.
Stedman RL (1968) The chemical composition of tobacco
and tobacco smoke. Chemical Reviews 68: 153.
Tso TC (1990) Production, Physiology and Biochemistry of
Tobacco Plant. Beltsville, MD: Ideals.
Waller WR and Feather MS (eds) (1983) The Maillard Reac-
tion in Foods and Nutrition. ACS Symposium Series 215.
Washington, DC: American Chemical Society.
A number of important parameters in toxicological
screening by HPLC include column packing material,
column dimensions, detection, standardization and
peak purity assessment. These topics will be treated
while the applicability will be demonstrated by pre-
sentation of selected examples of general screening
and speciRc detection of a limited number of com-
pounds.
Column Packing Materials
Underivatized Silica
UnmodiRed silica can retain drugs by a weak cation
exchange mechanism and was used for broad-spec-
trum drug screening as early as 1975. The main
problem, however, with the use of underivatized sil-
ica is the substantial variability of this material. Dif-
ferent brands of silica and even different batches
 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY
of the same brand of silica packing material often
result in different retention of selected basic
drugs. As a consequence, chromatographic condi-
tions (same batch of same brand of packing, eluent
composition, temperature control) need to be exactly
deRned and strictly followed before reproducible re-
tention times or retention factors (k) can be obtained
in one laboratory or in different laboratories (the
latter being even more difRcult). The impact of all
these parameters on retention is more substantial in
an adsorption system than under reversed-phase con-
ditions. Due to these difRculties, the use of un-
derivatized silica in the application of adsorption
chromatography to systematic toxicological analysis
(STA) remains rather limited.
Bonded-phase Packing Material
Bonded-phase chromatography, and more especially
reversed-phase chromatography on octyl- or octa-
decylsilica, is by far the most popular liquid chroma-
tographic technique used in STA. In the early 1980s
valuable methods for basic drugs on modiRed silica
began to appear. Also practical solutions to the tail-
ing problem were established and reRnements were
under investigation. Free silanol functions are known
to have a marked inSuence on retention behaviour
of different drugs. These silanol effects can be
reduced by changing the pH of the eluent or by
addition of competing aliphatic bases (amine modi-
Rers) as surface masking agents. Various manufac-
turers have launched specially prepared columns
claimed to be free of silanol effects and providing
more reproducible retention times. This is mainly
achieved by deactivation of the free silanols by vari-
ous endcapping procedures and by elimination of the
trace metals from the silica support.
Alternatively, polymeric stationary phases have
also been introduced. However, although the ability
to run these packing materials at pH values even
higher than 9 permits the analysis of basic drugs as
un-ionized compounds without tailing, these poly-
meric phases have only a limited application in sys-
tematic toxicological analysis.
Recently, again in an effort to eliminate chro-
matographic problems due to residual silanol groups
and to prevent the incorporation of buffer salts
in the eluent, an alumina-based packing material
coated with polybutadiene has also been used for
broad-spectrum drug screening purposes. The in-
herent absence of silanol functions on this packing
material simpliRes the retention mechanism, elimin-
ates the need for addition of amine modiRers and
prevents irreversible adsorption of co-extracted
impurities. In addition, aluminum oxide as well
4389
as the polybutadiene coating are stable in the pH
range of 2 to 12. This polybutadiene coating has
hydrophobic properties comparable to reversed-phase
packing materials. Consequently the same solvent
mixtures can be used as in reversed-phase chromato-
graphy. By incorporation of NaOH in the eluent
(0.0125 mol L\1), basic drugs can be chromato-
graphed without tailing. Of course, this high
pH results in poor retention of phenolic compounds
(e.g. morphine) or carboxylic acids (e.g. benzoylec-
gonine). The latter compounds need to be chromato-
graphed on a second and classical reversed-phase
packing material. This approach of using two
different and complementary packing materials
is certainly not unique in systematic toxicological
analysis.
Chromatographic Conditions
Due to the large differences in polarity of the
compounds encountered in broad-spectrum screening
and in view of simultaneous chromatography of par-
ent drugs and metabolites in nearly all reversed-phase
chromatographic systems, gradient elution is used.
The few systems based on adsorption chromatogra-
phy apply isocratic elution.
In LC-MS the choice of the solvent composition is
limited. The use of nonvolatile mobile phase constitu-
ents (e.g. phosphate buffers) is absolutely pro-
hibited. This limits the practical use of LC-MS by
excluding techniques like ion pair and ion exchange
liquid chromatography.
OfSine sample preparation procedures based
on liquid}liquid extraction or solid-phase extraction
are not really the subject of this article. However,
automation of solid-phase extraction coupled directly
with injection and chromatographic analysis and on-
line enrichment based on the use of two or more high
pressure columns or cartridges (column switching)
are already commercialized for broad-spectrum
screening in toxicological analyses and therefore
worthy of mention.
Detection Systems
Besides retention data, spectral information is essen-
tial for the positive identiRcation of an unknown
substance. Therefore, detection systems not providing
spectral information (e.g. Rxed wavelength UV detec-
tion, electrochemical detection) have found only lim-
ited application in toxicological laboratories such as
for repetitive analysis of a small group of structurally
similar compounds (e.g. epinephrine, norepinephrine
in the case of electrochemical detection).
4390 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY
Photodiode Array Detection
The introduction of diode array and fast-scanning ab-
sorption detectors allowed the acquisition of UV (and
visible) spectral data during the chromatographic pro-
cess. This combination of the discriminatory power of
the chromatographic retention parameters (which is
lower for HPLC than GC) with that of the UV spectral
data increased the overall reliability of an HPLC analy-
sis in the area of toxicology. Standard reference spectra
can be stored in a database tagged with parameters of
retention in order to restrict the search into a window
around each retention parameter.
The major problem in the identiRcation of un-
known compounds by a UV spectral match is the lack
of Rne structure in the UV spectrum of many com-
pounds. The identiRcation of metabolites is also dif-
Rcult because biotransformations do not always
result in a drastic change of the UV spectrum. Several
studies from the area of chemometrics have provided
models for UV spectral matching methods used for
toxicological drug analysis. Peak maxima, calcu-
lation of differences between normalized spectra
and between Rrst-derivative spectra are thereby essen-
tial data for estimations of similarity or dissimilarity.
Ideally, each toxicological laboratory should build
up its own library of UV spectra recorded under
stringent chromatographic conditions. Analysis of
unknown samples and recording unknown UV
spectra should then be performed under exactly the
same chromatographic conditions (column, eluent
composition, gradient, pH, etc.) because at least
the last three of these parameters can affect the
observed absorbance. Another source of library
variability is detector-to-detector variation. Because
different photodiode array detectors use different
numbers of diodes, a UV library development on one
system may not be able to meet the same criteria on
another instrument.
Co-elution of drugs with other drugs or with endo-
genous co-extracted substances remains one of the
major causes of errors in HPLC analysis. Erroneous
conclusions can be drawn if a co-eluting compound
mimics the UV spectrum of a known compound or
when the co-elution of two compounds results in
a spectrum that does not match any library spectrum.
Therefore, before running a library search, peak
purity assessment is essential. This can be done either
manually by comparison of the spectrum at dif-
ferent positions of the emerging peak, or alternatively
some computer programs automatically indicate the
peak purity under each peak in the chromatogram.
The software of the more sophisticated systems even
allows peak deconvolution of two co-eluting com-
pounds, resulting in the speciRc UV spectrum and
quantitative contribution of each compound. Other
software systems claim to be able to determine the
individual drugs from a UV spectrum even if this is
a result of up to six compounds eluting at the same
retention time. However, it should be notiRed that in
the latter case previous information on the probable
co-eluting drugs is essential. This can be obtained
from other chromatographic techniques.
Mass Spectrometric Detection
The combination of liquid chromatography and mass
spectrometry (LC-MS) offers a major improve-
ment regarding drug identiRcation compared with the
above-mentioned HPLC-DAD combination.
The improved resolution and the higher separation
efRciency together with the desire to interface
HPLC with mass spectrometric detection have been
the major driving forces behind the development of
capillary LC. Unlike GC, interfacing problems be-
tween LC and MS are still a challenge for researchers.
Since the early 1970s interfaces have been construc-
ted each applying a different technique to elimin-
ate the chromatographic eluent, which of course
cannot be introduced directly into the high vacuum
region of a mass spectrometer. At least seven major
interfacing techniques exist, i.e. moving belt (MB),
particle beam (PB), direct liquid introduction (DLI),
fast atom bombardment (FAB), thermospray (TS),
electrospray (ES) and atmospheric pressure chemical
ionization (APCI).
It is beyond the scope of this contribution to give an
extensive overview of these different techniques.
However, the respective advantages and/or disadvan-
tages of a number of these techniques, especially in
view of their application to broad-spectrum screen-
ing, will be presented. Both the DLI and the MB
techniques have only historical interest and have vir-
tually disappeared from the area of toxicological
analysis. Particle beam and FAB proved to be valu-
able for speciRc applications such as the detection of
steroids (PB) or the detection of compounds with high
molecular mass (FAB). However, in other applica-
tions sensitivity is often a problem for these two
techniques. Of primary interest for toxicological
analysis are the three remaining techniques: thermo-
spray, electrospray and atmospheric pressure chem-
ical ionization. Because TS is able to handle Sow rates
of conventional HPLC systems (1}2 mL min\1) it be-
came the Rrst popular HPLC-MS interface to be used
in many Relds with a high sensitivity. Because the ion
production in this technique is dependent on the sol-
vent composition, the application of TS with gradient
elution can result in difRculties.
Electrospray operates without heat in the spray
ionization step which makes this technique suitable
 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY 4391

for thermolabile compounds, such as sulfate conju-
gates of drugs.
Both ES and APCI have found a wider use during
the last decade with APCI having an excellent sensi-
tivity especially for hydrophobic compounds. Elec-
trospray has the advantage of being applicable to
a wide range of analyte polarity.
Looking to the number of applications and keeping
in mind that ES and APCI have not yet been exploited
to their full potential, these two techniques together
with TS are the most interesting techniques for tox-
icological analysis. The three techniques are based on
a relatively soft ionization process so the mass spectra
obtained sometimes lack the fragment ions necessary
for conRrmation of the identity of an unknown com-
pound.
Quadrupole mass spectrometers are used most fre-
quently because of their ruggedness, however, ion
trap instruments are becoming more and more com-
mon in STA. Coupling to an ion trap spectrometer is
interesting for a variety of reasons, e.g. economical
aspects, sensitivity and the ability to run MS-MS
experiments. Other techniques, such as collision-
induced dissociation in tandem mass spectrometric
conRgurations are also becoming available.
Reliability of Retention
As already mentioned, the efRciency of a HPLC
system is considerably less than that of a capillary GC
set-up so the risk of co-elution is greater. In addition,
the retention behaviour of a compound (together
with the spectral information, a pivotal criterion for
identiRcation of an unknown substance) is often im-
precise. Batch-to-batch variation and variation of
packing materials between different manufac-
turers result in inconsistent retention data. In addition,
coating of the active sites by irreversible adsorption
and loss of the bonded-phase by ageing of the column
can also contribute to changes in retention. Of course
this problem can be overcome by injection of the
authentic standard directly after the tentative identi-
Rcation of a compound. This procedure, however, is
time consuming and presupposes prior identiRcation
even without a perfect match of the retention time
and the availability of a pure standard. Several studies
have evaluated the use of homologous hydrocarbon
series, multiple drug reference standards and nitroal-
kanes to minimize the effect of irreproducible
HPLC retention data. Although these relative reten-
tion procedures improve the reproducibility, it is dif-
Rcult to obtain linear relative retention scales by using
a homologous series of compounds during gradient
elution, the latter being the most popular technique
for liquid chromatographic toxicological screening
purposes. The use of multiple drug standards instead
of non-drug compounds such as nitroalkanes results in
a more effective correction for retention shifts.
Both principles can also be combined by calculation of
retention indices of compounds from their retention
times by linear interpolation between standard drugs,
whose retention indices have been previously deter-
mined on a nitroalkane scale.
Applications
A number of recently developed broad-spectrum
screening procedures based on HPLC-DAD
are brought together in Table 1. They all use

Table 1 Operating conditions of HPLC-DAD in toxicological broad-spectrum screening

Column Eluent Flow rate Standardization Number of Year

(mL min\1) compounds

Superspher 100 RP18 4
m; Acetonitrile : triethylamine 1 1-Nitroalkanes 383 1994

125;4 mm phosphate; gradient 18 standard drugs
Symmetry C8 5
m; Phosphate buffer (pH 3.8) : 1 to 1.5 None 600 1997
250;4.6 mm acetonitrile; gradient
Supelcosil LC-DP 5
m; Acetonitrile : phosphoric 0.6 None 272 1995
250;4.6 mm acid : triethylamine; isocratic
Lichrospher 100 RP8 5
m; Acetonitrile : phosphoric 0.6 None 280 1995
250;4 mm acid : triethylamine; isocratic
Hypersil C18 5
m; Acetonitrile : phosphate buffer 1 None '300 1997
150;4.6 mm (pH 3.0) : sodium octyl sulfate :
triethylamine; gradient
Aluspher Methanol : water containing 1 None '150 1995
RP-select B 5
m; 0.0125 mol L\1 NaOH; gradient
125;4 mm
Spherisorb Acetonitrile : phosphate buffer 1 p-Methylphenyl- 130 1993
S5 ODS-2 5
m; (pH 3.1); gradient phenylhydantoin
150;3.8 mm
4392 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY
Figure 1 Chromatogram of a real postmortem whole blood sample. Peaks: (1) benzoylecgonine; (2) 2
-methylbenzoylecgonine;
(3) cocaine; (5) 2
-methylcocaine; (6) 3,4-methylenedioxy-N-ethylamphetamine. Levels: (1) 0.1; (2) 0.65; (3) 0.1; (5) 0.65; and
(6) 1.3
g mL\1. (Reproduced with permission from Clauwaert et al., 1997.)
reversed-phase packings on either a silica or an
alumina matrix. All except two procedures apply
gradient elution and standardization of the retention
is rather an exception (two procedures out of seven).
The authors present lists of a large number of drugs
and toxicologically relevant compounds (ranging
from 130 to 600) and state that this is not a limitation
but that other compounds can also be added to these
lists.
It is not possible to give a similar table showing the
operating conditions of LC-MS applied to broad-
spectrum screening in forensic sciences. Screening the
literature rapidly demonstrates that to date LC-MS
has only been applied to selected compounds or
groups of compounds such as steroids, thiourea pesti-
cides, mycotoxins, tricyclic antidepressants and 10
illicit drugs (all by TS), diuretics, non-steroidal anti-
inSammatory drugs, carbamate pesticides (by ES) and
-agonists, carbamate pesticides and alkaloids by
APCI. For detailed information on the operating con-
ditions of these various applications we refer to the
specialized literature in this Reld.
Application
The applicability of HPLC coupled to photodiode
array detection as well as to mass spectrometric de-
tection in the Reld of forensic sciences will be demon-
strated by the analysis of cocaine and some of its
metabolites by both techniques. Cocaine, benzoyl-
ecgonine and cocaethylene have been determined by
HPLC-DAD using 2 mL of blood, serum or urine
under reversed-phase gradient conditions. The quant-
itative limit, deRned as that concentration that can be
determined with an acceptable reproducibility
()6%), is 50 ng mL\1 for benzoylecgonine and
cocaine and 25 ng mL\1 for cocaethylene (using
2 mL of body Suid) (Figure 1).
On the other hand, cocaine, benzoylecgonine, ec-
gonine methylester, ecgonine and norcocaine have
been quantiRed in urine (1 mL) with an LC-MS sys-
tem based on step-gradient elution of a large
(250;7.6 mm) steric exclusion column followed
by atmospheric pressure chemical ionization}mass
spectrometry. The detection limits (signal-to-noise
ratio"3) under selected ion monitoring (SIM) mode
conditions were 320, 200, 200, 20 and 60 ng mL\1
for ecgonine, benzoylecgonine, ecgonine methyl ester,
cocaine and norcocaine, respectively. Unfortunately,
quantitative limits were not reported for this method
(Figure 2).
Both systems used solid-phase extraction for
sample preparation. Due to their non-UV-absorbing
properties ecgonine and ecgonine methyl ester
were not detected in the HPLC-DAD system.
This system was, however, able to detect and to
chromatograph other toxicologically relevant com-
pounds while for the LC-MS system this is not
mentioned.
Conclusion and Perspectives
Besides further optimization of both the liquid
chromatographic and the spectrometric parts of the
described conRgurations, a great challenge for the
future is the automation of those systems. Complete
automation of an analytical procedure including
online sample pretreatment is always advantageous
 III / TOXICOLOGICAL ANALYSIS: LIQUID CHROMATOGRAPHY 4393

Figure 2 Mass chromatograms and mass spectra of extracts obtained from suspected urine. Peaks. (1) ecgonine; (2) benzoylec-
gonine; (3) ecgonine methyl ester; (4) cocaine. Levels: (1) 3.2; (2) 4.8; (3) 3.6; and (4) 0.8
g mL\1, respectively. (Reproduced with
permission from Nishikawa et al., 1994.)

in STA. Because the optimum performance of HPLC-
DAD and more especially of LC-MS are determined
by the simultaneous optimization of a large number
of interrelated parameters (Sow, pressure, temper-
ature, voltage 2), expert systems should be opti-
mized allowing fully automated tuning and control.
The different detection principles as well as the
comparable sensitivity demonstrate the complement-
ary character of both HPLC-DAD and LC-MS. In the
future, both techniques will undoubtedly gain in
interest and will play an essential function in forensic
sciences.
Further Reading
Binder SR (1996) Analysis of drugs of abuse in biological
Suids by liquid chromatography. Advances in Chromato-
graphy 36: 201}271.
Bogusz M and Erkens M (1994) Reversed-phase high-per-
formance liquid chromatographic database of retention
4394 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION
indices and UV spectra of toxicologically relevant
substances and its interlaboratory use. Journal of
Chromatography A 674: 97}126.
Clauwaert KM, Van Bocxlaer JF, Lambert WE and
De Leenheer AP (1997) Liquid chromatographic deter-
mination of cocaine, benzoylecgonine, and cocaethylene
in whole blood and serum samples with diode-array
detection. Journal of Chromatographic Science 35:
321}328.
Gaillard Y and PeH pin G (1997) Use of high-perfor-
mance liquid chromatography with photodiode-array
UV detection for the creation of a 600-compound li-
brary. Application to forensic toxicology. Journal of
Chromatography A 763: 149}163.
Hoja H, Marquet P, Verneuil B, LotR H, PeH nicaut B
and Lacha( tre G (1997) Applications of liquid chro-
matography}mass spectrometry in analytical toxi-
cology: a review. Journal of Analytical Toxicology 21:
116}126.
Koves EM (1995) Use of high-performance liquid
chromatography-diode array detection in forensic
TOXINS: CHROMATOGRAPHY
See III / MARINE TOXINS: CHROMATOGRAPHY; NEUROTOXINS: CHROMATOGRAPHY
TRACE ELEMENTS BY COPRECIPITATION:
EXTRACTION
K. Terada, Kanazawa University, Kanazawa,
Japan
Copyright ^ 2000 Academic Press
In separation by precipitation, contamination with
other elements by coprecipitation is undesirable.
However, since the publication by Bonner and Kahn
of a summary on the separation of carrier-free radio-
active tracers by coprecipitation in 1951, this tech-
nique has found wider application to the separation
and preconcentration of trace elements in various
kinds of samples, such as natural water, treated
wastewater, high purity metals and geological and
biological materials.
In modern textbooks, coprecipitation is recommen-
ded for separation and preconcentration of a single
trace element or a group of trace elements when the
concentration is too low to be directly precipitated or
the amount is too small to be handled. In general,
toxicology. Journal of Chromatography A 692:
103}119.
Lai C-K, Lee T, Au K-M and Chan AY-W (1997) Uniform
solid-phase extraction procedure for toxicological drug
screening in serum and urine by HPLC with photodiode-
array detection. Clinical Chemistry 43: 312}325.
Lambert WE, Meyer E and De Leenheer AP (1995) System-
atic toxicological analysis of basic drugs by gradient
elution on an alumina-based HPLC packing material
under alkaline conditions. Journal of Analytical Toxi-
cology 19: 73}78.
Nishikawa M, Nakajima K, Tatsuno M, Kasuya F, Igarashi
K, Fukui M and Tsuchihashi H (1994) The analysis of
cocaine and its metabolites by liquid chromatogra-
phy/atmospheric pressure chemical ionization}mass spec-
trometry (LC/APCI-MS). Forensic Science International
66: 149}158.
Sato K, Kumazawa T and Katsumata Y (1994) On-line
high-performance liquid chromatography}fast atom
bombardment mass spectrometry in forensic analysis.
Journal of Chromatography A 674: 127}145.
coprecipitation of trace elements is carried out
with inorganic and organic precipitants attaining
high degrees of concentration, so that subsequent
determinations can be performed by using the
precipitate itself.
Mechanism
Depending on the nature of the solid phase produced
in a solution and the experimental conditions,
coprecipitation occurs by different mechanisms.
Although the various types of coprecipitation cannot
be distinguished clearly, they may be classiRed
according to the following mechanisms: (i) the
formation of mixed crystals and mixed chemical com-
pounds, (ii) surface adsorption and the occlusion
and (iii) mechanical inclusion of trace components
into the other compounds during crystal format-
ion. However, these processes often proceed
concurrently, making the precipitation process quite
complicated.
 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION
Isomorphous Mixed-Crystal Formation
The processes of coprecipitation by isomorphous
mixed-crystal formation have been well studied, and
the distribution of trace elements is found to be
governed by either the Berthelot}Nernst or the
Doerner}Hoskins law.
Berthelot}Nernst distribution law (homogeneous dis-
tribution). If digestion is continued throughout the
precipitation process, equilibrium will be established
between the trace elements in the interior of the
crystals and the solution, resulting in homogeneous
distribution of the trace element in the precipitate.
Then, the following equation applies:
(Trace element)ppt (Trace element)soln
"D
(Carrier)ppt (Carrier)soln
The higher the value of D, the higher the enrichment
of trace elements.
Doerner}Hoskins law (logarithmic distribution)
When the ions cannot reach the interior of the crys-
tals, equilibrium will be established between the trace
elements in the solution and those on an extremely
thin surface layer of the crystals. This results in logar-
ithmic distribution of the impurities, and the follow-
ing equation is applicable:
To Co
log "
) log
Ts Cs
where T and C represent trace element and carrier,
subscript o and s denote the concentration in the solu-
tion before and after the precipitation, respectively.
In practice, coprecipitation of the trace elements
may occur between the above two limiting distribu-
tion laws.
Surface Adsorption
The surface of a precipitate is particularly reactive.
Ions at the surface of a crystal are incompletely coor-
dinated and hence are free to attract other ions of
opposite charge from the solution. The surface ad-
sorption of ions on ionic precipitates has been de-
scribed by the Paneth}Fajans}Hahn rule which
demonstrates that adsorption generally increases with
the growing surface area of the crystal and with the
decrease in solubility of the compounds of the trace
elements which the elements form with oppositely
charged ions of the crystal. However, there are many
exceptions to this rule. For example, in spite of the
low solubility of PbCl2 or PbI2, they do not coprecipi-
tate with HgCl2 or HgI2.
4395
Two other factors, the dissociation constant of the
adsorbed compounds and the deformation ability of
the ions, are important for adsorption. The smaller
the dissociation of the adsorbed electrolyte, the larger
the adsorptivity. The adsorptivity also increases with
increased deformability of the adsorbed electrolyte.
The deformability usually increases with the size of
the ion.
Another mechanism of adsorption presented by
Pauling is an ion exchange process. When the radius
of an ion in the solution is similar to an ion on the
surface of the crystal, they are exchangeable with
each other. This is more effective when an ion in the
solution forms slightly soluble compounds with an
ion of opposite charge in the crystals. Thus, lead (II)
ions can be adsorbed on the surface of a barium
sulfate precipitate even in the absence of excess sul-
fate ion in the solution, according to the following
exchange reactions:
BaSO4(surface)#Pb2#PPbSO4(surface)#Ba2#
Occlusion and Mechanical Inclusion
When an ion adsorbed on a crystal surface from
the solution is trapped by subsequent crystal layers,
the ion will be occluded in the interior of the
precipitate. This situation can be prevented with col-
loidal precipitates rather than with crystal ones, espe-
cially in a rapid precipitation process. For example,
freshly precipitated hydroxides or sulRdes contain
a certain amount of impurities, most of which
are released upon ageing of the precipitates. Thus,
coprecipitation by occlusion generally gives a
poorly reporducible yield of the trace elements to be
coprecipitated.
Coprecipitation with Inorganic
Precipitants
Coprecipitation of trace elements with inorganic
precipitants is usually carried out using colloidal
precipitates with a large surface area such as metal
hydroxides and sulRdes. Among various hydrated
oxides, coprecipitation with those of iron(III) and
manganese(IV) have been commonly used and are the
most studied, but many other hydroxides, such as
Al(OH)3, Be(OH)2, La(OH)3, Th(OH)4 and Zr(OH)4,
and mixtures of metal hydroxides, such as Fe(OH)3
and Ti(OH)4, have also been employed.
Coprecipitation techniques are commonly used to
separate and concentrate trace elements from very
dilute solutions, such as natural water. Since the
solubilities of the metal hydroxides or sulRdes are
mainly governed by the pH value of the solution,
4396 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION

Figure 1 Relation between co-precipitation recoveries of metals with iron(III) hydroxide and pH of the solution.

control of pH is essential for an effective coprecipita- The second postulated mechanism involves the
tion of trace metals. Figure 1 shows coprecipitation chemical sorption of the trace metal ion Mm# on the
yields of some metal ions with iron(III) hydroxide. It surface of the hydrated metal oxide, followed by
can be seen that removal of many metal ions from the adsorption of hydroxyl ions:
a solution may be possible at pH 9}10. However, it
should be noted that the coprecipitation yield is S#Mm#"SMm#
also affected by the amounts of precipitants used,
the coexisting salts and the ageing time of precipita- SMm##OH\"SMOH(m\1)#
tion. Since most metals form sparingly soluble
hydroxides, coprecipitation by hydrated metal SMOH(m\1)##OH\"SM(OH)(m
2 \
2)#
oxides is usually of low selectivity, so that different
trace metals are likely to be coprecipitated simulta-
The third possible mechanism requires the adsorp-
neously.
tion of hydrolytic complexes of the trace metal ion,
Three possible mechanisms relating to the adsorp-
rather than the metal ion itself, on the surface of the
tion of the trace metal ion on the hydrated metal
hydrated metal oxide:
oxide surface prior to coprecipitation have been
suggested.
n \
Mm##nH2O"M(OH)(m n)#
#nH#
The Rrst involves ion exchange between adsorbed
hydrogen on the hydrated oxide surface and the trace
n \
S#M(OH)(m n \
"SM(OH)(m
n)# n)#
metal ion M in solution according to the equation:

kn Table 1 shows examples of preconcentration of

nSH#Mm# & SnM(m\n)##nH#
where n is the number of molecules of the hydrated
metal oxide and the surface area per molecule is S.
M represents the metal cation of charge m. Then the
distribution coefRcient D is given by:
[M]surface (mol kg\1) KnSHn
D" " which are sparingly soluble in aqueous solution and
[M]solution (mol dm\3) (H#)n
By taking logarithms the equation becomes:
log10D"npH#log10(Kn)#nlog10(SH).
trace elements by coprecipitation with inorganic
precipitants.
Coprecipitation with Organic
Collectors
Organic collectors are mainly complexing agents
form complex compounds with the desired metal
ions. The mechanisms of coprecipitation of trace ele-
ments with organic collectors have been described by
Minczewski et al. According to them:
 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION 4397

Table 1 Coprecipitation of trace elements with inorganic precipitants

Precipitants Trace elements Sample, experimental conditions, comments

Hydroxides
Al(III) Cr(III), (VI), Mo, W Natural waters. Quantitative precipitation: Cr(III) (pH 5}9), Mo (pH 4}5), W (pH
6}8), Cr (VI) (pH 5}6, but it is not quantitative)
Ce, Eu, La, Lu, Nd, Sm, Hot spring and crater waters. 10 mg Al to 2 dm3 sample. Al precipitation is carried
Tb, Th, Tm, Yb, U out at near boiling temperature with 14% NH3 solution to reach pH 6.5}7.5
Th, U Hot spring and crater lake waters. Al2(SO4)3 (25 mg) is added to 1 dm3 sample.
pH is adjusted to 7}8 with 14% NH3 solution
Li Geothermal waters: pH should be high ('12.5}13) and a long mixing time is
required. The recovery yield is increased by removal of Ca ions and polymerized
silica
Be(II) As High purity iron steel. Sample (1 g) is digested with HNO3. 5 cm3 BeSO4
(1 mg Be/cm3) is added in the presence of EDTA (mask matrix elements). As
coprecipitates as BeNH4AsO4 with Be(OH)2. Perfect recovery is obtained be-
tween 1.0 and 3.5 mol L\1 HCI with relative standard deviation (RSD) of c. 13%
for 1.0
g g\1 As. Detection limit is 0.3
g g\1 of solid sample
P High purity iron steel. Same procedure as As. Coprecipitation recovery is 98.7%.
Arsenic is removed from the solution as AsBr3 for Mo-blue spectrophoto-
metry of P
Bi(III)#In(III) Co, Cu, Fe, Mg, Ni High purity Ti metal. 0.5 g sample is dissolved in 6 mol L\1 HCI (20 cm3)#HF
(0.5 cm3). Bi(III) and In(III) are added (10 mg and 20 mg, respectively). After
addition of 4 cm3 of H2O2, pH is adjusted to 9.5 with 7.5 mol L\1 NaOH (pH'11
for Mg). Ageing time 30}60 min. Perfect recoveries are obtained for all metals.
RSD: 0.22%. 0.28%, 2.8%, 0.05% and 0.84%. Detection limit: 0.10, 0.5, 1.8,
0.08, 0.36
g g\1 for Co, Cu, Fe, Mg, Ni, respectively
Fe(III) Cd, Co, Eu Seawater. Coprecipitation yield in the radionuclide levels: Cd (85% at pH 9.0, Fe
35 mg dm\3), Co (95% at pH 9.0, Fe 35 mg dm\3), Eu (c. 100% at pH 9.0 Fe
10 mg dm\3, at pH 6.0}9.0, Fe 35 mg dm\3)
Mo Seawater. To 500 cm3 sample, 9 mol L\1 H2SO4 (1.0 cm3) and 0.1 mol L\1 FeCl3
(3.0 cm3) are added. c. 96.5% coprecipitation yield of Mo is obtained at pH 4.0; it
decreases with increasing pH value
Cr(III), (VI) Urine. Cr(III) is found to be precipitated at pH 10, while Cr(VI) remains in the
solution. Cr(VI) is only coprecipitated at 4}7
Cr(III) Seawater. 4 cm3 of 2 mol L\1 HCl, 4 mg Fe(III) are added to 2 dm3 sample, heat
to 50}603C, 60 cm3 borate buffer (19.07 g borate#4 g NaOH in 1 dm3) is added
to solution to pH c. 7.5. Recovery for Cr '99% at concentration 0.4
g dm\3.
Precision is $0.02
g dm\3
As, Cr, Ge, P, Sb, Se, Water sample. Optimum pH ranges are 5}7 for Sb and Se, 5}8 for As and W,
Te, W 5}10 for Cr, Ge and Te, 6}7 for P. Preconcentration factor is 50 for all except Se,
where it is only 5
Cu, Mn, Ni, Pb, Zn Natural waters. 2 mg Fe(III) is added to 200 cm3 sample; pH adjusted to
9 (NaOH). Detection limit is c. 1
g dm\3. ICP-AES is employed
Se Seawater, silicates, marine organisms. 20 mg Fe(III) is added to 5 dm3 seawater,
pH is adjusted to 5}6. After 2 h standing, another 20 mg Fe(III) is added to
solution, pH to 4}6 with aq. NH3. RSD is 6.0% for 0.5
g Se dm\3
V Seawater, natural water, biological materials, sediments, rocks. 15 cm3
1.0 mol L\1 HCl and 30 mg Fe(III) are added to 3 dm3 seawater; pH is adjusted to
5}6 with aq. NH3. Precipitate is dissolved in 10}20 cm3 2 mol L\1 HCl. Coeffi-
cients of variation are 2.8% for seawater, 1.3% for silicate rocks, 2.5% for marine
plants. Quantitative recovery is attained for 1.8
g V dm\3
Ag, Cd, Ce, Cr, Cs, Er, Low level waste solution. Effect of pH was studied. Sorption of Cs# and Rb# is
Eu, Gd, La, Mn, Rb, not strongly pH-dependent, but coprecipitation is low (20%)
Sr, Yb
Zn Quantitative recoveries are obtained for Ag (pH'8), Cd, Mn, Zn (pH&10),
Cr(III) (pH 9}10), Ce, Er, Eu, Gd, La, Yb (pH&10). Sr (pH 11}11.5, 65%).
Freshly precipitated Fe(OH)3 can be used for the decontamination of radio-
nuclides
Te Hair. Sample (2}4 g) is digested with a mixture of HCl#HClO4, heated to
evolving fumes, then boiled with 20% HCl to reduce Te to Te(IV). 5 mg Fe(III) is
added, pH is adjusted to 9, and centrifuged. Precipitate is dissolved with 3.3 cm3
conc. HCl, then diluted to 10 cm3. Recovery is 96.2$2.4% for 0.2
g Te

(Continued)
4398 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION

Table 1 Continued

Precipitants Trace elements Sample, experimental conditions, comments

Hydroxides
Ga(III) Al, As, Cd, Co, Cr, Seawater. 5 mg Ga is slowly added to 1 dm3 seawater. pH is adjusted to 9.0
Cu, Fe, La, Mn, Ni, Pb, (NaOH), ageing for 24 h. Precipitate is washed with H2O (remove Na, Mg, K, Ca),
Ti, Y, Zn dissolved with 2.5 cm3 1 mol L\1 HCI, diluted to 5 cm3. 200-fold concentration is
achieved. Quantitative recoveries are attained at pH 9.0 for Al, Co, Cr, Cu, Fe, La,
Mn, Ni, Ti, Y, Zn. As(III), Cd, Pb ((90%, pH 10)
In(III) Cu, Fe, Ni Ti alloy. Sample (0.2 g) is dissolved in HCl, 3 cm3 H2O2 is added to mask Ti.
10 mg In(III) is added, pH is adjusted to 9.0 (NaOH). 100% recoveries at pH
8.5}9.0. Detection limit: Cu 0.8, Fe 8.1, Ni 1.4
g g\1 alloy
Cd, Co, Cr, Cu, Fe, Natural waters. 10 mg In(III) to 100 cm3 sample. pH to 9.5. Precipitate is separ-
Mn, Ni, Pb ated by centrifuge, dissolved in 1 cm3 2.5 mol L\1 HBr. All are quantitatively
coprecipitated. Determination limits for GF-AAS are Cd 0.003, Cu 0.02
g dm\3
La(III) As, Bi, Sb, Se, Te Mo metal. 1 g Mo is dissolved in a mixture of 2 cm3 conc. HNO3#6 cm3 conc.
HCl. 50 mg La and small amount filter paper pulp to solution. pH is adjusted to
10.0 (NaOH). After filtration, precipitate is dissolved four times with 2 cm3, boiling
6 mol L\1 HCl. Coprecipitation and dissolution are repeated. Recoveries at pH
10.0, As 96.8, Bi 112.0, Sb 92.4, Se 100.1, Te 106.0% for each 5
g
Co, Fe, Mn, Ni, Zn Seawater. 5 mg La to 1 dm3 sample, pH adjusted to 9.8 with 1 mol L\1 Na2CO3.
Precipitation at 803C for 30 min and later aged for several hours
Mn(IV) Bi, Pb, Sb, Sn Ni matrix. Ni is dissolved with 25 cm3 (1#1) HNO3, pH adjusted to 2}3 with
(1#1) aq. NH3. Volume is made up to 200 cm3 by adding 0.008 mol L\1 HNO3.
MnO2 is formed from MnSO4#KMnO4 in the presence of acids with 2-min boiling
followed by standing for 30 min. Precipitate is treated with 50 cm3 10 mol L\1 HCl.
100% coprecipitation for Bi, Sb, Sn in 6.5 mg Mn precipitated, while Pb in
30 mg Mn, each in 1}100
g dm\3
Mo Seawater, silicates, biological materials. 1.0 dm3 water is pH adjusted to 2 with
dil. HCl. Each 2 cm3 ethanol and 1 mol L\1 KMnO4 is added and stands overnight.
Precipitate is dissolved in saturated aqueous solution of SO2. Quantitative cop-
recipitation is in the range of pH 1.3}5.5 for Mo level(5
g dm\3 seawater
Ga Aqueous solution. Coprecipitation conditions are studied. To 200 cm3 solutions
containing Ga (0.5}1000
g) 5 cm3 5% MnSO4 is added and adjusted pH to
1.5}1.6 with HCl or H2SO4, to boiling and slowly added 2.5 cm3 1.25 mol L\1
KMnO4, then stands for 5}7 min. Precipitate is dissolved in 10 cm3 5% HCl
containing 8}10 drops 3% H2O2. Ga is perfectly recovered
As Cu, Fe, Ni metals. Sample (0.1}2.0 g) is dissolved in 20}30 cm3 HNO3(1#1). pH
to 1.0}2.0 with aq. NH3 (except for Fe). As is coprecipitated by adding 60 mg
KMnO4#10 mg Mn(II). Precipitate is dissoved in mixture (HNO3#H2O2). Detec-
tion limit: 20 ng cm\3
Te(IV) Se High salt waste water. After boiling 100 cm3 of sample (1 mol L\1 HCl) for 30 min
in the presence of 5 g hydrazinium sulfate p.p.b. level Se(IV,VI) is quantitatively
coprecipitated with 25}500
g Te(IV) and completely collected on nitrocellulose
membrane filter (pore size 0.2
m). Heavy metals (high level) do not interfere
Ag, Au, Pd, Pt, Rh Cu, Fe, Ni-ores. 1 g sample is dissolved with aqua regia (10 cm3) at 1803C for 5 h
and diluted to 100 cm3. 10 cm3 is evaporated, leached with 0.5 mol L\1 HCl, Ag is
trapped on Dowex 50 w;8 column and eluted with 0.5 mol L\1 HCl. Au, Pd, Pt,
Rh in 2 mol L\1 HCl are coprecipitated by adding 5 mg Te (TeCl3) and SnCl2
(28%)
Th(IV) Mo Seawater. To 500 cm3 seawater (1 cm3 9 mol L\1 H2SO4) is added 3}4 cm3
0.1 mol L\1 Th(NO3)3, adjusting pH to 6.0 (aq. NH3) and standing for 30 min.
Precipitate is collected on millipore filter, dissolved in HCl. Precipitation yields:
99.5 (pH 6.0), 81.5 (pH 7.5), 61.6 (pH 8.5)
Zr(IV) Bi Sea-, spring-, river waters. To sample ZrOCl2 is added, pH to 9.0 with aq. NH3.
Precipitate is dissolved in 4 mol L\1 HCl. Quantitative recovery is obtained at
Zr'10 mg in the pH range 8.8}9.2.'0.5 mg Al, As(III), Cu, Sn interfere
Cd, Cu, In, Pb Sediments. To sample solution containing 0.04 g sediment, 1 cm3 ZrOCl2 is
added and pH is adjusted to 8.8. Precipitate is dissolved in 25 cm3 4 mol L\1 HCl.
0.01
g g\1 In in sample solution is quantitatively recovered with '5 mg Zr at pH
8.4}8.8. Cu and Pb are quantitatively collected at pH 8.25}9.0. 0.05 mg As, Bi
interfere with Cu determination; 0.05 mg Sn(II), 0.1 mg As(III), Tl(I) interfere with
Pb determination. Optimum pH for Cd is 9.0

(Continued)
 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION 4399

Table 1 Continued

Precipitants Trace elements Sample, experimental conditions, comments

Hydroxides
Sb Seawater, algae, silicates. 1 dm3 water is added 3 cm3 6 mol L\1 HCl and
300}400 mg K2Cr2O7, heating to 75}853C for 1 h. After cooling, 150 mg ZrCl4
completes dissolution; pH is adjusted to 5.0$0.3, ageing at least 90 min.
Recoveries of Sb 99.2% (pH 3.0), 99.1% (pH 5.0), 93.7% (pH 7.5). Standard
deviation: 0.003}0.009
g dm\3 for 0.08}0.42
g dm\3 Sb

Sul\des
Bi(III) As, Sb, Se Water samples. Coprecipitation is carried out in 1.2 mol L\1 HCl solution. Min-
imum amounts: As(III) 10 ng, Sb(III) 50 ng, Se 20 ng
Cd(II) Cr, Cu, Fe, Mn, Pb, V BaCl2 solution. Coprecipitation conditions are studied for 1
g of Cr, Fe, Mn, Pb,
V, 0.1
g of Cu. CdS is precipitated from Cd(CH3COO)2 with Na2S. Optimum
conditions: pH 7}8, BaCl2 15%
In(III) Co, Cr, Cu, Mn, Zn Calcite. 0.3 g sample is dissolved with HF, HNO3 and HClO4, diluted to 200 cm3.
30 mg In is added; pH adjusted to 9.0. In2S3 is precipitated by adding 0.1 g
thioacetamide. Coprecipitation recoveries: Co 89.4%, Cr 94.5%. Cu 88.8%,
Mn 94.9%, Zn 89.1% for each concentration of 25 p.p.m.
Pb(II) Cu Tap water, iron and steel. 30 cm3 tap water is acidified with 0.5 cm3 6 mol L\1
HCl. After adding 1.0 mg Pb, PbS precipitates by passing H2S. Precipitate is
dissolved in conc. HCl (1 cm3). RSD: 0.7% for 5
g dm\3

Sulfates, phosphates
Pb(II) Cr(III)(VI) Seawater. Na2SO4 (0.2 mol L\1 8 cm3) added to 800 cm3 sample; pH adjusted to
3; 8 cm3 Pb(NO3)2 added dropwise. Precipitate filtered on Nuclepore. Cr(VI) only
coprecipitates with PbSO4 at pH 3. To another 800 cm3 sample 3 cm3 0.2 mol L\1
Pb(NO3)2 and 0.2 cm3 1 mol L\1 NH4H2PO4, pH adjusted to '6 with aq. NH3.
Both Cr(III), Cr(VI) are quantitatively coprecipitated with Pb3(PO4)2. In seawater,
molar ratio of Pb to PO34\ is kept at 3 for perfect recovery of Cr (VI). 0.08
g dm\3
Cr is detectable
Bi(III) Am, Cm, Np, Pu Sequential separation by coprecipitation with BiPO4, by using suitable oxidizing
and reducing agents. Cm(III) is separated from Am, Np, Pu, U in their (VI) valency
states by adding K2S2O8-Ag# before coprecipitation. Next, Am(III) is separated
from Np, Pu, U by adding C2H5OH. Pu(IV) is separated from Np and U by adding
NaNO2. Last, Np(IV) is coprecipitated by addition of H2O2. Recoveries: Cm
(98.5%), Am (97.6%), Pu (94.7%) and Np (96.0%)

Fluorides
Ca(II) Cu, Fe Surface adsorption was shown to be the main cause for coprecipitation. Cu(II) is
coprecipitated as counter ion to excess F\, whereas Fe(III) coprecipitates as
FeF36\ in competition with matrix fluoride
U Aqueous solution. To sample 5 cm3 1 mol L\1 Ca(NO3)2, 30 cm3 0.3 mol L\1
NH4F solutions are added dropwise. Applicable to coprecipitation of
0.01 ng dm\3 U
Th Uranium ore. 1 g sample is digested in 50 cm3 of mixture (8 mol L\1
HCl#0.01 mol L\1 (NH4)2SiF6) by heating for 4 h. To solution 10 mg La carrier,
HF is added to precipitate LaF3. Precipitate is dissolved with 16 mol L\1 HNO3.
Recovery of 234Th tracer is 85% with RSD of $12% for 1 p.p.m. level

Oxalates
Ca(II) La, Lu, Tb Biological materials. Substoichiometric precipitation of rare earth elements
was studied. To 8 cm3 of solution containing 0.0125 mol L\1 Ca(II),
2.0;10\7}6.0;10\5 mol L\1 radioactive RE(III) and 1.25;10\3 mol L\1
CCl3COOH!0.1 mol L\1 oxalic acid. At pH 2}5 complete coprecipitation is
attained
Ce, Pm, Sr, Y Urine. 500 cm3 is wet-ashed. Resulting salt is treated with 30% H2O2, dissolved in
H2O, pH is adjusted to 3.0 (aq. NH3). Oxalate is removed by dissolving precipitate
in hot HNO3, heating in the presence of 70% HClO4. Residue is treated with H2O2
to reduce Ce(IV) to Ce(III) followed by dissolution with H2O. Recoveries: Ce 87%,
Pm 89%, Sr 100%, Y 64%

(Continued)
4400 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION

Table 1 Continued

Precipitants Trace elements Sample, experimental conditions, comments

Salts and metals

AgCN Pd Pure metals. Highly selective for Pd in 107}109 fold excess of Ag, Al, Bi, Cd, Co,
Cu, Fe, Mn, Ni, Pb, Sn, Th, Tl, U, Zn. Coprecipitation of Pd is not influenced. Pd is
coprecipitated with AgCN. RSD 4}16%; detection limit: 10\7% Pd
As Se, Te Geological and biological samples. After digestion of samples by mineral acids
(HNO3#HClO4). Na3AsO3 (c. 1.5 mg As) is added. As(III) is reduced to elemen-
tal As by phosphorous acid at 803C for 15 min. Ageing at room temperature for at
least 8 h to complete flocculation. Detection limit is 0.1 p.p.m. of Se and Te

1. The sparingly soluble organic compound, such as
a bulky organic cation, forms an ion pair with the
anionic complex.
2. An insoluble salt formed between the organic
anion and the metal cation is coprecipitated
together with the excess of the reagent, e.g.
metal-8-quinolinate in excess of 8-hydroxyquino-
line.
Table 2 Coprecipitation of trace elements with organic collectors
3. A soluble chelate compound of a trace metal can
be coprecipitated with the precipitate formed be-
tween the excess of the reagent and a bulky differ-
ent organic reagent cation.
4. An inner complex of the metal ion to be
separated is coprecipitated with a large excess of
the organic reagent such as 1-(2-pyridylazo)-2-
naphthol.

Collectors Trace elements Samples, experimental conditions, comments

Organic carriers
-Benzildioxime (-BD) Ni Seawater. 500 cm3 samples, 1 mg -BD, pH&9.5. Ageing time can be minimal.
Even 0.2 p.p.b. Ni can be determined
DDTCA (diethyldithio- As, Cd, Cu, Fe, Mn, Water samples. Ni, Cu are completely precipitated between pH 1 and 11. Cd,
carbamic acid) Ni, Pb, Se, Zn Fe(III), Pb, Zn begin to precipitate at pH 1}2 but complete precipitation is only
obtained above pH 4 (pH 5 was used). Complete recovery of As is only obtained
at pH 5.0}5.5. For very pure water, metal carrier should be used. Citrate is
a powerful masking agent for Fe
Co, Eu, Mn, Zn Natural water samples. To 250 cm3 sample, 20 cm3 2% (w/v) NaDDTC solution
and 5 cm3 buffer solution (pH 5) are added. Coprecipitation capacity:
900
mol L\1. Recoveries: Co 97}98%, Eu 88}100%, Mn 85}98%, Zn
82}100%
Cu, Fe, Hg, Zn Saline water. To 250 cm3 sample, 400 mg freshly prepared NaDDTC is added
at pH 4.0
DDTCA#dibenzylidene- As, Cd, Cr, Cu, Fe, Mn, Industrial wastewater, river water. Concentration range 1}50
g. pH 5.0}5.5
D-sorbitol (DBS) Pb, Sb, Zn (acetate buffer). 100 mg NaDDTC, 17 mg DBS as flocculant. 94}100% recov-
ery for Mn
Diethylammonium Cd, Cr, Cu, Hg, Ni, Pb Drinking, wastewaters. To 500 cm3 sample is adjusted pH to 5.0}5.5 (acetate
N,N -DDTC buffer), then diethylammonium N,N -DDTC is added to make 2%. Recovery
ranges: Cd 84}94%, Cr 86}102%, Cu 94}106%, Hg 100}108%, Ni 99}110%,
Pb 88}92%
DBDTCA (dibenzyldithio- As(III),(V), Cd, Fe, Zn Fresh water. pH 2. 100 cm3 samples, 10 mg of Na-salt of DBDTCA in methanol
carbamic acid) added. As(III) coprecipitates but not As(V) which precipitates after reduction to
As(III) with KI# Na2S2O3. Recovery of As(III) 100% in pH range 1}3 but drops
drastically for higher pH. 2}3 mg of DBDTCA is sufficient. High recoveries are
obtained for Cd, Fe and quite high (87.5%) for Zn
DBDTCA# Se(IV) Fresh water and seawater. 500 cm3 sample is adjusted to pH 2. 10 mg of
phenolphthalein Na-salt of DBDTCA and 100 mg phenolphthalein in methanol are added. With-
out phenolphthalein, recovery is 97% but decreases with ageing time. pH
should be(4. In the presence of phenolphthalein ageing does not reduce the
yield
Dithizone Ag, Bi, Cd, Cu, Hg, Dilute HCl and HNO3 solutions. After adjusting acid concentration, 0.1 g ascor-
Pb, Pd, Zn bic acid added to reduce Fe(III), finally dithizone is added. Recoveries depend
on the acid concentration. (HCl, M, recovery, %) Bi (10\2!5;10\2, 95), Cd
((0.002, 95), Cu ((2, 95), Hg ((1.5, 95), Pb ((0.001, 95), Pd ((1, 95), Zn
(3;10\4, &40)
(Continued)
 III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION 4401

Table 2 Continued

Collectors Trace elements Samples, experimental conditions, comments

Dithizone# Ag, Co Surface water. 4 dm3 water to be made 0.5 mol L\1 H2SO4 solution; About
phenolphthalein 28 mg dithizone and 300 mg phenolphthalein are added. Phenolphthalein helps
in collecting the precipitate. At pH 1 only Ag is coprecipitated. Co is recovered
quantitatively at pH 6.5}8. Dithizone in glacial CH3COOH, phenolphthalein in
ethyl alcohol
2-Mercaptobenzimidazole Ag, Au, Hg, Sn, Ta Seawater. Recoveries were studied at pH 1, 3, 5. To 20 dm3 seawater (pH 1) 5 g
(MBI) MBI in 100 cm3 ethanol were added and aged for 2 days (0}53C). High recove-
ries are obtained at the following pH ranges: Ag (1}5), Au(1), Hg(1}5) Sn(5),
Ta(1}3)
Nioxime (1,2-cyclohexane- Ni Aqueous solution. Selective precipitation of Ni at
g level at pH 7.0}9.5. Large
diondioxime) amounts of Co, Cu, Fe are masked with Na-tartrate and Ca-EDTA. 32 mg
nioxime is used
Oxine (8-hydroxyquinolinol) Ce, Fe, Pr, Pu Aqueous solution. Fe was co-crystallized with oxine produced in situ by hydroly-
sis of 8-acetoxyquinoline. Ce, Pr, Pu also quantitatively recovered
Cd, Cu, Mn, Pb, Zn Seawater. The metal ions are quantitatively precipitated in pH range 7.0}8.5
with oxine alone (5 cm3 of 2% solution) kept at 703C for 3 h. Detection limits
(ng dm\3): Cd 1.4, Cu 10, Mn 5, Pb 10, Zn 6. Recoveries'98%
Al Aqueous solutions. 100 cm3 sample, pH 6.8}6.9 (0.07 mol L\1 phosphate buf-
fer). Slow addition of 160 mg oxine. Ageing 1 h. Sensitivity 0.1}0.02 p.p.m.
PAN (1-(2-pyridylazo)- Cr(III), Cu, Hg, Mn, Seawater. 0.5}4 dm3, metal ions are precipitated by adding 20 mg of PAN
-2-naphthol)) Ni, Zn ethanol solution and heating at 70}803C for 10 min at pH 6.5}10 for Cu, Ni, Zn;
high pH&10 for Cr(III), Mn. It is preferred that pH should be &9 in order to
decrease coprecipitation of Ca. Alkali and alkaline earth metal ions do not
interfere
Organic carriers
U Seawater, tap water, digestate of biological samples. Coprecipitation is most
effective at pH 4.5}6.5 with recovery of 85}94%. In the presence of 1,2-
cyclohexylenedinitrilo tetra acetic acid (CyDTA) as a masking agent. The
method is highly selective for U. Detection limit: 3}4 ng dm\3 for 500 cm3
samples and 5
g kg\1 for 0.5 g biological samples
Thionalide (2-mercapto- AS(III), Cu, Sb(III)(V) Seawater. 0.005}0.25 mol L\1 H2SO4. As is only precipitated quantitatively at
N-2-naphthylacetamide) H2SO4'0.2 mol L\1 As(V) is reduced to As(III) by ascorbic acid. 0.015 mol L\1
#Oxine H2SO4 is used for total As precipitation. 8 cm3 2% thionalide in acetone is used.
Sb and Cu are coprecipitated with oxine at pH 6}9 while As is not coprecipitated
at all
96
TPAC (tetra- TcO\
4 Aqueous perchlorate solution. Recovery is constant at pH in the range 0.5}13.
phenylarsonium chloride) Precipitation yield is highly affected by TPAC and ClO\ 4 concentrations. At
253C, '90% yields are obtained at TPAC'0.02 mol L\1. Coprecipitation of
Tc(IV) is very low

Metal#organic carriers
Fe (DBDTC)3 U Natural waters. To 500 cm3 sample. 20
g Fe(III) and 2 cm3 0.1 mol L\1
KH2PO4(pH 4) are added prior to pH adjustment to 4.0$0.02. DBDTC (1%,
1 cm3) is added, stirring (15 min), ageing (15 min). Detection limit is 0.4 p.p.b.
Fe (TMDTC)3 (tetra- Cd, Co, Cu, Ni, Pb Mineral water. 0.5 mg Fe(III) to 200}250 cm3 sample, pH to 2}3, CO2 is boiled
methylenedithiocarbamate) out, 50 mg TMDTC is added. The choice of Fe(III) is due to its high concentra-
tion in mineral water. High recoveries are obtained '95%
Co(PDC)3 (pyrolidine Cd, Cu, Ni Seawater. 100 cm3 samples are adjusted to pH 2 (6 mol L\1 HCl). 50
g Co (as
dithiocarbamate) CoCl2), 10 mg APDC are added and aged for 5 min. Precipitate is dissolved in
acetone
Pb(PDC)2 Co, Cu, Hg Dead Sea surface water. To 1 dm3 sample 2.5 mg Pb is added, pH to 3.6, and
20 mg APDC is added. Determination by X-ray fluorescence
Al-oxinate or In-oxinate Co, Cu, Mo Agricultural sample digestate. Al-oxinate perfectly recovers Co, Mo, but not Cu.
Addition of thionalide or tannic acid or both leads to quantitative recovery for
both Al- and In-oxinates. To a 500 cm3 sample 15 mg Al, 500 mg oxine are
added; pH to 4.5; 200 mg tannic acid, 20 mg thionalide
Mg-oxinate Al, Cd, Co, Cu, Mn, Aqueous solutions. To 100 cm3 solution, 20 mg Mg2#, 100}200 mg oxine are
Ni, Pb, Zn added at pH 9.
Cr(III), Mo, V Ageing at 703C for 1 h. Recoveries: 100% except for Cr (83%), Mo (64%),
V (70%). Cr recovered at 98% at pH 10.5
4402 III / TRIGLYCERIDES / Liquid Chromatography
5. A chelate of the trace metal is adsorbed and cop-
recipitated with a water-insoluble organic com-
pound. Several metal dithizonates can be
coprecipitated with phenolphthalein.
6. The metal ions are coprecipitated by means of
colloidal}chemical sorption on a mixture of insol-
uble organic reagents.
Typical examples of the coprecipitation of trace
metals with organic collectors are listed in Table 2.
See also: II/Extraction: Analytical Extractions; Analytical
Inorganic Extractions.
Further Reading
Alfassi ZB (ed.) (1994) Determination of Trace Elements.
Weinheim: VCH Verlagsgesellschaft.
TRIGLYCERIDES
Liquid Chromatography
V. Ruiz-Gutierrez and J. S. Perona,
Instituto de la Grasa (CSIC), Seville, Spain
Copyright ^ 2000 Academic Press
Synopsis
High performance liquid chromatography (HPLC)
has become a useful tool for the analysis of triglycer-
ides from all sources. This article reviews develop-
ments for the analysis of molecular species of tri-
glycerides, including stationary phases, mobile
phases, sample solvents, detection and identiRcation.
It also points out the advantages of silver-ion
HPLC and emphasizes the need for stereospeciRc
analysis in the complete determination of triglyceride
molecular species because currently this is not pos-
sible by reversed-phase HPLC. Finally, the applica-
tion of HPLC to triglycerides from fats and oils is
described.
Introduction
The goal of chromatographic analyses of lipids is the
resolution of all classes and molecular species for the
purpose of a complete identiRcation and characteriza-
tion of all the components of a fat or an oil. This
characterization is not complete without the
Alfassi ZB and Wai CM (eds) (1992) Preconcentration
Techniques for Trace Elements. Boca Raton, FL: CRC
Press.
Bonner NA and Kahn M (1951) Radioactivity Applied to
Chemistry. New York: John Wiley.
Kolthoff IM, Sandell EB and Meehan EJ (1969) Quantitat-
ive Chemical Analysis, 4th edn. New York: Macmillan.
Minczewski J, Chwastowska J and Dybczynski R (1982)
Separation and Preconcentration Methods in Inorganic
Trace Analysis. Chichester: Ellis Horwood.
Mizuike A (1983) Enrichment Techniques for Inorganic
Trace Analysis. Berlin: Springer-Verlag.
Walton AG (1967) The Formation and Properties of Pre-
cipitates. New York: Interscience.
Zolotov YA and Kuzmin NM (1990) Preconcentration of
Trace Elements. Amsterdam: Elsevier.
determination of their triglyceride (TG) molecular
species proRle. Once the fatty acid composition of
a determined fat or oil is clear, the knowledge of how
these fatty acids are distributed within the glycerol
molecule is of major interest.
Fractionation of TGs has been carried out by dif-
ferent chromatographic techniques. Argentation thin-
layer chromatography (Ag-TLC) has been employed
to separate TG fractions, with subsequent analysis of
their fatty acid methyl esters. Direct gas chromatogra-
phy, using fused-silica capillary columns coated with
high-temperature polar stationary phases has also
been used for this purpose with rather poor results.
The introduction of chemically bonded phases and
high performance liquid chromatography (HPLC) in-
creased the usefulness of liquid chromatography for
the separation of TGs. The Rrst paper dealing with
the HPLC of triacylglycerols (TGs) was published in
1975 by Pei et al. Simple TGs of medium-chain length
were separated on a reversed-phase column. Other
workers then began to use HPLC for the analysis of
long-chain TGs, on silicic acid columns, reversed-
phase columns, or both. The Rrst fractionation of
natural TGs by HPLC on reversed-phase columns
was performed independently in 1977 by Plattner
et al. and Wada et al. The later authors were the Rrst
to establish a parameter, termed the partition number
(PN; PN"CN!2ND, where CN is the total num-
ber of carbons and ND is the number of double bonds
in the fatty acids constituting the TG molecule) for

characterizing TG molecules. They found that TGs
on reversed-phase columns eluted in increasing order
of PN. Today, reversed-phase high performance
liquid chromatography (RP-HPLC) is the most fre-
quently employed technique for separating complex
mixtures of TGs, as it allows a good resolution of
mixtures into molecular species, based on properties
such as molecular weight, degree of unsaturation,
polarity and molecular conRguration. Nevertheless,
despite notable success, the progress in RP-HPLC has
not been easy, due to difRculties encountered in the
process of separation, detection and identiRcation.
One of the main difRculties in the HPLC analysis of
TGs is the formation of the so-called critical pairs,
that is, molecules found to have close behaviour on
reversed-phase columns in spite of the difference in
chain lengths, number of double bonds and geometri-
cal conRguration. Critical pairs, therefore, have been
deRned as those structures, with the same PN. This
problem has not been solved in natural fat analysis
yet. However, a long time ago standards of critical
pairs of TGs were separated. El-Hamdy and Perkins
were able to separate two geometrical isomers:
triolein (54 : 3 ccc) from trielaidin (54 : 3 ttt), which
differ only by the conRguration of the double bonds.
The second difRculty is the establishment of
a chromatographic system capable of simultaneously
resolving TGs with large differences in carbon chain
lengths. The separation of short-chain, medium-chain
and long-chain TGs in the same chromatogram,
involves the utilization of elution gradients and some-
times yields different responses in different parts of
the chromatogram.
The third difRculty is the detection of molecules at
the column outlet. Refractive index and ultraviolet
detectors have been employed, but the analysis of
complex mixtures of TGs requires speciRc detectors.
The emergence of the evaporative light-scattering de-
tector (ELSD) and the application of mass spectro-
metry (MS) to HPLC has been decisive for the
analysis of TGs.
The last major problem is the identiRcation of
chromatographic peaks. As very few pure standards
are commercially available and as many critical pairs
remain unresolved, this is one of the most difRcult
aims to attain. Again, HPLC}MS looks like a useful
tool for this purpose, although several authors have
developed other systems for TG identiRcation.
Nomenclature
The proposal of Hirshmann has now been universally
adopted for structural assignments. An sn- preRx is
included in the names of all glycerols. Each fatty acid
in the glycerol molecule is identiRed by listing the
sn-1, sn-2 and sn-3 position in order. A rac preRx
III / TRIGLYCERIDES / Liquid Chromatography 4403
indicates that the middle fatty acid in the abbrevi-
ation is attached at the sn-2 position, while the re-
maining two acids are equally divided between the
sn-1 and sn-3 positions, yielding a racemic mixture of
two enantiomers. A
preRx indicates that the
middle fatty acid esteriRes the
- or sn-2 position.
Mobile Phase
The selection of the mobile phase is one of the most
important factors regarding TG liquid chromato-
graphic analysis. Plattner et al. brieSy examined the
effect of solvent composition upon triglyceride separ-
ations. Later, Pauls compared seven binary solvent
mixtures for the analysis of olive oil triglycerides.
They achieved the best critical pair separation
with the use of acetonitrile as weak solvent.
n-Propionitrile has also been proposed as an eluent
but disadvantages include high cost and toxicity. Re-
cently, Hirano and Takahasi have established three
factors for the selection of mobile phase solvents in
order to obtain optimum column efRciency. Solvents
should be low in molecular weight and viscosity but
high in solubility of TGs. These factors must be
balanced to ensure high column efRciency.
The function of the organic modiRer is to improve
the solubility of the compounds in the mobile phase,
so as to provide changes in their polarity, and thus
increase peak selectivity. An increase in the solvent
strength of the mobile phase is directly related to an
increase in both retention time and resolution of TGs,
including critical pairs. Among the organic modiRers
tried, Pauls et al. showed that chloroform and tet-
rahydrofuran had the greatest solvent strength for the
elution of the critical pair POO}OOO (52 : 2}54 : 3,
PN"48), while the best resolution for the pair
LOO}LPO (54 : 4}52 : 3, PN"46) was achieved
with dichloromethane. The dependence of resolution
upon solvent composition is a function of the extent
to which a solvent can shift retention per double bond
compared to the extent to which it shifts retention per
carbon unit. The most commonly employed binary
solvent mixture for TG analysis is acetone in acetonit-
rile as the weak solvent. However, acetone is incom-
patible with UV detectors as it absorbs at the same
wavelengths as TGs.
The analysis of TGs by RP-HPLC has been per-
formed for a long time with isocratic elution, due to
the general use of refractive index (RI) detection. This
system has provided good results for simple oils, but
the analysis of complex fat mixtures, i.e. animal fats,
requires gradient elution conditions. The goal is to
achieve a good resolution for poorly retained TGs
(saturated molecular species with short-chain fatty
acids) and, at the same time, to elute, in a reasonable
4404 III / TRIGLYCERIDES / Liquid Chromatography
separation time, the most retained TGs (saturated
molecular species with long-chain fatty acids). This
permits the resolution of complex mixtures of TGs,
such as those from Rsh oils, containing long-chain
polyunsaturated fatty acids, and from milk fats, with
a broad range of PN values.
Acetone, n-propanol, methyl tert-butyl-ether or
dichloromethane, give good results when used in
gradient conditions with acetonitrile. The gradient
systems can be linear or nonlinear. Nonlinear gradi-
ents, and step gradients have shown better separ-
ations of critical pairs.
Sample Solvent
The sample solvent is of great importance when the
sample is a complex mixture of TGs with a wide
range of polarity, because it is enormously difRcult to
Rnd an appropriate solvent for all the TGs. More-
over, the selected solvent must permit an appropriate
contact between the solute and the stationary phase
for chromatographic separations. Tsimidou and
McRae studied the inSuence of the injection solvent on
the RP-HPLC of TGs. They found that chloroform
produced inferior resolution under all conditions,
which was accentuated by the injection of large vol-
Figure 1 RP-HPLC of fish oil triglycerides. HPLC conditions: Waters 2690 liquid chromatograph equipped with a Spherisorb ODS-2
column (250;4.6 mm), coupled to a Eurosep DDL-31 light-scattering detector; solvent, a two-step gradient of 20}80% acetone in
acetonitrile at flow rate 1 mL min\1.
umes. Acetone was recommended, but it is not suitable
for high-molecular-weight saturated TGs. Mobile
phase has also been suggested as an ideal solvent, but
others have employed hexane obtaining better results.
Stationary Phase
Reversed-phase columns are used for separating
homologous series of compounds, such as TGs. Pre-
vious studies have shown that octadecylsilane (ODS)
stationary phases on spherical particles have the best
selectivity for TGs, with little variation among the
columns of different manufacturers. Columns with
a particle size of 3
m have the highest intrinsic efR-
ciency; however, until recently, their use was re-
stricted because of the high operating pressure
needed.
Most RP-HPLC analyses are carried out without
column thermostating. However, various workers
have shown that an increase in temperature
affects retention and selectivity, yielding poorer sep-
arations. Although lower temperatures give better
separations, elution times are increased signiRcantly.
Moreover, highly saturated TGs may precipitate out
of the mobile phase. For these reasons, the choice of
column temperature must represent a compromise
 III / TRIGLYCERIDES / Liquid Chromatography 4405

between good solubility of saturated TGs concomi-
tant with good selectivity of critical pairs.
In 1996, Hirano and Takahashi discussed the the-
oretical aspects of improving resolution of TG mo-
lecular species via RP-HPLC when working at low
temperatures. They analysed Rsh oils (Figure 1), with
a low melting point, establishing a critical temper-
ature (!153C), below which there is no improvement
in resolution. Similar results had been obtained be-
fore, through lowering temperature only to 153C.
chromatograms, and unlike UV, allowed utilization
of acetone. Subsequently, the inSuence of nebulizer
gas pressure, temperature, mobile phase composition
and Sow rate on the response of the detector was
investigated. Regardless of the exponential response
of the detector, which depends on solute concen-
tration, nowadays ELSD is the most commonly

Detectors
When most separations were made by isocratic
elution systems. Refraction index (RI) detection
was extensively employed, but complex mixtures of
TGs require gradient elution, making RI detection
impossible. Moreover, it had low sensitivity and
different responses for saturated and highly un-
saturated TGs.
The UV detector is compatible with gradient elution
and has been used for HPLC analyses of TGs. The
absorption region from 200 to 230 nm (ester bond) is
used to detect TGs. However, many solvents also ab-
sorb at these wavelengths, causing baseline drift with
gradient elution systems. In addition, different TGs
have nonuniform molar extinction coefRcients, and
consequently calculation of their response factors with
standards is needed for quantitative analysis. Other
workers have used Same ionization detection (FID)
and attained good sensitivity and baseline stability
with elution gradient. Nurmela and Satama tested FID
for TGs. They found a variable response for different
TGs, although the variation was smaller than with UV
detectors. In addition, a nonlinear response of the
detector was observed for injections (5
g. This may
be a shortcoming, because only a small portion of the
solvent eluted from the column can be introduced into
the FID.
The introduction of the mass or evaporative light-
scattering detector (ELSD) has brought a major ad-
vance in the detection of lipid classes upon HPLC
separation. ELSD, being sensitive only to the mass of
vaporized analyte, is not limited by the absorption
characteristics of the individual components and/or
the nature of the eluents. For this reason, it is compat-
Figure 2 RP-HPLC of butter triglycerides. (A) Refractive index
ible with gradient elution and volatile solvents do not
detection; Spherisorb-5-ODS-2 and isocratic elution of acetone in
give baseline drift, as they are removed before detec- acetonitrile. (B) Light-scattering detection: same conditions as
tion of the analyte by evaporation. The only require- (A). (Reproduced with permission form Robinson JL, Tsimidou
ment is that the compounds to be detected must be M and Macrae R (1984) Journal of Chromatography 303: 386.
much less volatile than the solvent. Copyright Elsevier Science). (C) Ultraviolet detection; two Lich-
rospher 100 CH-18/2 columns and isocratic elution of acetone in
ELSD was described for the Rrst time at the end of
acetonitrile; tricaprylin (8 : 0)3, and triarachidin (20 : 0) are internal
the 1970s. In 1984, Robinson and Macrae, compared standards. (Reproduced with permission from Nurmela KVV and
ELSD with UV and RI detectors for the analysis Satama LT (1988) Journal of Chromatography 435: 139. Copy-
of butter TGs (Figure 2). ELSD provided better right Elsevier Science.)
4406 III / TRIGLYCERIDES / Liquid Chromatography
employed detector for TG molecular species
determination.
Identi\cation of Molecular Species
In spite of their usefulness, all detectors described
above have the shortcoming of poor limited struc-
tural identiRcation. Mass spectrometry (MS) has be-
come necessary for a complete identiRcation of TG
species.
Several methods for mass spectrometry of TGs
have been proposed but some drawbacks have been
found. Electron impact ionization methods generally
result in spectra containing low molecular weight
fragments, with no quasimolecular ions present. Elec-
trospray ionization (ESI) provides only quasimolecu-
lar ions with no fragmentation. Unfortunately, this
lack of fragmentation can result in ambiguity in struc-
tural assignments for TGs with identical molecular
weight. Information on both the molecular weights
and the fatty acyl residues of TGs have been achieved
by combination of RP-HPLC and atmospheric pres-
sure chemical ionization MS. Desorption chemical
ionization (DCI) and positive ion chemical ionization
(PICI) have also been successfully used for TG struc-
tural characterization.
When MS is not possible, some authors have used
the equivalent carbon number (ECN) for the tentative
identiRcation of TGs. The ECN of each TG in the
sample is the ECN of the hypothetical saturated
TG having the same retention time. When carbon
numbers (CNs) are plotted against ECN, straight
parallel lines are found for different unsaturated
TGs. Thus a theoretical prediction can be achieved,
which has become a useful tool for TG identiR-
cation.
The linear relationship between the retention fac-
tor (k) and PN values of the TG, was Rrst established
by Wada et al. in 1977. Then HersloK f et al. estimated
theoretically the ECN for unsaturated TGs, on the
basis of their relative retention times, from an experi-
mental linear relationship between relative retention
time and carbon number. This ECN is analogous to
PN (ECN"CN!aND), with the difference that in
this case the value of a depends on the chromato-
graphic system used for measurements. However, a
generally takes values close to 2 and when a"2, the
values for the ECN and NP are equal. Takahashi et al.
calculated the value for a from the relationship bet-
ween log k, CN and ND (log a"q#bCN#cND).
The value of a is the quotient between the constants
b and c. These equations are calculated under iso-
cratic conditions, and are not appropriate for gradi-
ent-elution systems. For this reason, some workers
have developed new relationships based on the same
parameters, as the equivalent chain length (L) or the
theoretical carbon number (TCN).
The chromatographic behaviour of TG molecules
in RP-HPLC depends not only on CN and ND but
also on the number of unsaturated fatty acids within
the molecules (NUFA), because TGs with the same
ECN are eluted in the order of the increasing con-
stituent saturated fatty acids. This leads to the equa-
tion for ECN (ECN"CN#a1ND#a2NUFA).
The TG prediction process becomes increasingly
complex when the fat contains a great number of
different fatty acids, since the number of possible
combinations can be extremely high. Therefore, and
as a second part of the prediction process from the
ECN, some authors have proposed the application of
the equations developed by Takahashi et al. These
workers developed a matrix model with CN and ND
as variables for each fatty acid esterifying the glycerol
molecule.
Silver-ion Chromatography
Silver-ion HPLC can be performed on a reversed-
phase column (silver ions in the mobile phase), on
a silver-loaded, cation-exchange column, or on a sil-
ver-loaded silica column. Silver-ion chromatography
separates TGs according to their degree of unsatur-
ation, the distribution of double bonds between the
fatty acyl residues within a single molecule, the con-
Rguration and position of double bonds within each
fatty acid and the stereospeciRc position in which
fatty acids are esteriRed. The mechanism of separ-
ation is based on the ability of the -electrons in the
double bonds of the fatty acids to interact with the
silver ions of the stationary phase.
Silver ions are incorporated into columns in two
different ways: by impregnating the silica-gel support
with a silver salt or by bonding silver ions to the phase
by means of an ion-exchange phase. The impregna-
tion of columns with silver ions is generally made
with silver nitrate in concentrations from 5% to
10%. The problem of short column life is avoided
with cation-exchange supports, such as macroreticu-
lar sulfonic acid resins or silica-gel supports with
chemically bonded methylsulfonic acid groups.
The mobile phase is an important factor affecting
the separation of TGs by silver-ion HPLC. However,
the nature of the interactions between the silver ions,
unsaturated solutes and solvents in the mobile phase
has not been fully elucidated. Some workers have
suggested using elution gradients combining chlorin-
ated hydrocarbons with acetone and acetonitrile.
Components separated by silver-ion HPLC are
commonly detected by evaporative light-scattering
detectors (ELSD) or FID, because they place fewer

limitations on the choice of solvents for the mobile
phase, but these detectors do not provide structural
information on molecular composition. For this rea-
son, mass spectrometry has recently been employed
for this purpose.
Christie et al. have made the greatest progress in
developing silver-ion chromatographic systems. Sub-
sequently, other authors have applied their method
for separation of TGs from different natural sources.
More useful information of the TG composition of
natural fats may be achieved by combining this tech-
nique with RP-HPLC. Silver-ion HPLC allows separ-
ation of TGs with the same degree of unsaturation;
the fractions obtained can then be analysed by RP-
HPLC with chain length as a factor for separation.
Stereospeci\c Analysis
For the complete TG characterization of a fat it is
necessary to know not only the fatty acids that consti-
tute a TG molecule but also the positions of attach-
ment. This is of importance because physicochemical
properties change depending on the position in which
a fatty acid is attached.
However, the stereospeciRc analysis of a fat is one
of the most difRcult tasks to undertake, since these
molecules are similar in physical and chemical prop-
erties. When positions sn-1 and sn-3 are occupied by
distinct acyl groups, the TG molecule will be asym-
metric and will have optical activity. However, when
the same fatty acid is allocated at both positions,
diastereomer forms are outlined. This is not rare,
since the main biosynthetic route in animal and plant
tissues is the sn-glycerol-3-phosphate pathway, and
enzymatic systems in this pathway can be speciRc to
certain fatty acids or to certain fatty acid combina-
tions.
Vander Wall and Coleman and Fulton indepen-
dently developed a theory of fatty acid distribution in
the glycerol molecule. They postulated that fatty
acids are distributed randomly in the sn-2 position
and randomly, but independently from sn-2, in the
sn-1 and sn-3 positions. They demonstrated that it is
possible to know the fatty acid distribution from data
obtained on the stereospeciRc fatty acid composition
of the distinct fractions collected after hydrolysis.
However, hydrolysis has revealed that fatty acids do
not follow a random distribution in TG molecules. In
fact, vegetable oils have C18 polyunsaturated fatty
acids at the sn-2 position, with saturated and C20 and
C22 polyunsaturated fatty acids at sn-1 and sn-3.
Oleic acid (C18:1) is distributed at the three positions.
Among animal tissues, ample differences can be
found. The majority of animal fats have saturated
fatty acids at the sn-1 position; however, there are
III / TRIGLYCERIDES / Liquid Chromatography 4407
fats like pig adipose tissue, with palmitic acid at sn-2,
or milk fat, with long-chain saturated fatty acids at
the sn-1 and sn-2 positions.
HPLC analysis, which gives the stereospeciRc dis-
tribution of TGs, uses as substrate mono- and diacyl-
glycerols, obtained after hydrolysis of TGs in the Rrst
step of the process. This hydrolysis is usually made
through a Grignard reaction with magnesium ethyl
bromide. Mono- and diacylglycerols are separated by
thin-layer chromatography (TLC) or by solid-phase
extraction (SPE). The products obtained may
be analysed by liquid}solid chromatography,
reversed-phase liquid chromatography or chiral-
phase liquid chromatography. By liquid}solid
chromatography 1,2-, 1,3- and 2,3-diacylglycerols
are separated through formation of (S)-(#)-1-(1-
naphthyl)ethyl urethane diastereoisomeric deriva-
tives. The combination of the total fatty acid com-
position obtained by gas chromatography and
liquid}solid chromatography permits calculation of
the stereospeciRc composition of the fatty acids in the
TGs of a natural fat.
Reversed-phase chromatography (RP-HPLC) has
been less widely employed for this purpose. SemporeH
and Bezard achieved separations of 3,5-dinitrophenyl
urethane (DNFU) derivatives of 1,2- and 2,3-diacyl-
glycerols with a octadecylsiloxane-bonded silica
(ODS) column and a mobile phase composed of
acetonitrile and acetone. By RP-HPLC Redden et al.
separated fractions containing all the molecular spe-
cies of 1,2-, 1,3- and 2,3-diacylglycerol.
Finally, greater success has been achieved using
chiral-phase HPLC. Acceptable separations have
been obtained for both DNFU derivatives of mono-
acylglycerols and diacylglycerols employing chiral
phases of (S)-2(4-chlorophenyl)isovaleroyl-D-phenyl-
glycine or N-(R)-1-(1-naphthyl)ethylaminocarbonyl-
(S)-valine chemically attached to an aminopropyl-
silane support. However, drawbacks include high
retention times and poor resolution. Recently,
new chiral stationary phases have been proposed,
with (R)-(#)-1-(naphthyl)ethylamine. These phases
provide improved resolution and reduction of separ-
ation times by using shorter columns. By this
method 1,2- and 2,3-diacylglycerols are separated
into fractions, which are subsequently analysed by
gas chromatography in order to determine their fatty
acid composition.
Applications of Triglyceride Analyses
by HPLC
Knowledge of the TG proRle could be a more appro-
priate tool to characterize oils and fats, avoiding the
4408 III / TRIGLYCERIDES / Liquid Chromatography

use of saponiRcation and formation of methyl esters.

HPLC has become as routine as gas chromatography
(GC), providing more complete information about
TG composition of fats and oils.
At a research and development level, detailed TG
structural information might facilitate understanding
of TG biosynthesis in plant and animal cell metabol-
ism, where the activity of acyltransferases are
involved. In this regard, knowledge of the TG mo-
lecular species of a dietary fat, as well as the TG
composition of organs and tissues, can provide signif-
icant information for nutritional purposes.

Vegetable Oils
Virgin olive oil presents a characteristic and unique
pattern of TGs, which may be used to determine
origin and to detect adulteration. Due to its relative
simplicity in TG composition and its relevance in
human nutrition, HPLC was soon employed in the
study of olive oil. In this work, isocratic mobile
phases and refractive index detectors were employed.
With these conditions, up to 10 TG molecular species
could be detected. Triolein (OOO) was found to be
the main TG, with the important presence of
dioleoyl-linoleoyl-glycerol (LOO) and dioleoyl-pal-
mitoyl-glycerol (POO). Later studies, carried out by
RP-HPLC with ELSD, showed that approximately
one-half of the total content of TGs corresponds to
OOO, while the corresponding percentage of POO is
close to 20% and LOO close to 10%. In spite of the
improvement achieved, with the utilization of gradi-
ent elution systems and ELSD, some critical pairs still
remain unresolved.
TG analysis has been extensively employed for the
characterization of edible oils. El-Hamdy and Perkins
determined the TG composition of olive and soybean
oils. The latter oil contained mainly trilinolein (LLL)
and dilinoleoyl-oleoyl-glycerol (LLO). Similar results Figure 3 RP-HPLC of virgin olive oil triglycerides. (A) Refrac-
were reported by other authors using isocratic condi- tive index detection. Hewlett-Packard HP-1050 liquid chromato-
graph equipped with a RP-18 column (250;4.6 mm), coupled to
tions. In the Rrst analysis of soybean oil using gradi-
a Hewlett-Packard HP-1047A refractive index detector; solvent,
ent elution and ELSD, 19 chromatographic peaks 50 : 50% acetone in acetonitrile at a flow rate of 0.9 mL min\1.
were detected, but could not be identiRed. A similar (B) Virgin olive oil with 2% sunflower oil. Same conditions as (A).
number of chromatographic peaks were resolved by (C) Light-scattering detection. Waters 2690 liquid chromatograph
Barron et al. and Hierro et al. using gradients and equipped with a Novapak column (150;3.9 mm), coupled to
a Eurosep DDL-31 light scattering detector; solvent, linear gradient
ELSD. Unexpectedly, LLO was not as abundant as
of 50 : 50% acetone in acetonitrile at flow rate 1 mL min\1.
was originally determined; in both studies, LLL was
the main TG, followed by LnLO and then LLO.
However, Rezanka et al. found signiRcant amounts of elution RP-HPLC with a light-scattering detector
LLO and low amounts of LLnO by RP-HPLC with (Figure 3). They analysed oils rich in oleic acid, such
MS. These differences might be due to different gradi- as olive and rapeseed oils, oils rich in linoleic acid
ents and mobile phases. (soybean and sunSower oils), oils rich in both oleic
Several other oils of interest have been character- and linoleic acids (peanut oil) and oil rich in saturated
ized by HPLC. Perrin and Prevot determined the TG fatty acids (palm oil). They also developed analyses of
composition of various vegetable oils by gradient lard and tallow with great success, identifying more
 III / TRIGLYCERIDES / Liquid Chromatography 4409

than 11 chromatographic peaks for each oil or fat.
More recently, a newly introduced oleic-rich oil, high
oleic sunSower oil, as well as two oils with similar
fatty acid composition, borage and primrose oil, have
been analysed, each showing a different TG distribu-
tion.
Animal Fats
Characterization of animal fats Animal fats are
more complex than vegetable oils. The great differ-
ence in the fatty acids contained in these fats causes
two basic problems. The Rrst one is the difference
in chain length and degree of unsaturation, which
makes it difRcult to achieve a good resolution
for all TGs. The second problem is that there is a
great number of different fatty acids in animal
fats, thus a greater number of TGs appear in
the chromatograms, and there are more critical
pairs.
Animal fats are employed for industrial purposes.
The prediction of the melting behaviour of a fat is
difRcult due to the complexity of the constituent TGs.
Although the amount of stearic or linoleic acids has
been proposed as a good predictor of the consis-
tency of a fat, determination of the TG species pro-
vides more information, since not only the fatty acid
composition but also the positions in which those
fatty acids are esteriRed, are responsible for its phys-
ical behaviour.
Lard is the cheapest animal fat, and commercial
shortenings, providing improved physical properties,
are usually prepared by its interesteriRcation. Al-Ras-
hood et al. analysed pig lard by RP-HPLC with RI
detection to characterize it before and after random-
ization. Lard has been analysed many times before.
Other interest has been focused on pig fat TG charac-
terization to determine the conditions used in pig
husbandry.
Because of its complexity, the structural elucida-
tion of milk and butter fat TGs is a formidable task.
The large number of fatty acids it contains has made
milk fat a particular challenge in terms of TG separ-
ation and identiRcation. Until the introduction of
ELSD, no satisfactory results had been obtained. The

Figure 4 RP-HPLC of rat liver triglycerides. Two columns (Spherisorb ODS-2 3
m) connected in series with a nonlinear elution
gradient of 20}100% (v/v) acetone in acetonitrile were developed at a rate flow of 1.0 mL min\1. (Reproduced with permission of
Perona JS and Ruiz-Gutierrez V. Journal of Liquid Chromatography and Related Technologies, in press. Copyright Marcel Dekker.)
4410 III / TRIGLYCERIDES / Liquid Chromatography

Rrst to analyse butter fat by HPLC with ELSD
were Robinson and Macrae in 1984; they also com-
pared the chromatograms obtained with those of
other detectors, such as UV and RI. As milk fat
needs gradient-elution systems, the latter detectors
offered poor resolution. FID, with linear and
non-linear gradients of acetone in acetonitrile
has also been used to give 62 peaks. Resolution
was enhanced when ELSD was applied, almost
always with acetone in the elution system. Using
this method, up to 111 peaks were separated with
a nonlinear gradient of acetone}acetonitrile as mobile
phase.
in the rat liver were very similar to those of TG in very
low density lipoproteins (Figure 5). Parren o et al.
studied plasma TG composition of a Catalonian
population by HPLC.
Adipose tissue is the most important extrahepatic
tissue in regulating lipid metabolism. Although it
contains up to 97% of TGs, little work has been done
to study its composition of TG molecular species.
Huang et al. have reported 18 molecular species of
TGs in rat adipose tissue using HPLC with UV detec-
tion (Figure 6).

Nutritional interest Not only industry is interested

in the evaluation of the TG content of foods. Medical
and nutritional beneRts can be achieved through
determination of molecular species of TGs.
Human milk, as well as cow or ewe milk fat, have
been subjected to analysis for both industrial and
nutritional purposes. The objective is to achieve the
substitution of the oils employed at present in milk
formulas for infants (coconut oil, corn oil) with
others closer in composition to human milk. The TG
structure seems to be an important factor for the
bioavailability and absorption process of fats in the
Rrst weeks of life. It has been suggested that unsatur-
ation of TG fatty acids does not affect pancreatic
lipase levels, whereas the chain length of the constitu-
ent fatty acids does appear to exert an effect. The
distribution of fatty acids within the glycerol mol-
ecule might also effect absorption, as it has been
shown to regulate TG hydrolysis to 2-monoacyl-
glycerol and fatty acids.
The physiological effects of TG structure and
composition of the diet are more relevant in the
intestine and liver, the most actively involved tissues
in TG synthesis and secretion. The speciRcity of
lipolytic enzymes for fatty acids acylated at the sn-1
position of the glycerol molecule affects the re-
synthesis of TGs either in enterocytes or hepatocytes.
After this re-synthesis, TGs are transported via lipo-
proteins, to peripheral tissues so that their con-
stituent fatty acids are incorporated into the
cellular lipid metabolism. However, until recently
little has been done on TG from these tissues or
lipoproteins.
Thirty-one molecular species of TG from rat liver
have been identiRed by RP-HPLC with an ELSD by Figure 5 RP-HPLC of rat liver (A) and very low density
Perona et al. (Figure 4). Oleic, linoleic or palmitic lipoprotein (B) triglycerides. Supelcosil LC-18 column (250;
acids formed the main TGs in the rat liver. Rat liver 4.6 mm), coupled to a Varex ELSD II light-scattering detector;
solvent, linear gradient of 10}90% isopropanol in acetonitrile
had also been investigated for TG molecular species
at flow rate 1 mL min\1. (Reproduced with permission from
by other workers using ELSD, resolving a lower num- Yang LY, Kuksis A, Myher JJ and Steiner G (1995) Journal
ber of TGs. Yang et al. observed that the fatty acid of Lipid Research 36: 125. Copyright Journal of Lipid
composition and the major molecular species of TG Research.)

Figure 6 RP-HPLC of rat adipose tissue triglycerides. Two
Supelcosil LC-18 (250;4.6 mm) columns in series, with UV de-
tector; solvent, 35 : 65% isopropanol in acetonitrile at a flow rate
of 2 mL min\1. Rats were fed a diet containing linoleic acid at
a level of (a) 0.01 g kg\1, (b) 24 g kg\1. (Reproduced with per-
mission from Huang YS, Lin X, Sminth RS et al. (1992) Lipids 27:
711. Copyright AOCS Press.)
Conclusion
The chromatographic analysis of TGs has undergone
great advances in the last few years. Among other
factors, the advent of reversed-phase HPLC and the
emergence of the evaporative light-scattering detector
(ELSD) allow the resolution of many TG applications
in vegetable and animal samples. Nevertheless, the
III / TRIGLYCERIDES / Liquid Chromatography 4411
shortcomings of TG identiRcation need to be im-
proved. Mass spectrometry (MS) is a helpful tool for
this purpose. However, it is still difRcult for many
researchers to incorporate such a technique in their
laboratories. The drawback of the incomplete resolu-
tion of critical pairs of TGs with the same ECN and
the stereo speciRc analysis of TGs are the two chal-
lenges which investigations will have to address in the
near future.
See Colour Plate 123.
See also: II/Chromatography: Liquid: Detectors: Ultra-
violet and Visible Detection; Mechanisms: Normal Phase;
Mechanisms: Reversed Phases. III/Lipids: Gas Chro-
matography; Thin-Layer (Planar) Chromatography. Oils,
Fats and Waxes: Supercritical Fluid Chromato-
graphy. Silver Ion: Liquid Chromatography; Thin-Layer
(Planar) Chromatography.
Further Reading
Aitzetmuller K (1997) Recent developments in the analysis
of food lipids and other lipids. Ol. Corps Gras, Lipides
4(1): 8}19.
Beare-Rogers JL (1983) Advances in Nutrition Research,
vol. 5. London: Plenum Press.
Breckenridge WC (1978) In: Kuksis A (ed.) Handbook of
Lipid Research, vol. I. New York: Plenum Press.
Christie WW (1987) High-Performance Liquid Chromato-
graphy and Lipids: A Practical Guide. Oxford:
Pergamon Press.
Geeraert E and Sandra P (1987) In: Kuksis A (ed.)
Chromatography of Lipids in Biomedical Research and
Clinical Diagnosis. Amsterdam: Elsevier Science Pub-
lishers.
Hammond EW (1982) In: Macrae R (ed.) HPLC in Food
Analysis. London: Academic Press.
Hammond EW (1993) In: Hammond EW (ed.) Chromato-
graphy for the Analysis of Lipids. Florida: CRC.
LitchReld C (1972) Analysis of Triglycerides. New York:
Academic Press.
Marini D (1992) In: Nollet LML (ed.) Food Analysis by
HPLC. New York: Marcel Dekker.
Nikolova-Damyanova B (1997) In: Christie WW (ed.) Ad-
vances in Lipid Methodology. 4. Dundee: The Oily
Press.
Ruiz-GutieH rrez V and BarroH n LJR (1995). Methods for the
analysis of triacylglycerols. Journal of Chromatography
B 671: 133d168.
4412 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography
Thin-Layer (Planar) Chromatography
P. E. Wall, Merck Ltd, Poole, Dorset, UK
Copyright ^ 2000 Academic Press
Introduction
Triglycerides (TGs) belong to the larger group of
natural products called lipids. A lipid is one of
a wide range of natural materials that are generally
based on fatty acids or closely related compounds, are
insoluble in water, but soluble in organic solvents.
Lipids that are solid at ambient temperature are
termed fats whilst those that are liquids are de-
scribed as oils. Lipids can be split into two groups;
neutral lipids, which include acylglycerols, fatty
acids, alcohols and waxes, and polar lipids, which
include phospholipids and glycolipids.
TGs make up a major part of the group of neutral
lipids and are found in an extensive range of animal
and vegetable fats, seed and plant oils. Lipids are
present in body organs and Suids. They also Rnd their
way into many other food products, e.g. frying oils,
salad dressings, margarine, butter, and various other
types of spreads.
Figure 1 Structures of different types of underivatized triacyl-
glycerides.
TGs are fully acylated derivatives of the trihydric
alcohol, glycerol. Hence more accurately they should
be described triacylglycerides, but quite often
they are commonly called triglycerols or triacyl-
glycerols. The structure of this group of lipids is
shown in Figure 1. Each arm of the glyceride is an
ester of a fatty acid. This chain can be fully saturated
or it can vary in unsaturation. Some natural triacyl-
glycerides have the same three ester groups, e.g. tri-
sterin (18 : 0), tripalmitin (16 : 0), triolein (18 : 1),
trilinolein (18 : 2), and trilinolenin (18 : 3). More
usually the fatty acid esters are different on each
glycerol backbone leading to many variations de-
pendent on the number of fatty acids available and on
the degree of unsaturation.
Degradation
Triacylglycerides are susceptible to hydrolysis with
the resulting products being free fatty acids (FFAs),
diacylglycerides (DGs), and monoacylglycerides
(MGs). If the fatty acid esters are formed from
unsaturated fatty acids, then the susceptibility to
oxidation and hydrolytic degradation is increased.
Unsaturated triacylglycerides undergo oxidative
breakdown involving the formation of free radicals.
This process can occur just in the presence of atmos-
pheric oxygen at ambient temperature, although the
process will be accelerated by increase in temper-
ature. The primary products are initially allylic
hydroperoxides that then undergo a series of complex
reactions to form volatile compounds including al-
dehydes, ketones, alcohols, esters, and short chain
fatty acids (see Figure 2).
Hydrolytic breakdown normally occurs at elevated
temperatures and is often catalysed by enzymes; e.g.
lipases. This degradation results in di- and monoacyl-
glycerides and long chain fatty acids (see Figure 3).
Thin-Layer Chromatography
Without doubt thin-layer chromatography (TLC) is
one of the simplest and most widely employed tech-
niques in the analysis of lipids. Over the past 30 years,
planar chromatography on a silica gel matrix has
proved to be the most practical method of distin-
guishing between lipid classes including glycolipids,
acylglycerols, phospholipids, sphingolipids, and ether
lipids. The continued interest in improving the separ-
ation capabilities for lipids using TLC is reSected in
the recently published literature.
 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography
Figure 2 One possible route for autoxidation of unsaturated
lipids. R"carbon chain length linked to the glyceryl backbone via
a COO linkage, R1"a saturated carbon chain. Other degradation
routes can occur and result in mixtures of aldehydes, ketones,
alcohols, esters, and acids. A route to aldehydes and alcohols is
shown.
Normal Phase Separations
Of all the stationary phase adsorbents available, silica
gel 60 has proved to be the adsorbent of choice for
the rapid separation of triacylglycerides and their
Figure 3 Hydrolytic reaction of a typical triacylglyceride. The
reaction is usually catalysed by the presence of lipase. The values
of n, m, and p are variable depending on the particular triacyl-
glyceride. In the majority of cases, n"14 (palmitic) or 16 (oleic),
and m and p are 4 and 7, respectively (linoleic).
4413
Table 1 Separation of acylglyceride classes on normal phase
silica gel. Mobile phase: n-hexane}diethyl ether}acetic acid
(70 : 30 : 1, v/v)
Glyceride RF value (approx.)
TG 0.70
FFA 0.45
1,2-DG 0.26
1,3-DG 0.23
MG 0.05
hydrolysis products, including any hydrolytic damage
that may have occurred as a result of lipolysis. Nor-
mal phase separations enable the resolution of neutral
lipids into TGs, DGs, MGs and FFA. The solvent
used is normally a mixture of diethyl ether and
hexane, pentane or a low boiling petroleum spirit.
The ratio is in the range of 15}30% v/v diethyl ether
in the saturated hydrocarbon. A modiRcation with
a small amount of formic or acetic acid (about 1%,
v/v) helps to improve resolution and is necessary
where any organic acids (fatty or otherwise) are sus-
pected as being present in the sample. This aids in
suppressing ionization of any FFA. The order of
retention on the chromatographic layer of sample
components tends to follow the expected adsorp-
tion/partition type mechanisms. The triacylglycer-
ides, being the least polar, exhibit the least retention
and hence migrate well up towards the solvent front
with any sterol esters that may be present in the
sample. Perhaps surprisingly these are closely fol-
lowed by the fatty acids. DGs migrate much more
slowly and monoglycerides show only the minimal
movement from the origin. One of the interesting
features of the normal phase separation of TG is the
ability to clearly resolve the 1,3 and 1,2 isomers of
DGs that may be present (see Table 1). In order to
attain adequate migration of the MG from the origin,
the TLC plate can be developed twice with diethyl
ether to a solvent distance of 20}30 mm with inter-
mediate drying. This enables sufRcient migration of
the MG from any more polar lipids on or near the
origin. Following this separation step, the standard
development can be carried out as before. Sometimes
the TG zone on the chromatogram may appear some-
what elongated and even a partial resolution of com-
ponents may be observed. This is due to the variation
in the saturation and fatty acid ester chain length of
the TG present.
For densitometric evaluation, silica gel 60 high
performance thin-layer chromatography (HPTLC)
plates can be used and samples applied using an
automated band applicator. After development and
detection with an appropriate reagent the chromato-
4414 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography
Table 2 Separation of acylglyceride classes on normal phase
silica gel impregnated with 5% w/v sodium carbonate. Mobile
phase: diethyl ether}n-hexane}methanol (65 : 35 : 3, v/v)
Glyceride RF value (approx.)
FFA 0.00
MG 0.18
1,3-DG 0.79
1.2-DG 0.85
TG 0.98
graphic tracks can be scanned at set wavelengths
using a spectrodensitometer. Using external stan-
dards on the layer, accurate quantiRcation of the
separated components can be obtained.
To a limited extent the enzymic hydrolysis of TG in
food products can be followed. Sodium carbonate-
impregnated (5%, w/v) silica gel plates are used.
Before the sample is applied to the layer, the enzy-
matic action is terminated by the addition of sodium
dodecyl sulfate (SDS). The chromatogram is de-
veloped for a very short period (about 1 minute) with
diethyl ether}methanol (97 : 3, v/v), which results in
all the acylglycerides migrating with the solvent
front and the fatty acids remaining at the origin.
A modiRcation to this solvent system; diethyl ether}n-
hexane}methanol (65 : 35 : 3, v/v) results in a separ-
ation of all the various acylglycerides from the fatty
acids (see Table 2).
Variations on the above mobile phases have been
developed depending on the type of separation re-
quired and the origin of the sample. Table 3 lists
a number of solvent mixtures that have proved suc-
cessful for various types of TG separations.
When lipid mixtures prove to be complex, two-
dimensional systems can be helpful in resolving the
large number of components. Although seldom used,
there are instances where two-dimensional TLC has
enabled the separation of mixed acylglycerides from
steryl esters, methyl esters and fatty acids. Sample
components can be resolved using a Rrst development
Table 3 Solvent mixtures recommended for the separation of acylglycerides on normal phase silica gel
Acylglycerides Adsorbent
TG, DG, MG, FFA (as classes) Silica gel 60
TG, DG, MG, FFA from plasma Silica gel 60
Human aortic lipids including unsaturated TG HPTLC silica gel
TG, FFA, amides and cholesterol Silica gel
TG containing oxygenated fatty acid methyl esters Silica gel 60
TG containing epoxy and hydroxyl fatty acids Silica gel 60
with n-hexane}diethyl ether (80 : 20, v/v). This is
followed by plate drying and development in the
second dimension at 903 to the Rrst using a solvent
mixture composed of n-hexane}diethyl ether}meth-
anol (70 : 20 : 10, v/v). If more polar lipid components
are present, then an alkaline-based solvent mixture
is recommended for the development in the Rrst
dimension (chloroform}methanol}0.88 ammonia
solution}water [65 : 30 : 2 : 2, v/v]) followed by an
acid-based one in the second dimension (chloroform}
methanol}acetic acid}water [100 : 15 : 15 : 3.5, v/v]).
Other Modi\cations to Normal Phase Separations
Orthoboric acid Orthoboric acid-impregnated sil-
ica gel layers are used in the TLC of TG to prevent
acyl migration from the 2 to the 1 or 3 position on the
glycerol backbone. The speed of migration is depen-
dent on the acyl moiety. It is therefore important
that this effect is prevented from occurring in an
analysis of DG and MG. Orthoboric acid does this by
weak interaction and complex formation with the
free hydroxyl groups on the acylglycerides.
Precoated TLC plates can be impregnated with
orthoboric acid (15% w/v) dissolved in water}meth-
anol (25 : 75, v/v). Either dipping or spraying the
plate in the solution gives satisfactory results. The
plates are dried after impregnation for 30 minutes at
1103C. Separations can then be performed with
methanoldchloroform (3 : 97, v/v) as solvent.
Silver nitrate Silver nitrate or argentation TLC has
been used extensively for the analysis of triacylglycer-
ides. The reason for its popularity is that silver nitrate
has a retarding effect on acylglycerides that contain
unsaturated fatty acid ester moieties. The silver ni-
trate forms complexes with varying strength of bond-
ing by interaction with the
double bonds. The more
double bonds present, the greater the complexation
and the less accessible the double bonds, the less the
complexation. Hence, polyunsaturated glycerides
and FFA will be more retained than their oligo-
Mobile phase
Diethyl ether}n-hexane}acetic acid (80 : 19 : 1, v/v)
Toluene}diethyl ether}ethyl acetate}acetic acid
(8 : 1 : 1 : 20, v/v)
n-Hexane}diethyl ether}acetic acid (65 : 35 : 1, v/v)
Toluene}diethyl ether}ethyl acetate}acetic acid
75 : 10 : 13 : 1.2, v/v)
n-Hexane}diethyl ether (30 : 70, v/v)
n-Hexane}diethyl ether (1 : 1, v/v)
 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography
Table 4 Solvent mixtures that have proved satisfactory for the separation of acylglycerides on silica gel impregnated with silver
nitrate
Sample containing acylglycerides
Soybean and fish oils
Palm oil, lard, beef tallow, cocoa butter and groundnut oil
Lard and cocoa butter
Triacylglyceride standards
Positional isomers of triacylglycerides, lard, and sunflower oil
Orange seed oil
unsaturated counterparts whilst any saturated com-
ponents remain unaffected. As accessibility of the
double bonds also has a bearing on the degree of
complexation, cis and trans isomers can be separated
and acylglycerides of fatty acids that only differ in the
positional location of the double bond can often be
resolved.
Impregnation of silica gel 60 plates can be achieved
with silver nitrate (10% w/v) dissolved in water}
methanol (15 : 85 v/v). Precoated TLC and HPTLC
plates are dipped in the silver nitrate solution for
10}20 s. After draining, the plates are dried in air
under fume extraction and then heated for activation
at 803C for 20 minutes.
Argentation TLC has proved to be of immense
importance in a number of research areas including
plant-derived oils and confectionery fats. In fact, the
technique has been proposed as a method for the
determination of 2-oleo-1,3-disaturated triacylglycer-
ides in cocoa butter as a part of the necessary analysis
in the manufacture of chocolate. As expected, the
separation of triacylglycerides follows the order of
the number of double bonds with the least un-
saturated being the least retained. However, if the
unsaturation is in the 2-position, then there is some
hindrance to the formation of the silver complex and
some differentiation in the separation between the 2-
and 1- or 3-position can be observed. As an example
of this, it is possible to separate 2-oleo-1,3-distearin
(SOS) and 3-oleo-1,2-distearin (SSO). As the interac-
tion of the silver ion with the 2-position isomer is
more sterically hindered, this is the one which is less
retained on the layer and hence has the slightly higher
RF value. Of course, not only do TG vary in the
amount and position of unsaturation, but also both
cis and trans isomers of the same fatty acid esters
occur. Examples of this are cis-9-octadecanoic acid
(oleic acid) and trans-9-octadecanoic acid (elaidic
acid). If any trans isomers are present, these are less
retained than the cis isomers. Structurally this would
be expected as the cis double bond is more accessible
to the large silver ion, and hence complexes more
readily. The general order of separation starting from
the least retained and representing the fatty acid
4415
Mobile phase
Diethyl ether}n-hexane (8 : 92, v/v)
Chloroform}cyclohexane (1 : 1, v/v)
Chloroform}benzene}diethyl ether (70 : 30 : 1.5, v/v)
Benzene}diethyl ether (85 : 15, v/v)
Chloroform}methanol (99 : 1,v/v)
Petroleum ether (40}603C)}acetone (100 : 7, v/v)
chains of the TG as 0, 1, 2, or 3 depending on the
number of double bonds is: 000, 001, 011, 002, 111,
012, 112, 022, 003, 122, 013. 222, 113, 023, 123,
223, 033, 133, 233, 333.
For the common C18 chain, the fatty acid chains
would be stearic acid (18 : 0), oleic acid (18 : 1), lin-
oleic acid (18 : 2), and linolenic acid (18 : 3). Whilst
C18 represents one of the most common chain lengths,
shorter and longer chain lengths do occur and this
increases the complexity of the problem. Palmitic
acid (16 : 0) occurs more, widely naturally than
stearic acid (18 : 0), being present in almost all veg-
etable fats, Rsh oils and milk fats. Fortunately for the
analyst, the unsaturated versions of the C16 chain
such as palmitoleic acid (16 : 1) are only minor com-
ponents of seed oils and animal fats and only take on
signiRcant proportions in Rsh oils. Typical solvent
mixtures used for the development of silver nitrate
chromatograms are given in Table 4.
Both symmetrical and unsymmetrical TG are pres-
ent in lard and cocoa butter and these can be
separated effectively with two-dimensional argenta-
tion TLC. Unsymmetrical TG occur where the carbon
chain on position 1-, 2- or 3- on the glycerol back-
bone vary in length. Examples of this are: POS (pal-
mitin (16 : 0), olein (18 : 1), and stearin (18 : 0) or
PPO (palmitin (16 : 0), palmitin (16 : 0), and olein
(18 : 1). The separation is carried out on a dual sta-
tionary phase plate. One section of the plate is coated
with a thin strip of reversed-phase silica gel, and the
rest is coated with a normal phase silica gel.
The sample is applied to the reversed-phase strip and
the chromatogram developed using acetonitrile}
acetone (80 : 20, v/v) as mobile phase. The normal-
phase silica gel portion of the plate is impregnated
with silver nitrate and the second dimension develop-
ment then proceeds with a mobile phase composed of
chloroform}benzene}diethyl ether (70 : 30 : 1.5, v/v).
Reversed-phase Separations
The resolution of TG on reversed-phase layers is
usually noticeably better than that on normal-phase
TLC. Although separation of acylglycerides, and FFA
4416 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography

Table 5 Stationary and mobile phase conditions for the separation of acylglycerides and free fatty acids on reversed-phase silica gel
plates

Sample containing acylglycerides Stationary phase Mobile phase

Most seed oils (e.g. sunflower,
olive, rapeseed oils)
Most seed oils
Most seeds oils, DG, MG, and FFA
HPTLC-silica gel RP18 glass plates
HPTLC silica gel RP18 glass plates
HPTLC silica gel RP18 glass plates
Dichloromethane}acetic acid}acetone (1)
(20 : 40 : 50, v/v)
Chloroform}acetonitrile}acetone (2)
(20 : 40 : 50, v/v)
Dichloromethane}ethyl
acetate}methanol}acetic acid (3)
(27 : 22 : 38 : 12, v/v)

into respective groups is possible using normal-phase
silica gel, reversed-phase layers will resolve individual
members of these groups into sharp, often well-
deRned, zones. However, it is only possible to detect
unsaturated acylglycerides and fatty acids on re-
versed-phase layers. This may initially be viewed as
a limiting feature of the technique, but as the separ-
ation number even with HPTLC layers in one dimen-
sion is rarely more than 20, there is always only
a Rnite length of chromatographic layer available in
which the separation can occur. Hence, as the
saturated lipids are undetectable, there is more separ-
ation capacity available for unsaturated compounds.
The reversed-phase separation of TG has resulted
in a method for the identiRcation of fatty oils. The
protocol is given in the BP98 appendix XN and the
EP97 (2.3.2) and shows a typical chromatogram ob-
tained on HPTLC RP18 layers for a number of seed
oils. The test method acts as an identitiRcation for
a wide range of oils as each has a TLC Rngerprint of
unsaturated acylglycerides unique to itself. Solvent
mixtures that give good separation reproducibility for
reversed-phase are given in Table 5. Solvent mixtures
1 and 2 are comparable, but solvent mixture 3 gives
similar resolution for the TG at lower RF values and
also good resolution for many of the DGs, MGs and
FFAs. This solvent mixture therefore has been used
effectively for investigations into the deterioration of
frying oils. Figure 4 shows a typical chromatogram of
a blended frying oil.
As the separation has occurred almost purely by
partition, it is possible to relate the positions on the
chromatogram of the acylglycerides to the degree and
the position of the unsaturation in the molecule. This
then enables the prediction of the position of acyl-
glycerides on the chromatogram and aids in the

Figure 4 Separation of unsaturated acylglycerides and FFA on HPTLC silica gel RP18 glass plates. Mobile phase: dichloro-
methane}ethyl acetate}methanol}acetic acid (27 : 22 : 38 : 12, v/v). Detection with 1% w/v phosphomolybdic acid in ethanol. Plate
heated to 1003C for 5 minutes. Scanned at 700 nm with a spectrodensitometer. Sample: degraded blended frying oil. Peaks 1}5 are
triacylglycerides, peaks 6}8 are diacylglycerides, peak 10 is a free fatty acid and peaks 11 and 12 are monoacylglycerides.
 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography 4417

identiRcation of unknowns. The acylglycerides are
separated according to the equivalent carbon number
(ECN). This is deRned as:
ECN"CN!2n
where CN"carbon number, n"number of double
bonds.
However, this does not take into consideration
the position of the double bonds. For this reason
in HPLC an adjustment has been made to this
equation:
ECN"CN!d1n1!d2n2!d3n3
where n1, n2, and n3 are the number of double bonds
attributable to oleic, linoleic and linolenic acids, re-
spectively.
The values d1, d2 and d3 are calculated by means
of reference triacylglycerides. They are: d1"2.60,
d2"2.35 and d3"2.17.

Figure 5 Separation of unsaturated triacylglycerides on an HPTLC silica gel RP18 layer impregnated with silver nitrate (5% w/v
solution). Mobile phase: dichloromethane}ethyl acetate}methanol}water}acetic acid (25 : 20 : 35 : 6 : 6, v/v). Detection with 1%
phosphomolybdic acid in ethanol. Plate heated to 1303C for 10 minutes. Scanned at 700 nm with a spectrodensitometer. Sample: Fresh
blended frying oil. Peaks 3 and 6 are triolein and trilinolein respectively. Other peaks are other unsaturated triglycerides, sterols, and
antioxidants unidentified.
4418 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography

Figure 6 Separation of unsaturated triacylglycerides on an HPTLC silica gel RP18 layer impregnated with silver nitrate (5% w/v
solution). Separation conditions as in Figure 5. (Corn oil) Peaks 3 and 6 are triolein and trilinolein respectively. Other peaks not
identified. (Soya oil) Peaks 2 and 5 are triolein and trilinolein respectively. Other peaks not identified. (Sunflower oil) Peaks 2 and 4 are
triolein and trilinolein respectively. Other peaks not identified. (Olive oil) Peaks 2 and 4 are triolein and trilinolein respectively. Other
peaks not identified.

Silver nitrate As with normal phase silica gel, it is
possible to modify reversed phase silica gel with silver
nitrate. Pre-coated reversed phase silica gel layers can
either be impregnated or, where applicable, silver
nitrate can be added to the mobile phase. A suitable
impregnating reagent can be prepared with silver
nitrate (5% w/v) dissolved in water}methanol
(10 : 90, v/v). The separations obtained indicate
a much stronger though limited resolving capability
than is possible with unmodiRed reversed-phase
layers (see Figure 5). The triacylglycerides are separ-
ated over a much wider RF range enabling more
marked differences to be observed in the chromato-
grams for a number of plant seed oils. A comparison
of these is shown in Figure 6. However, as mentioned
previously the technique does have limitations. DGs,
MGs, and FFAs will all be found at or near the
solvent front if the mobile phase has been adjusted to
give maximum resolution for the triacylglycerides.
Detection is also not as sensitive as for the corre-
sponding reversed-phase layers (about a four-fold re-
duction).
Detection Methods
Detection of acylglycerides relies upon the use of
chemical reagents as any UV absorbance is weak and
none of these neutral lipids show any natural Suores-
cence (either in the visible or UV spectrum). Most
chemical methods rely on reduction or charring tech-
niques for acylglycerides. However, these can still be
quite sensitive with the limit of detection usually
being in the nanogram range. FFAs are much more
reactive and hence a much wider range of detection
reagents are available. The same applies to any degra-
dation products due to oxidation where aldehydes,
ketones or esters may have formed.
 III / TRIGLYCERIDES/Thin-Layer (Planar) Chromatography 4419

Detection on Normal Phase Layers
Visualization of both saturated and unsaturated acyl-
glycerides is easily achieved on normal phase silica gel
layers. However, it should be borne in mind that if the
detection reagent is a charring one, then care must be
taken when commercial pre-coated plates or sheets are
used, particularly with sulfuric acid or chlorosulfonic
acid. This is because in order to obtain good reproduci-
bility and abrasive resistance, the pre-coated layers
contain a small percentage of a polymeric organic
binder. Unfortunately this can also char along with the
sample components limiting the contrast between
chromatographic zones and the background. How-
ever, if the temperature and duration of heating are
carefully controlled, good results can be obtained.
Iodine vapour gives very good sensitivity, giving
yellow-brown zones on a pale yellow background.
The unsaturated compounds are stable for a much
longer period of time than the saturated ones. The
interaction with the
double bonds forms an iodine
complex that is much more stable than the adsorption
of iodine by the saturated compounds, which is re-
versible. These results can be made much more per-
manent by spraying the plate with soluble starch
solution that forms dark blue complexes on the zones
where iodine has been adsorbed.
Phosphomolybdic acid reagent (1}5%, w/v solu-
tion in ethanol) is probably the most popular reagent
for lipid detection and gives a limit of sensitivity of
50}200 ng, depending on the glyceride. Zones appear
after heating as blue-grey on a yellow background.
This yellow background can be destained by expo-
sure to ammonia vapour.
Other reagents that have been used to good
effect are listed along with the above in Table 6.
Detection on Reversed-phase Layers
Of all the reagents listed in Table 6, the Rrst four can
also be used on reversed-phase layers. However, they
can only be used to detect unsaturated acylglycerides
and FFAs. Sensitivity, however, is on a par with
normal phase layers with both iodine vapour and
phosphomolybdic acid giving the best results. Char-
ring reagents are best avoided as the background
easily chars as well due to the fact that it is bonded
with an aliphatic carbon chain.
Detection on Argentation-modi\ed Phases
On commercial pre-coated layers, phosphomolybdic
acid gives results comparable with those obtained on
normal or reversed-phase plates. There is usually
a lack of background staining which improves the
contrast. The use of ammonia vapour though is not to
be recommended as this reacts with the excess silver
nitrate and a brown speckled background appears.
For the reversed-phase plates, heating is required at
a higher temperature (1503C for 10 minutes) to detect
the unsaturated zones.
Some charring techniques have been used for
home made normal phase silver nitrate modiRed
layers but these involve the use of very aggressive
chemical reagents.

Table 6 Detection reagents suitable for the visualization of acylglycerides and free fatty acids on normal silica gel layers (non-
commerical)

Detection reagent Acylglyceride detected Observation

Iodine vapour Both saturated and unsaturated Yellow/brown zones

Phosphomolybdic acid spray or dip Both saturated and unsaturated Blue-grey zones on yellow background
followed by heating at 100}1203C for and FFA
10 minutes
Manganese (II) chloride/sulfuric All acylglycerides and FFA Brown zones on white background
acid [(0.2 g manganese chloride in
water (30 mL) } methanol (30 mL) plus
sulfuric acid (2 mL)] Heat at 100}1203C
for 10 minutes
Copper (II) acetate/sulfuric acid All acylglycerides and FFA Brown-grey zones on a white background
(copper acetate (3% w/v) in phosphoric
acid (10% v/v)) Heat at 100}1203C
for 10 minutes
Sulfuric acid (10}20% v/v) Heat Both saturated and unsaturated Black or grey zones
at 120}1503C for 15 minutes and FFA
Berberine solution (10 mg/100 mL ethanol) All acylglycerides Yellow fluorescent zones under UV at
360 nm
4420 III / TRIGLYCERIDES/Thin Layer (Planar) Chromatography
Future Developments
It seems unlikely that major developments will occur
in the future with improvement of the separation
method of acylglycerides on normal and silver nitrate
impregnated silica gel. However, the use of the newer
commercially available smaller particle size (&4
m)
and spherically shaped particles will result in an im-
provement in resolution, sensitivity and scanned peak
shape of chromatographic zones. Automated multiple
development (AMD) has already proved to be an
excellent analytical tool for focusing zones in lipid
separations but is still very much in its infancy with
a big potential available for acylglyceride separations.
There is no doubt that reversed-phase HPTLC pro-
vides a reliable method for following the breakdown
of oils and fats in use. The commercial possibilities
here have yet to be fully exploited. There is still much
work to be done in developing reliable, but simple,
and rapid methods of analysis for triacylglyceride
breakdown. Presently available HPTLC procedures
not only have the potential to analyse and quantify
the total FFA, but also to separate these and deter-
mine them individually. Some quantitative work on
acylglycerides has already been accomplished, but in
the future it should be possible to quantify far more.
One of the present drawbacks has been the lack of
availability of pure standards, particularly for many
of the unsaturated acylglycerides. This is not alto-
gether surprising as many are unstable and need to be
kept deep frozen to avoid degradation.
The use of TLC for the analysis of triacylglycerides
has further potential in the quantiRcation of other
organic species that may be present in oils and fats.
Some oils naturally contain tocopherol, which acts as
an antioxidant, and other oils may have this, or other
antioxidants, added to extend their life. Sterols can
also be present. As TLC requires little sample prep-
aration before application to the chromatographic
layer, the technique is usually quite easy (and many
samples can be analysed at the same time). Many of
these other compounds can be separated and deter-
mined quantitatively. The future of TLC for the anal-
ysis of triacylglycerides shows considerable potential.
See also: II /Chromatography: Thin-Layer (Planar):
Densitometry and Image Analysis; Layers; Spray Re-
agents. III/Impregnation Techniques: Thin-Layer
(Planar) Chromatography. Lipids: Gas Chromato-
graphy; Liquid Chromatography; Thin-Layer (Planar)
Chromatography. Oils, Fats and Waxes: Supercritical
Fluid Chromatography. Silver Ion: Liquid Chromatogra-
phy; Thin-Layer (Planar) Chromatography. Triglycerides:
Liquid Chromatography. Appendix 17: Thin-Layer
(Planar) Chromatography: Detection.
Further Reading
Dobson G, Christie WW and Nikolova-Damyanova
B (1995) Silver ion chromatography of lipids and fatty
acids. Journal of Chromatography B 671: 197}222.
Gunstone F (1996) Fatty Acid and Lipid Chemistry. Lon-
don: Blackie Academic and Professional.
Hammond EW (1993) Chromatography for the Analysis of
Lipids. London: CRC Press.
McSavage J and Wall PE (1998) Optimization of a mobile
phase in reversed-phase HPTLC for the separation of
unsaturated lipids in vegetable oils degraded during fry-
ing. Journal of Planar Chromatography 214}221.
Myher JJ and Kuksis A (l995) General strategies in
chromatographic analysis of lipids. Journal of
Chromatoaraphy B 671: 3}33.
Nikolova-Damyanova B and Amidzhin B (1991) Den-
sitometric quantiRcation of triglycerides. Journal of
Planar Chromatography 397}401.
Olsson NU (1992) Advances in planar chromatography for
the separation of food lipids. Journal of Chromatogra-
phy 624: 11}19.
Ritchie AS and Lee MH (1987) A note on: Triglyceride
analysis using silver nitrate and 2-phase 2-dimensional
thin-layer chromatography. Recent Advances in Thin-
layer Chromatography. London: Plenum Press.
Ruiz-GutieH rrez V and Barron LJR (1995) Methods for the
analysis of triacylglycerols. Journal of Chromatography
B 671: 133}168.
Salia SK and Das SK (1996) A simple densitometric method
for the estimation of polar and non-polar lipids by thins
layer chromatography with iodine vapour, visualization.
Journal of Liquid Chromatography and Related Tech-
nologies 19: 3125}3134.
Touchstone JC (1995) Thin-layer chromatographic proced-
ures for lipid separation. Journal of Chromatography
B 671: 169}195.
Traitler H, Jacolet C and Winter H (1990) Triacylglycerol
structure elucidation planar chromatographic separ-
ation of randomly formed diacylglycerols. Journal of
Planar Chromatography 177}180.
 III / ULTRASOUND-ASSISTED METAL EXTRACTIONS
ULTRASOUND-ASSISTED METAL
EXTRACTIONS
C. Bendicho and I. Lavilla, Universidad de Vigo,
Vigo, Spain
Copyright ^ 2000 Academic Press
Introduction
Ultrasonic energy has been used for a wide variety of
applications in industry, medicine and science. In the
analytical chemistry Reld, most applications lie in
the ability of ultrasound to extract compounds from
the solid matrix. Solid}liquid extraction with the use
of ultrasonic energy (i.e. ultrasound-assisted extrac-
tion) has been successfully applied for many years as
a sample pretreatment method to extract organic
compounds from matrices to which they are weakly
bound (e.g. environmental samples). Sonication
methods have been compared to other methods for
pretreatment of solid samples (e.g. Soxhlet extrac-
tion, accelerated solvent extraction and supercritical
Suid extraction), being competitive to them owing
to its simplicity, efRciency and ease of use. More-
over, sonication methods do not involve the use of
high temperatures, pressures or concentrated and
harmful chemicals. Usually, ultrasound has been ap-
plied to the sample with the use of ultrasonic cleaning
baths. Ultrasonic cleaning baths are readily available,
large numbers of samples can be simultaneously
treated and low-cost instrumentation is involved, but
they lack the capability of transmitting sufRcient
ultrasonic power to produce the desired effects on the
sample.
More recently, ultrasound-assisted extraction has
been applied to the separation of inorganic com-
pounds and metal ions from the matrix, to facilitate
their analytical determination, and to avoid tradi-
tional sample pretreatment methods such as dry or
wet ashing, which involve tedious and time-consum-
ing treatments with corrosive reagents.
Other application areas of ultrasound-assisted
extraction include the selective extraction of different
physicochemical forms of elements for speciation. In
this case, advantage is taken of the nondestructive
character of ultrasound treatments which, under suit-
able conditions, maintain the integrity of the extrac-
ted species.
Finally, ultrasound can accelerate many sequential
extraction schemes which are traditionally applied
for metal partitioning in environmental samples such
as soils, sludges and sediments.
4421
Ultrasound-Assisted Extraction for
Metal Determination
Intensive sample pretreatment of biological, environ-
mental and industrial samples is frequently a neces-
sary requirement for elemental analysis, so that
ideally a matrix-free solution is obtained. Typically,
dry ashing or wet ashing methods involve the use of
high temperatures or corrosive reagents, usually un-
der pressure, which demands very stringent safety
conditions. However, a simple analyte separation
without matrix decomposition is enough for many
analytical techniques. Thus, atomic absorption spec-
trometry (mainly with the use of electrothermal
atomization) allows analytical determinations to
be carried out with minimum sample pretreatment
owing to the low dependence of the analytical signal
on the accompanying matrix as compared with other
techniques for elemental analysis.
Thus, in the authors laboratory, a number of ele-
ments have been quantitatively extracted from a large
variety of matrices when probe-type sonicators oper-
ated under optimized conditions are employed. Toxic
metals such as Cd and Pb can be easily extracted from
mussel tissue and other biological samples, the exact
extraction conditions depending on the metal. Cad-
mium could be quantitatively extracted from
a sample mass of 10 mg slurried in a 1.5 mL volume.
The sample has to be previously ground, the particle
size being a critical parameter in the case of Pb since
extraction efRciency diminishes for a particle size
large than 150
m. For Cu and Cd, extraction can be
achieved for a particle size larger than 200
m. The
presence of an acidic medium is an essential require-
ment for quantitative extraction to be attained. For
analytical techniques such as electrothermal atomic
absorption spectrometry (ETAAS), nitric acid is re-
commended since unlike hydrochloric acid it does not
form volatile compounds with analytes, which are the
origin of interferences. In addition, nitric acid com-
bined with the ultrasonic action promotes matrix
oxidation so that analyte extraction is facilitated.
Minimum acid concentration used for quantitative
extraction depends again on the analyte to be extrac-
ted. A nitric acid concentration as low as 0.05% v/v is
sufRcient for quantitative extraction of Cd, whereas
Pb requires at least 1% v/v nitric acid. Additional
parameters controlling the amount of ultrasonic
power delivered to the sample such as sonication time
4422 III / ULTRASOUND-ASSISTED METAL EXTRACTIONS

and vibrational amplitude of the probe (expressed as Table 1 Percentage of metal extracted from certified reference
a percentage of the nominal power) should be opti- materials using ultrasound irradiated with a probe ultrasonic
processor
mized for best performance. Metals which are easy to
extract, such as Cd, require very short sonication Certified sample % Extraction
times, typically less than 1 min while stronger bound
metals such as Pb require 3}5 min. When using Cd a Cu b Cr c Pb c
a 100 W probe sonicator, at least a 10% amplitude is
BCR 278 Mussel tissue 101.8 82.4 42.0 94.2
necessary for extraction of Cd, while a 60% ampli-
NRCC DORM-2 Dogfish 93.0 93.2 2.7 }
tude is required for Pb. Sample mass is also an impor- muscle
tant variable; extraction is usually quantitative for NRCC DOLT-2 Dogfish liver 91.6 } 32.9 }
a mass of less than 20 mg suspended in 1.5 mL vol- BCR 60 Aquatic plant 101.4 102.4 30.2 101.2
ume. Although the preparation of suspensions in lar- BCR 145 R Sewage sludge 56.3 } 46.6 104
BCR 320 River sediment 75.4 } 15.0 69.0
ger volumes with larger amounts of ground material
NRCC TORT-2 Lobster } } 69.3 }
is also feasible, preparation of suspensions in hepatopancreas
autosampler cups is a more convenient way for BCR 482 Lichen } } 23.7 }
ETAAS when sample homogeneity is not a limiting GBW07605 Tea leaves } } - 95.5
factor. An experimental design applied to the extrac- a
Capelo JL, Lavilla I and Bendicho C (1998) Journal of Analytical
tion process of Cd and Pb conRrmed that soft sonica-
Atomic Spectrometry 13: 1285}1290.
tion conditions (minimum sonication time and b
Capelo JL, Filgueiras AV, Lavilla I and Bendicho C (1999)
amplitude) along with maximum particle size (e.g., Talanta 50: 905}911.
'200
m) could be used for quantitative solid}liquid c
Capelo JL, Lavilla I and Bendicho C (1999) Journal of Analytical
extraction of Cd provided that maximum acid con- Atomic Spectrometry 14: 1221}1226.
centration was used (e.g., 3% v/v). On the other
hand, Pb needed maximum sonication time, ampli-
tude and acid concentration together with minimum background absorbance caused by the small amount
particle size. The concentration of nitric acid proved of matrix released can be easily handled by the back-
to be the most critical factor for achieving quantitat- ground correction system (Table 1).
ive extraction. The use of other analytical techniques for detection
A study carried out with Pb as target analyte and after ultrasound-assisted extraction has also been re-
certiRed reference materials has shown the import- ported. For instance, Ashley has studied the extrac-
ance of using the appropriate ultrasonic processor tion of Pb from several standard reference materials
so that quantitative extraction is attained. Thus, an (SRMs) such as lead-based paint, urban particulate
ultrasonic cleaning bath is not suitable since only and river sediment followed by anodic stripping vol-
a fraction of the analyte is brought into solution tammetry (ASV). Analytical results were satisfactory
even using long sonication times (e.g., 60 min). When after ultrasonic extraction for 30 min using a 10%
comparing two probe-type sonicators (50 versus v/v nitric acid solution. ASV has been also used for
100 W), quantitative extraction was observed with determination of Pb in workplace air samples col-
the 100 W sonicator for all biological materials at- lected in the Reld using cellulose ester membrane
tempted. The explanation for the above results could Rlters. The Rlters were subjected to ultrasound under
lie in the greater ability of probe-type sonicators to the conditions given above for SRMs. An advantage
cause cavitation in the liquid medium, which results of ultrasound-assisted extraction methods over
in a more efRcient disruption of solid particles, so methods involving matrix decomposition (e.g.,
facilitating metal extraction. microwave-assisted digestion) is the ability to use
Incomplete extraction was observed for Pb and Cd them in the Reld, hence facilitating on-site analysis
from sediments, thereby indicating that matrix} with portable instruments.
analyte binding plays an important role in the The use of diluted acids for extraction can also
solid}liquid extraction process. This may be due to offer a simpliRed methodology for determination of
particle disruption being more difRcult with hard metals by Same atomic absorption spectrometry
materials such as sediments, so that unless the analyte (FAAS). In a comparison of Rve methods for pretreat-
is adsorbed on the surface the fraction of analyte ment of plant samples, Matejovic and Durackova
occluded inside the solid particles will not be brought found that extraction of metals could be accomp-
into solution, hence resulting in incomplete extrac- lished with 1 M hydrochloric acid in an ultrasonic
tion. bath. After sonication the extracts were Rltered so
Usually, the ultrasonic action will cause the matrix that no particulate material could clog the nebulizer.
to be partly extracted into the liquid medium, but In this case, the use of a nonoxidizing and complexing
 III / ULTRASOUND-ASSISTED METAL EXTRACTIONS
acid such as hydrochloric acid is perhaps more conve-
nient than other acids, since it avoids a Rnal evapor-
ation step to remove the excess of acid as is necessary
when concentrated acids are used for mineralization
in conventional digestion procedures. Incomplete re-
lease of P bound into organic compounds and Fe was
observed with this procedure.
Leaching of heavy metals from aquatic plants used
as environmental biomonitors has been performed
by ultrasound-assisted extraction with a 1% w/w
HCl#15% w/w HNO3 mixture. In this case, two
consecutive extractions were needed to quantitatively
extract Mn, Cu and Zn, the recovery of Cu being only
about 75%, with RSDs lower than 2.5%. In order to
obtain good analytical performance when applying
ultrasound-assisted extraction, all the variables
inSuencing the process should be borne in mind:
concentration of the suspension (i.e. sample mass and
extraction volume), particle size, sonication time,
sonication amplitude, type of acid and its concentra-
tion, and temperature. This last variable is seldom
considered for its inSuence on ultrasonic extractions.
Since most ultrasonic cleaning baths warm up slowly
during operation, many applications reported with
these devices for extraction use a pre-heated liquid so
that temperature is constant, hence improving repro-
ducibility. On the other hand, acoustic cavitation is
diminished on increasing the temperature above
503C, and consequently extraction efRciency is also
diminished. Thus, some workers have found only
partial extraction for some elements when using
a pre-heated ultrasonic bath or allowing the bath to
warm up during operation to a temperature higher
than 503C. Other workers have reported quantitative
extraction of metals such as Cd, Cu, Pb and Mn from
powdered biological samples when sonication is car-
ried out at 403C. Other extractants succesfully em-
ployed for solid}liquid extraction with an ultrasonic
cleaning bath include dilute HCl, HNO3 and H2O2.
Some procedures employing ultrasonic baths for
sample pretreatment were aimed at complete diges-
tion of the sample by the use of concentrated acids,
and therefore cannot be regarded as extraction
procedures.
Applications of Ultrasound-Assisted
Extraction for Element Speciation
Ultrasound extraction shows advantageous features
for element speciation. Organometallic species can be
extracted without changes in their integrity under
suitable extraction conditions. Both organic and
aqueous extraction media have been used for separ-
ation of organometallic and inorganic species from
4423
the solid matrix, most applications using ultrasonic
cleaning baths for extraction.
A recent application of ultrasound-assisted extrac-
tion with the use of a probe-type sonicator has been
reported for mercury speciation in combination with
Sow injection}cold vapour}atomic absorption spec-
trometry (FI-CVAAS) for detection. In this case,
a 400 mg portion of sample and 1}7 mL of 0.5}7 M
acid were placed in a centrifuge tube and sonicated at
a Rxed ultrasound amplitude for 1}5 min. Selective
extraction of methylmercury required less than 5 mL
of 2 M HCl, the extraction being quantitative
('95%) when the HCl volume was higher than
2 mL. The extraction could be accomplished using
ultrasound amplitude in the range 20}70% for
2}5 min. The optimization procedure was addressed
to selectively extract methylmercury from slurried
biological samples such as mussel tissue; inorganic
mercury extraction required higher HCl concentra-
tions. Both mercury species could be extracted with
5 mL of 5 M HCl and sonicating at 20}70% ampli-
tude for 3}5 min. Methylmercury was determined
using sodium tetrahydroborate(III) as reducing agent
whereas inorganic mercury was determined by selec-
tive reduction with stannous chloride in the extracts
containing both species. The limits of detection were
11 and 5 ng g\1 for methylmercury and inorganic
mercury, respectively. The repeatability (between-
batch precision), was in the range 5}10% for both
mercury species.
In a study on As extraction, similar distributions
of arsenicals (e.g., arsenobetaine, arsenocholine
and dimethylarsinic acid) were found in a compar-
ison between accelerated solvent extraction and
sonication. Nonpolar As is extracted with acetone
whereas polar As is extracted with 50% w/w meth-
anol.
Cr(VI) has been extracted from industrial hygiene
samples with an ultrasonic cleaning bath at 40}503C
for 1 h using alkaline solutions containing 0.05 M
(NH4)2 SO4}0.05 M NH3. The Cr(VI) was separated
from other cations present in the extract by retention
with an anion-exchange resin. Elution of Cr(VI) from
the resin was performed with a buffer solution at pH
8. The eluate was acidiRed with HCl and the complex
between Cr(VI) and 1,5-diphenylcarbazide was mea-
sured by Sow injection}UV/VIS detection. Deter-
mination of total Cr following ultrasonic extraction
was also feasible using a prior oxidation step with
Ce(IV) so that Cr(III) is converted into Cr(VI). This
simple and effective preparation method compared
favourably with other methods employing intensive
treatments leading to matrix decomposition (e.g.,
acid digestion) for determination of total Cr in Sy ash,
paint chips, etc.
4424 III / ULTRASOUND-ASSISTED METAL EXTRACTIONS

Sequential Extraction of Metals from
Environmental Samples
The bioavailability and mobility of trace metallic and
metalloid elements in the environment depend on the
chemical form of the element and the type of binding
to the matrix. Sequential extraction schemes, al-
though far from being perfect, have the ability to
extract elemental species from particular solid phases
in sediments, soils and sewage sludge. However, ap-
plication of these schemes entails a difRcult experi-
mental task owing to the large number of slow and
tedious stages. For instance, the Tessier scheme ap-
portions metal distribution in four different stages:
(1) exchangeable, (2) associated to carbonates, (3)
associated to Fe and Mn oxides and (4) associated to
organic matter and sulRdes. For dissolving a particu-
lar solid phase, chemical extractants are applied suc-
cessively to the solid sample, each follow-up
treatment being more drastic in chemical action or
different in nature from the previous one. Thus, for
the phases mentioned above, an MgCl2 solution, an
NaOAc solution, an NH2OH.HCl solution, and an
HNO3#H2O2 solution are used sequentially. The
Tessier scheme requires an overall operation time of
about 18 h. Ultrasonic energy from a probe-type soni-
cator has been employed for acceleration of the se-
quential chemical extraction of Cu, Cr, Ni, Pb and Zn
from sediment and sewage sludge samples. Conven-
tional and ultrasound-accelerated Tessier extraction
schemes offered similar partitioning patterns for the
two Rrst fractions (i.e., exchangeable and carbonate-
bound) when applied to a sewage sludge sample.
However, signiRcant differences in metal extractabil-
ity were observed for some metals when applying the
ultrasound-accelerated Tessier scheme to river sedi-
ments. On the other hand, a good agreement for the
total extractable contents (i.e., sum of metal contents
found in each stage) was seen for Ni, Pb and Zn in
sewage sludge and Cr, Ni, Pb and Zn in river sedi-
ment, meaning that the ultrasound methodology
could be useful for fast screening of extractable

Table 2 Analytical results obtained by applying the conventional and the modified Tessier sequential extraction schemes for metal
partitioning in a river sediment and a sewage sludge

Fraction Element River sediment a Sewage sludge b

Conventional Ultrasound Recovery d Conventional Ultrasound Recovery d

method method (%) method method (%)
(X$SD) c (X$SD) c (X$SD) c (X$SD) c

Exchangeable Cu 2.17$0.05 1.90$0.1 87.6 18.4$0.18 18.2$0.12 98.9

Cr ND ND } ND ND }
Ni 12.2$0.3 12.1$0.4 99.3 9.51$0.18 9.24$0.23 97.2
Pb 9.83$0.16 9.73$0.34 99.0 10.9$0.26 10.7$0.26 97.6
Zn 14.2$0.3 14.0$0.4 98.6 96.7$2.1 96.2$3.7 99.5

Carbonate-bound Cu 15.5$0.47 4.21$0.27 27.2 8.16$0.11 8.1$0.12 98.7

Cr ND ND } ND ND }
Ni 14.1$0.57 14.0$0.21 99.3 6.35$0.09 6.16$0.25 97.0
Pb 41.1$0.37 40.6$0.83 98.9 13.7$0.24 13.6$0.19 99.8
Zn 70.8$1.34 69.2$1.43 97.7 80.0$1.1 78.6$1.6 98.2

Fe}Mn oxide-bound Cu 7.71$0.35 18.7$0.28 242 10.3$0.16 26.1$0.27 253

Cr 7.06$0.11 2.85$0.1 40.4 ND ND }
Ni 6.0$0.23 6.0$0.3 100 4.58$0.30 4.42$0.14 96.3
Pb 165.4$3.7 134$1 81.2 19.7$0.71 19.2$0.16 97.7
Zn 130$3 106$1 81.3 397$3 393$3 99.1

Organic matter-bound Cu 152$2 149$4 98.3 165$3 46.3$0.43 28.0

Cr 3.92$0.04 ND 0.0 8.31$0.23 ND 0.0
Ni ND ND } 6.00$0.12 5.97$0.32 99.5
Pb 5.60$0.36 34.6$0.63 618 16.0$0.46 15.6$0.47 97.4
Zn 15.4$0.24 32.2$1.0 210 90.0$2.0 58.4$2.0 64.9

a
PeH rez-Cid B, Lavilla I and Bendicho C (1999) International Journal of Environmental Analytical Chemistry 73: 79.
b
PeH rez-Cid B, Lavilla I and Bendicho C (1999) Fresenius Journal of Analytical Chemistry 363: 667.
c
Average of three determinations (expressed as
g g\1)$standard deviation.
d
The recovery was calculated in the following way: [metal leached using the accelerated method/metal leached using the conventional
method];100.
ND, non detected.
 III / ULTRASOUND-ASSISTED METAL EXTRACTIONS
metals in solid environmental samples. The operation
time per sample was 20 and 28 min for sewage sludge
and river sediment, respectively, when ultrasound
was used for the Tessier scheme (Table 2).
The sequential extraction scheme, proposed by the
Community Bureau of Reference (BCR), now the
Standards, Measurement and Testing Programme,
consists of three stages: acid-soluble, reducible and
oxidizable. The reagents employed are a HOAc solu-
tion, an NH2OH.HCl solution and an H2O2 solution,
respectively. Despite using a stage less than the Tes-
sier scheme, its operation time is much longer (about
51 h per sample). Application of the BCR scheme to
sewage sludge showed that a drastic shortening in
time from 51 h to about 22 min per sample could be
achieved by the use of ultrasonication. In this case,
a much better agreement between the conventional
and the ultrasound-accelerated BCR schemes was
found in all fractions, so that information concerning
extractable metal contents from sewage sludge was
virtually the same.
Conclusions
Ultrasound-assisted extraction can be used as an
alternative to traditional sample preparation methods
for elemental analysis and speciation where matrix
separation rather than complete matrix elimination is
performed. Sonication methods usually involve mild
treatments which meet an important requirement for
speciation, i.e., extraction of the species of interest
without changes in their integrity. As a result of the
decreased amount of matrix released during sonica-
tion treatments, matrix interferences can also be re-
duced. Additionally, ultrasonic treatments provide
a signiRcant speeding up of those methods requiring
long and tedious extractions (e.g., sequential extrac-
tion of metals from solid environmental samples). So
far, analytical results obtained on applying ultra-
sound for sample preparation are very promising, and
new developments are expected on the topics ad-
dressed in the present work. On-line solid}liquid ex-
traction with the use of ultrasound will require
specially designed ultrasonic cells to further simplify
sample treatment.
See also: II /Extraction: Analytical Inorganic Extractions.
III/Microwave-Assisted Extraction: Environmental
Applications.
Further Reading
Amoedo L, Capelo JL, Lavilla I and Bendicho C (1999)
Ultrasound-assisted extraction of lead from solid sam-
ples: a new perspective on the slurry-based sample prep-
4425
aration methods for electrothermal atomic absorption
spectrometry. Journal of Analytical Atomic Spectro-
metry 14: 1221}1226.
Ashley K (1998) Ultrasonic extraction of heavy metals from
environmental and industrial hygiene samples for their
subsequent determination. Trends in Analytical Chem-
istry 17: 366}372.
Capelo JL, Lavilla I and Bendicho C (1998) Ultrasound-
assisted extraction of cadmium from slurried biological
samples for electrothermal atomic absorption spectro-
metry. Journal of Analytical Atomic Spectrometry 13:
1285}1290.
El Azouzi H, Cervera ML and de la Guardia M (1998)
Multi-elemental analysis of mussel samples by atomic
absorption spectrometry after room temperature sonica-
tion. Journal of Analytical Atomic Spectrometry 13:
533}538.
Lavilla I, Capelo JL and Bendicho C (1998) Determination
of cadmium and lead in mussels by electrothermal
atomic absorption spectrometry using an ultrasound-
assisted extraction method optimized by factorial de-
sign. Fresenius Journal of Analytical Chemistry 363:
283}288.
Lavilla I, PeH rez-Cid B and Bendicho C (1998) Leaching of
heavy metals from an aquatic plant (Lagarosiphon ma-
jor) used as environmental biomonitor by ultrasonic
extraction. International Journal of Environmental Ana-
lytical Chemistry 72: 47}57.
Luque de Castro MD and da Silva MP (1997) Strategies for
solid sample treatment. Trends in Analytical Chemistry
16: 16}23.
Mamba S and Kratochvil B (1995) Application of ultra-
sound to dissolution of environmental samples for
elemental analysis. International Journal of Environ-
mental Analytical Chemistry 60: 295}302.
Matejovic I and Durackova A (1994) Comparison of
microwave digestion, wet and dry mineralization, and
solubilization of plant sample for determination of cal-
cium, magnesium, potassium, phosphorus, sodium,
iron, zinc, copper and manganese. Communications in
Soil Science and Plant Analysis 25: 1277}1288.
McKiernan JW, Creed JT, Brockhoff CA, Caruso JA and
Lorenzana RM (1999) A comparison of automated and
traditional methods for the extraction of arsenicals from
Rsh. Journal of Analytical Atomic Spectrometry 14:
607}613.
Minami H, Honjyo T and Atsuya I (1996) A new solid-
liquid extraction sampling technique for direct deter-
mination of trace elements in biological materials by
graphite furnace atomic absorption spectrometry. Spec-
trochimica Acta, Part B 51: 211}220.
PeH rez-Cid B, Lavilla I and Bendicho C (1998) Speeding up
of a three-stage sequential extraction method for metal
speciation using focused ultrasound. Analytica Chimica
Acta 360: 35}41.
PeH rez-Cid B, Lavilla I and Bendicho C (1999) Analytical
assessment of two sequential extraction schemes for metal
partitioning in sewage sludge. Analyst 121: 1479}1484.
4426 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY
Rio-Segade S and Bendicho C (1999) Selective reduction
method for separate determination of inorganic and
total mercury in mussel tissue by Sow-injection cold
vapor technique. Ecotoxicology and Environmental
Safety 42: 245}252.
VENOMS: CHROMATOGRAPHY
See III / NEUROTOXINS: CHROMATOGRAPHY
VETERINARY DRUGS: LIQUID
CHROMATOGRAPHY
H. F. De Brabander and K. De Wasch, University of
Ghent, Merelbeke, Belgium
Copyright ^ 2000 Academic Press
For the analysis of residues of veterinary drugs, liquid
chromatography (LC) is of increasing importance:
some of these molecules are polar, heat-sensitive
and/or difRcult to analyse by gas chromatogra-
phy}mass spectrometry (GC-MS). Moreover, LC is
the method of choice for components of high molecu-
lar mass. Since the introduction of benchtop LC-MS
instruments, there has been an increasing number of
publications on the application of this technique in
the Reld of residue analysis.
Equipment
In LC a large variety of packed columns are in use but
most residue separations are carried out with some
kind of reversed-phase material based on modiRed
silicas (RP-18, RP-8, etc.). Hitherto, in our labora-
tory, a particle size of 5
m with column dimensions
150;2.1 mm has been commonly used. For a lab-
oratory involved in residue analysis under accredita-
tion, the daily reproducibility of the chromatogram
from column to column is very important (see section
on quality criteria, below). In the future, column
material of smaller particle sizes (3
m) may be used
routinely, allowing faster separation, higher sample
throughput and better limits of detection.
The nature of the mobile phase depends on the
column used. In most cases a mixture of water and an
organic solvent such as methanol or acetonitrile is
used. Special LC grades of solvents are necessary. For
analysis of residues, gradient elution is a must. In
Rio-Segade S and Bendicho C (1999) Ultrasound-assisted
extraction for mercury speciation by the Sow-injec-
tion}cold vapor technique. Journal of Analytical Atomic
Spectrometry 14: 263}268.
most cases the column has to be cleaned from inter-
fering components after each run by a gradient. As
well as organic solvents, a number of chemicals may
be added to the mobile phase (buffers and chelating
agents) but the compatibility of these products with
the detector should be checked. For LC-MS only
volatile components (e.g. triSuoroacetic acid) can be
used and this limitation sometimes hinders the trans-
formation of an LC into an LC-MSn method.
Autoinjection is a must for the routine analysis of
residues of veterinary drugs, not only for higher
sample throughput but also for reproducibility in the
validation of the results. However, particular atten-
tion should be drawn to the danger of cross-contami-
nation with such injectors, especially in combination
with LC-MS which has low detection limits.
Detectors
For screening purposes universal detectors such as UV
and light-scattering detectors are used. However, for
the conRrmation of suspect samples more is required
than just retention time and detector response. Since
the results of laboratory analysis may have a serious
impact on individuals and companies, false positives
must be avoided at any price. For example, a sample
of poultry feed, analysed by ion chromatography,
was suspected to contain KSCN (a thyreostatic drug).
Both the retention time and co-chromatography met
the quality criteria. However, the presence of KSCN
was so unlikely that the efSuent was collected and
mixed with Fe3# (to give a red colour with
SCN). This test was negative. Later on, it was found
that the sample contained acetylsalicylic acid, which
is often used in poultry rearing, and that the two
molecules are not separated in the chromatographic
system used.
 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY
More analytical evidence could be gathered by us-
ing a diode array detector (DAD). However, at low
concentrations of the analyte and/or dirty samples,
interferences are very likely. With a Suorescence de-
tector more speciRc analysis at lower detection limits
can be performed but in most cases some kind of
derivatization of the analyte is needed.
The mass spectrometric detector is very important
in residue analysis: the most common interfaces are
electrospray (ES) and atmospheric pressure interface
(API).
Quanti\cation
Quantitative analysis is necessary for residues of legal
veterinary drugs having a maximum residue limit
(MRL). The method used must have limit of quantiR-
cation of (at least) half the MRL. The validation of
quantitative method is very time-consuming and
expensive. Therefore, qualitative LC is often used
for analysis of residues of illegal substances (with a
so-called zero tolerance). However, quantitative
methods always have a qualitative aspect (a value for
the correct substance) while qualitative methods
always contain a quantitative background (e.g. the
estimation of peak intensities). This quantitative as-
pect is reSected in the so-called action limits: levels of
residues which an efRcient laboratory should be able
to reach (e.g. 2 p.p.b. for anabolics). In our laborat-
ory, qualitative data (residue present or not) are
transferred into quantitative data as follows: a large
number of samples (e.g. 50 urines of different origin)
are spiked with several anabolics at a certain level
(e.g. the action limit) and analysed. The percentages
detected spikes are calculated. A 95% detection
levels is statistically accepted. So, it could be stated to
the inspection services: if a sample contains the
spiked level, the residue will be detected with a 95%
probability. Higher or lower levels will be detected
with higher or lower probabilities. It should also be
mentioned that quantiRcation of one signal (e.g. a UV
Figure 1 MS (ABCDEF, analyte; pqt, xyz, uvw and pqrs, interferences); MS-MS on ABCDEF; MSn: formation of granddaughter and
grandgranddaughter ions.
4427
absorbance) is easier than quatiRcation of a complex
signal (e.g. a mass spectrum). However, complex sig-
nals give much more information. Internal standards
play an important role in quantiRcation in residue
analyses. For LC-MS the availability of deuterated
standards is often a limiting factor.
Generally, it is important to convince customers
(e.g. inspection services) that very reliable quantita-
tive analysis of many samples with low detection limits
in a short time for a very low price is not possible.
Special Features of LC
LC-MSn
The Rrst benchtop LC-MS-MS machine based on
a modiRcation of an ion trap was introduced in 1996.
In tandem MS an ion (e.g. the molecular ion) may be
chosen as parent ion, isolated and concentrated in the
trap, while all other ions are ejected. Afterwards the
speed of the ion is increased: the ion collides with He
present in the trap and fragments. The fragment ions
(daughter ions) are measured. The daughter ions are
theoretically derived from the parent ion only, but in
practice some interference is still present (Figure 1).
With quadrupoles, MS-MS is normally the end of
the story. In an ion trap one daughter ion may be
concentrated and fragmented over and over again. In
theory, MSn opens the way to a signiRcant reduction
of the clean-up of the sample. However, fewer and
fewer ions of the analyte are present and the signal-
to-noise ratio competes with the ability of the appar-
atus to detect ions. In practice MS2 is only needed for
analysis of most residues with LC-MS.
LC as Clean-up in Residue Analysis
Some hyphenated techniques are claimed to be so
speciRc that they only need minimum sample clean-
up. In our experience this is not yet true for the
analyses of all residues (e.g. anabolics in complex
4428 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY
matrices at the p.p.b. (
g kg\1) level). The clean-up
of the primary extract needs special attention. LC
puriRcation adds a considerable value to the speciR-
city of the method and inSuences the reliability of the
results in a positive sense. By fraction collection, very
clean extracts are obtained and the limit of detection
is substantially decreased.
Immunological methods can also be coupled to LC
to eliminate interfering substances.
Quality Criteria for the Use of LC
in Residue Analysis
Minimum quality criteria for the identiRcation
of residues using different analytical techniques
have been published in the European Commission
(EC) directive 93/256. For LC the EC has
speciRed the following quality criteria for methods
of analysis which may be used for conRrmatory pur-
poses:
1. The analyte should elute at the retention time
which is typical for the corresponding standard
analyte under the same experimental conditions.
2. The nearest peak maximum in the chromatogram
should be separated from the designated analyte
peak by at least one full peak width at 10% of the
maximum height.
3. The absorption maximum in the spectrum of the
analyte should be at the same wavelength as those
of the standard analyte within a margin deter-
mined by the resolution of the detection system.
For diode array detection this is typically within
$2 nm.
4. The spectrum of the analyte above 220 nm should
not be visually different from the spectrum of the
standard analyte for those parts of the two spectra
with a relative absorbance *10%. This criterion
is met when the same maxima are present and no
observed point in the difference between the two
spectra is more than 10% of the absorbance of the
standard analyte.
5. For conRrmatory purpose, if the method is not
used in combination with other methods, then
co-chromatography in the LC step is mandatory.
Discussion of the Quality Criteria
1. Quality criterion 1 is same for any chromato-
graphic procedure: the retention times of the two
peaks, formed by the analyte and the standard,
should correspond. Otherwise the analyte clearly
differs from the standard. A window of 3% is
a reasonable quality criterion. Where a great devi-
ation occurs, co-chromatography may be used (see
point 5).
2. Criterion 2 requires a resolution of one between
two peaks. However, this quality criterion is not
clearly described in the EC document. Here the
question might be put whether the criterion should
only be required for peaks with the same max-
imum wavelength. For example, an analyte with
maximum absorbance of 430 nm may in practice
be readily distinguished from an interfering com-
pound with a maximum of 310 nm, even if they
partly co-elute. In LC-MSn, this criterion will the-
oretically not be valid if deuterated standards,
which nearly co-elute, are used.
3 and 4. These criteria match only LC-DAD. For
LC-MSn criteria have not yet been described.
5. In criterion 5, co-chromatography is required for
proper identiRcation of an analyte. The usefulness
of co-chromatography may be questioned: co-
chromatography may prove that the peak in ques-
tion is not the analyte but not that the peak is
without any doubt the analyte. Moreover, it is
important that the concentration of standard
analyte added is of the same magnitude as that of
the sample.
Examples of LC Methods in Residue
Analysis
In this section some examples of LC and LC-MSn
methods for residues of some illegal growth pro-
moters, legal drugs and feed additives are discussed.
More extensive information can be found in the Fur-
ther Reading section.
LC Methods for Illegal Growth Promoters
Thyreostatic drugs The use of these drugs in cattle
results in a spectacular weight gain, arising mainly
from an increased Rlling of the gastrointestinal tract
and an augmented water retention. In our laboratory
a speciRc thin-layer chromatography (TLC) method
for the determination of thiouracil and analogous
compounds has been established. For additional con-
Rrmation, the Rnal extract of the TLC method could
also be analysed by LC-MSn yielding speciRc MS2 and
MS3 spectra (Figure 2).
Anabolic steroids The use of anabolic steroids
as growth promoters in the fattening of animals
is prohibited in all EU member states. GC-MS is
the method of choice for a large number of
these components. But some compounds such as
stanozolol and its most important metabolite
in cattle (16
-hydroxystanozolol; Figure 3) are
difRcult.
 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY 4429

Figure 2 (A) Chromatogram and MS2 spectra of some thyreostats. Thyreostats: 4(6)-R-thiouracil (R"H (TU); methyl (MTU);
n-propyl (PTU); phenyl (PhTU)); TAP, 1-methyl-2-mercaptoimidazole (tapazole); DMTU, (4(5,6)-dimethyl-2-thiouracil). (B) MS1, MS2
and MS3 spectrum of the thyreostat tapazole.

Recently, GC-MS, LC-MS, MS-MS and MSn
methods for this metabolite have been described and
compared, in a collaborative study between three
Belgian and three Dutch laboratories. It was observed
that the spectra obtained on different types of LC-MS
systems are clearly different: from one diagnostic ion
(in a single quadrupole) to a lot of diagnostic ions
with LC-MSn. This illustrates the difRculty of work-
ing out quality criteria for LC-MSn analysis.

-Agonists During the 1980s the
-agonists found
illegal application in animal breeding (extra weight
gain together with a repartition between muscle and
fatty tissue). An LC method with post-column
4430 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY
Figure 3 16
-Hydroxystanozolol: the most important metab-
olite of stanozolol in cattle.
derivatization (with a diazotization mixture) for the
determination of clenbuterol and analogues has been
described. Later, the very speciRc detection for ani-
lines was replaced by MSn detection: it is easier to
switch from one analyte to another with an LC-MS
system than with a post-column derivatization de-
tector. Moreover, deuterated clenbuterol can be used
for quantiRcation. In Figure 4 a chromatogram of
some
-agonists (not all are represented here) and an
example of an MS2 spectrum (tulobuterol) are given.
Corticosteroids Corticosteroids are also abused in
cattle fattening. The weight gain is probably due to
secondary effects of the corticosteroids, such as water
retention. For the analysis of residues of corticos-
teroids, GC-MS with negative ion chemical ioniz-
ation (NCI) detection is still the method of choice.
However, for the identiRcation of newly used cor-
ticosteroids in injection sites, LC-MSn offers more
Figure 4 Chromatogram of some
-agonists and MS2 spectrum of tulobuterol.
identiRcation power than GC-MS (no derivatization;
different MSn spectra).
Anti-infection Agents
This broad range of chemicals is used for both thera-
peutical and/or growth-promoting reasons. Screening
for residues of antibacterials in slaughtered animals is
carried out in most states by microbial inhibition tests
on kidney tissue. In the case of a positive test, the
identity and (in the case of legal drugs) the concentra-
tion of the substance should be determined. It is in
this aspect that LC and LC-MS methods are mostly
used.
Sulfonamides Several LC methods for the deter-
mination of sulfonamides have been described. In our
laboratory an LC method from the literature was
quickly transformed into an LC-MSn method. In
Figure 5 a chromatogram and MS2 spectra of some
sulfonamides are given. Currently, six sulfonamides
are monitored in one run.
Antibiotics For antibiotics such as penicillins,
cephalosporins, quinolones, macrolides and tetracyc-
lines a lot of LC and some LC-MS methods have been
described. For tetracyclines, for example, ligands (e.g.
oxalic acid) have to be added to the mobile phase to
prevent extreme tailing. Post-column derivatization
(e.g. with ZrCl4) followed by Suorescence detection is
 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY 4431

Figure 5 Chromatogram and MS2 mass spectra of some sulfonamides.

very speciRc for these molecules yielding very low
limits of detection: 0.5}1.5
g kg\1 in comparison
with 2}5
g kg\1 with LC-MS.
Antiparasitic Agents
An example of a potent antiparasitic veterinary drug
is ivermectine (a macrocyclic lactone disaccharide).

Figure 6 Chromatogram and MS2 mass spectra of some tranquillizers.

4432 III / VETERINARY DRUGS: LIQUID CHROMATOGRAPHY
Figure 7 Formulas of carbadox, olaquindox and metilolaquindox.
The drug is effective in low dosages and therefore
requires methods with low detection limits (MRL:
15
g kg\1 in porcine liver). Since the molecule
has a high molecular mass, LC is the method of
choice. HPLC-UV methods for screening of iver-
mectine residues in animal tissues and milk have
been described. For conRrmation the molecule
can be derivatized (with methylimidazole}acetic
anhydride) and analysed by LC with a Suorescence
detector.
Tranquillizers
Tranquillizers may be used illegally to prevent stress
during the transport of pigs and bulls to the abattoir.
A large number of LC methods have been published
for the determination of residues of these compo-
nents. In our laboratory one method was transferred
into an LC-MSn method with which seven tranquilli-
zers could be determined in a short time. In Figure 6
mass chromatograms and the MS2 spectra of these
components are given. The data, given in Figures 4}6
were obtained with the same apparatus. This is an
illustration of the ease of switching from one analyte
to another.
Feed Additives
Some components are not considered as veterinary
drugs but as feed additives. Examples are the
quinoxalines, carbadox and olaquindox. However,
residues of these components may be present in edible
tissue as well. Also nonregistered equivalents of these
components could be used: as an example, the pres-
ence of metilolaquindox was suspected in animal
feed: this was possibly a modiRcation of olaquindox
with a methyl group. LC analysis gave a chromato-
gram containing a large peak different from carbadox
and olaquindox. However, LC-MS analysis gave a mo-
lecular mass less than carbadox and olaquindox. By
combining MS with NMR a structure for this molecule
was proposed (Figure 7). This example illustrates the
important of MS in residue analysis.
Conclusion
The demands for speciRcity, reliability, speed and
turnover in residue analysis of veterinary drugs are
continuously increasing. LC, especially with MS de-
tection, is a reliable analytical technique which
should be able to cope with these stringent demands.
In comparison with GC and GC-MS, a large range
of analytes can be covered and in most cases there
is no need for derivatization. It is also easy to
switch an LC-MS system from one analyte to an-
other. The lower yield of the LC-MS interfaces
and the poorer separation power of LC columns in
comparison with GC may be regarded as points to be
improved.
The use of illegal alternatives to registered drugs or
feed additives poses two important problems for rou-
tine inspection: Rrst of all there is no target compon-
ent. The situation is comparable with the search for
a unknown needle in an unknown haystack. Second-
ly, no analytical standards of the molecule are avail-
able. MS (and MSn) is able to give more information
about a suspect peak. The future of LC in residue
analysis will depend largely on the possibilities of
identiRcation of illegal substances abused and quali-
tative and quantitative analysis of legal veterinary
drugs with LC-MSn.
See Colour Plate 124.
See also: II/Chromatography: Liquid: Detectors: Mass
Spectrometry; Detectors: Ultraviolet and Visible Detection;
Mechanisms: Reversed Phases. III/Forensic Sciences:
Liquid Chromatography.
Further Reading
Crosby NT (1998) Determination of Veterinary Residues in
Food. Lancaster: Technomic.
Heitzman RJ (ed.) (1994) Veterinary Drug Residues. Ox-
ford: Blackwell ScientiRc.
March RE and Hughes RJ (eds) (1992) Quadrupole Storage
Mass Spectrometry. New York: Wiley Interscience.
Nollet L (ed.) (1992) Food Analysis by HPLC. New York:
Marcel Dekker.

Oka H, Nakazawa H, Harada K-I and MacNeil JD (eds)
(1995) Chemical Analysis for Antibiotics Used in Agri-
culture. Arlington: AOAC International.
OKeeffe M (ed.) (2000) Residue Analysis in Food } Prin-
ciples and Applications. Amsterdam: Harwood Aca-
demic Publishers.
Proceedings of the International Symposium on Analysis
of Anabolizing and Doping Agents. I: Ghent, 1988:
VIRUSES: CENTRIFUGATION
L. L. Bondoc Jr., BioPort Corporation, Lansing,
MI, USA
Copyright ^ 2000 Academic Press
Viruses have proved to be detrimental as well as
beneRcial. They are notoriously infectious agents that
are at the root of several major diseases in man,
domesticated animals, and agricultural crops. How-
ever, their attenuated or noninfectious forms have
been used as vaccines, enabling the development of
immunity against particularly devastating diseases.
Recently, replication-deRcient viruses have been used
as agents for gene delivery and as potential vaccine
carriers, as they have evolved efRcient mechanisms of
infectivity.
Viruses are particulate in nature and are made up
essentially of DNA or RNA, wrapped in a predomi-
nantly protein coat. They range in size from 20 to
2000 nm (0.02}2
m) and in molecular weight from
4;106 to 2;109 Da. Many viruses possess an envel-
ope that is typically derived from the host cellular
membrane.
Initial isolation of viruses usually involves centrifu-
gation, particularly density gradient centrifugation
(DGC). For almost half a century DGC has been
regarded as the most rapid, and reliable preparative
procedure for the isolation of highly puriRed and
concentrated virus preparations for subsequent
physicochemical and biological characterization. As
such, it is used as a benchmark against which alterna-
tive methods can be evaluated. To date the technique
has permitted the isolation and subsequent character-
ization of a plethora of viruses belonging to at least
39 major families.
Centrifugal Separations
Although signiRcant improvements in centrifugation
hardware have led to increased operational efRcien-
III / VIRUSES: CENTRIFUGATION 4433
J. Chromatogr. 489 (1989). II: Ghent, 1990: J.
Chromatogr. 564 (1991). Ghent, Belgium.
Proceedings of the International Symposium of Hormone
and Veterinary Drug Analysis. I: Ghent, 1992: Anal.
Chim Acta 275 (1993). II: Bruges, 1994: The Analyst
119 (1994). III: Bruges, 1998: The Analyst 123, 12
(1998).
cies, the theory behind centrifugation and the
variations of the technique as applied to viruses
are well characterized. In a suspension of particles,
the rate at which particles sediment when subjected to
a centrifugal force depends on the nature of the par-
ticles, the nature of the medium, and the magnitude
of the centrifugal force. For spherical particles, the
sedimentation rate or velocity of the particle depends
on a variety of factors as indicated in eqn [1], one of
the many forms of the Svedberg equation:
dr/dt"[2r2p(p!m)2r]/9 [1]
where dr/dt is the velocity of the particle; rp is the
radius of the spherical particle; p is the density of
the particle; m is the density of the medium;  is
the angular velocity; r is the radial distance of the
particle from the axis of rotation; the product 2r is
proportional to the centrifugal force; and  is the
viscosity of the medium. It is possible to deRne a par-
ticle in terms of its behaviour in a centrifugal Reld by
manipulation of eqn [1] to yield a simpliRed version
of the Svedberg equation (eqn [2]) that uses the sedi-
mentation coefRcient, s, where:
s"(dr/dt)/2r [2]
For most biological macromolecules, the magnitude
of s is about 10\13 s, so this value is used as the unit of
sedimentation, the Svedberg (S). The sedimentation
coefRcient for viruses varies between 40 and 4500 S,
while for globular proteins it is 2}5 S.
Types of Separations
For a particular viral preparation, the most effective
centrifugal separation procedure is one that yields
a concentrate with signiRcant recovery of bioactivity
4434 III / VIRUSES: CENTRIFUGATION
and high quality based on several measures of purity.
There are three types of centrifugal separations avail-
able for viruses: (1) differential centrifugation; (2)
rate-zonal centrifugation; and (3) DGC or isopycnic
centrifugation. Differential centrifugation separates
particles according to size as well as density (from
eqn [1]), since denser particles will form pellets at
a faster rate than less dense particles of the same
mass. By choosing an appropriate centrifugal force
and centrifugation time, it is possible to clarify a viral
suspension from contaminating fermentation debris
by Rrst pelleting the contaminants at a given g force
and leaving the virus in suspension, then pelleting the
virus at a higher g force. Viruses that are unstable
when pelleted can be sedimented on a cushion or plug
of material (e.g. caesium chloride, sucrose, potassium
tartrate, Nycodenz (Nyegaard & Co.), and glycerol)
that has a density higher than that of the viral
particles. The major problems with this mode of
separation are the low yields and low resolution
from contaminants. Differential pelleting is often
used for the initial processing of heterogeneous
mixtures, to obtain fractions that are enriched in
the virus particles of interest prior to further puri-
Rcation.
In rate-zonal centrifugation particles move at dif-
ferent rates depending upon their mass. To avoid the
co-sedimentation of particles of different sizes, sam-
ples are typically layered as a narrow zone on top of
a density gradient. The gradient is used to facilitate
the layering of the sample and to minimize convection
currents in the liquid column during centrifugation
that would otherwise disrupt the particle zones as
they move down the tube. Rate-zonal separations are
ideal for particles of uniform size but not for particles
of the same type that are heterogeneous in size. Fur-
thermore, even though separation conditions can be
optimized, it is not yet possible to recover the separ-
ated fractions as fractionated species. For viral prep-
arations this mode of centrifugation is primarily used
for characterizations, such as molecular weight deter-
mination and the determination of possible interac-
tions with other molecules.
The most common method of centrifugal separ-
ation for viruses is DGC, or isopycnic centrifugation.
In this process the particles move until their density is
the same as that of the surrounding medium. The
particles are separated purely on the basis of their
density, and their size only affects the rate at which
they reach their isopycnic positions. Because the sep-
aration is an equilibrium process, run times are gener-
ally much longer than for rate-zonal or differential
centrifugation. Prolonged centrifugation does not af-
fect the separation as long as the gradient remains
stable and the activity and integrity of the particles
are not adversely affected by centrifugation. Mater-
ials are typically spun at 25 000}200 000 g for up to
20 h. Samples are loaded either on a pre-formed
gradient or on self-forming gradient media.
Centrifugation Media
For isopycnic separations, the choice of media is
important. There are several desirable characteristics
for a medium, the most important being that the
maximum density of the gradient is greater than that
of the particles to be separated. In general viruses
have buoyant densities in the range 1.1}1.5 g cm\3.
However, as a result of different levels of hydration of
viral particles in different media, the densities of the
particles can vary depending on the medium being
used. The physico-chemical properties of the solu-
tions of the gradient medium should be known, and it
should be possible to determine the precise concentra-
tion of the medium using one or more of these proper-
ties (e.g. refractive index or densitometry). The
medium should be inert and safe to use, and should
not interfere with monitoring of the zones of frac-
tionated material within the gradient (e.g. by ultra-
violet or visible absorbance, radioactivity counting,
protein determination, etc.). It should be easy to sep-
arate the sample material from the gradient medium
(by dialysis, ultraRltration, or centrifugation) without
loss of the sample or sample activity. Ideally the
medium should also form solutions of low ionic
strength with low viscosity and be iso-osmotic with
the virus.
Gradient media for DGC are either ionic or
nonionic. Commonly used ionic media include cae-
sium salts (e.g. caesium chloride), potassium salts,
rubidium salts and sodium salts (e.g. sodium chlor-
ide). These materials are used to form solutions with
maximum buoyant densities of 1.4}2.6 g mL\1.
Gradients of caesium salts, especially caesium chlor-
ide, are used almost exclusively for virus puriRcation.
They can be pre-formed using any of the standard
techniques or they can be formed in situ by centrifu-
gation. Solutions containing caesium salts are highly
ionic, and while they are nonviscous, they all have
high osmolarities. Gradients formed from these salts,
differ with respect to their solubility, maximum den-
sity, activity and steepness, all of which can affect the
banding of materials.
Nonionic gradient media can be subdivided into
carbohydrates, iodinated gradient solutes, colloidal
silica suspensions and proteins. Sucrose, a disacchar-
ide, has been widely used for the isopycnic fractiona-
tion of viruses. Its popularity is due to its inertness
towards biological materials, ready availability, low
cost, and stability. The main disadvantages of sucrose

include its high osmotic strength, high viscosity, hy-
pertonicity for solutions more concentrated than 9%
(w/v) and rather low buoyant density of 1.03 g mL\1.
Sucrose gradients must be pre-formed for isopycnic
fractionations.
To circumvent the problems that arise from frac-
tionating osmotically sensitive particles in high osmo-
tic strength sucrose solutions, several polysaccharides
have been used as gradient media. These include
glycogen, dextrans, and Ficoll (Pharmacia). Ficoll
is produced by the chemical copolymerization of
sucrose molecules with epichlorohydrin to give
a polymer with a molecular weight of 400 kDa. Ficoll
solutions below 20% (w/v), equivalent to a buoyant
density of 1.07 g cm\3, have a relatively low
osmolarity, although at higher concentrations the os-
molarity rises sharply. Gradients of Ficoll, which
have a higher viscosity and better stability than suc-
rose gradients, must be prepared using a gradient
mixer.
Most iodinated gradient media used in the separ-
ation of viruses are derivatives of triiodobenzoic acid
to which hydrophilic groups have been attached to
increase water solubility. The ionic forms of these
compounds include the sodium or N-methyl-
glucamine salts of metrizoate, diatrizoate, and
iothalamate, and the nonionic forms include met-
rizamide and Nycodenz. These materials form stable
solutions at buoyant densities up to 1.45 g cm\3. Iod-
inated compounds have several advantages, including
much lower osmolarities and viscosities than sucrose
at all densities. Gradients of these media can be pre-
formed or generated in place.
Colloidal silica gradients have been used for several
years, but only one preparation, namely Percoll
(Pharmacia), has been developed for centrifugation.
In this particular preparation the silica particles are
coated with polyvinylpyrrolidone, which minimizes
their interaction with biological material and also
stabilizes the colloid against freezing and thawing and
the presence of salts. Its solutions are isoosmotic and
its low viscosity facilitates the rapid banding of vi-
ruses. However, Percoll is precipitated at low pH and
solutions of high ionic strength destabilize the col-
loidal suspension. Gradients of Percoll readily self-
form, or can be pre-formed using a simple mixer.
Percoll forms suspensions at buoyant densities up to
1.13 g cm\3. The removal of Percoll from virus solu-
tions can be problematic because Percoll particles
(17}30 nm in diameter) are very close in size of some
viruses.
Proteins have a hydrated buoyant density of
approximately 1.27 g cm\3 and can be used as gradi-
ent media, but no applications to viruses have been
reported.
III / VIRUSES: CENTRIFUGATION 4435
Types of Gradients
Pre-formed gradients for DGC of viruses can be con-
tinuous or discontinuous. Continuous gradients may
be linear, convex, or concave and are usually pre-
pared using a dedicated gradient former. Discontinu-
ous or step gradients are prepared by successively
layering solutions of different density. Pre-formed
gradients must be handled very carefully prior to
centrifugation to avoid gradient disruptions caused
by vibration or temperature variations. The virus
sample itself, which must have a density less than that
of the top of the gradient, is gently layered onto the
gradient before centrifugation is started. To minimize
changes in the density proRle at the top of the gradi-
ent, the sample volume should be small compared
with the gradient volume.
For self-forming gradients the initial sample
volume is not a concern, as the sample is either
mixed with a concentrated solution of the gradient
solute or solid gradient solute is added to give the
correct initial density. The duration of centrifugation
for self-forming gradients is longer than that for
pre-formed gradients since time is required to form
the gradient.
Rotors
DGC can be carried out in all the available types of
rotors. Preparative centrifuge rotors are classiRed into
four main types, namely swing-out (swinging bucket),
Rxed angle, vertical, and zonal. In swing-out or hori-
zontal rotors, the tubes of sample solutions are placed
in individual buckets that move out perpendicular to
the axis of rotation as the rotor rotates. This creates
a long migration path to separate viruses along the
density gradient and requires a long period to achieve
signiRcant separation. Horizontal rotors can be
spun to attain maximum speeds corresponding to
100 000 g or more.
In Rxed angle rotors the tubes are at a Rxed angle
(varying from 143 to 403) to the axis of rotation, and
when the rotor rotates the solution reorients in the
tubes. This reorientation enhances the loading capa-
city of the isopycnic gradients. Rotors with shallow
angles are more efRcient at pelleting because the sedi-
mentation pathlength is shorter. Fixed angle rotors
are designed to operate up to very high centrifugal
forces ('600 000 g).
As the name suggests, in vertical rotors the tubes
are held in a vertical position, and centrifugal forces
similar to those for Rxed angle rotors can be achieved.
When the vertical rotor turns, the solution begins to
reorient through 903. Vertical rotors thus have short
sedimentation pathlengths, so the diameter of the
4436 III / VIRUSES: CENTRIFUGATION
tube and the capacities of the gradients in these
rotors are higher than in horizontal and Rxed angle
rotors.
Zonal rotors are often used for gradient separation.
Although the sample is pumped into a hollow rotor
chamber, the working principle of these rotors is
similar to that for vertical rotors, as it is the gradient
solution that reorients during the run, before the
sample is introduced under centrifugation. There
are two types of zonal rotors, namely batch and
continuous Sow, that differ based on the volume
of sample they can be used to process. Batch type
centrifugation is typically used for 10}200 mL
samples while for larger sample volumes of 100 L
or more, continuous Sow centrifugation is required.
For volumes 200 mL to 100 L, vertical rotors can be
used.
Recovery
Recovery of viral particles from DGC is performed
either manually or automatically. After centrifu-
gation density gradients can be recovered or unloaded
from the bottom, middle or the top of the tube. The
methods for unloading gradients from the bottom of
the tube include bottom puncture of the tube or
withdrawal using a narrow tube inserted through
the gradient to the bottom. Targeted bands from
within the gradient can also be unloaded by punctur-
ing the tube at the appropriate position. The methods
for unloading gradients from the top of the tube
include direct unloading from the top or collection
by upward displacement of the gradient by introduc-
ing a dense, preferably immiscible liquid at the
bottom of the centrifuge tube. Automated collection
systems with Sow-arrest or volumetric monitoring
are available commercially. Automated recovery
systems require a heavy displacement solution such
as 65% sucrose or Maxidens (Nyegaard & Co.),
an inert, nonviscous organic liquid immiscible
with aqueous gradients. Great care must be taken
in fractionating gradients after centrifugation since
resolution is easily lost at this stage. All operations sh
o uld be designed to minimize disturbance of the
gradient.
Conclusion and Future Developments
DGC is still the method of choice for the initial,
relatively quick isolation of novel viral particles.
The gradient medium, gradient shape, type of rotor,
and mode of recovery are determined empirically,
for a particular viral preparation, to yield the
highest possible recovery of bioactive puriRed
virus and permit subsequent characterization and
use.
The advent of gene therapy using viruses for
gene delivery or as vaccine carriers has encour-
aged the development of scaleable procedures for
virus isolation and puriRcation based on centrifu-
gation. With this impetus gradient media, rotor
designs, and modes of recovery continue to be
improved.
See also: II/Centrifugation: Theory of Centrifugation.
Further Reading
Bondoc LL Jr and Fitzpatrick S (1998) Size distribution
analysis of recombinant adenovirus using disc centrifu-
gation. Journal of Industrial Microbiology & Biotech-
nology 20: 317}322.
Brakke MK (1951) Density gradient centrifugation: a new
separation technique. Journal of the American Chemical
Society 73: 1847}1848.
Cantor CR and Schimmel PR (1980) Biophysical Chem-
istry. Part II: Techniques for the Study of Biological
Structure and Function. San Francisco: W.H. Freeman
and Company.
Croyle MA, Anderson DJ, Roessler BJ and Amidon GL
(1998) Development of a highly efRcient puriRcation
process for recombinant adenoviral vectors for oral gene
delivery. Pharmaceutical Development and Technology
3: 365}372.
Foster GD and Taylor SC, eds (1998) Plant Virology Proto-
cols. Totowa: Humana Press.
GrifRth OM (1986) Techniques of Preparative, Zonal, and
Continuous Flow Ultracentrifugation, 5th edn. Palo
Alto: Beckman Instruments.
Mushahwar DC, Erker JC, Muerhoff AS et al. (1999)
Molecular and biophysical characterization of TT virus:
evidence for a new virus family infecting humans. Pro-
ceedings of the National Academy of Sciences U.S.A. 96:
3177}3182.
Myers TM, Smallwodd S and Moyer SA (1999) IdentiRca-
tion of nucleocapsid protein residues required for Sendai
virus nucleocapsid formation and genome replication.
Journal of General Virology 80: 1383}1391.
Payment P and Trudel M (1993) Methods and Techniques
in Virology. New York: Marcel Dekker Inc.
Rickwood D, ed. (1984) Centrifugation: A Practical Ap-
proach, 2nd edn. Oxford: IRL Press Limited.
Soeda E, Krauzewicz N, Cox C et al. (1998) Enhance-
ment by polylysine of treatment of transient, but
not stable, expression of genes carried into cells
by polyoma VP1 pseudocapsids. Gene Therapy
5: 1410}1419.
 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography
VITAMINS
Fat-Soluble: Thin-Layer
(Planar) Chromatography
W. E. Lambert and A. P. De Leenheer,
Universiteit Gent, Gent, Belgium
Copyright ^ 2000 Academic Press
Introduction
Thin-layer chromatography (TLC) is a very widely
used chromatographic technique allowing the separ-
ation of simple mixtures followed by a qualitative
identiRcation or a semiquantitative visual analysis of
the samples. All this can be performed in an inexpen-
sive and simple way without requiring highly sophis-
ticated instrumentation.
On the other hand, high performance thin-layer
chromatography (HPTLC) is a highly instrumental
technique allowing fast and very efRcient separations
with quantitative results of accuracy and precision
rivalling those obtained by the far more popular tech-
niques such as high performance liquid chromatogra-
phy (HPLC) and gas chromatography (GC). The
small particle size (5
m) and the more uniform layer
of the stationary phase of the commercially pre-
coated HPTLC plates are responsible for this in-
creased efRciency and sensitivity.
This article focuses on the speciRc separation of
fat-soluble vitamins by TLC. Strategies from sample
preparation, stationary phases, mobile phases and de-
tection modes will be discussed for each vitamin separ-
ately. However, from the recent reviews published
biennially in Analytical Chemistry it can be seen that
the number of new applications of TLC to the analysis
of fat-soluble vitamins is diminishing all the time.
The chemistry (stability) of the different com-
pounds, will be treated because of its importance in
TLC analyses.
Figure 1 Structure of vitamin A. Related compounds include
retinol (R"CH2OH), retinal (R"CHO), retinoic acid (R"COOH)
and retinyl palmitate (R"CH2OCO(CH2)14CH3).
4437
Vitamin A
The structures of vitamin A and of some related
compounds are presented in Figure 1. The parent
compound, all-trans-retinol or vitamin A, is an iso-
prenoid structure with Rve conjugated double bonds
resulting in an absorption maximum at 325 nm (in
n-hexane or ethanol), in a high molar extinction co-
efRcient and in a sensitivity of the compound towards
isomer formation and/or oxidation. The formation of
isomers is catalysed by light and iodine while the
relative amount of the isomers depends on the
wavelength and on the solvent used. The four exocyc-
lic double bonds can theoretically result in the forma-
tion of 16 isomers. All have been characterized. An
increase in the number of cis bonds generally results
in a lower absorption maximum as well as a decrease
of the molar extinction coefRcient relative to the
all-trans isomer. Vitamin A and the Vitamin A-re-
lated compounds are also sensitive towards oxidation
and peroxidation by contact with air. The presence
of transition group metals is known to catalyse this
reaction.
To prevent degradation it is imperative to take
special precautions when working with vitamin A-
related compounds, for example, storing the samples
at very low temperature, working under subdued
light, avoiding drastic reagents and contact with air
or peroxide-containing organic solvents.
The lability of these compounds makes research in
the vitamin A Reld a real analytical challenge. Espe-
cially during the TLC process, special precautions are
necessary, as will be described below.
Chromatographic Conditions
TLC on polar inorganic nonmodiRed sorbents such as
alumina and silica remains very popular. Silica plates
can be activated by heating at 1203C for 1 h in an
attempt to enhance resolution, while spraying the
plates with a solution of an antioxidant has been
reported to prevent degradation of the compounds on
the plates.
As with what is known from liquid chromatogra-
phy, chromatographic systems based on silica plates
with eluents of hexane, petroleum ether or cyclo-
hexane, with a variable amount of a more polar
solvent such as 8% diethyl ether, 50% diethyl ether
or 20% ethyl acetate, offer the best separation of
the geometric isomers of vitamin A compounds. In
a similar way, high performance silica gel thin-layer
4438 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography

Table 1 Representative RF values of geometric isomers of ever, is less efRcient than that obtained on para-
vitamin A compounds fRn oil-impregnated kieselguhr. Separation on a
C18 phase and on a silica phase (on one single plate)
Compound a b c d
has been used in a two-phase two-dimensional TLC
Retinol determination of all-trans- and 13-cis-retinoic acid in
All-trans 0.09 0.14 0.21 cream samples. The reversed-phase step served to
9-cis 0.17 0.23 separate the retinoic acid isomers from the cream
13-cis 0.23 0.28
excipients, while the silica sorbent was ideally suited
11-cis 0.12 0.28 0.28
for the separation of the two isomers from each other.
Retinal Generally, on reversed-phase TLC plates, methanol
All-trans 0.27 0.47 0.46
9-cis 0.52 0.50
or acetonitrile can separate retinol from retinyl acet-
11-cis 0.47 0.58 0.53 ate while dichloroethane with acetonitrile can separ-
13-cis 0.60 0.55 ate the long chain retinyl esters.
Retinoic acid
All-trans 0.34 Detection
13-cis 0.39
Quenching the Suorescence of the indicator Rxed on
a, Silica gel: hexane}ether (92:8, by vol.); b, silica gel: the thin-layer plate itself (F254) is a very common and
hexane}ether (50:50, by vol.); c, silica gel: cyclohexane}tol-
uene}ethylacetate (50:30:20, by vol.); d, silica gel: diethyl
nondestructive way to localize spots on a TLC plate.
ether}cyclohexane}acetone}glacial acetic acid (40 : 60 : 2 : 1, Of course, this can also be applied to vitamin A com-
by vol.). pounds. Retinol and retinyl esters on the other hand
can be identiRed by the yellow-green Suorescence
they exhibit under 366 nm UV light. Other tech-
plates eluted with diethyl ether}cyclohexane} niques to visualize vitamin A compounds include
acetone}glacial acetic acid allows the separation of absorption of iodine vapours (with the formation of
all-trans- and 13-cis-retinoic acid in gel formulations brown spots) or destructive procedures such as spray-
(Table 1). Separation of vitamin A from the lipophilic ing with sulfuric acid.
vitamins is also possible on silica plates eluted with Other spray reagents include SbCl3 or SbCl5 solu-
mixtures of benzene}petroleum ether}acetic acid. tions in chloroform, a 5% solution of phosphomolyb-
Under these conditions the water-soluble vitamins dic acid in ethanol and a mixture of equal volumes of
remain at the origin. Very often, classical TLC on a 1% aqueous solution of potassium permanganate
silica plates serves as a kind of clean-up step before and a 5% aqueous solution of sodium carbonate.
ofSine quantiRcation, e.g. for the quantiRcation of After heating the plate coloured spots appear for
vitamin A in fruits and vegetables. With the introduc- vitamin A. The same reagents are often applied to
tion of the smaller HPTLC plates, more efRcient sep- visualize vitamins D and E.
arations together with shorter development times are As an alternative a large array of dyes has been
possible. This has allowed the quantitative deter- evaluated as visualizing agents for fat-soluble vit-
mination of retinol and of
-tocopherol in plasma amins, including vitamin A. The different dyes (ani-
with tocopheryl acetate as an internal standard. line blue, alkaline blue, brilliant green, neutral red,
In isolated cases kieselguhr plates or talc, starch or bromocresol green, bromothymol blue, thymol blue,
cellulose thin layers have been impregnated with 10% phenol red, helasol green, brilliant cresyl blue and
parafRn oil in cyclohexane. This was applied to bromophenol blue) are used as a solution of 50 mg of
a study of the hydrophobicity of a number of vitamin the dye either in 100 mL of water or in 100 mL of
A-related compounds and for a separation of vitamin a 2% aqueous sodium hydroxide solution. Evalu-
A-acetate from vitamin A-palmitate. For these studies ation of the plates is then performed 20 min after
the impregnated plates were eluted with mixtures of spraying or after acceleration of the reaction by heat-
methanol}water (95 : 5, by vol.) and of acetone}con- ing the plates at 1103C for 15 min.
centrated acetic acid (30 : 20, by vol.). For quantitative measurements, densitometric
Both the elution order of the compounds under evaluation can be applied to vitamin A compounds.
investigation and the composition of the elution sol- In this way, absorbance can be measured by diffuse
vents clearly demonstrate a reversed-phase type of reSectance at 290 nm using a mercury lamp, while
retention under these conditions. UV spectra can be recorded between 200 and 400 nm
Reversed-phase stationary phases such as RP-2 or with a deuterium lamp. Detection limits for retinol
C18 are also used in the analysis of vitamin A-related using this technique are around 160 ng mL\1 using
compounds. The separation on the RP-2 phase, how- 200
L plasma. By using tocopheryl acetate as an
 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography
Figure 2 Structure of (A) vitamin D2 (ergocalciferol) and (B) vitamin D3 (cholecalciferol).
internal standard, the coefRcient of variance (inter
plate and intra plate) can be kept below 12.5%.
Using the dyes as a visualizing agent, the highest
sensitivity on a silica plate can be obtained with
bromophenol blue (in 2% aqueous sodium hydroxide)
without heating the plate. Under these conditions
3
g can be visualized. The same reagent applied on
a partition-type TLC plate (silica gel impregnated
with a 5% solution of parafRn oil in chloroform)
resulted in a Rvefold decrease in sensitivity towards
vitamin A.
Vitamin D
Vitamin D and its structural analogues are a group of
9,10-seco-steroids: their basic structures are shown in
Figure 2. The D2 series (ergocalciferol) is of vegetable
origin and has a side chain derived from ergosterol
containing an additional C22}23 double bond and
a C24 methyl group, whereas vitamin D3 (cholecal-
ciferol) is formed in the skin of humans and animals
and has a side chain derived from cholesterol. The
conjugated system of three bonds results in a molar
extinction coefRcient of 18 300 L mol\1 cm\1 with
a max at 264 nm. Both analogues are derived
photochemically from their respective precursor
(provitamin D). Indeed, irradiation of this provitamin
D results in various photolysis products such as
tachysterol, lumisterol and pre-vitamin D. Pre-vit-
amin D then undergoes spontaneous rearrangement
to vitamin D. In the human body vitamin D3 is exten-
sively metabolized. The liver converts it to 25-hy-
droxy vitamin D3 while further hydroxylation in the
kidney yields, among others, 1,25-dihydroxy vitamin
D3. As with vitamin A, protection from light and
from air is of great importance for the analysis of
vitamin D by TLC.
TLC and, recently, HPTLC have found several
applications in vitamin D analysis, including differen-
4439
tiation of vitamin D analogues, separation of vitamin
D from other lipids (e.g. sterols, other fat-soluble
vitamins), determination of the purity of radiolabel-
led vitamin D derivatives and analysis of vitamin
D metabolites as a part of radioligand assays.
Chromatographic Conditions
Polar inorganic sorbents such as silica gel or, occa-
sionally, alumina have been applied to the separation
of vitamin D analogues. In this way, provitamin D3,
tachysterol3, lumisterol3 and pre-vitamin D3 were sep-
arated on silica gel and the eluting order of the com-
pounds could be correlated with the increasing
planarity of the compounds. Vitamins D2 and D3 can
be separated on their basis of their hydrophobicity;
the double bond in the hydrocarbon chain of
vitamin D2 results in lower hydrophobicity compared
with vitamin D3. This was proven by comparing
the RF values of the two compounds both
in an adsorption system (silica gel eluted with ben-
zene}methanol, 9 : 1) and in a partition system. For
the latter experiment, kieselguhr plates impregnated
with a 10% solution of parafRn oil in benzene were
eluted with different mixtures of methanol}water or
acetonitrile}water. In spite of the extra methyl func-
tion in the side chain of D2 the double bond makes
this compound more polar than D3, as demonstrated
by their RF values in the latter system (Table 2).
For the separation of vitamin D from other lipids,
e.g. in foods, in most cases silica gel plates are ap-
plied. In cases where vitamin D has to be separated
from sterols (cholesterol, -sitosterol, stigmasterol or
lanosterol), however, alumina may be the preferred
stationary phase because silica gel can show too high
an adsorption strength towards free sterols.
In particular cases, multiple development of the
plates may be necessary, e.g. when vitamin D has to
be determined in cod liver oil. Despite the availability
of column chromatographic procedures or, more
4440 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography

Table 2 Representative RF values of vitamins D2 and D3a

Eluent Compound

D2 D3

Methanol}water (v /v)
100 : 0 0.78 0.74
95 : 5 0.56 0.48
90 : 10 0.36 0.33 Compound R1 R2 R3
85 : 15 0.18 0.15
80 : 20 0.05 0.04
-Tocopherol CH3 CH3 CH3
Acetonitrile}water (v/v) -Tocopherol CH3 H CH3
100 : 0 0.60 0.55 -Tocopherol H CH3 CH3
95 : 5 0.50 0.41 -Tocopherol H H CH3
90 : 10 0.38 0.29
85 : 15 0.29 0.25 Figure 3 Structure of tocopherols.
80 : 20 0.23 0.17
75 : 25 0.12 0.09
green, can also be used to visualize vitamin D-related
a
Kieselguhr impregnated with 10% paraffin oil in benzene. compounds.
As already mentioned, for quantiRcation purposes
of vitamin D-related compounds, TLC is often
recently, of HPLC, TLC is still used in the clean-up of incorporated as a clean-up step before ofSine
extracts of lipid-rich matrices such as foods, tissues, measurement either by gas chromatography}mass
oils or multivitamin preparations. The same even spectrometry (GC-MS) or radioimmunoassay (RIA).
holds true for applications of TLC in the sample This clean-up function offers a certain future for TLC
preparation step for separation of the different and HPTLC, especially for laboratories specializing
metabolites of vitamin D. Of course, in biological in RIA and lacking HPLC equipment.
matrices the content of the vitamin D metabolites is
too low to allow visualization by spray reagents.
Typically, the areas corresponding to the compounds
Vitamin E
of interest are then scraped off and wetted with Vitamin E is a collective term for tocopherols and
a small volume of solvent to make it directly amen- tocotrienols, a series of potent antioxidants derived
able to a quantitative radioligand determination. from 6-chromanol by substitution with a saturated
Silica gel and silica gel impregnated with silver (tocopherols) or partially unsaturated (tocotrienols)
nitrate have also been used to monitor the purity isoprenoid side chain and one to three methyl func-
of radiolabelled vitamin D. The advantage of the tions (Figure 3). The principal form is
-tocopherol
application of TLC for this type of study is based on (5,7,8-trimethyltocol) which in nature occurs in the
the fact that TLC offers a total picture of all impu- 2R, 4R, 8R conRguration. Tocol can be regarded as
rities, whereas in liquid chromatography it can never the unsubstituted parent molecule, while
-, - and -
be totally excluded that some impurities are not and -tocopherol form a homologue series of tri-,
eluted. di- and monosubstituted tocols, respectively. The
Polar organic sorbents (e.g. cellulose) or nonpolar dimethyltocols (- and -tocopherol) are positional
bonded phases are infrequently used in vitamin D isomers.
analysis by TLC. However, separation on kieselguhr All vitamin E derivatives have strong reducing
plates impregnated with parafRn oil, described properties, with
-tocopherol being the most biolo-
above, clearly demonstrates that nonpolar bonded gically active homologue. By scavenging free radicals
phases are worth evaluation for the separation of and other oxidative species,
-tocopherol is known to
D2 from D3. protect membrane lipids from peroxidation. Other
functions described for vitamin E remain more con-
Detection
troversial. In the absence of air, vitamin E derivatives
Although not very sensitive, UV absorbance or Su- are quite stable to heat and alkali. However, in the
orescence quenching (the latter on plates containing presence of air they are rapidly oxidized by alkali and
a Suorescence indicator) are universal procedures metal ions. Vitamin E derivatives absorb light in the
that are valid for the detection of vitamin D. UV region (max 292}295 nm;  3530 L mol\1 cm\1)
Iodine vapours, 0.005% aqueous solution of and they are natively Suorescent (ex 205 and 295 nm;
fuchsin or a 0.05% aqueous solution of bromocresol em 330 nm).
 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography 4441

Table 3 TLC conditions for vitamin E-related compoundsa

Compounds Mobile phase Visualization Comments

-Tocopherol in rat liver 1D: Benzene}ethanol (99 : 1, v/v) 20 h at 110}1203C

-, -, -Tocopherol in feeds,
oils

-Tocopherol in pig organs

-, -, -Tocopherol and

-Tocopherol3 in algal lipids

-, -, -, -Tocopherol and

-, -, -, -tocopherol3 in
cereals and plant oils
2D: Hexane}ethanol (9 : 1, v/v)
Petr.ether}diethyl ether}acetic
acid (90 : 10 : 1, v/v)
1D: Benzene}ethanol (99 : 1, v/v)
2D: Hexane}ethanol (9 : 1, v/v)
Hexane}isopropylether (85 : 15,
v/v)
1D: Chloroform
2D: Hexane}isopropylether
(80 : 20, v/v)
0.004% 2,7-dichlorofluorescein
Ethanolic bathophenan-
throline-FeCl3
15 min at 1003C 10% copper(II)
sulfate phosphoric acid 10 min
at 1903C
-Tocopherol and -tocopherol
co-migrate
-Tocopherol and -tocopherol
co-migrate
-Tocopherol and
-tocopherol3 co-migrate

a
All separations were done on silica plates.

volume) offer the best separation. Of the four tocoph-

Chromatographic Conditions
For TLC separation of vitamin E derivatives, silica gel
plates have been widely used. Within the group of
tocopherols migration is correlated with the degree of
ring methylation. However, for the separation of -
from -tocopherol (two dimethyl tocols), often two-
dimensional TLC is necessary with an eluent based on
petroleum ether and diisopropyl ether for the second
TLC run (Table 3).
Resolution of the naturally occurring tocopherols
and tocotrienols also requires two-dimensional TLC.
The separation between -tocotrienol and -tocoph-
erol, in particular, remains an analytical challenge.
Both capillary GC and HPLC have now replaced TLC
approaches, but the solvents used in HPLC often rely
on solvent systems applied in earlier TLC separations.
Traditionally, TLC on silica gel or on alumina has
also played an important role in the clean-up of
extracts of biological materials for the spectro-
photometric analysis of tocopherols/tocotrienols in
the presence of a large excess of interfering lipids. The
whole procedure, however, often included saponiRca-
tion, extraction, column chromatography and two
successive TLC runs before the Rnal spectro-
photometric measurement.
Both silica gel and alumina lend themselves to
separation of tocopherols from their decomposition
products (
-tocopherylquinone,
-tocopherylhydro-
quinone) from other fat-soluble vitamins or from
other lipophilic antioxidants such as butylated hy-
droxytoluene, butylated hydroxyanisole, ethoxyquin,
gallate esters and ascorbyl palmitate.
More recently, reversed-phase chromatographic
conditions have been evaluated for the separation of

-, -, - and -tocopherol. Kieselguhr G plates im-
pregnated with a 10% solution of parafRn oil in
benzene and eluted with methanol}water (9 : 1, by
erols considered, the difference between the RF values
of - and -tocopherol was small.
Alternatively, reversed-phase C18 plates have also
been applied to the separation of
-tocopherol from
other antioxidants or from the other tocopherols.
A new and interesting trend consists of the separation
of D and L enantiomers of tocopherol on chiral
plates (stationary phase, chiral plate solvent: pro-
panol}water}methanol (8.5 : 1.0 : 0.5, by volume) ac-
tivated by heating at 1003C for 15 min). Because of the
different biological activities of both enantiomers, this
type of separation should be further investigated.
Detection
The commonest mode of detecting tocopherols and
tocotrienols on TLC plates is based on quenching the
Suorescence of supports impregnated with a Suor-
escent indicator. Alternatively, tocopherols and
tocotrienols can be visualized by nonspeciRc proced-
ures such as charring preceded by spraying with sul-
furic acid, perchloric acid, nitric acid or 10%
copper(II) sulfate in 8% phosphoric acid. More speci-
Rc visualization procedures are based on the reducing
properties of the vitamin E-related compounds. In
this way, ferric ions are reduced to ferrous ions which
react with
,
-dipyridine or bathophenanthroline to
form a red-coloured complex (Emmerie}Engel reac-
tion). Phosphomolybdic acid and a 20% antimony
pentachloride solution in chloroform both produce
characteristic colour reactions allowing -tocopherol
to be distinguished from -tocopherol, or all four
tocopherols from each other. QuantiRcation of vit-
amin E-related compounds after TLC separation
can be performed either off-plate or on-plate. Off-plate
methods include scraping the areas of interest from the
plate and eluting the compounds with an organic sol-
vent, followed either by a colorimetric measurement or
4442 III / VITAMINS / Fat-Soluble: Thin-Layer (Planar) Chromatography
Figure 4 Structure of vitamin K and related compounds.
by GC determination. On-plate quantiRcation is based
on densitometry of the coloured spots obtained with
chromogenic spray reagents, on the native UV absorb-
ance or on the native Suorescence properties of the
compounds of interest.
Vitamin K
All K-vitamers are derivatives of the same 2-methyl-
1,4-napthoquinone nucleus. The number of isoprene
units or the number of carbon atoms in the side chain
can be used to characterize the molecules. Accord-
ingly, MK-4 contains four isoprene units, or K1(20) has
20 carbon atoms in the side chain (Figure 4). Three
molecules, each representative of a particular group
of K-vitamers, are of special importance:
1. Phylloquinone (Vitamin K1(20)) is synthesized by
green plants and is found in chloroplasts of photo-
synthetic plants. Epoxidation of the double bond
between carbons 2 and 3 of the naphthoquinone
nucleus results in K1(20)-epoxide
2. Menaquinone-n, also called MK-n, is character-
ized by a propenyl side chain often containing
a large number of isoprene units (up to 13), with
n indicating the number of units. Menaquinones
(ranging from MK-4 to MK-13) are synthesized by
bacteria (e.g. Escherichia coli and Staphylococcus
aureus)
3. Synthetic vitamin K3 (menadione or MK-0) does
not occur in nature. In the body, menadione ex-
hibits vitamin K activity by virtue of its in vivo
conversion to menaquinones, chieSy MK-4, by
microorganisms or by alkylating enzymes.
Phylloquinone and the other K-vitamers are de-
stroyed in alkaline media and are sensitive to daylight
(isomer formation). The K-vitamers are easily re-
duced but are fairly stable towards oxidizing condi-
tions and heat. Both vitamin K1(20) and MK-n show
a characteristic UV absorption spectrum with maxi-
ma at 244, 249, 263, 270 and 331 nm (in methanol).
Their molar extinction coefRcient at 249 nm is of the
order of 20 000 L mol\1 cm\1.
Chromatographic Conditions
TLC procedures for vitamin K can be divided into
three main types:
1. adsorption chromatography on silica plates for the
separation of cis-trans isomers
2. argentation chromatography (also on silica layers)
to separate saturated and unsaturated homologues
of vitamin K
3. reversed-phase chromatography for the separation
of methylated and demethylated K-vitamers
These three systems are complementary and will be
treated below.
One major advantage of TLC on silica gel is that
silica gel has little or no tendency to catalyse the
degradation of vitamin K. This is in contrast to
alumina-based separations. Separations on silica are
mainly based on differences in polarity, which makes
the procedure the method of choice for the isolation

of vitamin K from other lipids. In this way, TLC
on silica plates developed with light petroleum
ether}diethyl ether (85 : 15, by volume) is included in
the sample preparation for the determination of vit-
amin K in lipid-rich animal tissues. Although no re-
cent publications have been found, TLC on silica
plates is especially suited for the separation of geo-
metric isomers (cis-trans isomers).
Silica plates have been impregnated with 5}20%
silver nitrate. Under these conditions lipids contain-
ing unconjugated double bonds in their side
chain form complexes with the silver ions and show
a higher retention than the saturated counterparts.
Consequently, separation between saturated (K1(20)),
partly saturated [MK-n (Hn)] and fully unsaturated
homologues (MK-n) becomes possible. On the other
hand, in argentation chromatography the resolution
between cis and trans isomers is completely lost.
Silver ions are not destructive for vitamin K, so sam-
ples can be eluted from the silica afterwards. How-
ever, for high molecular weight menaquinones,
irreversible adsorption to argentation TLC plates has
been reported.
Unlike in argentation TLC, where retention is cor-
related to the degree of unsaturation, in reversed-
phase TLC the retention is based on the length of the
side chain. Both techniques are thus perfectly com-
plementary for the separation of menaquinones.
In addition to silica plates and argentation TLC,
reversed-phase TLC has been applied to vitamin K-
related compounds. Typical eluents consist of water
and an organic solvent such as methanol, acetonitrile
or tetrahydrofuran. However, because of wettability
problems with aqueous solvents, often nonaqueous
reversed-phase conditions are used with dichlorome-
thane and methanol (70 : 30, by vol.) as eluting solvent.
Detection
As with the other fat-soluble vitamins, Suorescence
quenching can be applied to localize the position of
vitamin K-related compounds on a TLC plate. More
sensitive but often destructive for the compounds of
interest include spray reagents such as 70% per-
chloric acid (5}10 min at 1053C), a 0.05% solution
Liquid Chromatography
M. H. Bui, Swiss Vitamin Institute,
University of Lausanne, Lausanne, Switzerland
Copyright ^ 2000 Academic Press
III / VITAMINS / Liquid Chromatography 4443
of rhodamine B in ethanol, a 0.2% anilinonaphtha-
lene sulfonic acid solution in methanol and a 10%
solution of phosphomolybdic acid in ethanol.
Again densitometry (based on reSectance, trans-
mission) has completely replaced visual inspection as
well as the ofSine quantiRcation after elution of the
bands. Densitometry allows internal standardization
and results in a higher degree of sensitivity and speed
of analysis.
General Conclusions
From the above overview it should be clear that TLC
is no longer the method of choice for the analysis of
fat-soluble vitamins. The major reason for this lies in
the great progress made in HPLC. Newer trends such
as HPTLC and densitometric scanning may give TLC
a new momentum but never to the extent that it will
again supersede HPLC as a routine technique for the
determination of fat-soluble vitamins in foods or
biological materials. Undoubtedly, however, modern
instrumental TLC can offer automation, improved
repeatability and more accurate quantiRcation com-
pared to classical TLC.
See also: II/Chromatography: Thin-Layer (Planar):
Spray Reagents. III/Vitamins: Liquid Chromatography.
Further Reading
De Leenheer AP, Lambert WE and Nelis HJ (1992) Modern
Chromatographic Analysis of Vitamins. New York:
Marcel Dekker.
De Leenheer AP and Lambert WE (1996) Lipophilic vit-
amins. In: Sherma J and Fried B (eds) Handbook of
Thin-layer Chromatography, 2nd edn, pp. 1055}1077.
New York: Marcel Dekker.
Friedrich W (1988) Vitamins. Berlin: Walter de Gruyter.
Poole CF and Poole SK (1994) Instrumental thin-layer
chromatography. Analytical Chemistry 66: 27A}37A.
Sherma J (1994a) Modern high performance thin-layer
chromatography. Journal of AOAC International 77:
297}306.
Weins C and Hauck HE (1996) Advances and develop-
ments in thin layer chromatography. LC-GC Inter-
national 9: 710}717.
Introduction
Vitamins are a group of organic compounds essential
to life in very low concentrations. They are either
4444 III / VITAMINS / Liquid Chromatography
insufRciently produced by the body or not at all.
Inadequate vitamin intake causes deRciency disorders
in both humans and animals. The various vitamins
are not related to each other chemically and have
quite different properties. Two main groups, the fat-
soluble and the water-soluble vitamins, may be distin-
guished.
Increased interest in vitamin research, together
with the requirements of food and pharmaceutical
quality control, have led to a proliferation of methods
for vitamin assay, especially by liquid chromato-
graphy (LC). Bioassay methods are no longer used,
but microbiological methods, physicochemical
methods and chromatographic procedures (thin-layer
chromatography, gas chromatography and liquid
chromatography) are commonly employed. Classical
open-column liquid chromatography is occasion-
ally used, but modern high performance liquid
chromatography (HPLC) is by far the technique of
choice for vitamin analysis and is the subject of this
article.
Vitamin analysis is performed to establish the vit-
amin status of humans or animals, to determine the
potency of foods and feeds, and to monitor the stor-
age stability of vitamin-containing pharmaceutical
preparations. Information on the physicochemical
and biochemical aspects of vitamins and vitamin in-
take is widely available in the literature (see Further
Reading).
Sample Preparation
Vitamin A, the carotenoids, and vitamins E, D and
K belong to the group of fat-soluble vitamins, which
are soluble in organic solvents. The water-soluble
vitamins B1, B2, B6, B12, C, biotin, folic acid, pan-
tothenic acid, niacin, choline and inositol are soluble
in water (Table 1). The structures of some fat-soluble
and water-soluble vitamins are shown in Figures 1
and 2.
Table 1 Different kinds of vitamins
Fat-soluble vitamins
Vitamin A (retionol)
Carotenoids
Vitamin D (ergocalciferol, vitamin D2; cholecalciferol, vitamin D3)
Vitamin E (
-, -, -tocopherols plus tocotrienols)
Vitamin K (phylloquinone, vitamin K1; menadione, vitamin K3)
Sample preparation prior to the Rnal chromato-
graphic analysis is highly dependent upon the nature
of the matrix. Minimal preparation is necessary for
the analysis of concentrated solutions. For complex
biological matrices more elaborate sample prepara-
tion procedures may be necessary. A recovery test
is highly recommended. This consists of adding
a known amount of pure vitamin, approximatively
equal to the estimated value in the sample, and pro-
cessing the fortiRed sample in the same way as the
sample itself. Loss of vitamin during analysis should
not exceed 6%.
Fat-Soluble Vitamins
For fat-soluble vitamin assays all manipulations must
be carried out in subdued light, in dark glass vessels,
and in a nitrogen atmosphere to avoid isomerization
and oxidation.
In foodstuffs, major interferences in assays for vit-
amin A, carotenoids, and vitamins E, D and K are
caused by the large excess of other lipids. The vitamin
A, carotenoids, and vitamins E and D contents are
measured generally after alkaline hydrolysis with
ethanolic KOH under a nitrogen stream at 60}803C
for 20}30 min in the presence of an antioxidant.
Pyrogallol, hydroquinone, ascorbic acid (vitamin C)
and butylated hydroxytoluene (BHT) are the most
common antioxidants used during this manipulation.
After saponiRcation the free retinol, carotenoids, vit-
amin E and vitamin D are extracted into n-hexane or
petroleum ether and evaporated to dryness. Vitamin
A, carotenoids and vitamin E are redissolved in an
organic solvent compatible with the chromatographic
method to be employed. Vitamin K needs milder
conditions for extraction from protein. Both vitamins
D and K may require further puriRcation before
chromatography.
In the analysis of serum, vitamin A or retinol is
liberated from its binding protein by denaturation
with acetonitrile, ethanol or methanol. An internal
Water-soluble vitamins
Vitamin B1 (aneurin, thiamin)
Vitamin B2 (riboflavin)
Vitamin B6 (pyridoxine, pyridoxal, pyridoxine, pyridoxol)
Vitamin B12 (cyanocobalamin)
Vitamin C (ascorbic acid dehydroascorbic acid)
Biotin
Folic acid (vitamin B9)
Pantothenic acid (vitamin B5, panthenol)
Niacin (nicotinamide, vitamin PP)
Choline
Inositol (myo-inositol)
 III / VITAMINS / Liquid Chromatography 4445

Figure 1 Chemical structures of some fat-soluble vitamins.

4446 III / VITAMINS / Liquid Chromatography
Figure 2 Chemical structures of some water-soluble vitamins.
standard, e.g. either retinyl acetate or tocol, is gener-
ally added for quantitation purposes. After protein
precipitation the free retinol is extracted with 1%
BHT in n-hexane solution, evaporated to dryness
under nitrogen and subjected to chromatographic
separation as above. Vitamin E and carotenoids are
extracted in the same manner. Vitamins A, E and
carotenoids can be simultaneously injected for LC
separation. The analysis of vitamin D and trace quant-
ities of vitamin D metabolites, e.g. 1,25-dihydroxy-
cholecalciferol and 25-hydroxycholecalciferol, and
also the analysis of vitamin K in human plasma,
require an additional step for prepuriRcation using
column extraction or a semipreparative LC system
prior to the Rnal HPLC separation.
Water-Soluble Vitamins
Water-soluble vitamins are in general more stable
than fat-soluble vitamins, although vitamin B2 and to
a lesser extent vitamin B12 (folic acid) are light sensi-
tive. All manipulations should therefore be performed
in subdued light. No special treatment of samples in

pharmaceutical preparations is required before chro-
matography. In biological Suids or foodstuffs pre-
puriRcation and/or derivatization of the compounds
are necessary before LC separation. The methods
include acid extraction followed by enzymatic hy-
drolysis with takadiastase, papain or acid phos-
phatase, sometimes with pre-column or post-column
chromatography. Trichloroacetic, perchloric and
metaphosphoric acids are usually preferred for acid
extraction. Depending on the aim of the investigation,
vitamins can be determined in their free forms or in
both free and phosphorylated forms. In the latter case
the enzymatic hydrolysis step is omitted.
In blood, plasma and food, vitamin B1 in protein-
free extract is oxidized by agents such as [Fe(CN)6]3\,
cyanogen bromide or mercuric chloride to thioch-
rome in a pre- or post-column chromatography reac-
tor. For pre-column chromatography two different
procedures are used. In the Rrst, the thiochrome ex-
tract is neutralized by concentrated phosphoric acid
to ensure a pH level compatible with the C18 column
used for the separation and to eliminate possible
pH-dependent alkaline degradation of thiochrome to
its disulRde. It is then centrifuged and the supernatant
injected into the HPLC. In the second procedure,
isobutyl alcohol is used to extract thiochrome after
alkaline oxidation. Aliquots of the extracts are then
chromatographed.
After acid extraction vitamin B2 is readily detected
owing to its intense Suorescence.
Since vitamin B6 is present in six chemical forms,
there are methods for the simultaneous separation of
the three free forms and the three phosphorylated
forms as well as methods for determining the sum of
all the forms. Pyridoxamine is transformed into py-
ridoxal by reaction with glyoxylic acid in the presence
of Fe2# as catalyst. The pyridoxal is then reduced to
vitamin B6 pyridoxol by the action of sodium
borohydride in an alkaline medium before LC separ-
ation. Semicarbazide is also used for post-column
derivatization of vitamin B6.
In multivitamin}multimineral preparations, vit-
amin B12 or cyanocobalamin is extracted with a mix-
ture of dimethyl sulfoxide (DMSO) and water, or
ammonium pyrrolidine dithiocarbamate and citric
acid in DMSO and water. The extract is centrifuged
and the supernatant is diluted with water before con-
centration and clean-up by solid-phase extraction us-
ing a quaternary amine and a phenyl column in series
before LC separation. There are few LC methods for
the determination of vitamin B12 in human plasma
and food.
In biological Suids and foodstuffs a treatment for
removing protein is a major requirement for vitamin
C assay. Protein precipitation may be done by organic
III / VITAMINS / Liquid Chromatography 4447
reagents (methanol or acetonitrile) or mineral acids
(perchloric, metaphosphoric acid, etc.). Aqueous
solutions of vitamin C are rapidly oxidized on expo-
sure to air. Stabilizers such as hydrogen sulRde and
1,4-dithio-DL-threitol have also been employed. The
deproteinization may be followed by an enzymatic
oxidation of ascorbic acid to dehydroascorbic acid,
which is transformed with 1,2-phenylenediamine to
its quinoxaline derivative for Rnal separation.
Biotin, or vitamin H, is very stable. However, the
limitation of HPLC lies in the lack of a suitable
detection system. There are applications of LC to
pharmaceutical products containing at least 300
g
biotin per tablet. In pharmaceutical products and feed
premix, biotin is extracted from the matrix with buf-
fer, followed by puriRcation and concentration by
solid-phase extraction and separation by LC. How-
ever, there are few LC methods for the estimation of
biotin in biological samples.
Folic acid (pteroylglutamic acid; also called vit-
amin M) and its derivatives are stable substances.
Folic acid may be determined simultaneously with
other water-soluble vitamins in pharmaceutical prep-
arations. In food products folates are extracted from
the matrix with buffer and enzymes (e.g. hog kidney
and chicken pancreas, or rat plasma conjugase,
-
amylase and protease together), followed by puriRca-
tion and concentration by solid-phase extraction or
with afRnity chromatography before Rnal separation.
In pharmaceutical preparations panthenol, pan-
thotenic and its salt (vitamin B5) are extracted with
a phosphate solution. The extract is centrifuged, Rl-
tered and separated by LC. There are few methods for
the determination of pantothenic acid and its salt in
food products and biological Suids.
Niacin (or nicotinic acid) and nicotinamide are the
two different forms of vitamin PP (so called for its
pellagra-preventive factor). Nicotinamide is the form
of the vitamin generally found in human plasma.
Plasma is deproteinated with acetone/chloroform, the
organic layer evaporated to dryness, and the meth-
anolic extract of the residue separated by a reversed-
phase HPLC. Isonicotinic acid is used as an internal
standard. Urine is puriRed by extraction with chloro-
form, the aqueous phase evaporated, and taken for
separation by LC. In foods vitamin PP is present
mostly in its phosphorylated forms. Hydrolysis is
necessary to break the ester bonds, releasing the total
vitamin PP content of the food for assay. In food
products niacin is extracted with buffer and enzyme.
The sample extracts are puriRed through an ion ex-
change column (e.g. Dowex 1-X8 resin) before
HPLC.
In multivitamin preparations 0.1 mol L\1 hydro-
chloric acid is used to extract the vitamins and
4448 III / VITAMINS / Liquid Chromatography

DMSO containing anhydrous citric acid is used to

disperse the multivitamin}multimineral preparation,
since vitamin B6 is not completely extracted by either
0.1 mol L\1 hydrochloric acid or DMSO owing to
adsorption of the vitamin to the minerals. The extrac-
tion of nicotinamide is not impaired by the addition
of citric acid to DMSO.
Choline in plant material is extracted with iso-
propanol containing internal standards and p-nitro-
benzylhydroxylamine hydrochloride for the forma-
tion of p-nitrobenzyl oximes. The extract is puriRed
by solid-phase extraction (C18 and ion exchange),
after which the choline fraction is benzoylated to
yield UV-absorbing derivatives. In biological samples
choline is extracted with formic acid in acetone con-
taining an internal standard. After puriRcation the
sample is separated by LC.
For the analysis of inositol mono- and diphosphate
isomers in foods the method involves extraction
of samples with hydrochloric acid and separation of
inositol phosphates by anion exchange chromato-
graphy.
Liquid Chromatography
Liquid chromatography is an extremely valuable
method for separation, identiRcation and quantitat-
ion of the different vitamins. Excellent separations
can be achieved in a reasonable time for routine
analysis.
Fat-Soluble Vitamins
For fat-soluble vitamins normal-phase and reversed-
phase chromatography are used.
In the normal-phase modes, silica and nonpolar
mobile phases containing n-hexane or petroleum
ether with a small percentage of a more polar solvent
are used. Addition of a small amount of water or
alcohol (e.g. ethanol) regulates the sorbent activity,
reduces peak tailing and gives better reproducibility
of retention times. Silica is the adsorbent of choice for
the separation of cis/trans isomers and diastereo-
isomers. Selectivity on silica is determined by the
number and the nature of the functional groups as
well as the overall steric conRguration (position of the
double bonds) of the molecule (Figure 3).
In the reversed-phase mode, hydrophobic column
packings (C18, C8, etc., bonded to a silica surface) are
used together with an aqueous buffered mobile phase
and a water-miscible organic solvent (i.e. methanol,
acetonitrile).
Retinol analysis in biological Suids and foods
is performed using both normal-phase and reversed-
phase chromatography. For normal-phase (or
liquid}solid) chromatography there is compatibility
Figure 3 Example of an HPLC separation of
-, -, - and
-tocopherol from a wheat germ oil sample with
-tocopherol-
acetate added. Peaks: 1,
-tocopherolacetate; 2,
-tocopherol; 3,

-tocotrienol; 4, -tocopherol; 5, -tocopherol; 6, -tocopherol.
Experimental conditions: stationary phase, Lichrosorb Si 60,
7
m; column dimension, 250;4.6 mm; mobile phase, n-
hexane/dioxane (97 : 3 v/v); flow rate, 1.0 mL min\1; injection
volume, 20
L; detection, fluorimetric with excitation at 295 nm
and emission at 330 nm. (Reproduced with permission from Fed-
eral Office of Public Health, 1989.)
between the sample extraction solvent and the LC
mobile phase; this avoids peak artefacts, especially
for lipid extracts. Geometrical isomers such as 11-cis,
13-cis, 9-cis and all-trans retinol are well resolved.
Retinol serum determination by reversed-phase
chromatography allows the use of retinol acetate as
internal standard which is well separated from re-
tinol, unlike the case for liquid}solid chromatogra-
phy. Meanwhile there is an additional step of
evaporation of the extraction solvent in the sample
preparation procedure before LC.
The most appropriate systems for the separation of
polar and nonpolar carotenoids include the use of
polymeric C18 or C30 bonded phases without end-
capping in conjunction with a moderate pore-size
packing column and a methanol-based mobile phase
(to obtain a good recovery). Cis-isomers of -caro-
tene are largely resolved from each other and from
other carotenes. Separation of lutein and zeaxanthine
is also obtained with this system. Better separations
of the xanthophylls are also observed. Accurate caro-
tenoid measurements require the right selection of
column and mobile phase and, due to the large degree
of variability in the purity of commercial carotenoid

Figure 4 Chromatogram of a human serum sample. Peaks:
RET, retinol; IS, retinol acetate (internal standard); LUT, lutein;
ZEA, zeaxanthin; -TOC, -tocopherol; -TOC, -tocopherol;
-
TOC,
-tocopherol; LYC, lycopene;
-CAR,
-carotene; -CAR,
-carotene; trans-CAR, trans--carotene; a, unidentified caro-
tenoid.
Experimental conditions: stationary phase. Lichrosorb RP-18,
7
m; column dimension, 250;4.6 mm and 15;3.2 mm guard
column; mobile phase, acetonitrile/tetrahydrofuran/methanol
(68 : 22 : 7 v/v/v) adjusted to 100% with ammonium acetate; flow
rate, 1.5 mL min\1; injection volume, 15
L; detection, programm-
able and variable-wavelength UV/Vis, 325 nm from 0 to 3.0 min,
450 nm from 3.0 to 4.9 min, 290 nm from 4.9 to 7.4 min, 470 nm
from 7.4 to 12.0 min and 450 nm from 12.0 to 15.0 min. (Repro-
duced with permission from Bui, 1994.)
preparations, special precautions must be taken dur-
ing calibration. Simultaneous determination of caro-
tenoids, retinoids and tocopherols in serum and foods
is performed on a C18 column using wavelength-
programmable UV-visible (Figure 4), Suorescence or
electrochemical detection.
The same separation conditions are used for the
separation of
-tocopherol and retinol in biological
Suids and foods.
-, -, - and -Tocopherol as well
as
-, -, - and -tocotrienol are well resolved only by
normal-phase chromatography. On an RP-18 column

-, - and -tocopherol isomers are all separated,
but not - and -tocopherols. Tocol as internal stan-
dard is also used for vitamin E determination in
serum.
Retinol and
-tocopherol in biological Suids, foods
and pharmaceutical preparations can be separated
simultaneously by normal- or reversed-phase HPLC.
In biological Suids analyses of vitamin D and its
metabolites are performed using solid-phase extrac-
tion (SPE) cartridge coupled to semipreparative HPLC
on a silica column and analytical HPLC on an oc-
tadecyl (C18) column with UV or electrochemical de-
tection (Figure 5). They can also be puriRed by one or
two preparative HPLC steps on silica and quantiRed
III / VITAMINS / Liquid Chromatography 4449
Figure 5 Chromatogram of HPLC-ECD for 24,25-dihydroxy-
cholecalciferol (24,25-(OH)2D3).
Experimental conditions: stationary phase, Nucleosil C18, 5
m;
column dimension, 300;7.5 mm; mobile phase, 5% (v/v) meth-
anol in acetonitrile with 0.025 mol L\1 HClO4; flow rate,
1.2 mL min\1; injection volume, 50
L; detection, dual-electrode
electrochemical detection: detector 1 (#0.20 V); detector
2 (#0.60 V). (Reproduced with permission from Masuda et al.,
1997.)
by HPLC on a C18 column with UV detection. Vitamin
D2 or D3 may be used as internal standard (Figure 6).
The analysis of vitamin K (phylloquinone,
menaquinone and epoxides), like vitamin D, uses
normal-phase semipreparative LC followed by an
analytical reversed-phase column with UV or electro-
chemical detection (Figure 7). Water-soluble mena-
dione sodium bisulRte or vitamin K3 in animal feeds is
determined by reversed-phase HPLC with UV detec-
tion. To improve vitamin K3 detection limits, post-
column reaction Suorimetric detection is used.
Menadione is hydrogenated by sodium borohydride
to 2-methyl-1,4-dihydroxynaphthalene, which is de-
tected Suorimetrically.
Water-Soluble Vitamins
Water-soluble vitamins are separated using ion ex-
change (IEC), normal-phase or reversed-phase chro-
matography.
Ion exchange chromatography is the preferred
method of separation for the analysis of strongly ionic
compounds. The chromatographic separation may be
optimized by altering the pH or ionic strength of the
mobile phase. Reversed-phase chromatography is the
method of choice for water-soluble vitamins. Rever-
sed-phase columns such as C18 and mobile phase NH2
have been employed. The mobile phase is a mixture of
methanol or acetonitrile with an acetate or phosphate
buffer. For ionic compounds the reversed-phase
ion pair mode is generally used. Unlike conventional
4450 III / VITAMINS / Liquid Chromatography

Figure 7 Analytical HPLC chromatogram of a kiwi fruit (A) with-

out purification and (B) with purification with semipreparative
HPLC.
Experimental conditions: stationary phase, Vydac 201 TP 548
5
m; column dimension, 250;4.6 mm; mobile phase, 96%
methanol/0.05 mol of sodium acetate buffer (pH 3); flow rate,
1.5 mL min\1; injection volume, 30
L; detection, dual-electrode
electrochemical detection: upstream electrode (!1.1 V); down-
stream electrode (0 V). (Reproduced with permission from Koivu
et al., 1997.)

level at pH'8, the mobile phase should contain

a buffer. Polymeric C18 packings are more suitable for
these high pH conditions (Figure 8).
C18 columns are used for the determination of ribo-
Figure 6 Analytical HPLC chromatograms of (A) standard mix- Savin and its derivatives, Savin mononucleotide
ture of ergocalciferol (D2) and cholecalciferol (D3); (B) D2 (internal (FMN) and Savin}adenine dinucleotide (FAD). The
standard, IS) and D3 in chicken sample; and (C) D2 (IS) and D3 in compounds are separated isocratically with a mixture
pork liver.
Experimental conditions: stationary phase, Zorbax ODS#
Vydac 201 TP 54 5
m; column dimension, 250;4.6 mm; mobile
phase, 4% water in methanol; flow rate, 1.0 mL min\1; injection
volume, 50
L; detection, UV at 264 nm. (Reprinted from Horvath
CsG (ed.) (1980) High Performance Liquid Chromatography, New
York: Academic Press. Copyright ^ 1980 by Academic Press.)

IEC, this technique can separate nonionic and ionic

compounds simultaneously. The chromatographic
separation may be optimized by altering the ion pair
reagent, pH and ionic strength of the mobile phase.
Water-soluble vitamins (vitamin B1, thiamin, B2,
riboSavin or riboSavin 5-monophosphate; B6, py-
ridoxine and nicotinamide) in commercial vitamin
preparations can be separated using either strong
cation exchange resins or reversed-phase chromatog-
raphy using an ion pair reagent (e.g. sodium alkane -
sulfonate, dioctyl sodium sulfosuccinate, tetrabutyl am-
monium phosphate) in the eluent with UV detection. Figure 8 Chromatogram of thiamin (thiochrome) and riboflavin
Similar LC methods are used for the separation of in a skimmed milk sample using fluorescence detection.
thiamin in foods and biological Suids, usually with Experimental conditions stationary phase, Ultrasphere ODS
5
m; column dimension, 250;4.6 mm; mobile phase, meth-
Suorescence detection. In pre-column procedures, sil-
anol/water (20#80) containing 0.005 mol L\1 tetrabutylammo-
ica, C18, NH2 and poly(styrene}divinyl benzene) nium phosphate pH 7.5; detection, fluorimetric with excitation at
phases are used. Since the intensity of thiochrome 360 nm and emission 425 nm. (Reproduced with permission from
Suorescence depends on pH and reaches a steady Augustin, 1984.)
 III / VITAMINS / Liquid Chromatography 4451

Figure 10 (A) Representative chromatogram of standard vit-

Figure 9 Chromatograms of B1, B2 and B6 vitamers in standard
and in whole blood.
Experimental conditions: stationary phase, Nova Pack
C18 5
m; column dimension, 125;4.6 mm; mobile phase, 20%
methanol in ion pair solution (at least 70
L of di-N-butylamine
solution per litre of the eluent); flow rate, 1.0 mL min\1, detection,
UV at 254 nm for B1; fluorimetric with excitation at 290 nm and
emission at 395 nm for B2 . (Reprinted from Setrell WH Jr and
Harris RS (eds) (1967) The Vitamins, 2nd edn. New York/London:
Academic Press. Copyright ^ 1967 by Academic Press.)
amin B6 vitamer, 4-deoxypyridoxine (dPN) and 4-pyridoxic acid
(4-PA). Peaks: A, pyridoxal phosphate PLP (8 pmol); B, 4-PA
(10 pmol); C, pyridoxamine phosphate PMP (5.5 pmol); D, py-
ridoxal PL (10 pmol); E, pyridoxine PN (10 pmol); F, dPN
(12 pmol); G, pyridoxamine PM (6 pmol). (B) Vitamin B6 vitamer
profile of human plasma.
Experimental conditions: stationary phase, Ultramex C18 guard
column (30;4.6 mm) 3
m and Ultramex C18 column
(150;4.6 mm) 3
m; mobile phase, (A) 0.033 mol L\1 phosphoric
acid containing 0.01 mol L\1 1-octanesulfonic acid adjusted to pH
2.2 with 6 mol L\1 potassium hydroxide; (B) 0.33 mol L\1 phos-
phoric acid in 10% (v/v) 2-propanol adjusted to pH 2.2 with
6 mol L\1 potassium hydroxide; flow rate, 1.2 mL min\1; injection
volume, 25
L; detection, fluorimetric with excitation at 328 nm
and emission at 393 nm. (Reprinted from Setrell WH Jr and Harris
RS (eds) (1968) The Vitamins, 2nd edn. New York/London: Aca-
demic Press. Copyright ^ 1967 by Academic Press.)
of methanol and water (or buffer solution) using
Suorimetric detection (Figure 9).
Simultaneous determination of the six chemical
forms of vitamin B6 in foods and biological samples is
performed by IEC or reversed-phase chromatography,

Figure 11 Chromatographic separation of vitamin B6 vitamers in yeast. (A) With deletion of the deamination and reduction steps; (B)
with deletion of the reduction step: (C) without deletion of either step.
Experimental conditions: stationary phase, Lichrospher 60 RP Select B octylsilyl 5
m; column dimension, 250;5 mm; mobile
phase, acetonitrile/0.05 mol L\1 potassium dihydrogen phosphate (4 : 96) containing 0.5;10\3 mol L\1 sodium heptanesulfonate
adjusted to pH 2.5 with phosphoric acid; flow rate, 1.0 mL min\1; injection volume, 20
L; detection, fluorimetric with excitation at
290 nm and emission at 395 nm. (Reprinted from Setrell WH Jr and Harris RS (eds) (1972) The Vitamins, 2nd edn. New York/London:
Academic Press. Copyright ^ 1972 by Academic Press.)
4452 III / VITAMINS / Liquid Chromatography
Figure 12 Chromatogram of multivitamin}multimineral tablets
at (A) 50 nm and (B) 360 nm after preconcentration and clean-up
by solid-phase extraction. Peaks: 1, vitamin B12; 2, vitamin B2.
Experimental conditions: stationary phase, Bondapack
C18 10
m; column dimension, 150;39 mm; mobile phase, (A)
methanol/water (10 : 90); (B) methanol/water (90 : 10); gradient
elution, linear gradient to 50% B for the first 15 min, followed by
100% B for the next 2 min and maintained isocratically for 10 min;
flow rate, 1.0 mL min\1; injection volume, 200
L; detection, UV
at 550 nm. (Reproduced with permission from Dalbacke and
Dahlquist, 1991.)
with or without ion pair reagents; detection is by
Suorimetry (Figure 10). The vitamin B6 content of
foods can also be determined by ion pair HPLC after
Figure 13 Chromatogram of the main folate forms found in fortified white bread.
Experimental conditions: stationary phase, Phenomenex Ultramex C18 5
m; column dimension, 250;4.6 mm; mobile phase,
33 mmol L\1 phosphoric acid, pH 2.3 with increasing acetonitrile; gradient elution, 5% (v/v) acetonitrile maintained isocratically for the
first 8 min, linear gradient to 17.5% (v/v) within 25 min; flow rate, 1.0 mL min\1; detection, UV at 280 nm. (Reproduced with permission
from Pfeiffer et al., 1997.)
pre-column derivatization of the free and phos-
phorylated vitamin into pyridoxol (Figure 11).
Vitamin B12 is separated from other water-soluble
vitamins in pharmaceutical preparations by reversed-
phase using a methanol/water gradient with detection
at 550 nm (Figure 12).
Reversed-phase chromatography is mostly used for
ascorbic acid determination. In foods total vitamin
C (ascorbic acid and its oxidized form, dehydroascor-
bic acid) are determined using ion pair chromatogra-
phy with UV detection. In biological Suids and foods
total vitamin C, as its quinoxaline derivative, is separ-
ated on a C18 column with Suorescence detection. The
determination of ascorbic acid in plasma can also be
achieved using a C18 column and electrochemical de-
tection. Another procedure for vitamin C determina-
tion consists of Rrst measuring the ascorbic acid
present, then reducing the dehydroascorbic acid, at
neutral pH, with dithiothreitol, and Rnally measuring
the total ascorbic acid. The dehydroascorbic acid is
determined by difference. The separation is on a C18
column with electrochemical detection.
After a clean-up procedure, biotin in pharmaceut-
ical products is assayed using a C18 column with
methanol/water as the mobile phase and UV detec-
tion. Extracts of folates (folate monoglutamates and
folic acid) in food and biological samples after puriR-
cation are separated by gradient elution and UV or
Suorescence detection (Figure 13).
Pantothenic acid is separated from other water-
soluble vitamins with an isocratic system on an
aminopropyl bonded phase using a mixture of
acetonitrile/phosphate buffer as mobile phase and UV
detection (Figure 14). Panthenol in multivitamin

Figure 14 Chromatogram of a high potency B complex tablet
extract. Peaks: A, niacinamide; B, vitamin B6; C, vitamin B2; D,
vitamin B12; E, unknown; F, pantothenic acid.
Experimental conditions: stationary phase, Hibar II Lichrosorb
NH2 10
m; column dimension, 250;4.6 mm; mobile phase,
0.005 mol L\1 monobasic potassium phosphate (pH
4.5)/acetonitrile (13 : 87 v/v) 0.01 mol L\1 1-octanesulfonic acid;
flow rate, 2.0 mL min\1; injection volume, 10
L; detection, UV at
210 nm. (Reproduced with permission from Hudson and Allen,
1984.)
preparations is determined on a C18 column with
a gradient system using an ion pair reagent (e.g.
sodium hexanesulfonate) and UV detection. Niacin
and nicotinamide are also separated on a C18 column
using an ion pair reagent with UV detection.
Isonicotinic acid is used as an internal standard.
Cation exchange chromatography has been used to
determine with UV or electrochemical detection. Sep-
arations of inositol mono- and diphosphate isomers
in foods is performed on an anion exchange column
using a sodium acetate in sodium hydroxide gradient
with electrochemical detection.
III / VITAMINS / Liquid Chromatography 4453
Future Developments
Liquid chromatography is the method of choice for
vitamin analysis in pharmaceutical products, foods,
feeds and especially in biological Suids. In biological
sample analysis LC affords separation of the vit-
amins, their related compounds and various metab-
olites for nutrition research. The main problem
encountered in biological materials is the detection
limit, particularly for water-soluble vitamins. There
are two main areas where developments are necessary
for future vitamin LC. First, automatic sample prep-
aration techniques involving vitamin puriRcation and
enrichment, e.g. automating solid-phase extraction,
need to be improved. Second, the coupling LC and
mass spectrometric (MS) detection needs to be further
developed. These techniques may become leading
methods in vitamin analysis in the future.
See also: II/Chromatography: Liquid: Mechanisms:
Normal Phase; Mechanisms: Reversed Phases. III/Carot-
enoid Pigments: Supercritical Fluid Chromatography.
Food Technology: Supercritical Fluid Chroma-
tography. Vitamins: Fat-Soluble: Thin-Layer (Planar)
Chromatography; Water-Soluble: Thin-Layer (Planar)
Chromatography.
Further Reading
Augustin J (1984) Simultaneous determination of thiamine
and riboSavine by liquid chromatography. Journal
of the Association of OfTcial Analytical Chemists 67(5):
1012}1015.
Ball GFM (ed.) (1988) Fat-Soluble Vitamin Assays in Food
Analysis. London and New York: Elsevier Applied
Science.
Ball GFM (ed.) (1994) Water-Soluble Vitamin Assays in
Human Nutrition. London: Chapman & Hall.
BoK tticher B and BoK tticher D (1987) A new HPLC-method
for the simultaneous determination of B1-, B2- and
B6-vitamers in serum and whole blood. International
Journal for Vitamin and Nutrition Research 57:
273}278.
Bui MH (1994) Simple determination of retinol,
-tocoph-
erol and carotenoids (lutein, all-trans-lycopene,
- and
-carotene) in human plasma by isocratic liquid
chromatography. Journal of Chromatography B 654:
129}133.
Dalbacke J and Dahlquist I (1991) Determination of vit-
amin B12 in multivitamin}multimineral tablets by high-
performance liquid chromatography after solid-phase
extraction. Journal of Chromatography 541: 382}394.
De Leenheer AP, Lambert WE and Nelis HJ (eds) (1992)
Modern Chromatographic Analysis of Vitamins, 2nd
edn. New York: Marcel Dekker.
Federal OfRce of Public Health (1989) Vitaminbestimmun-
gen in Lebensmitteln und Kosmetica Kapitel 62. Bern,
Switzerland: The Federal OfRce of Public Health.
4454 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography
Gaby SK, Bendich A, Singh VN and Machlin LJ (eds)
(1991) Vitamin Intake and Health. New York: Marcel
Dekker.
Hudson TJ and Allen RJ (1984) Determination of pan-
tothenic acid in multivitamin pharmaceutical prepara-
tions by reversed-phase high performance liquid
chromatography. Journal of Pharmaceutical Sciences
73: 113}115.
Koivu TJ, Piironen VI, Henttonen SK and Mattila PH
(1997) Determination of phylloquinone in vegetables,
fruits and berries by high performance liquid
chromatography with electrochemical detection. Jour-
nal of Agricultural and Food Chemistry 45: 4644}4649.
Masuda S, Okano T, Kamao M, Kanedai Y and Kobayashi
T (1997) A novel high-performance liquid chromato-
graphic assay for vitamin D metabolites using a coulo-
metric electrochemical detector. Journal of Pharmaceut-
ical and Biomedical Analysis 15 (9}10): 1497}1502.
Mattila PH, Piironen VI, Uusi-Rauva EJ and Koivistoinen
PE (1995) Contents of cholecalciferol, ergocalciferol,
Water-Soluble: Thin-Layer (Planar) Chromatography
J. C. Linnell, Royal Free and University College Medi-
cal School, London, UK
Copyright ^ 2000 Academic Press
As a tool, chromatography has long been important
for the separation of vitamins from complex mixtures
and their initial isolation and identiRcation would
have been greatly hampered without the use of paper,
column or thin-layer chromatography (TLC). While
more sophisticated chromatographic techniques are
now widely available, TLC has great advantages in
terms of its simplicity and Sexibility of use.
The vitamins classiRed as water-soluble are all
compounds important in human metabolism either as
coenzymes or their precursors which the body cannot
make for itself (Figure 1). The recommended daily
allowance (RDA) of each vitamin ranges from hun-
dreds of milligrams to just a few micrograms a day
(Table 1). These compounds have few properties in
common apart from their water-solubility, but this
fact alone makes TLC an excellent technique for their
separation, particularly in pharmaceutical prepara-
tions and food products. Even at physiological con-
centrations, TLC is widely used after extraction of the
vitamins from tissues or body Suids. This generally
needs to be under acid conditions. Since most of these
compounds are unstable at high pH. Some are in
addition very light-sensitive. Following TLC separ-
ation, special methods of detection may also be
required, since tissue levels of most water-soluble
vitamins are low or very low.
and their 25-hydroxylated metabolites in milk products
and raw meat and liver as determined by HPLC. Journal
of Agricultural and Food Chemistry 43: 2394}2399.
Packer L and Fuchs J (eds) (1993) Vitamin E in Health and
Disease. New York: Marcel Dekker.
Pfeiffer CM, Rogers LM and Gregory III JF (1997) Deter-
mination of folate in cereal-grain food products using
trienzyme extraction and combined afRnity and rever-
sed-phase liquid chromatography. Journal of Agricul-
tural and Food Chemistry 45: 407}413.
Reitzer-Bergaentzle M, Marchioni E and Hasselmann
C (1993) HPLC determination of vitamin B6 in foods
after pre-column derivatization of free and phos-
phorylated vitamers into pyridoxol. Food Chemistry 48:
321}324.
Sebrell WH Jr and Harris RS (eds) (1972) The Vitamins,
2nd edn. New York/London: Academic Press.
Sharma SK and Dakshinamurti K (1992) Determination of
vitamin B6 vitamers and pyridoxic acid in biological
samples. Journal of Chromatography 578: 45}51.
Thiamin (Vitamin B1)
Thiamin occurs in plant and animal tissues and the
richest sources are seeds and nuts, peas and beans,
cereals and yeast. Fish and meat, notably pork, are
also good sources. Thiamin is commonly available as
its monohydrochloride, but it also forms acid salts
and esters with nitric and phosphoric acids. Meta-
bolically, thiamin is required as the coenzyme
thiamin pyrophosphate for the mitochondrial meta-
bolism of glucose and pyruvate.
Thiamin may be extracted from tissues, foodstuffs
or pharmaceutical preparations with aqueous alcohol
mixtures at a pH of 4}6 and separated from closely
related compounds and metabolites by TLC on
cellulose or silica gel. Various mobile phases have
been successfully used, including isopropanol}
water}trichloracetic acid}ammonia (71 : 9 : 20 :
0.3) and butan-1-ol}acetic acid}water (40 : 10 : 50).
Thiamin may be separated from its hydrolysis and
oxidation products by TLC/densitometry and other
chromatographic techniques have been reviewed.
Sandwich-type chambers afford rapid separation of
thiamin from other water-soluble vitamins by TLC
on silica gel GF254 and the spots then located under
UV light. An alternative technique for the quantitat-
ion of thiamin in pharmaceutical products involves
the use of high performance TLC (HPTLC) and post-
separation derivatization with a hexacyanoferrate
(III)-sodium hydroxide reagent and Suoroden-
sitometry, sensitive down to 500 pg per spot. Other
modiRcations include the use of a Rbreoptic probe
 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography 4455

Figure 1 Structural formulae of water-soluble vitamins.

4456 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography
Table 1 Recommended daily allowancesa water-soluble
vitamins
Vitamin RDA
Ascorbic acid 30}75 mg
Nicotinic acid 15}20 mg
Pyridoxine 1}3 mg
Riboflavin 1.5}2.0 mg
Thiamin 1}2 mg
Folic acid 300
g
Cobalamin 1}2
g
a
There is no quoted RDA for biotin or pantothenic acid.
to improve measurement of thiamin in the nanogram
range.
Ribo]avin (Vitamin B2)
RiboSavin and other Savinoids occur in dairy pro-
duce, meat and to a lesser extent in cereals. Flavins
are stable to heat and acid but are destroyed by
exposure to light. Ultraviolet irradiation of riboSavin
in acid or neutral solution gives rise to the Suorescent
compound lumichrome, whereas in alkaline solutions
irradiation produces lumiSavin. Flavins are required
in the body as their coenzymes Savin mononucleotide
and Savin adenine dinucleotide, which are involved in
redox reactions involving one- and two-electron
transfers and linked to many energy-dependent pro-
cesses in the body.
Pharmaceutical preparations containing riboSavin
may be analysed by applying concentrated ethanolic
extracts to silica gel TLC plates developed in
butanol}benzene}acetic acid}water (8 : 7 : 5 : 3) or
butanol}acetic acid}water (9 : 4 : 5). Foods, tissue
samples and urine each require particular methods of
sample preparation and these methods and the sol-
vent systems successfully employed have been re-
viewed elsewhere. A dark room is required for sample
preparation and chromotography of Savins to pre-
vent photolytic degradation. The Suorescent property
of Savins provides a convenient means of detection
and spots may be located under radiation at 254 and
366 nm. HPTLC followed by Rbreoptic Suorimetry
has been used to measure riboSavin in vitamin mix-
tures and can detect 48}320 ng. Separation is also
effective on mixed-layer plates of GDX-102 and silica
gel G (1 : 1) pre-coated with hexadecyltrimethylam-
monium bromide, developed in 60}70% ethanol.
Nicotinic Acid (Vitamin B3)
Nicotinic acid (niacin) and various nicotinamides are
sources of the coenzyme nicotinamide adenine dinu-
cleotide, synthesized in the mitochondria and vital for
oxidative energy production in many metabolic reac-
tions. Niacin is normally acquired from a balanced
diet of meat, Rsh, whole cereals and yeast. Peas,
beans, nuts, fruit and vegetables are also good sources
of this vitamin.
Analysis of powdered preparations containing
nicotinic acid has been achieved on silica gel plates
impregnated with zinc acetate, developed in
butanol}benzene}acetic acid}water (8 : 7 : 5 : 3) or
butanol}acetic acid}water (9 : 4 : 5) to provide a
self- indicating system. An overpressure chromato-
graphic procedure using HPTLC silica gel plates and
a mobile phase of butan-1-ol}pyridine}water
(50 : 35 : 15) is also effective. This method uses
photodensitometric detection to separate nicotinam-
ide from other vitamins and the method is fast, accu-
rate and speciRc. Other methods based on HPTLC
and Rbreoptic Suorometric quantitation have been
described in which nicotinic acid is converted to a Su-
orescent derivative before chromatography. After
separation, the plate is scanned by a bifurcated Rbre-
optic which transmits the excitation radiation and
collects the signal emitted from the plate. Good cali-
bration curves have been obtained in the range
10}100 ng nicotinic acid.
Pantothenic Acid (Vitamin B5)
Pantothenic acid is required in the formation of
acetyl coenzyme A which holds a key position in
many metabolic pathways. Only the natural dex-
trorotatory form is active. Pantothenic acid is found
in most foods of plant and animal origin and good
sources include liver, kidney, wheat germ, royal
jelly, peanuts, spinach, cheese and peas. There is
no quoted RDA, though most diets provide at least
10 mg per day.
Panthenol and pantothenic acid have been identi-
Red and quantiRed in pharmaceutical preparations by
extraction with ethanol or benzyl alcohol and separ-
ated by TLC on silica gel plates developed in propan-
2-ol}water (85 : 15). Spots are measured by spectro-
densitometry. Postaire has applied the over-pressure
derivatization technique following separation of cal-
cium pantothenate from other hydrophilic vitamins
on silica gel HPTLC layers developed in butan-1-
ol}pyridine}water (50 : 35 : 15).
Pyridoxine (Vitamin B6)
Pyridoxine is found chieSy in animal tissues; py-
ridoxal and pyridoxamine occur in plant tissues. To-
gether these three forms of the vitamin are of vital
importance in the body for the synthesis of pyridoxal
5-phosphate which acts as coenzyme to amino
 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography
transferases, facilitating more than 60 amino group
transfers and other reactions, including formation of
neurotransmitters. The RDA is 1}3 mg but may in-
crease on a high protein diet. Good sources are yeast,
liver, peanuts, bananas, grapes and pears, beef and
Rsh.
Chromatographic analysis of the vitamin B6 com-
plex, including sample preparation and pre-TLC
extraction, have been well reviewed. Separation of
pyridoxine from other water-soluble vitamins in
pharmaceutical preparations can be improved by
impregnating silica gel plates with zinc acetate to
provide a self-indicating system after separation.
Impregnation of plates with hexadecyltrimethyl-
ammonium bromide has similarly been used to
improve the TLC analysis of vitamin B6 in foods.
Postaire has reported better separation and resolu-
tion of B6 from other compounds using the over-
pressure layer technique than by HPTLC.
Cobalamin (Vitamin B12)
Vitamin B12 is the generic name for a group of
vitamins known as cobalamins. The basic molecule
consists of a corrin ring enclosing a central cobalt
atom subtending axial ligands which determine
the form and function of each individual cobalamin.
Cyanocobalamin (CNCbl) was the Rrst form of
the vitamin isolated in 1948, independently by
two groups. Both relied heavily on chromatography
for the Rnal separation and puriRcation of CNCbl.
Its complex three-dimenstional structure was elucid-
ated in 1956 by Dorothy Hodgkin using elegant X-
ray crystallographic techniques.
The cobalamin molecule can only be synthesized by
microorganisms, but all mammalian cells are equip-
ped to covert the vitamin into its coenzymes. Co-
balamin is without known function in plants and, if
present, is only associated with the metabolic activity
of microorganisms. Hence, unlike folate, dietary
sources of the vitamin are exclusively animal in origin
and include Rsh, meat } particularly liver and kidney
} eggs and milk. Cobalamin is acid- and heat-stable
but, like other hydrophilic vitamins, is destroyed by
exposure to high pH. Notable features of cobalamin
are that it is a much larger molecule (mol wt of
OHCbl is 1346) than any other B-group vitamin and
tissue levels are lower than any other, with total
amounts in the body amounting to only a few milli-
grams. The low RDA of 1}2
g is a reSection of the
efRcient means employed by the body to retain the
vitamin. The low tissue levels of cobalamins
naturally cause analytical problems and this has led
to the development of enhanced methods of detec-
tion, discussed below.
4457
Figure 2 Photolysis of methylcobalamin (MeCbl) in extracts of
normal human plasma exposed to daylight. Most of the MeCbl
was converted to hydroxocobalamin (OHCbl) in 2 min.
In humans, the two coenzyme forms of vitamin
B12 are adenosylcobalamin (AdoCbl) and methyl-
cobalamin (MeCbl) and each is required in speciRc
reactions involving, respectively, isomerization and
transmethylation. Both coenzymes are very light-sen-
sitive and are readily converted to hydroxocobalamin
(OHCbl) by exposure to white light, as may be
demonstrated (Figure 2). MeCbl was Rrst synthesized
in the laboratory by Lester Smith and detected in
human plasma by Lindstrand in 1963 as an unidenti-
Red zone on paper chromatograms. Using large
quantities of liver, this compound was isolated using
chromatographic methods and characterized as
MeCbl.
The bulk of cobalamin in the body occurs as
AdoCbl in cells and MeCbl in plasma, but other
forms detected include OHCbl, CNCbl and
suphitocobalamin, which may be a breakdown prod-
uct of glutathionylcobalamin, possibly an important
metabolic intermediate. A variety of adsorbents may
be used for cobalamin TLC but none has been found
to better a mixed layer of Whatman CC41 micro-
granular cellulose and silica gel G (Figure 3). A sensi-
tive two-dimensional TLC method has been
developed which allows small blood and tissue sam-
ples to be used (Figure 4) to investigate cobalamin
metabolism in health and a wide range of diseases,
including cobalamin deRciencies and genetic errors of
B12 metabolism. The sensitivity of the method relies
on the bioautography organism which is a selected
strain of Escherichia coli, which has a cobalamin
growth response down to 1}2 pg. Growth zones are
4458 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography
Figure 3 Separation of an aqueous mixture of four cobalamins by TLC on a gradient layer of cellulose (CC41) and silica gel (SGG),
showing the influence of varying adsorbent mixtures on separation of the cobalamins. The mobile phase was butan-2-ol}water}0.880
ammonia (75 : 25 : 2).
Figure 4 Separation of cobalamins extracted from normal
human plasma. The adsorbent was cellulose CC41}silica gel
G (3 : 1) developed first in butan-2-ol}water}0.880 ammonia
(75 : 25 : 2) and second, in water saturated with benzyl alcohol.
The second development was at right angles to the first, after
air-drying the plate. Extraction and chromatography were in dark-
ness or by red light. Cobalamin zones were detected bioauto-
graphically by over-layering the chromatogram with agar seeded
with a cobalamin-senistive Escherichia coli mutant and a
tetrazolium growth indicator. The sandwich was incubated at
353C for 18 h. Methylcobalamin (MeCbl) is the main form present
in healthy subjects, with smaller amounts of adenosylcobalamin
(AdoCbl) and hydroxocobalamin (OHCbl).
enhanced by inclusion of 2,3,5-triphenyltetrazolium
chloride in the agar medium which is converted to the
red dye fomazan during growth of the organism. The
red zones corresponding to cobalamins separated on
the TLC plate are then scanned by densitometer or
computer and quantitated. A 10}100-fold increase in
sensitivity is gained if radiolabelled cobalamins are
separated and the bioautogram growth zones excised
and their radioactivity measured.
Folic Acid
Folic acid (pteroylglutamic acid) and related com-
pounds are present at high concentrations in liver, but
spinach, broccoli, peanuts and fresh fruit are also
good dietary sources. Folates are important for the
synthesis of tetrahydrofolate, which is important with
cobalamin for a series of 1-carbon transfer reactions
leading to DNA synthesis, failure of which leads to
megaloblastic anaemia.
Chromatographic analysis of folate compounds in-
cluding methotrexate and other antifolates has been
reviewed. Process impurities in the reduced folate
compound leucovorin calcium may be monitored
using a TLC method with Suorescence detection.
An overpressure layer TLC procedure (OPLC)
has been used to improve the separation of folic
acid from other water-soluble vitamins with good
recovery and resolution. The method uses silica gel
layers developed in butan-1-ol}pyridine}water
(50 : 35 : 15) at a rate of 0.25 mL min\1 for baseline
separation. Quantitation is achieved without derivat-
ization.
 III / VITAMINS / Water-Soluble: Thin-Layer (Planar) Chromatography
Biotin
Good sources of biotin are liver, pork, nuts,
chocolate, pulses, cereals and royal jelly; biotin
is widely distributed among all types of food and
dietary deRciency is rare. However, biotin is
inactivated by avidin, which is present in raw
egg white, and severe eczema has been reported
from this type of deRciency. This does not arise
if cooked eggs are included in the diet, since
heat deactivates avidin. Biotin acts as coenzyme
to carboxylase enzymes, for example in the
catabolism of propionate to methylmalonate.
Biotin is stable in acid and neutral solutions
and hence may be extracted at low pH before
chromatography.
Biotin is separable from other water-soluble
vitamins by TLC on silica gel or cellulose layers
developed in neutral or acidic butanol}water mix-
tures. Various detection reagents have been used
for biotin, including iodine vapour, 1% potassium
permanganate, 1% dimethylaminobenzaldehyde
in hydrochloric acid and p-dimethylaminocinnamal-
dehyde in a mixture of methanol and sulfuric acid,
which is speciRc for biotin, yielding intense
orange zones with an absorbance maximum at
533 nm. More recently, TLC, HPTLC and OPLC
techniques have been compared, using Rve different
mobile phases. Biotin tends to be resolved poorly
from pantothenic acid by HPTLC but this is im-
proved by OPLC, although in the systems investi-
gated this led to less than perfect separation of biotin
and folic acid.
Ascorbic Acid (Vitamin C)
Ascorbic acid occurs abundantly in fresh fruit, espe-
cially blackcurrants, citrus fruit and strawberries, and
in most fresh vegetables; good sources are broccoli
and peppers. It is destroyed by heat and is not well
stored in the body. Ascorbic acid is a good reducing
agent and facilitates many metabolic reactions and
repair processes.
In pharmaceutical preparations and fruit juices,
ascorbic acid is readily separated from other
compounds by TLC on silica gel and quantitated
directly by absorption at 254 nm. Serum and plasma
may be deproteinized with twice the volume of meth-
anol or ethanol. Various ascorbic acid compounds
in plant extracts and foods have been separated
on cellulose layers and detected by spraying
with 2,5-dichlorophenol indophenol. Heulandite, a
natural zeolite (particle size 45
m) has success-
fully been employed as an adsorbent and ascor-
bic acid and other hydrophilic vitamins have
4459
separated within 5 cm by ascending chromatography
in dimethylformamide. HPTLC and OPLC methods
have been developed to improve the separation of
ascorbic acid from other water-soluble vitamins, with
some success.
Conclusion
TLC is a Sexible and well-established technique for
the separation of water-soluble vitamins, limited
only by the stability of the compounds to be separ-
ated, the resolving power of the TLC system and
the sensitivity of the detection method. In complex
biological systems these factors assume greater
importance as vitamin concentrations are lower and
metabolites may interfere with the separation. A
preliminary extraction step or use of a short clean-
up column can help remove salts and other interfer-
ing substances and may increase the concentration
of vitamins to be chromatographed. Recovery
experiments will monitor any selective losses at this
stage.
The introduction of HPTLC and OPLC with opti-
mized solvent systems has undoubtedly increased the
resolving power for a number of vitamins. Gradient
or two-dimensional TLC can increase this still fur-
ther. Ultimately, it is the means of detection which
determines the sensitivity of the system. Fluorimetry
has become the method of choice for those vitamins
forming Suorescent derivatives, but there are alterna-
tives. One is to overlayer the chromatogram with an
agar medium seeded with a microorganism whose
growth is sensitive to the vitamin. This can detect as
little as a few pg of the vitamin. Even higher sensitiv-
ity can be achieved using radioactive vitamins detec-
ted autographically or with phosphorimagers. In
future, the development of an immunoassay tech-
nique similar to Western blotting is likely to allow the
most sensitive quantitation of vitamins separated by
HPTLC.
See also: II/Chromatography: Thin-Layer (Planar):
Densitometry and Image Analysis; Instrumentation;
Modes of Development: Forced Flow, Overpressured
Layer Chromatography and Centrifugal; Spray Reagents.
III/Vitamins: Liquid Chromatography.
Further Reading
Argekar AP and Kunjir SS (1996) Simultaneous determina-
tion of isoniazid and pyridoxine hydrochloride in phar-
maceutical preparations by high-performance thin-layer
chromatography. Journal of Planar Chromatography 9:
390}394.
4460 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
Bates CJ (1997) Vitamin analysis. Annals of Clinical Bio-
chemistry 6: 599}626.
Bhushan R and Ali I (1987) TLC resolution of constituents
of the vitamin B complex. Archives of Pharmacology
320: 1186}1187.
DeLeenheer WL and DeRuyter G (1985) Modern
Chromatographic Analysis of the Vitamins. New York:
Marcel Dekker.
Diaz A, Paniagua A and Sanchez F (1993) Thin-layer
chromatography and Rberoptic Suorometric quantita-
tion of thiamine, riboSavin and niacin. Journal of
Chromatography A 655: 39}43.
Linnell JC, Hussein HA-A and Matthews DM
(1970) A two-dimensional chromato-bioautographic
method for complete separation of individual
plasma cobalamins. Journal of Clinical Pathology 23:
820}821.
Linnell JC and Bhatt H (1995) Inherited errors of
cobalamin metabolism and their management. In:
Wickramasinghe S (ed.) Megaloblastic Anaemia:
Baillie` res Clinical Haematology } International
Practice and Research, pp. 567}601. London: Baillie` re
Tindall.
Postaire E, Cisse M, Le Hoang M and Pradeau D (1991)
Simultaneous determination of water-soluble vitamins
by over-pressure layer chromatography and photoden-
VOLATILE ORGANIC COMPOUNDS
IN WATER: GAS CHROMATOGRAPHY
M. C. Tombs, North West Water Limited,
Warrington, UK
Copyright ^ 2000 Academic Press
Introduction
An important class of substances for which it
is increasingly necessary to analyse in environ-
mental waters comprises a wide range of volatile
organic compounds (VOC). These include aromatics
such as methylbenzene (toluene) and the dimethyl-
benzenes (xylenes), and the environmentally
persistent halogenated solvents such as tetrach-
loromethane and trichloroethene. Many of these
compounds are Rnding their way onto national
and international lists of proscribed or regulated
compounds, and as a result there is a requirement
for robust methods of analysis to monitor both the
sitometric detection. Journal of Pharmacology Science
80: 368}370.
Quadros EV, Hamilton A, Matthews DM and Linnell JC
(1978) Isolation of 57Co-cobalamin coenzymes at high
speciRc activity from Streptomyces griseus. Journal of
Chromatography 160: 101}108.
Sherma J and Fried B (1996) Handbook of Thin-layer
Chromatography. New York: Marcel Dekker Inc.
Stahl E (1969) Thin Layer Chromatography: a Laboratory
Handbook. New York: Springer-Verlag.
Surmeian M, Ciohodaru G, Ionescu MS and Cosofret
VV (1995) Derivative UV-spectrophotometric determin-
ation of binary mixtures of procaine hydrochloride with
benzoic acid, pyridoxine hydrochloride and 4-
aminobenzoic acid from pharmaceutical preparations.
Revue Roumaine de Chimie 40: 111}117.
Tseng M-C, Tsai M-J and Wen K-C (1996) Quanti-
tative analysis of acetominophen, ethoxybenzamide,
piroxicam, hydrochlorthiazide, caffeine, chlorz-
oxasone and nicotinamide, illegally adulterated in
Chinese medicinal pills. Journal of Food and Drug
Analysis 4: 49}56.
Zempleni J, McCormick DB and Mock DM (1997) Identi-
Rcation of biotin sulfone, bisnorbiotin methyl ketone
and tetranorbiotin-1-sulfone in human urine. 65:
508}511.
environment itself and potential sources of discharge
to it.
In the aqueous environment, there are a number
of sample types that an analyst may be required
to examine, each presenting their own problems
and challenges and requiring slightly different ana-
lytical solutions. Drinking waters, for example,
are a relatively straightforward matrix, often
with a clearly deRned quality standard imposed,
such as the requirements of the European Union
Drinking Water Directive (see Further Reading).
River waters and marine waters may also be
required to meet exacting environmental quality
standards (EQS), which are frequently much
lower than those set for drinking waters where
the presence of haloforms, for example, is an
accepted by-product of the disinfection process.
Monitoring of wastewater efSuents is fundamental to
environmental quality management, since these are
 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
a major source of VOCs in the environment. Such
efSuents are frequently complex mixtures of many
different compounds present in a wide range of con-
centrations, and as such offer special challenges to the
analyst.
Quantitative analytical methods must therefore of-
fer good precision and accuracy and be able to with-
stand the rigorous inspection required by the
legislative environment, in order to demonstrate satis-
factory compliance with the regulations.
This article discusses some of the methods available
for the analysis of VOCs in these matrices and is
illustrated with examples taken from routine drinking
water and wastewater quality analysis. The
chromatograms are reproduced here by courtesy of
North West Water Laboratory Services.
Sampling Techniques
The Rrst stage of any analysis, whether carried out in
the Reld or remotely in the laboratory, is the collec-
tion of a representative sample and the preservation
of that sample intact until it reaches the analyst.
Without doubt the best approach is to sample straight
into the container to be used for the analysis, but this
may present logistical problems with handling either
very small containers or carrying out precise measure-
ments of volume. It is easy enough to do this in
a clean, well-equipped laboratory, but it becomes
a much more challenging task on a cold, wet river
bank or windswept beach!
The problem is that with the analytes being so
volatile, their concentration in the matrix can change
signiRcantly between sampling and analysis if the
sample is not correctly taken. One method widely
used with good results is to use a pre-cleaned screw-
cap septum vial, made from borosilicate glass and of
around 20 or 40 mL capacity. The vial is rinsed sev-
eral times with the sample before being Rlled so that
the meniscus stands proud of the brim. A thick sep-
tum faced with polytetraSuoroethylene (PTFE) is
then slipped sideways over the top of the vial, ensur-
ing that no air bubble remains trapped within, and
the septum cap is then Rrmly screwed down, sealing
the sample in the vial. A vial with a leaking seal will
obviously cause sample to be lost, but will also allow
preferential evaporation of VOCs. Similarly, a vial
containing an air bubble will also damage sample
integrity by allowing dissolved VOCs to equilibrate
between the aqueous and vapour phases. Any sub-
sequent sample taken from the vial for analysis will
therefore contain a lower concentration of VOCs
than the original.
Use of a septum will allow the withdrawal of a sub-
sample from the vial, using a syringe and an air bleed
4461
Table 1 Common options for the analysis of VOCs
Introduction Separation Detection
Solvent extraction FID
Direct aqueous injection ECD
Headspace GC MS
Purge-and-trap ELCD
PID
Note: FID, flame ionization detector; ECD, electron-capture de-
tector; MS, mass spectrometry; ELCD, electrolytic conductivity
detector; PID, photoionization detector.
needle, without opening it and risking the possible
loss of volatiles. Similarly, a suitable extraction sol-
vent may be added by a displacement technique.
Some laboratories use these approaches; others will
open the vial and rapidly transfer the required volume
to another closed container. Either way, taking fur-
ther subsamples should be avoided, as the concentra-
tion of VOC in the sample will already have begun to
change. Further information on sample collection is
to be found in a 1987 HMSO publication.
Methods of Analysis
Modern capillary gas chromatography (GC) lends
itself particularly well to the low-level analysis of
VOCs, offering a good separation of the analytes and
high sensitivity. There is a variety of sample introduc-
tion techniques in common use and a wide choice of
columns is available to the analyst. Several different
detector systems can be used, dependent on the
analytes and the sensitivity and speciRcity required.
The actual method of analysis chosen will depend
on several factors. These include the analytes them-
selves, the sample matrix, the resources available to
the analyst and the level of conRdence required in the
results. For example, an analyst interested in a rough-
and-ready assessment of the presence of aromatic
solvents at levels in excess of 1 mg L\1 might choose
to use direct aqueous injection (DAI) with a Same
ionization detector (FID) as the simplest way of ob-
taining the information required. Conversely, an ana-
lyst investigating a complex industrial wastewater
and providing evidence for prosecuting an illegal dis-
charge may prefer the precision of a headspace
sample introduction technique and the conRrmatory
information which may be obtained by using a mass
spectrometer (MS) as a detector. The commonest
options are set out in Table 1.
Sample Introduction Techniques 1:
Solvent Extraction
Solvent extraction is a useful technique for dealing
with relatively clean samples, such as drinking waters
4462 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
Table 2 Solvent extraction performance data for selected com-
pounds
Compound Recovery RSD LOD
(%) (%) (
g L\1 )
Trichloromethanea 85 6.1 0.19
Tetrachloromethanea 79 11.7 0.009
Tetrachloroethenea 104 3.1 0.012
Benzeneb 105 nd 1.58
Methylbenzeneb 99 nd 0.13
a
Chlorinated compounds determined at a concentration of
2.5
g L\1 (tetrachloroethene 5
g L\1) with four degrees of free-
dom. 20 mL sample extracted with 2.5 mL petroleum ether
30}403C. Analysis by packed column GC-ECD. From HMSO
(1987).
b
Aromatic compounds determined at a concentration of 10
g L\1.
1 L sample extracted with 10 mL pentane, cleaned up with florisil
and concentrated to 1 mL. Analysis by 50 m;0.2 mm OV-1 capil-
lary column with FID. From HMSO (1987). RSD, relative stan-
dard deviation; LOD, limit of detection; nd, not determined.
or high-quality river waters. The pentane or hexane
extraction solvent may be added to the sample vial by
displacement, as described above, and the vial is then
shaken or rolled for up to 30 min. A sample of the
solvent may then be withdrawn } again without
opening the vial } and analysed by GC using conven-
tional sample inlet techniques such as split/splitless or
on-column injection. Typical sample/solvent ratios of
between 5 : 1 and 20 : 1 give some sample pre-
concentration, but the injection volume of around
1}2
L restricts ultimate sensitivity.
The technique is well-suited to the analysis of
chlorinated hydrocarbons, using an electron-capture
detector (ECD), but may also be used in conjunction
with most other types of detector, including mass
spectrometers. Its main drawback is the time and
effort required to carry out the extraction. Some
performance data are listed in Table 2.
Sample Introduction Techniques 2: Direct Aqueous
Injection
Perhaps the simplest of all sample introduction tech-
niques, direct aqueous injection has been the subject
of several papers. It has a number of advantages, not
the least of which is convenience: samples collected in
the manner described above require no further hand-
ling between collection and Rnal analysis. As the
name of the technique suggests, a 1
L aliquot of
sample is taken from the vial and injected directly
into the instrument, using either an on-column or
split/splitless injector.
This technique has been applied to the analysis of
trihalomethanes and is said to be reliable and
offers good precision and recoveries. Recoveries of
100% and peak area standard deviations of between
1.9 and 5.2% with approximately 14 degrees of free-
dom at the 10}100
g L\1 level for the four chlorine-
and bromine-containing trihalomethanes have been
quoted. This compares with recoveries of between 60
and 90% for pentane extraction. The technique has
been shown to be applicable to other chlorinated
hydrocarbons, including 1,1,1-trichloroethane and
tetrachloromethane.
The technique certainly works well with small
numbers of samples, but experience in a laboratory
handling upwards of 30 analyses daily suggests that
the robustness of the analytical system becomes an
important factor. Passing relatively large quantities of
water vapour through an ECD shortens its useful life,
and therefore the alternative inlet techniques de-
scribed here are to be preferred where large numbers
of samples are involved.
Another potential problem with the technique is
that there is no initial clean-up of the sample and it is
therefore only appropriate for relatively clean sam-
ples such as drinking waters. With other sample
matrices there is a risk that signiRcant quantities of
nonvolatiles (including inorganic salts) can build up
at the front of the column, reducing its life; similarly,
the presence of less-volatile contaminants remaining
on the column may interfere with subsequent ana-
lyses.
Despite this, DAI may be used successfully where
a minimum effort, rough screening method is re-
quired, e.g. for an industrial wastewater. The use of
an FID allows other, nonhalogenated compounds
such as aromatics to be detected and estimated at
milligram per litre levels. This analysis may be sufR-
cient to meet some needs, but could also be used as
a pre-screening technique to identify appropriate di-
lution factors for headspace or purge-and-trap analy-
sis. An example of a chromatogram of a standard
solution of aromatic compounds in water is shown in
Figure 1.
Figure 1 Aromatic compounds in water by DAI. Conditions:
1
L injection; injector temperature 2303C; 30 m;0.25 mm DB-1
column; temperature program 703C, hold 2 minP903C at 203C
min\1P2603C at 353C min\1, hold 5 min; FID temperature
2803C. Chromatogram shown is from a standard solution in water
containing 10 mg L\1 each compound.
 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
Figure 2 Analysis of VFA in sewage sludge supernatant water
by DAI. Conditions: 1
L 20 : 1 split injection; injector temperature
2503C; 12 m;0.53 mm BP-21 column; temperature program
603CP2503C at 203C min\1, hold 3 min; carrier gas, nitrogen
3 mL min\1; FID temperature 3003C. (A) Chromatogram shown is
from a standard solution in water containing approximately
800 mg L\1 each compound. (B) VFA in digested sludge. Acetic
acid concentration 100 mg L\1; others (10 mg L\1. (C) VFA in
a partially digested sludge. Acetic acid concentration approxim-
ately 800 mg L\1.
DifRcult-to-extract analytes such as alcohols,
ketones or volatile fatty acids (VFA) may also be
estimated by this technique (Figure 2). A speciRc ap-
plication is the analysis of VFA in sewage sludge, the
results of which are used to monitor the performance
of sludge digesters in wastewater treatment plants.
Samples of sludge are centrifuged and the supernatant
water Rltered through a 1
m membrane, in order to
prevent particulate matter blocking either the injec-
4463
tion syringe needle or the capillary column itself.
AcidiRed with phosphoric or formic acid, the samples
are then analysed by direct aqueous injection GC-
FID. The analytical range required of the application
is typically 1}1000 mg L\1 for each compound, al-
though ethanoic (acetic) acid will predominate in
samples from a stable digester.
A simple packed column application of direct aque-
ous injection is the analysis of methane in water.
A 1
L sample is injected directly onto a 1 m Chro-
mosorbTM 101 column for isothermal analysis at
703C, with an FID. Calibration is normally carried
out using a standard gas mixture.
Sample Introduction Techniques 3: Headspace
Analysis
Headspace analysis is a clean, reliable method of
introducing volatile analytes to a GC column, and is
especially useful where complicated matrices such as
industrial wastewaters containing many other con-
taminants must be analysed. Involving the analysis of
just the vapour above a sample of water, the method
provides instant clean-up by ensuring that only vol-
atile materials are introduced into the GC sample
inlet, resulting in a clean chromatogram and en-
chanced column life.
The technique is a practical application of
Henrys law, which states that the vapour pressure
of a solute is proportional to the amount of solute
present in a solution at equilibrium with its vapour.
Thus if the concentration of an analyte in the vapour
phase can be measured, it can be correlated by a
suitable calibration with its concentration in the
sample.
The two methods of introducing samples to the GC
are known as static and dynamic headspace. In the
former, the vapour in equilibrium with the sample in
a vial is analysed, usually at an elevated temperature;
in the latter the vapour is Rrst enriched by actively
purging the sample with an inert gas.
Given that the solubility of gases decreases with
increasing temperature, raising the temperature of the
sample will favour the vaporization of the analytes,
enriching the headspace, and this effect is used to
enhance analyte recovery. It does mean, however,
that temperature must be rigorously controlled both
during analysis and from sample to sample, if repro-
ducibility is to be assured.
Static headspace Samples may be collected for this
analysis in two ways. A vial can be Rlled as described
previously, or a Rxed amount (typically 5}10 mL) of
sample may be accurately measured and sealed } with
an internal standard, if one is to be used } in a vial of
about 20}25 mL capacity, which is to be used for the
4464 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY

analysis. The latter option would allow the sample to

be presented to the instrument unopened, minimizing
the requirement for sample preparation in the labor-
atory, but practical considerations in the Reld mean
that the former is often preferred.
Sealed in its headspace vial, the sample is placed in
a thermostatted heater and allowed to equilibrate
with the air space above it. Some headspace sampling
devices will also agitate the sample to accelerate this
equilibration. Once equilibrium is established, the
vial is pressurized with carrier gas passing through
a sampling needle penetrating the septum. On reach-
ing the required pressure, the Sow is reversed, carry-
ing sample vapour to the GC inlet. The volume
transferred is controlled either by reversing the Sow
for a Rxed time or by the use of a sample loop. The
liquid sample therefore does not come into contact
with any part of the GC itself.
Calibration of the system is carried out by prepar-
ing standard solutions of the analytes of interest in
water, and treating them in exactly the same way as
samples, sealing the same volume in a vial and sub-
jecting them to the whole procedure described above.
Key to the process is consistency: each sample and
standard must be treated exactly alike, and auto-
mated headspace samplers facilitate this. In order to
achieve reproducible results, it is not even necessary
for the samples to achieve equilibrium; providing they
are consistently treated, i.e. by equilibrating at exact-
ly the same temperature and for the same length of
time, reproducibility is assured.
A suitable internal standard can be added to the
headspace vial before sealing and then used either
directly to calibrate the individual analysis or to aid
an external calibration process. Figure 3 Analysis of trihalomethanes and chlorinated hydro-
carbons in drinking water by headspace GC-ECD. Conditions:
The technique has a good linear range and sensitiv-
5 mL sample, equilibrated for 10 min; 30 m;0.53 mm DB-624
ity and provides a robust and reliable method of column; isothermal at 803C. ECD temperature 2503C; carrier gas:
introducing both clean and dirty water samples to nitrogen 7.5 mL min\1. (A) Chromatogram shown is from a stan-
a GC with very little sample preparation (Table 3). dard solution in water containing 91.5
g L\1 tricholoromethane
As might be expected, attainable recoveries measured (highest concentration component) and 2
g L\1 tetrach-
loromethane (lowest). (B) Trihalomethanes in drinking water de-
against standard solutions are close to 100% and
rived from a surface source. Concentrations: trichloromethane
63
g L\1; bromodichloromethane 7
g L\1; dibromochloro-
Table 3 Headspace performance data for selected compounds methane 1
g L\1. Note the almost total absence of other chlorin-
ated hydrocarbons. (C) Trihalomethanes in blended drinking
Compound Recovery RSD LOD water derived from surface and underground sources. Concentra-
(%) (%) (
g L\1 ) tions: trichloromethane 18
g L\1; bromodichloromethane
12
g L\1; dibromochloromethane 2
g L\1; tribromomethane
Trichloromethane 99.95 3.4 1.42 (2
g L\1. Note the reduced concentration of chlorinated com-
Tetrachloromethane 98.30 4.18 0.04 pounds and the increase in brominated compounds.
Tetrachloroethene 96.84 4.43 0.44

Data determined at a concentration of 122
g L\1 (tri-
chloromethane); 3
g L\1 (tetrachloromethane); and 10
g L\1
(tetrachloroethene) with approximately 17 degrees of freedom.
5 mL sample equilibrated for 5 min at 803C. Analysis by
30 m;0.53 mm DB-624 capillary column GC-ECD. Data pro-
vided by North West Water Laboratory Services.
interferences and column degradation are minimized.
It may be used with any detector type, including mass
spectrometers. An example of the analysis of
trihalomethanes and chlorinated hydrocarbons in
drinking water is shown in Figure 3.
 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
Absolute recoveries are dependent on equilibration
time and temperature. Increasing the equilibration
time to the point where the sample and its vapour are
fully equilibrated will maximize recovery; raising the
temperature will increase the partial pressure of vol-
atile compounds in the headspace. Systems typically
operate at temperatures of up to 803C: any higher and
the increased vapour pressure of the water matrix
itself interferes, negating the beneRts.
One drawback is the inability to reanalyse samples,
because once a vial has been subsampled, the integrity
of the sample itself is destroyed and equilibration
with the new headspace will alter the composition of
the sample. This means that a subsequent repeat
analysis cannot be carried out if conRrmation of a
result is required: a fresh sample must be collected.
Although this may present a problem to the environ-
mental analyst, a technique known as multiple head-
space extraction (MHE) has been described (see
Further Reading) where a sample is equilibrated with
successive volumes of gas. Analysis of each successive
headspace volume will allow the distribution of the
analytes to be determined, providing a measure of an
important physical property. Headspace analysis has
been used for this purpose almost since it was Rrst
developed.
The extrapolation of MHE data to the zero equili-
bration level provides a measure of the original con-
centration of an analyte. This may be particularly
useful in situations where for some reason it is not
possible to calibrate the analytical system with either
the analyte or matrix of interest, or where an abso-
lute recovery must be determined.
Dynamic headspace Dynamic headspace or purge-
and-trap sampling is an effective way of achieving
high sensitivity in the analysis of VOC. This is parti-
cularly important in the analysis of environmental
samples for comparison with stringent EQS, which
frequently test the limits of analytical methodology.
Unlike static headspace, where the sample is simply
allowed to equilibrate, in purge-and-trap the sample
is purged with an inert gas (usually helium) in order
to drive the volatiles out into the vapour phase. The
vapour is then caught in a cold trap or adsorbed on an
appropriate support before being thermally desorbed
and passed to the GC inlet. The method retains the
advantage of the headspace technique in terms of
presenting a clean sample to the GC, but potentially
offers much greater sensitivity (Table 4).
Whilst the technique has the ability to improve
sensitivity, the overall range of an analysis may not be
increased, since this is dependent on the dynamic
range of the detector. This means that with the gen-
eral tendency to analyse suites of compounds to-
4465
Table 4 Purge-and-trap performance data for selected com-
pounds
Compound Recovery RSD LOD
(%) (%) (
g L\1 )
Trichloromethane 94.10 3.80 0.002
Tetrachloromethane nd nd 0.004
Tetrachloroethene 98.80 1.30 0.005
1,2-Dichlorobenzene 77.70 8.50 0.020
Methylbenzene 99.11 0.77 0.001
1,2-Dimethylbenzene 98.38 1.46 0.004
Recovery and RSD data determined at a concentration of
4
g L\1 with approximately two degrees of freedom. 25 mL
sample purged at 353C with helium, 50 mL min\1 for 10 min.
Analysis by packed column GC-FID. From Driss and Bouguerra
(1991). LOD, limit of detection. 5 mL sample purged with helium,
40 mL min\1 for 2 min. Analysis by 60 m;0.53 mm DB-624 capil-
lary column GC with electrolytic conductivity detector and photo-
ionization detector. From Mehran, Nickelsen, Golkar and Cooper
(1990) Journal of High Resolution Chromatography 13: 429}433.
gether, those compounds for which low limits of
detection are required can be analysed satisfactorily,
but those present in higher concentrations in the same
sample may well exceed the range of the detector! For
example, purge-and-trap GC-MS analysis of some
drinking waters easily achieves the required limit of
detection of 0.3
g L\1 for tetrachloromethane, but
exceeds the linear range of the detector for the tri-
chloromethane present in a much higher concen-
tration.
The optimization of a purge-and-trap method has
been examined and both purge gas volume and tem-
perature have a signiRcant effect on analyte recovery.
Keeping the purge gas Sow rate constant, but extend-
ing the purge time from 10 to 20 min, greatly en-
hanced the recoveries of all the analytes examined,
although the recoveries of compounds with a higher
solubility in water (e.g. tribromomethane) were
still poor. DifRculties have been reported with the use
of very short purge times: 1 min gave rise to repro-
ducibility problems due to the mode of operation
of the equipment, whereas 2 min resulted in an
acceptable performance and a much faster method.
Extended purge times may risk compromising recov-
eries of highly volatile compounds, since these can be
purged efRciently in a short time. Recovery is then
dependent on the efRcacy of the trap in retaining
them until the chromatographic separation is ready to
begin.
Elevated temperatures also speed recovery. Results
obtained from purging for 20 min at 253C have been
found to be comparable with those from a 10-min
purge at 403C. The disadvantage of using a higher
temperature is that more water vapour is carried over
into the analytical system. Without effective control
4466 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY

Table 5 Effect of purge time and temperature on recovery of selected compounds

Compound Recovery (%)

10 min purge 20 min purge 303C 403C

Trichloromethane 78.06 91.81 82.01 87.63

1,2-Dichloroethane 60.55 88.10 62.48 64.75
Tetrachloroethene 98.03 99.10 97.94 100
1,2-Dichlorobenzene 59.86 66.82 60.94 65.41
Methylbenzene 85.73 93.12 88.63 94.05
1,2-Dimethylbenzene 81.86 84.28 84.99 89.66

Data determined at a concentration of 4
g L\1 with approximately two degrees of freedom. 25 mL sample purged with helium,
50 mL min\1 10/20-min purge data determined at 253C. Temperature data determined with a purge time of 10 min. Analysis by packed
column GC-FID. From Driss and Bouguerra (1991).

this will interfere with the analysis } particularly with recoveries may readily be optimized by temperature
moisture-sensitive equipment such as ECD or mass control.
spectrometers. For this reason, purge temperatures Although the effect may be used to improve the
signiRcantly greater than 403C are not widely used. performance of a method, it is important to remem-
Table 5 shows the effect of purge time and temper- ber that the samples themselves may be subject to
ature on the recovery of selected compounds. some variability. Table 6 provides the evidence to
Waters containing detergents or other foaming show why a seawater sample could not be analysed
agents may prove difRcult to analyse effectively by using a method set up and calibrated for use with
this technique. It is, however, suitable for use with drinking water, or vice versa: the performance of the
most types of detector. method will differ signiRcantly between the two ma-
trices. For the same reason, it is important that the
Matrix Modi\cation
ionic strength of both samples and standards is con-
The use of matrix modiRers is common practice in sistent. If there is any doubt then an excess of salt (e.g.
water analysis, and they are often used to enhance the around 2 g mL\1) should be added to all samples and
performance of some methods for the analysis of standards to ensure consistency.
VOC } in particular the headspace and purge-and-
Analytical Columns
trap methods. The addition of modiRers such as so-
dium chloride or sodium sulfate to samples prior to Today, most applications for VOC analysis use capil-
analysis will } by modifying the activity coefRcient of lary columns, the length and Rlm thickness of which
the VOC solutes and the vapour pressure of the sol- will depend on the complexity of the analysis. Up to
vent } enhance the relative concentration of the about 20 compounds can be satisfactorily resolved by
analyte in the headspace above the sample. This effect a 25}30 m column in about 10 min, whereas
is most noticeable for the less soluble or less volatile a 50}60 m column is more appropriate for samples
compounds such as the dimethylbenzenes or the dich- containing 60 or more analytes, taking 30 min to 1 h
lorobenzenes, although it may be of limited use with to achieve an acceptable separation.
other compounds, such as trichloromethane, where Although the superior resolution of the capillary
column means that most separations can be achieved
Table 6 Effect of ionic strength on recovery of selected com- using standard nonpolar or moderately polar phases
pounds such as DB-1 from J & W or BP-5 from SGE, there is
an increasing number of columns tailored for speciRc
Compound Purging efficiency analyses. These include J & Ws DB-624 phase,
designed to substitute for the packed column speciRed
0% NaCl 10% NaCl 20% NaCl
in US Environmental Protection Agency (US EPA)
Trichloromethane 83.63 93.98 98.52 method 624, for purgeable organic compounds. Such
1,2-Dichloroethane 63.2 67.97 76.17 columns are designed to optimize the separation and
Benzene 90.12 97.22 99.50 the time required for the analysis of the compounds of
1,3-Dichlorobenzene 72.45 81.59 91.5
interest.
1,2-Dimethylbenzene 86.43 96.72 98.91
The direct aqueous injection technique requires the
Purge-and-trap recovery data determined at a temperature of use of bonded-phase columns to ensure that the water
353C. From Driss and Bouguerra (1991). passing through it does not destroy the column.
 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY 4467

Detectors

Perhaps the commonest detectors used for VOC analy-

sis in the environmental industry are the ECD and the
mass spectrometer, primarily because of the keen
interest in levels of organochlorine compounds in the
environment. However, the FID also Rnds application
in the analysis of hydrocarbons } including aromatics
} providing reliable detection and sensitivity down to
around 100
g L\1 without sample pre-concentra-
tion. It may also be used with mixtures of hydrocar-
bons and some chlorinated solvents, and although
limits of detection for the latter are relatively high, the
robustness of the detector may make it an appropriate
choice for the analysis of an industrial wastewater,
for example. As previously described, the FID can be
used for the analysis of relatively high concentrations
of VOCs by the direct aqueous injection technique.
Where low levels of halogenated solvents are to be
determined, by far the best option is to use an ECD,
which has a high speciRcity and sensitivity for many
halogenated compounds. This will work with all of
the sample introduction techniques previously de-
scribed, although complex industrial wastewaters
may contain compounds that contaminate the de-
tector. In such cases, selective introduction tech-
niques such as headspace or purge-and-trap are to be Figure 4 Analysis of VOCs by headspace GC-MS. Conditions:
preferred, as these will eliminate or substantially re- 5 mL sample, equilibrated for 10 min; 30 m;0.25 mm DB-5MS
duce the contaminants introduced to the system. column; temperature program 403C, hold 5 minP2003C at
153C min\1P2503C at 503C min\1; carrier gas: helium
When a wide-ranging screen coupled with speciR- 3 mL min\1; Ion TrapTM detection, EI mode, 1 s scan, mass range
city and reasonable sensitivity is required, then a mass 45}220 amu. (A) Chromatogram shown is from a 100
g L\1
spectrometer may be used. Small bench-top instru- standard solution in water. (B) VOCs in trade effluent from a road
ments are increasingly found in environmental labor- hauliers premises. Conditions as above except carrier gas ap-
atories, and many are employed in just this kind of proximately 1.5 mL min\1. Approximate concentrations: trichloro-
ethene 1150
g L\1; toluene 1700
g L\1; m- and p-xylenes
activity, coupled to headspace GC systems. Such 500
g L\1; o-xylene 250
g L\1; 1,2,4-trichlorobenzene
a conRguration provides a good response to a variety 20
g L\1; 1,2,3-trichlorobenzene 40
g L\1. Other compounds,
of compound classes, is robust enough to handle including the internal standard, are also present, but cannot be
samples of badly contaminated industrial waste- seen on this scale.
waters, and yet has sufRcient sensitivity to analyse
clean river waters to the levels required by most EQS.
Additionally, the ability to produce a recognizable ing water. The convenience the ECD offers over the
mass spectrum lends conRdence to the identiRcation ELCD coupled with the greater speciRcity and sensi-
of analytes. This is of particular importance when tivity of the PID relative to the FID means that a use-
collecting evidence for the prosecution of an illegal ful application for the detectors in tandem is the
discharge. Examples of analysis of chlorinated and analysis of both halogenated and aromatic com-
aromatic hydrocarbons by headspace GC-MS are pounds in the same sample.
shown in Figure 4. Of course, providing the sample introduction tech-
Other detector types in use, particularly in the nique is compatible (as indeed the headspace methods
USA, include the electrolytic conductivity detector inevitably will be), any detector type can be used to
(ELCD) and the photoionization detector (PID) speci- meet the speciRc requirements of the analysis.
Red in some EPA methods. The latter can be up to
a hundred times more sensitive than a FID when used
for the analysis of some aromatic compounds, but in
Conclusion
contrast to both the ECD and ELCD it will not detect This article has summarized the main methods
the lighter haloalkanes such as those found in drink- of analysing for VOC in common use today and
4468 III / VOLATILE ORGANIC COMPOUNDS IN WATER: GAS CHROMATOGRAPHY
has brieSy described some of the advantages and
disadvantages of each. It is hoped that the data illus-
trating the performance of the methods will assist
readers in selecting an appropriate technique for their
own application.
It is difRcult to see where VOC analysis will go in
the future, although possible developments include
the more widespread application of automation to
the dynamic headspace technique, enabling the unat-
tended analysis of large batches of samples. Con-
tinued development of membrane and other direct
inlet techniques for mass spectrometry and the
shrinking size and price of MS-MS instruments may
ultimately render the time-consuming chromato-
graphic separation itself superSuous, offering the
prospect of analysis in seconds rather than tens of
minutes. However, for the time being the availability
of suitable membranes permitting the migration of
VOC restricts the application of this technique.
The increasing sensitivity of detection systems
could well prove to be of little beneRt to the analyst,
as it may only serve to encourage the setting of even
lower quality standards!
Whatever happens with the equipment and meth-
odology that is employed, it is clear that continued
growth in legislation controlling these substances in
the environment will lead to an ever-increasing work-
load for the analytical laboratory.
See also: II/Chromatography: Gas: Column Techno-
logy; Detectors: General (Flame Ionization Detectors
and Thermal Conductivity Detectors); Detectors: Mass
Spectrometry; Detectors: Selective; Gas-Solid Gas Chro-
matography; Multidimensional Gas Chromatography;
Sampling Systems; Theory of Gas Chromatography.
Extraction: Analytical Extractions; Solid-Phase Extrac-
tion; Solid-Phase Microextraction. III/Gas Analysis: Gas
Chromatography.
Further Reading
Carmichael D and Holmes W (1990) Screening of
trihalomethanes by direct aqueous injection using elec-
tron capture detection. Journal of High Resolution
Chromatography 13: 267}269.
Driss MR and Bouguerra ML (1991) Analysis of volatile
organic compounds in water by purge-and-trap and gas
chromatography techniques. International Journal of
Environmental Analytical Chemistry 45: 193}204.
EC (1980) European Community Directive 80/778/EEC
relating to the Quality of Water Intended for Human
Consumption. Brussels: EC.
Grob K (1984) Further development of direct aqueous injec-
tion with electron-capture detection in gas chromato-
graphy. Journal of Chromatography 299: 1}11.
Grob K and Habich A (1983) Trace analysis of halo-
carbons in water; direct aqueous injection with elec-
tron-capture detection. Journal of High Resolution
Chromatography and Chromatographic Communica-
tions 6: 11}15.
HMSO (1980) Determination of Volatile Fatty Acids in
Sewage Sludge 1979. London: HMSO.
HMSO (1987) Determination of Very Low Concentrations
of Hydrocarbons and Halogenated Hydrocarbons in
Water 1984}5. London: HMSO.
HMSO (1988) The Determination of Methane and Other
Hydrocarbon Gases in Water 1988. London: HMSO.
Kolb B and Ettre LS (1991) Theory and practice of
multiple headspace extraction. Chromatographia 32:
505}513.
Kolb B and Ettre LS (1997) Static Headspace}Gas Chromato-
graphy Theory and Practice. New York: Wiley-VCH.
Mehran MF, Nickelsen MG, Golkar N and Cooper WJ
(1990) Improvement of the purge-and-trap technique
for the rapid analysis of volatile organic pollutants in
water. Journal of High Resolution Chromatography 13:
492}433.
Shinohara A, Sato A, Ishii H and Onda N (1991) Capillary
headspace-gas chromatography for the characterization
of the Savour of fresh vegetables. Chromatographia 32:
357}364.
Temmerman I and Quaghebeur D (1990) Analysis of
trihalomethanes by direct aqueous injection (THM-
DAI). Journal of High Resolution Chromatography 13:
379}381.
Umbrett GR (1977) Trace analysis by gas chromatography.
In: Grob RL (ed.) Modern Practi ce of Gas Chromatog-
raphy, pp. 365d420. New York: Wiley.
US EPA (1984) Method 624: Purgeables. Methods for the
Chemical Analysis of Waters and Wastes, EPA-600
series, vol. 49, no. 209. Cincinnati: US Environmental
Protection Agency Environmental Monitoring and Sup-
port Laboratory.
 III / WATER TREATMENT / Overview: Ion Exchange 4469

WATER TREATMENT

tions for water quality have become progressively

Overview: Ion Exchange more stringent, and regulations to enforce these have
become more strict. Hence the choice of resin types
for a particular application becomes increasingly
J. Irving, Purolite International Limited, Pontyclun,
complex.
Mid Glamorgan, Wales, UK
Copyright ^ 2000 Academic Press
Applications of Ion Exchange
in Water Treatment
Introduction
A wide variety of new water treatment applications
Ion exchange resins are used for many water treat-
employ ion exchange resins in limited volume and
ment applications. Of these applications, in terms
there is limited use in niche areas for many special
of the volume of resins used, water softening and
resin types. However, reverse osmosis (RO) is
demineralization of water are the most signiRcant.
increasingly being used instead of ion exchange
Water softening has been practiced commercially for
where treated water quality requirements are not
a century or more, making use of a wide range of
particularly high. The use of RO followed by an
natural and synthetic products. As the variety of uses
ion exchange polishing process is often used for
for puriRed water has increased, so has the need to
the production of high purity water, for example
soften and demineralize water. Demineralization has
in the manufacture of silicon chips for the computer
only been practiced since the discovery of synthetic
industry. Table 1 lists the major water treatment pro-
anion exchange resins in the 1920s. Their useful-
cesses in which ion exchange resins are used.
ness increased greatly with the invention of strongly
basic anion exchange resins, which can remove weak-
ly acidic compounds such as silica and carbon diox- Principles of Ion Exchange Applicable
ide, as well as mineral acids. This process of ion
exchange can be used as a simple method to produce to Water Softening
water of very high purity. In general, as industrial The properties and theoretical principles of ion ex-
and domestic requirements have grown, speciRca- change resins are fully covered elsewhere. This article

Table 1 Major water treatment processes

Process Resins used Significant property Application areas

Softening SAC Hardness selectivity Domestic, industrial processes,

food processing, etc.

Partial softening WAC Temporary hardness Potable water, beverage industry,

Dealkalization WAC removal and industrial processes, laundry, glass
alkalinity removal washing

Demineralization WAC, SAC Removal of cations Industrial water processes

WBA, SBA Removal of anions Steam generation, food processing
industry, process water for pharma-
ceutical use, etc.

Nitrate removal SBA Nitrate selectivity Potable water, food and beverage
processing

Metals removal WAC chelate resins Selectivity for heavy metals Wastewater treatment

Sorption Macronets WBA, SBA Selectivity for organics Organic scavenging

Key: WAC, weak acid cation resin; WBA, weak base anion resin; SAC, strong acid cation resin; SBA, strong base anion resin; chelate,
chelating ion exchange resin; macronet, special resin with adsorption properties.
4470 III / WATER TREATMENT / Overview: Ion Exchange
discusses only those principles that directly relate
to operating performance of the ion exchange resins
that are currently used in water treatment.
Ion Exchange Equilibria
Water softening is a very efRcient process. Water
containing hardness ions (calcium and magnesium) is
passed through a cation exchange resin in the sodium
form. In dilute solution, the hardness ions are selec-
tively held:
Na"([CaR]/[NaR]);([NaS ]/[CaS]);CS/CR [1]
KCa 2 2
where K is a simpliRed selectivity coefRcient describ-
ing the equilibrium. [Ca] is the calcium concentra-
tion, [Na] is the sodium concentration, and C is the
overall ionic concentration. Subscripts R and S rep-
resent resin and solution phase, respectively. This
equation takes no account of activity coefRcients, but
nevertheless can be used, certainly for comparative
purposes, and usually gives fairly accurate predic-
tions. Clearly the larger the value of KCa
Na, the greater
the fraction of calcium residing in the resin phase.
Tables of selectivity coefRcients have been compiled
for a wide variety of cations and anions.
Selectivity for Hardness in Softening and Resin
Regeneration
The more dilute the ionic concentration, the higher
the calcium (and magnesium) fraction in the resin
phase, and the less calcium is needed in the solution
phase to satisfy the equation. However, when consid-
ering regeneration, the ionic concentration in solution
is much higher, so as CS/CR tends to a value greater
than 1, so less calcium is needed in the resin phase to
satisfy the equation. The poorer selectivity of the
resin sites for calcium present in high ionic concentra-
tions ensures that the regeneration stage is very
efRcient. The principle of this equation applies
to all comparisons between monovalent ions such as
sodium and divalent ions such as calcium. Since mag-
nesium, the other main contributor to total hardness,
is also divalent, the principles described apply equally
to magnesium. In order that the regeneration process
is reasonably efRcient, there must be an excess of
sodium ions.
Reverse Osmosis
One of the advantages of RO is that no regenerant
chemicals are required. The disadvantages of RO is
that capital costs are higher and pumping costs
can also be high. In addition, the volume of the reject
waste can be large, even though the ionic concentra-
tion within the reject is quite low. There is also a risk
of membrane fouling. Generally when the total dis-
solved solids (TDS) in the treated water are high, RO
is preferred. However, as both processes are constant-
ly changing in efRciency, the commercial breakeven
cost point is constantly changing.
Water Quality and Regeneration Ef\ciency
Clearly the use of smaller quantities of regenerant
would make the ion exchange softening process more
efRcient and competitive. It has been demonstrated
that the ion exchange process must be carried out in
a column if the efRciency is to be optimized. For
example, regenerating calcium from a resin with so-
dium chloride solution simply by adding the regener-
ant to the resin while stirring in a beaker is very
inefRcient. It suffers from the disadvantage that all
the calcium displaced from the ion exchange resin has
the opportunity to re-enter the resin and re-occupy
the ion exchange sites. On the other hand, in a col-
umn operation, the displaced ion is carried away, and
cannot return to the same beads at the regeneration
entry point. As more fresh regenerant is added this
same principle applies to the exchange process further
down the column. It has been shown that, as the
regenerant contact time in the column was increased
to 24 h or more, the efRciency of regeneration de-
creased to the point where some 25}30% fewer sites
were regenerated with the same quantity of regener-
ant. The Rnal lower value tended asymptotically to
that obtained under batch equilibrium conditions.
Counter]ow and Co-]ow Regeneration
A second advantage of column operation is that those
beads situated at the point of entry of the regenerant
will come in contact with a vast excess of pure regen-
erant. This ensures that almost all of the hardness
ions loaded in the previous cycle are carried away
from that part of the resin bed. In a stirred batch
system, the ratio of regenerant ions to hardness ions
would be approximately equal in every bead, depend-
ing on the quantity and concentration of regenerant
used. It follows that since the water being treated is in
counterSow to that of the regeneration, this allows
the treated water to pass the most highly regenerated
and rinsed resin at the point of exit, thus ensuring
near zero leakage of hardness. Of course, the resin
bed must not be disturbed for this advantage to apply.
In co-Sow regeneration, any regenerant at the treated
water outlet has previously been in contact with the
ions to be removed, hence this part of the column is
least efRciently regenerated. In the following exhaus-
tion cycle the sodium in the water at the outlet can
back-exchange for the hardness residual at the col-
umn outlet. This results in a signiRcant hardness leak-
 III / WATER TREATMENT / Overview: Ion Exchange
age. To optimize the operating efRciency, the use of
regenerant chemical should be cut to a minimum.
This has the added advantage that less excess regener-
ant will pollute the environment. From the above
arguments, it is easy to see why counterSow regenera-
tion is more efRcient.
Diffusion in the Regeneration Stage
Returning to eqn [1], it can be shown that reduction
of the regenerant quantity to a minimum can create
signiRcant disadvantages in the softening process.
When the regenerant enters the column, it will inevi-
tably suffer some dilution with the water it displaces.
As has already been explained, the lower the ionic
concentration in solution, the higher the selectivity
for the divalent hardness ions, so any dilution will
reduce the efRciency of the regeneration process.
However, this dilution at the start of the regeneration
process is not a signiRcant disadvantage, because the
regenerant is in contact with resin heavily loaded with
hardness ions and thus some of these ions are easy to
remove. However, as the regeneration proceeds, the
concentration of regenerant increases and the beads
are increasingly invaded by sodium chloride. This is
termed Donnan invasion. In fact, the Rrst molecule
of sodium chloride is not effective in achieving any
regeneration of divalent hardness because only one of
the two hardness bonds is released. The positive
charge of the single displaced calcium ion is satisRed
with the negatively charged chloride ion.
R}Ca2#}R#Na>Cl\"R\}Na##R\}Ca2#}Cl\
[2]
where R is the matrix and functional group of the
strong acid cation (SAC) resin.
A second molecule of sodium chloride is clearly
needed to free the hardness (calcium) ion and so
allow it to start the diffusion process towards the
outside of the beads.
R}Ca2#}Cl\#Na#Cl\"R\Na##Ca2#Cl\
2 regenerated with acid.
[3]
The general direction of the regenerant is towards the
centre of the beads as the diffusion of the regenerant
proceeds from the solution surrounding the beads. Any
hardness ion has to move against the regenerant cur-
rent to leave a particular bead. Once the regenerant
has completed its diffusion path to the centre of each
bead, then the outward diffusion of the regenerated
hardness can presumably proceed more rapidly.
As the regeneration draws to a close, the regenerant
is followed by the displacement rinse. Dilution of the
regenerant by the displacement water will cause selec-
4471
tivity reversal, thus any hardness ion still situated
inside the beads, and travelling towards the outside, will
promptly displace two regenerated sodium sites. If the
regenerant quantity is cut to a minimum, then the
release of the calcium will be slower (see eqn [2]) and
the proportion of diluted regenerant to concentrated
regenerant will be greater. It follows that the shorter
difusion path offers a more effective regeneration result-
ing from improved column efRciency.
The use of beads of a narrow size range and core shell
beads with a limited diffusion path offers a superior
performance; this has already been seen.
Resin Kinetics
Beads of narrow size range can also present a larger
surface area of exchange to the water being treated,
and therefore the kinetics of exchange may be im-
proved. Larger beads with a shell core formation also
provide ease of access to each individual ion exchange
site. These features can be important where high Sow
rates are used. Other signiRcant factors such as resin
bed depth, operating temperature, efRcient distribu-
tion and collection of the water being treated are also
important.
Principles of Ion Exchange Applied
to Demineralization
The principles applicable to the softening process also
apply to demineralization. It is useful to discuss some
of these in more detail.
Resin Selectivity
Demineralization requires an ion exchange process of
at least two stages. In the Rrst essential stage, the
cations in the water to be treated are replaced with
hydrogen by passing the water through strong acid
cation (SAC) resin in the hydrogen form. When the
resin becomes exhausted to the extent that the water
is not treated to the required quality, it must be
The Rrst stage of the demineralization process may
be improved to combine the use of both a SAC resin
and a weak acid cation (WAC) resin. The latter is
positioned upstream. This is one of the many ways in
which the efRciency of this regeneration process can
be varied. Returning to the properties of SAC resins,
the process is unlike softening in that there are no
advantages from changes in ionic concentration when
regenerating sodium. The selectivity for sodium, as
compared with hydrogen, may be simpliRed by the
following equation.
H "([NaR]/[HR]);([HS]/[NaS])
KNa [4]
4472 III / WATER TREATMENT / Overview: Ion Exchange
where H is the hydrogen ion; all other symbols as for
eqn [1].
It has been reported that KNa
H varies according to
the cross-linking of the SAC resin.
Counter]ow Regeneration
The advantages to be obtained in water quality from
operation in the counterSow mode are, perhaps, even
more important than for softening. The difference
between the selectivity of sodium and hydrogen is
very small (in fact the coefRcient lies between 1 and
2). Hence any residual sodium located near the outlet
of the bed is easily displaced. Low sodium leakage is
clearly an essential parameter where high water pu-
rity is required. It follows that either the whole resin
bed must be highly regenerated, or the resin bed has
to be operated in the counterSow mode. This ensures
that the treated water at the bed outlet is in contact
with highly regenerated resin containing only minute
traces of sodium.
Mixed Bed
The use of a mixture of SAC resins and strong base
anion (SBA) resins, regenerated separately before
mixing, has generally been considered the best way to
achieve treated water of the highest purity. The pas-
sage of water alternately via cation and anion resins
affords the more or less continuous neutralization of
acids and bases produced by contact with the pre-
vious bead of opposite charge. The disadvantage is
that the component resins have to be separated before
regeneration. Incomplete separation will cause the
offending beads to be regenerated with the wrong
regenerant, resulting in a signiRcant deterioration in
treated water quality. Techniques of rinse recycle and
efRcient counterSow regeneration are now producing
water qualities close to that of mixed bed polishing.
Effects of Sodium Leakage
One further problem occurs if the sodium leakage is
high. The treated water from the cation resin outlet
(decationized water) passes through the anion resin,
regenerated to the hydroxide form. In general, this is
a very efRcient reaction, because the process is one of
neutralization of acids, so the exchange reaction is
essentially non-reversible, and the only by-product of
the neutralization reaction is water. However, any
sodium leakage from the cation resin must be accom-
panied by an anion to preserve electroneutrality. At
the start of the cycle, when the anion resin is freshly
regenerated, the anion that accompanies sodium will
usually be hydroxide. This is not particularly desir-
able, but is easily removed by the following mixed
bed resins. However, as the resin bed exhausts, the
least selective anion, in exhaustion, will accompany
the sodium. This anion is usually silica. Silica can
cause deposits in superheaters, boilers, turbines and
condensers, so accelerating corrosion. Thus it is clear
that sodium leakage carries a two-fold danger.
Anion Leakage
The very high selectivity of the anions of mineral
acids for ion exchange resins usually prevents high
mineral anion leakage, even though regeneration to
remove all of these is rarely complete. As with soften-
ing, the efRciency of regeneration is crucial if good
capacity and low leakage are required. The use of
narrow size range resins can shorten the diffusion
path in the regeneration process.
Regeneration Ef\ciency
The regeneration efRciency of cation resins has al-
ready been discussed. In the demineralization process,
the regeneration efRciency of anion resins is of even
more importance.
E During the actual water treatment process, the
removal of anions is essentially an acid}base neu-
tralization. Hence it is driven rapidly more or less
to completion. However, the regeneration stage
involves the exchange of the loaded ions for the
hydroxide ion. Hence this stage is a true ion ex-
change process where the ions being removed are
in competition with the ion being Rxed on the
resin. It follows that the extent of regeneration is
the limiting step, dictating the operating capacity
of the resin. In fact, when operating SBA resins at
recommended Sow rates the operation capacity is
normally only 5}10% below the available regen-
erated capacity. In other words, the chromato-
graphic proRle is extremely sharp.
E Type I SBA resins do not regenerate easily. In fact
the selectivity coefRcient KCl OH is approximately
15}20 for gel Type I SBA resins, and even higher
for equivalent macroporous types.
E Type II and acrylic SBA resins are more easily
regenerated, but have the disadvantage that they
remove silica less efRciently (especially the Type II);
both types have poor thermal stability.
E Recently strong base resin types have been com-
pared in their operating performance and related
properties. A Type III resin has been developed that
is comparable with Type II or acrylic resins in its
ease of regeneration, while having thermal stability
and silica removal similar to that of a Type I resin.
E Weak base anion (WBA) resins regenerate quite eas-
ily, and can be used in conjunction with SBA resins to
improve overall regeneration efRciency. However,
 III / WATER TREATMENT / Overview: Ion Exchange
Figure 1 Regeneration efficiency of chloride form SBA Type I
clear gel anion at 65 g L\1 NaOH.
they do not remove silica or carbon dioxide, so
the proportion of WBA to SBA must be controlled
to balance the needs of the process. Also the fresh
regenerant must be used to regenerate the resins in
the order SBAPWBA, otherwise the advantage is
lost.
The mechanisms for regenerability of anion resins
have been discussed in the literature. BrieSy, the high
selectivity for chloride, sulfate, and nitrate arises from
the fact that these ions are less hydrated than is the
hydroxide ion. Hence the hydroxide ion prefers to
remain in the aqueous phase. As Figure 1 shows, the
lower the moisture retention of the resin, the more
difRcult is the regeneration process.
These mechanisms also offer good explanations
for the variation in thermal stability between resin
types. It has been shown that nucleophilic attack on
the nitrogen of the active group by the hydroxide ion
is responsible for the thermal degradation. It follows
that the more hydrated the hydroxide ion, the lower
the electron charge density and the lower the rate of
degradation. This is conRrmed by experimental data.
This subject is further complicated by the differences
in selectivity of chloride and sulfate ion in the regen-
eration process.
Principles of Dealkalization
The removal of alkalinity is a less common process
than softening or demineralization, hence it will only
be dealt with brieSy. It has long been recognized that
hardness associated with bicarbonates (termed tem-
porary hardness) is more of a problem than that
associated with mineral acids (permanent hardness).
When water is heated the bicarbonate decomposes to
carbon dioxide and insoluble calcium carbonate. This
forms the scale that is found in the equipment used to
heat or transport water. The removal of scale may be
achieved by using a weak acid cation resin regen-
erated to the hydrogen form.
4473
The reaction:
2RCOOH#CaCO3 8 Ca(RCOO)2#CO2#H2O
[5]
proceeds quite easily, while the reaction
RCOOH#CaCl2 { Ca(RCOO)2#2HCl [6]
does not, because the ionized acid produced inhibits
further reaction.
The WAC resin is easily regenerated with
a stoichiometric amount of acid. This makes the pro-
cess chemically efRcient but it is kinetically slow. It
also has the advantage that the total dissolved solids
are reduced by the removal of the hardness. The
volatile carbon dioxide can be removed by deaer-
ation. The main disadvantage is that acid is needed
for regeneration.
An alternative process uses a SBA resin in the
chloride form. Here the bicarbonate is exchanged for
chloride directly. Thus the offending temporary hard-
ness is exchanged for permanent hardness. There is
no reduction in total dissolved solids and the operat-
ing capacity is lower, but the advantage is that the
SBA resin may be regenerated with common salt.
Principles of Nitrate Removal
The World Health Organization (WHO) limit for
potable water is 50 ppm of nitrate. Many water sour-
ces have higher levels, partly because of the use ni-
trate-based fertilizers to boost crop yields. Although
this practice has been curtailed in recent years, the
problem will remain for many years to come. Ion
exchange resins, regenerated with sodium chloride,
selectively remove nitrate in preference to bicarbon-
ate, but not in preference to sulfate. This means that
the sulfate in the water is also needlessly removed.
The exchange of both nitrate and sulfate can result in
the chloride level increasing above WHO limits.
Where sulfate levels in the water are signiRcant, the
use of a specially developed nitrate-selective (over
sulfate) resin can give increased capacity and better
overall water quality.
Physical Stability of Ion Exchange
Resins
Ion exchange resins have to be physically strong to
last for the expected life span of 4}6 years, depending
upon the resin type, temperature of operation and
regeneration, ability to resist irreversible fouling from
trace contaminants in the water to be treated, and
adherence to the recommended operating conditions.
4474 III / WATER TREATMENT / Overview: Ion Exchange
This means that they have to be mechanically strong
and also able to withstand rapid changes in swell-
ing/shrinking arising from changes in ionic form and
changes in ionic concentration. It is most important
that the ion exchange plant is designed to accommod-
ate the anticipated changes in bed depth, and resins
should be free to adjust to the volume changes occur-
ring naturally through the process. Tests have been
developed to evaluate osmotic shock resistance; these
are mentioned in other articles.
The elimination of the larger beads in the resin bed
will reduce the chances of breakdown resulting from
osmotic shock, because the build-up of differential
stress with changes in ionic concentration is less in
smaller beads. Up to now partially activated beads,
which have the advantages of a smaller diffusion path
described above, have not been physically stable be-
cause of the stresses developing between the swelling
and shrinking of the activated part of the resin, com-
pared with the properties of the inert core. There are
indications that this difRculty is now being overcome.
Plant Design
Water analyses vary considerably, depending on the
source of the water supply and the geographical area.
Thus it has been impossible over the years to standar-
dize on one type of plant or on which are the most
suitable resins. Table 2 gives on example of the water
analysis information needed to design a treatment
plant. Changes in total concentration of dissolved
salts, the proportion of sodium and silica (as already
discussed), the temperature of the water, the ratio of
alkalinity to total anions, and the proportion of sul-
fates, chlorides and nitrates, all have an inSuence on
the operating capacity and treated water quality.
Table 2 Typical water analysis information
Influent water specification
Origin Mains water
Pretreatments None
Temperature 83C
Organic matter 20.000 mg L\1 KMnO4
Cation Concentration Anion
(meq L\1)
Ca 0.900 HCO3
Mg 0.140 CO3
Na 0.300 Cl
K SO4
Fe NO3
TC 1.340 TA
TC, total cation concentration; TA, total anion concentration. Note that TC should equal TA. In general if there is a difference the
analysis is probably incorrect. If the pH is not neutral, the hydrogen and hydroxide ions should be included in the balance.
The presence of iron can severely affect the perfor-
mance of softeners. Iron can slowly accumulate and
cause irreversible fouling. It also acts as a catalyst,
promoting oxidation of resins that causes an increase
in moisture retention and swelling of the resin.
This in turn can cause stress within the resin bed,
not only on the resin itself, but also on the internal
collectors and distributors. The performance of de-
mineralizers can also be affected, especially if only
sulfuric acid is available for resin regeneration and
resin cleaning.
Natural organic matter can vary both in quantity
and quality from area to area, hence the choice of
resins and preferred cycle times, the quantity of so-
dium hydroxide used for regeneration and its temper-
ature as well as cleaning regimes, can vary from
region to region.
Basic Principles
Developments in ion exchange resins since the 1940s
have been accompanied by developments in engineer-
ing equipment and processes, from the invention of
styrene-divinyl benzene resins to advances in tech-
niques for TOC removal. The latter is now of utmost
importance for the production of ultrapure water.
The basic principles of sizing an ion exchange
vessel are quite simple. Let us suppose that it is
necessary to treat F m3 of water per hour, and it is
required to remove C eq m\3. Then the load per hour
is FC mol, and FCh is the total load presented for an
exhaustion cycle of h h. It is therefore possible to
calculate the volume of resin needed, provided the
operating capacity (O eq m\3) for the particular resin
is known for the speciRed operating conditions. If we
have V m3 of resin, then for the resin plant to treat the
Concentration Other Concentration
(meq L\1) (meq L\1)
0.580 CO2 0.010
0.000 SiO2 0.060
0.450
0.310
0.000
1.340
 III / WATER TREATMENT / Overview: Ion Exchange 4475

water efRciently, the following equation must be are now available. In certain cases they are suitable
satisRed. both for new plant and to revamp or modify existing
plant. They may also be used to check the current
FCh"OV
performance of a particular ion exchange line in op-
where h is the time in hours between successive regen- eration, and to evaluate possible changes in operating
erations. parameters. These programs or computer printouts
Many factors can affect the operating capacity in can be obtained from various resin manufacturers
a particular situation. This must always be lower than and from original engineering manufacturers to help
the total volume capacity of the resin. The most experts in these calculations. They can also serve as
important of these factors are the regeneration level useful training for those engineers learning their
and the particular analysis of the water being treated. trade, and for water treatment chemists who need to
In certain cases other factors such as Sow rate or cycle understand the operations of a particular plant in
time, operating and regenerant temperature, treated which they have an interest.
water quality and cycle end point, and resin bed Table 3 gives an example of data provided by
depth may also need to be taken into account. Fur- a plant design computer program on the design re-
thermore, the plant itself will need to treat some extra quirements for a speciRed Sow rate and treated water
water in order to operate successfully. This extra load quality. Table 4 gives an example of the engineering
needs to be allowed for in the design. The more data provided by a computer program. Its use can
concentrated the feed water the more extra load is save many hours in calculation time and allow explo-
placed on the resin beds. Indeed, the water may con- ration of the many options available before making
tain such a high level of dissolved salts that treatment a Rnal choice.
may not be economic. In such cases RO often pro-
Choice of Resins
vides a satisfactory alternative, or as a pretreatment.
When all these factors are carefully considered, the The optimization of any given process requires con-
optimization of the plant becomes quite complicated, siderable skill on the part of the design engineer. It is
especially when taking into account the various pro- important to understand the strengths and weak-
cess options and the possible choices of resin types nesses of each resin type so that the correct type and
and their various combinations. Such an exercise re- particle size are chosen. The design program is ex-
quires a vast experience of water treatment. To help tremely useful to balance and match the conditions of
design engineers and water treatment plant operators regeneration to produce the correct quality and
to make suitable design plans, computer programs quantity of treated water. The combined experience
of the customer and the engineer, together with the
expertise and support of resin specialists, is generally
Table 3 Design requirements regarded as the best approach to determining the
optimized conditions of operation and choice of the
Operating conditions
Flow rate per line 2.1 m3 h\1 correct equipment.
Running time 9.8 h
Net run 21 m3
Resin Life
Treated water quality
Provided that the design has been optimized, in all
Achieved Specified End point but the most difRcult cases the resin life should be in
the range of 4}6 years, depending on the type of resin.
Conductivity 0.70 1.00 2.00
In fact SAC resins have been known to last very much
(
S cm\1)
Silica leakage (ppb) 16 20 500 longer, provided they are used at the optimum tem-
Sodium leakage 0.038 perature and are kept free of contaminants and poten-
(ppm) tial oxidizing agents. There must also be the provison
that regenerate quantities, concentrations and Sow
Residual CO2 after 0.59 meq L\1
rates are designed to avoid stress on the resins. In
SAC filter
many cases regular checks on the state of resins can be
Process options beneRcial. This will prevent build-up of chemical con-
Ion exchange Demineralization taminants and highlight any maloperation before real
process damage is done.
Plant layout SACPSBA
No. of lines (as required) See Colour Plates 125, 126.
Resins chosen SAC SBA
See also: I/Ion Exchange.
4476 III / WATER TREATMENT / Overview: Ion Exchange

Table 4 Calculation of full plant design details for an ion exchange plant

Filter

SAC SBA

Ion exchange load

Gross run (m3) 21 21
Ionic load (eq) 28 30

Resin data
Resin type SAC SBA
Resin grade } }
Theoretical capacity (eq L\1 R) 1.15 0.62
Operational capacity (eq L\1 R) 0.48 0.56
Resin volume (L) 58 53
Flow rate (BV h\1) 36.9 40.2
Organic load (g L\1 KMnO4) 7.880

Regeneration data
Regeneration mode CTF : FB CTF : FB
Regenerant HCl NaOH
Concentration % 5.0 4.0
% of Theory 345 180
Level (g L\1 R) 61.0 40.0
Total (kg 100%) 4 2
Excess (eq) 69 24
Temperature (3C) 25
Dilution water (m3) 0.1 0.0
Slow rinse (m3) 0.1 0.2
Fast rinse (m3) Recycling Recycling

Plant size data

Bed depth (changing from supplied form as shown)
Supplied form (mm) 586 538
Exhausted form (mm) 562
Regenerated form (mm) 609 634
Vessel diameter (mm) 365 365
Cross-section (m2) 0.10 0.10
Cylindrical height (mm) Per design Per design

Hydraulic data
Linear velocity (m h\1) 21.2 21.2
Pressure drop (kPa) 19.0 14.8

Design factor 0.42 0.90

(Limitations caused by flow rate)

Note: BV, bed volume; R, resin.

Further Reading
Abrams MI and Benezra L (1967) Encyclopedia of Polymer
Science and Technology, pp. 692}742. Chichester: John
Wiley & Sons.
Chu B, Whitney DC and Diamond RM (1962) Journal of
Inorganic Nuclear Chemistry 24: 1405}1415.
Dale J and Irving J (1992) Comparison of strong base resin
types. In: Slater MJ (ed.) Ion Exchange Advances,
pp. 33}40. London: Elsevier Applied Science.
Diamond RM and Whitney DC (1966) Resin selectivity in
dilute to concentrated aqueous solutions. In: Marinsky
JA (ed.) Ion Exchange, vol. 1, pp. 277}349. New York:
Marcel Dekker.
Dorfner K (1991) Introduction to ion exchange and ion
exchangers. In: Dorfner K (ed.) Ion Exchangers. Berlin
and New York: Walter de Gruyter.
Harland CE (1994) Some engineering notes. In:
Ion Exchange: Theory and Practice, 2nd edn,
pp. 261}276. Cambridge: Royal Society of Chem-
istry.
Helfferich F (1962) Ion exchange equilibria. In: Ion
Exchange, pp. 151}248. New York: McGraw-
Hill.
 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
Newell PA, Wrigley SP, Sehn P and Whipple SS (1996)
An economic comparison of reverse osmosis and ion
exchange in Europe. In: Greig JA (ed.) Ion Exchange
Development and Applications, pp. 58}66. Cambridge:
Royal Society of Chemistry.
Anion Exchangers: Ion Exchange
W. H. Ho] ll, Karlsruhe Nuclear Research Center,
Karlsruhe, Germany
Copyright ^ 2000 Academic Press
Introduction
Anion exchange resins consist of a polymeric matrix
to which different functional groups are attached.
Most weakly basic anion exchangers contain tertiary
amino groups; in a few cases primary and secondary
groups are also encountered. In many cases weakly
basic anion exchangers are not monofunctional but
possess a variety of amino groups. Strongly basic
resins contain quaternary ammonium groups. Stan-
dard commercially available exchangers con-
tain either }N#(CH3)3 groups (type 1 resins) or
}N#(CH3)2C2H4OH groups (type 2 resins). Both
weakly and strongly basic exchange resins are avail-
able in gel-type or macroporous modiRcations.
Properties and Relds of application mainly depend
on the dissociation properties of the functional
groups in which dissociation plays the most impor-
tant role. By means of the mass action law, dissoci-
ation constants of the protonated amino and
ammonium groups can be estimated. The respective
numerical values are in the range of pKa'13 for
strongly basic resins and 5}8 for weakly basic resins.
Therefore, strongly basic resins are protonated over
the entire pH range but weakly basic exchangers are
protonated at pH values below 5}8, depending on the
type. As a consequence, strongly basic resins will
exchange anions in both acid and alkaline solutions.
In addition, these exchangers can adsorb weak acids
and even ionize very weakly dissociated acids. Weak-
ly basic resins, however, can operate only in acidic
media and are unable to convert neutral salts to the
respective hydroxides (e.g. NaCl to NaOH). Further-
more, they cannot normally adsorb weak acids.
The uptake of anions by resins is subject to speciRc
interactions between counterions and co-ions and the
distribution of exchangeable ions depends on the
properties of both the exchanger and the ions. Conse-
quently, a favoured sorption of certain types of
anions occurs. The sequence of afRnities is given
either qualitatively by the selectivity series or quantit-
4477
Nolan J and Irving J (1984) The effect on the capacity of
strong base anion exchange resins of the ratio of chloride
to sulphate in the feed water. In: Naden D and Streat
M (eds) Ion Exchange Technology, pp. 160}168. Chi-
chester: Ellis Horwood.
atively by, for example, separation factors. For weak-
ly basic anion exchangers the sequence of most com-
mon anions in fresh water is:
OH\SO24\'NO\
3 'Cl\
Due to the dissociation properties of the functional
groups, hydroxyl ions are strongly preferred. This is
important for the conversion of these resins to the
hydroxyl (or free base) form in the regeneration step,
in which only slightly more than the stoichiometric
amount of OH} bearing solutions is required.
For strongly basic anion exchangers the selectivity
series is:
SO24\'NO\
3 'Cl\'HCO\
3 'OH\
For these resins hydroxyl ions are the least prefer-
red among the standard anions. Conversion of the
resins to the hydroxyl form therefore requires com-
paratively large excess amounts of sodium hydroxide.
For elimination of nitrate anions from drinking
water the preferred sorption of sulfate ions causes
considerable disadvantages. Extensive research dur-
ing the 1980s has shown that this drawback can be
overcome by introducing functional groups which are
more hydrophobic and bulkier. By these means, the
ability to adsorb sulfate ions is considerably de-
creased. The so-called nitrate-selective resins which
are now commercially available contain triethyl in-
stead of trimethyl groups and, therefore, exhibit a re-
versed preference for nitrate and sulfate ions.
The rate of exchange depends mainly on the inter-
nal interdiffusion of exchanging ions. On the com-
pletely ionized strong base resins this diffusion is
rather quick and the overall rate of exchange mainly
depends on the particle size distribution. With weakly
basic exchangers, however, the poor dissociation and
further speciRc interactions between functional sites
and diffusing ions considerably slow the rate of ex-
change. For these resins, both the uptake of acids and
conversion to the free base form by means of sodium
hydroxide strongly depend on the concentration of
the liquid phases.
Strongly basic anion exchangers in the hydroxyl
form are subject to considerable degradation of
4478 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
functional groups at temperatures above 403C, lead-
ing to a loss of strongly basic functionality. Weakly
basic exchangers are very stable.
Applications
Water Demineralization
The most important application of anion exchangers
in water treatment has been in demineralization pro-
cesses. Demineralization consists of the subsequent
cation exchange for hydrogen ions as the Rrst step
and either adsorption of acids by a weakly basic
exchanger or real anion exchange for hydroxyl ions
on a strongly basic exchanger as the second step.
As a result, demineralized water is produced: this may
be used for various purposes, such as boiler feed
water.
The efSuent from the preceding cation exchange
step consists of a mixture of strong mineral acids
and carbonic acid; this is equivalent to the cations
originally dissolved. Furthermore it contains silica
and dissolved organic compounds, neither of which
are retained by the cation exchangers. If no silica
removal is needed, weakly basic anion exchangers
in their free base form are used to remove strong
mineral acids. This process develops as the uptake
of acids:
R3N#(HCl, H2SO4) 0 R3NH##(Cl\, SO24\) [1]
(Parantheses are used to express the stoichiometry of
the exchange; R3 denotes the matrix of the weakly
basic exchanger.) In technical Rlter columns the
uptake of sulfuric and hydrochloric acid results in
the development of zones with different predominant
loadings: Rrstly, a sulfate-rich zone close to the
inlet; secondly, a chloride zone; and Rnally, a zone
in which the resin is still in the free base form.
During the Rlter run the sulfate and chloride zones
become larger and the boundaries between the zones
are shifted towards the column outlet. Consequently,
chloride ions are the Rrst to appear in the column
efSuent.
Theoretically, the strong acids should be com-
pletely eliminated. In practical installations, however,
the leakage of sodium from the strongly acidic cation
exchanger cannot be avoided and it is in the range of
2}50
g L\1, depending on the level of regeneration.
By this means, the feed solution for the anion ex-
changer contains some neutral salt which cannot be
eliminated. Consequently, an equivalent amount of
chloride ions remains in the efSuent. Apart from this
sodium leakage-caused slip, the presence of chloride
ions may also be due to the hydrolysis of the hydro-
chloride form of weak base resins:
R3NH#Cl\ 0 R3N#HCl [2]
Weak acids like carbonic acid are only adsorbed to
a small extent and hydrogen carbonate ions are rap-
idly replaced by anions of strong acids. Carbonic acid
or CO2 is therefore eliminated either by physical
degassing or by means of a strongly basic exchanger.
Since the capacity of weakly basic resin increases with
increasing total ionic concentration (to which carbon-
ic acid contributes), degassing of CO2 is often placed
after the weakly basic exchanger Rlter.
Regeneration of weakly basic anion exchangers is
usually carried out by means of sodium hydroxide,
although other chemicals like Ca(OH)2 or am-
monium hydroxide have also been proposed. To
avoid the precipitation of calcium sulfate, the regen-
erant solutions of anion exchangers generally require
calcium-free water. Since the strong acids are easily
neutralized, regeneration is efRcient. The operating
capacity depends on the composition of the raw
water and the resulting loading of the anion ex-
changer. If equal amounts of caustic are applied in the
regeneration step, the effective capacity decreases by
about 10% if the raw water contains exclusively
sulfate instead of only chloride ions.
Unlike with weakly basic resins, the removal of
acids by a strongly basic resin develops as a neutral-
ization reaction:
R4N#OH\#(HCl, H2SO4)
0 R4N#(Cl\, SO24\)#H2O [3]
(R4 denotes the matrix of the strong base resin.)
Because of the high pH value in the resin phase,
strongly basic exchangers can ionize and adsorb ions
from carbonic acid and even from silica:
R4N#OH\#H2CO3 0 R4N#HCO\
3 #H2O [4]
R4N#OH\#H4SiO4 0 R4N#H3SiO\
4 #H2O
[5]
For boiler feed water, for example, the removal of
both carbonic acid and silica is of great interest.
Usually, most of the carbonic acid is removed by
physically degassing. By this means a large part of the
capacity of the strongly basic anion exchanger is
saved and the resin has to eliminate only the remain-
ing traces of carbonic acid. The removal of silica is
far more complicated than shown in eqn [5] above.
 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
Figure 1 Interdependence of sodium and silica leakage.
During the accumulation in the resin phase, silica
starts to polymerize. This is favoured by long cycles
as well as by high concentrations in the feed water.
Regeneration using NaOH may, therefore, become
a dissolution rather than a true ion exchange process.
Warm solutions (approximately 503C) are consider-
ably more effective than solutions at ambient temper-
ature. In addition, regeneration may be adversely
inSuenced by the mode of regeneration and the pos-
sible displacement of hydrogen carbonate ions. The
leakage of silica is closely related to the leakage of
sodium in the preceding cation exchanger (Figure 1).
Regeneration of strongly basic resins is generally
carried out by means of NaOH at concentrations
between 1 and 5%. As with weakly basic exchangers,
the resulting operating capacity depends on the com-
position of the feed water, the relative amount of
NaOH, temperature and, to a certain extent, the rate
of Rltration during regeneration. The relative quanti-
ty of NaOH required is considerably higher than for
weak base resins and lies between 150 and 250% of
the stoichiometric amount. For the slightly weaker
basic resins of type 2, the operating capacities are
higher than for type 1 resins.
There can therefore be savings in chemicals and
cost if the anion exchange step is split into two stages.
In the Rrst stage a weakly basic resin is applied for the
elimination of strong acids. After physical degassing
to remove carbon dioxide, the strongly basic ex-
changer is exclusively used to eliminate the remainder
of the carbonic acid and silica. Both resins are regen-
erated together: the NaOH solution Rrst passes
through the strongly basic resin which requires a con-
centrated regenerant without impurities. The efSuent
4479
then passes the weakly basic resin. Due to the strong
preference for hydroxyl ions, considerable levels of
impurities can be tolerated.
Demineralization of water by subsequent or
simultaneous cation/anion exchange has been
realized in countless installations for both fresh
and waste water treatment. It has become extremely
important for boiler feed water and for ultrapure
water for the electronics and pharmaceutical indus-
tries. Typical conditions to be met are conductiv-
ities below 0.2
S cm\1 and silica concentrations
(5 mg L\1.
Partial Demineralization of Drinking Water
In Europe numerous water supplies distribute drink-
ing water with elevated total hardness caused by the
presence of sulfate and calcium ions. Treatment of
such water, therefore, requires the elimination of sul-
fate ions parallel to diminishing hardness. In such
applications complete demineralization is not re-
quired. Consequently, considerable leakage of the ion
exchangers can be tolerated. The condition of simul-
taneously diminishing the concentrations of alkaline
earth and sulfate ions is met by the carbon dioxide
regenerated ion exchange (CARIX) process. This
process uses a mixed bed consisting of a weakly acidic
resin in the free acid form and a strongly basic resin in
the hydrogen carbonate form. In contact with cal-
cium- and sulfate-bearing raw water, both kinds of
ions are replaced by carbonic acid:
RCCOOH RCCOO\(Ca2#)
#(Ca2#, SO24\) 0
R4N#HCO\
3 R4N#(SO24\)
#H2CO3 [6]
The advantage of this process is its reversibility: for
regeneration, carbon dioxide is dissolved in untreated
raw water under a pressure of typically 0.4}0.5 MPa.
The carbonic acid solution is pumped across the
mixed bed and simultaneously regenerates both
resins. Although carbon dioxide is applied in large
excess, this does not contribute to the salt content of
the waste water which exclusively contains the ions
removed during the service cycle. Carbonic acid is
a weakly effective regenerant for both resins and
generates an effective capacity of +50% of the total
capacity on the cation exchanger and only 20% on
the anion exchanger. As a consequence, complete
demineralization is not possible and considerable
leakage is observed. The possible throughput between
two regenerations is small. Within certain limits the
4480 III / WATER TREATMENT / Anion Exchangers: Ion Exchange

1
Figure 2 Concentration histories of chloride, nitrate and sulfate in the Carix process. Feed alkalinity (HCO\
3 ): 6.5 mmol L\ . Circles,

3 ; diamonds, SO4\; squares, HCO\

2
Cl\; triangles, NO\ 3 .

CARIX process can be adapted to the objective Nitrate Removal

of treatment by adjusting the volume ratio
For many drinking water supplies the presence of
Vanion : Vcation of the two resins in the range of 3 : 1
nitrate in ground and surface water at elevated con-
to 1 : 3. With respect to both the removal of sulfate
centrations has become a serious problem because of
and regeneration by means of carbonic acid, type
its health effects. As the simplest possibility, nitrate
2 and acrylic anion exchangers show the best
can be eliminated from water by means of strong base
results.
anion exchangers in the chloride form:
So far, the process has been realized in three full-
scale plants in water works in Germany. Each of these R4N#Cl\#(NO\
3 , SO4\, HCO3\)
2 2

plants consists of one or two sets of three Rlters which

operate in a merry-go-round mode. The total 0 R4N#(NO\
3 , SO4\, HCO3\)#Cl\
2 2
[7]
throughput for each Rlter between two regenerations
is 35}50 bed volumes. After half of the total through- Using conventional strong-base anion exchangers,
put, a second Rlter starts its service cycle. After the the main difRculty of this method arises from the
full throughput, the Rlter is regenerated and waits for
Table 1 Partial demineralization by means of the CARIX
the next service cycle while the third one is started. process
The product water is the mixed efSuent of the two
operating Rlters. Typically, about 10% of the raw Parameter Raw water Product water
water is needed for regeneration. From the waste
Total hardness 5.4 mmol L\1 2.3 mmol L\1
water, 90% of the unspent carbon dioxide is re- Hydrogen carbonate 6.6 mmol L\1 3.4 mmol L\1
covered. Thus, the consumption of CO2 amounts to Sulfate 1.6 mmol L\1 0.35 mmol L\1
0.4}0.45 kg m\3. Nitrate 0.6 mmol L\1 0.45 mmol L\1
Figure 2 shows a typical breakthrough history for Chloride 1.5 mmol L\1 1.3 mmol L\1
Conductivity 930
S cm\1 470
S cm\1
standard anions. Average product water concentra-
pH Value 7.30 7.80
tions are listed in Table 1.
 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
preferred uptake of sulfate ions. Consequently,
the nitrate uptake capacity strongly depends on the
sulfate/nitrate concentration ratio in the raw
water. As an example, for a type 2 resin the nitrate
uptake capacity decreases by 60% if the sulfate
concentration in the feed increases from 50 to
220 mg L\1.
During the uptake of sulfate, nitrate and hydrogen
carbonate by a chloride-loaded conventional anion
exchanger, loading zones develop in the Rlter column:
a sulfate-rich zone, a nitrate-rich section and a mixed
chloride/hydrogen carbonate zone followed by the
resin in its original chloride form.
Due to the development of the zones and the dis-
placement of less preferred anions, continuous
change occurs in the efSuent composition. When the
hydrogen carbonate-rich zone has reached the outlet
there is still removal of nitrate species, but an increase
in the concentration of hydrogen carbonate and a cor-
responding increase in the pH in the efSuent occur.
The column run has to be stopped when the nitrate-
rich zone reaches the Rlter outlet.
Regeneration is carried out using brine solutions of
2}10% NaCl:
R4N#(SO24\, NO\
3 )#NaCl
Figure 3 Operating capacity of a type 2 strongly basic anion exchanger depending on the concentrations of nitrate and sulfate in the
raw water. Continuous line, 0.36 mmol L\1; dashed line, 0.72 mmol L\1; dotted and dashed line, 2.1 mmol L\1. Resin: Lewatit M600.
4481
0 R4N#Cl\#(Na2SO4, NaNO3) [8]
The total quantity of NaCl used depends on the
tolerable nitrate leakage in the service cycle. To
achieve efSuent concentrations 42 mg L\1, the re-
quired amount is 5200 g L\1 resin. With smaller
quantities the leakage becomes larger.
The operating capacity of conventional anion ex-
changers depends on both the nitrate and sulfate
concentrations of the feed water. Figure 3 shows the
respective interdependence for a commercially avail-
able type 2 resin.
The objective of treatment normally does not con-
sist of the complete elimination of nitrate but only of
its reduction below the maximum permitted concen-
tration. A tolerable nitrate concentration in the prod-
uct water can therefore be achieved by two different
methods: the Rrst uses an exchanger which has been
regenerated with a large amount of NaCl. Since the
leakage of the column is small, only part of the water
has to be treated and can be blended with a by-pass of
untreated water.
This kind of nitrate removal process was Rrst real-
ized in 1975 in Long Island in a pseudo-continuous
installation. This plant consisted of a closed-loop tube
including sections for back-washing, regeneration
4482 III / WATER TREATMENT / Anion Exchangers: Ion Exchange

Table 2 Data of operation of Ecodenit plant dilute solutions, HCrO\4 is the predominant species.
Above pH 6.5 mostly CrO24\ is found. With respect to
Parameter Feed water Product water Waste water ion exchange processes it is important that dimeriz-
concentration concentration concentration
(mg L\1 ) (mg L\1 ) (g L\1 ) ation occurs at elevated concentrations:

4 0 Cr2O7\#H2O
2
Nitrate 70 25 25 2HCrO\ [9]
Sulfate 50 10 10
Hydrogen 65 55 55 The respective pH conditions may exist in the
carbonate anion exchanger phase. Therefore, the resins can be
Chloride 50 120 120 considered to be at least partly loaded with Cr2O27\
species.
Chromate species are strongly preferred over chlor-
and rinsing of the exhausted resin. After exhaustion
ide and sulfate ions under normal conditions. How-
the resin material in the contacting section is pulsed
ever, at acidic pH this preference vanishes for both
into the back-wash part, whereas regenerated and rin-
weakly and strongly basic anion exchangers when
sed resin material enters the contacting section. The
chromate is the trace component. Consequently,
plant was designed for a maximum throughput of
Rxed-bed experiments using standard acidic chro-
277 m3 h\1. The nitrate concentration was diminished
mate-bearing waste waters exhibit a gradual increase
from 100 to 43 mg L\1 as the tolerable maximum.
in the efSuent concentration. An increase in the con-
In the second possibility the amount of NaCl re-
centration of competing sulfate ions yields only a neg-
quired during regeneration is smaller. As a conse-
ligible decrease in chromate capacity. In contrast to
quence, the regeneration is less efRcient and the
this, an increase in the chloride concentration leads to
nitrate leakage becomes larger. Thus, all the water
a considerable decrease in chromate capacity for
has to be treated. This principle is shown in the
a given liquid-phase concentration. For weakly basic
Ecodenit process developed in France. This process
resins the chromate capacity decreases with increas-
operates with NaCl quantities of 4100 g L\1 of
ing pH because of the deprotonation of the functional
resin and uses co-Sow regeneration. The Rrst full-
groups. Below pH 5 the capacity is constant and
scale plant for the treatment of 160 m3 h\1 is in
depends only on the background composition of the
service in northern France. This plant consists of
solution.
three Rlters operating in a merry-go-round mode.
Using strongly basic exchangers chromate can be
Each Rlter contains 8 m3 of a strongly basic resin.
removed by the following exchange processes:
Results of the operation of the full-scale plant are
summarized in Table 2.
Both principles are combined in a third modiRca- R4N#OH\#(CrO24\) 0 R4N#(CrO24\)#(OH\)
tion. Such a plant went into service in 1983 in Cali- [10]
fornia. The plant was to treat 115 m3 h\1 blended
with 45 m3 h\1 raw water. It consists of three col- R4N#Cl\#(Cr2O27\) 0 R4N#(Cr2O27\)#(Cl\)
umns which are also operated in a merry-go-round [11]
mode. For each Rlter the throughput between two
regenerations amounts to 250 bed volumes. The Weak base anion exchangers are applied, for
nitrate concentration is decreased from 71 to example in the sulfate form:
11.5 mg L\1 in the mixed efSuent of the service Rlters
and to 27}36.5 mg L\1 in the blended product water.
R3N#(SO24\)#CrO24\ 0 R3N#(CrO24\)#SO24\
Since 1983, numerous nitrate elimination plants have
come into service: most apply nitrate-selective ex- [12]
changers.
R3N#(SO24\)#Cr2O27\ 0 R3N#(Cr2O27\)#SO24\
Chromate Removal [13]

Chromic acid and chromates are widely used in many AcidiRcation of the feed solution leads to the
applications. Anion exchange offers an ideal oppor- formation of hydrogen dichromate species, which
tunity for the removal and recovery of chromates. doubles the capacity of the resins, e.g.:
Both strongly and weakly base resins can be applied.
In aqueous solutions chromate ions exist in differ- R3N#(Cr2O27\)#Cr2O27\#H2SO4
ent ionic forms. The speciation mainly depends on the
pH value. In the acidic region (1(pH(6.5) and for 0 R3N#(HCr2O\
7 )2#SO4\
2
[14]
 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
Figure 4 Elimination of Cr(VI) from cooling tower blowdown. Effluent concentration histories of loading (circles) and polishing
(squares) columns. Feed water composition: Cl\(500 mg L\1), SO24\ (200 mg L\1), Cr(VI) (10 mg L\1), HCO3 (100 mg L\1),
Ca2##Mg2# (5.5 mmol L\1), Na# (250 mg L\1). Resin: Amberlite IRA 94.
After exhaustion, treatment with NaOH will con-
vert hydrogen dichromates to chromates and, sub-
sequently, yield half of the chromate loading on
elution. Removal of the remainder of the chromate
from the strongly basic exchanger requires concen-
trated sodium sulfate solutions:
R4N#(CrO24\)#Na2SO4
0 R4N#(SO24\)#Na2CrO4
In contrast to strongly basic exchangers, weakly
basic ones can be completely regenerated by means of
NaOH. However, before the following service cycle,
the resins have to be reconverted to the sulfuric acid
form:
R3N#(CrO24\)#2NaOH 0 R3N#Na2CrO4
R3N#H2SO4 0 R3N#(SO24\)
Chromate species elimination from cooling tower
blowdown has been achieved by application of
a styrene weakly basic resin in a merry-go-round
system with three columns. In the plant, column 1
serves as the crude loading column which eliminates
most of the chromate. The relatively low efSuent
concentrations are further decreased by the second
(polisher) column. The third column is regen-
erated/conditioned or waits for service. When the
maximum tolerable efSuent concentration of the
polisher column is exceeded, the sequence is switched
4483
so that the polisher column now acts as the loading
column and the freshly regenerated Rlter is applied
for polishing. The previous loading column is regen-
erated and conditioned. A typical example is given in
Figure 4.
Removal of Organic Substances
Natural organic substances normally have an anionic
nature at pH values above neutral. Apart from their
elimination in a suitable pretreatment step, they can
[15]
be adsorbed by both weakly and strongly basic anion
exchangers. Uptake of organic substances is due to
both ionic and van der Waals forces. They may pres-
ent a problem for anion exchangers because of a pos-
sibly irreversible adsorption by styrene-based gel-type
resins.
Both types of exchangers, therefore, can act as
scavengers for the removal of organic compounds
[16] prior to further ion exchange steps. In general, strong-
ly basic exchangers exhibit a better elimination of
humic substances. For demineralization of industrial
[17] water they are applied in the chloride form as scaven-
ger units at the head of the deionization train. The
basic problems associated with the application of
anion exchangers as scavengers in demineralization
plants are the additional exchanger unit and the ex-
change of the organics and hydrogen carbonate ions
for chloride. Thus, the efSuent composition may be
less favourable for the application of weakly acidic
exchange resins.
Depending on the total concentration of organic
matter in the feed water and to the desired level in the
efSuent, different resin types are recommended. The
combined application of weakly and strongly basic
4484 III / WATER TREATMENT / Anion Exchangers: Ion Exchange
resins has been successfully applied in the Stratabed
concept, in which two layers of the respective resins,
both of the polystyrene type, are used. The applica-
tion of this concept requires the use of exchange
resins with a particle size distribution which allows
the maintenance of the stratiRed bed in each column.
Since regeneration is carried out using warm NaOH
solution, precipitation of silica is avoided. Favourable
conditions are met: waters of poor alkalinity and
chloride and sulfate are the major part of the anions
present. Acrylic anion exchangers allow a reversible
uptake of humic acids. Although the capacity of
acrylic resins for organics is poorer, they are highly
effective. They can be used in all applications except
in condensate polishing with extremely high Sow
rates. Under these conditions the elastic properties of
these resins exclude their use.
At low levels of organic matter in the feed water the
application of macroporous weakly basic resins in
their free base form in the de-ionization train allows
an efRcient removal. Furthermore, these resins are
less susceptible to irreversible fouling.
In drinking water treatment, macroporous styrene-
based strong base resins were employed in Hanover
in Germany as a single treatment step for the removal
of organic matter from ground water. Dissolved or-
ganic carbon was to be reduced from 6.5 to about
3 mg L\1. After 5000 bed volumes, the resin was
regenerated by means of an alkaline NaCl solution
which could be reused seven times. The plant was
shut down in 1995.
WATER-SOLUBLE VITAMINS: THIN- LAYER
(PLANAR) CHROMATOGRAPHY
See III / VITAMINS/WATER-SOLUBLE: THIN-LAYER (PLANAR) CHROMATOGRAPHY
WAXES: SUPERCRITICAL FLUID
CHROMATOGRAPHY
See III / OILS, FATS AND WAXES: SUPERCRITICAL FLUID CHROMATOGRAPHY
Apart from the elimination of natural organic
matter, anion exchangers may also be applied to the
removal of organics from different types of waste
water. Weakly basic resins have been used for the
removal of phenol. Sorption is carried out at rather
low Sow rates of 2}8 BV h\1. NaOH or organic
solvents are used for regeneration. Macroporous
weakly basic resins in the sulfuric acid form are ap-
plied for decolorization of kraft bleach liquors. After
exhaustion they are regenerated using NaOH and
conditioned by means of sulfuric acid.
See also: I/Ion Exchange. III/Water Treatment: Over-
view: Ion Exchange. Resins as Biosorbents: Ion
Exchange.
Further Reading
Bayer AG (1974) Lewatit-Lewasorb Manual. Leverkusen:
Bayer.
Bolto BA and Pawlowski L (1987) Wastewater Treatment
by Ion Exchange. London: E. & F. Spon.
Dorfner K (1991) Ion Exchangers. Berlin: Walter de
Gruyter.
Harland CE (1994) Ion Exchange, Theory and Practice.
Bath: Bath Press.
Kunin R (1996) Amber-hi-lights. Littleton, CO: Tall
Oaks.
Mitsubishi Chemical. Diaion Manual of Ion Exchange
Resins and Synthetic Adsorbents. Mitsubishi Chemical.
SenGupta AK (ed.) (1995) Ion Exchange Technology.
Lancaster and Basel: Technomic.

WHISKY: DISTILLATION
D. S. Pickerell, Makers Mark Distillery,
Loretto, KY, USA
Copyright ^ 2000 Academic Press
Introduction
Grain fermentation yields a water-based liquid mix-
ture commonly referred to as distillers beer. This
beer will typically contain between 5 and 9% by
weight ethyl alcohol, 6}8% by weight residual grain
solids, and a very small quantity of other compounds
known as fusel oils. These fusel oils, also known as
congeners, are primarily higher alcohols that are sol-
uble in ethyl alcohol but only partially soluble in
water. The congeners contribute to the taste and
aroma of whisky and are not typically removed in
a single-column distillation.
All separation technologies exploit some difference
between items in a mixture or solution in order
to cause them to separate. These differences may
be physical, chemical or electrical in nature. In
particular, distillation takes advantage of the differ-
ence in boiling points to separate soluble liquids from
one another. Not all liquid solutions may be eco-
nomically separable by distillation for a variety of
reasons. For example, one or more of the liquid
components may not appreciably volatize, or the
change in the concentrations of the components be-
tween the gas phase and the liquid phase may be so
small that the process becomes impractical. It may
even happen that there is no change in the composi-
tion whatsoever.
In general, during distillation of completely
miscible liquids, the component with the
higher boiling point moves toward the bottom of
the still while the component with the lower boil-
ing point moves toward the top. In whisky produc-
tion, water boils at a higher temperature while
ethyl alcohol boils at a lower temperature. As a
result, distillation has an added beneRt as the
separation technique of choice, because the grain
residue is naturally carried to the bottom of the
still along with the water. If the still is
properly designed, the concentration of alcohol
in the still bottoms should be negligible and the
discharge from the bottom of the still will contain
all of the unwanted grain residues and the excess
water.
III / WHISKY: DISTILLATION 4485
Vapour^Liquid Equilibrium
In order to understand what happens during the dis-
tillation process, we need to address the topic of
vapour}liquid equilibrium. For the purpose of this
discussion, we will consider the case of distilling
a mixture of water and ethyl alcohol at a constant
pressure of 1 atm. Figure 1 shows the vapour}liquid
equilibrium curves for this mixture. It should be
noted that the concept of a single boiling point is
invalid for this type of solution. The lower line is
referred to as the bubble point line. At a given concen-
tration of ethyl alcohol in a liquid mixture of ethanol
and water, the bubble point line indicates the temper-
ature at which the Rrst bubble of vapour forms as the
solution is heated. The upper line is called the dew
point line. At a given concentration of ethanol in
a vapour mixture of ethanol and water, the dew point
line indicates the temperature at which the Rrst drop
of condensate is formed as the mixture is cooled.
In order to explain the distillation process, a rather
simplistic approach is employed by assuming an
absolutely ideal system with no inefRciencies. For
actual distillation system design, a much more thor-
ough analysis would need to be done. For illustrative
purposes, let us assume we have a liquid solution
consisting of 40% by weight ethanol and 60% by
weight water in a pot at 823C. Figure 2 shows the
vapour-liquid equilibrium of this solution, which is
currently at point A. Let us further assume that we
will add heat to this pot in an effort to bring the
temperature up to 993C, as represented by point
B (Figure 3). The solution will heat up until the
Figure 1 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.
4486 III / WHISKY: DISTILLATION
Figure 2 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.
temperature reaches about 833C where the heating
line intersects the bubble point line at point M. At this
point, the Rrst bubble of vapour forms, but because
the ethanol vaporizes more easily than the water at
this point, the vapour phase is enriched in ethanol.
The concentration of ethanol in this Rrst bubble of
vapour is found at point N, about 75% ethanol by
weight (Figure 4). As the mixture continues to heat
up, eventually point P is reached at about 873C. At
this point, the mixture is boiling. The liquid still in the
pot has a concentration of about 17%, as indicated
by point 0, while the total vapour concentration is
represented by point R at about 64% ethanol (Fig-
ure 5). The solution can continue to be heated until
the heating line intersects with the dew point line at
point T. At this point there is only one drop of liquid
left in the pot and its concentration is found at point
S to be about 2% ethanol by weight. If all of the
vapour from this experiment was collected, its con-
centration would be found at point T } approxim-
ately 40% by weight ethanol } right back where we
Figure 3 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.
Figure 4 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.
started from, only hotter. The vapour could then be
superheated to 993C at point B, but no further cha-
nges in ethanol concentration would occur.
It should be noted that, during the distillation pro-
cess, once the bubble line is reached, the concentra-
tion of ethanol in the liquid phase moves along the
bubble point line from left to right, constantly de-
creasing until the supply of liquid is exhausted. Sim-
ilarly, the concentration of ethanol in the vapour
phase also decreases, along the dew point line, as the
liquid in the pot is exhausted. As a result, if we were
going to distill ethanol from water in a batch process
with a lower limit of acceptable proof, we would
have to stop the process before all the ethanol
could be recovered. Ideally, we would like to be
able to recover all of the ethanol from the solution at
some speciRed constant proof. If point N is the target,
we could devise a process where we continually
replenish the liquid in the pot at 40% ethanol and
a rate equal to the rate that product is taken off by
condensation.
Figure 5 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.

Figure 6 Vapour}liquid equilibrium: ethanol}water, 760 mmHg.
Continuous line, bubble point line; dashed line, dew point line.
If a concentration greater than that represented by
point N is desired, a single pot cannot accomplish the
task (Figure 6). Suppose, however, that we set up an
apparatus whereby we constantly feed the Rrst pot as
described earlier, but now we condense the product
vapour and put it in another pot where it can be
distilled a second time. The concentration of the va-
pour from the second pot is represented by point X,
about 82% by weight ethanol. It can be seen that
adding more pots to this scheme would result in
higher and higher concentrations of ethanol in the
product. There is, however, a limit to this approach.
As the bubble point and dew point lines get closer
together, the increase in ethanol concentration per
added pot decreases. Eventually, these two lines
touch. The point at which these lines touch is called
an azeotrope. Azeotrope is a Greek word meaning
to boil together. Literally, at this point additional
separation by conventional two-phase distillation is
impossible because the liquid- and vapour-phase con-
centrations are identical. In fact, even getting close to
the azeotrope requires more advanced distillation
practices than those commonly used in whisky produc-
tion.
By anology, it can be seen that the problem of
recovering the residual ethanol from the still bottoms
can also be accomplished through the use of addi-
tional distillation stages. In practice, however, the
distillation column is a more efRcient method of ac-
complishing these distillation tasks than multiple pot
stills. Single malt Scotch whisky makes use of mul-
tiple pot stills in the production of their Rnal distillate
in a manner similar to that described above.
The Continuous Beer Still
The Rrst distillation element in a multicomponent
whisky distillation system is commonly referred to as
III / WHISKY: DISTILLATION 4487
Figure 7 Cut-away of typical distillation column.
the beer still. The beer still consists of a cylindrical
shell and number of evenly spaced trays connected by
pipes called downcomers. Figure 7 shows a cut-away
view of the inside of a typical beer still. The liquid in
the still moves across the trays and down the down-
comers. The vapour in the still moves up the column
through holes in the tray and through the liquid. The
pressure of the vapour under each tray must be great
enough to allow the vapour to pass through the holes
and through the liquid to the next level up the tray
without allowing the liquid to drip through the holes
(Figure 8). Each time the vapour passes through the
liquid, the vapour gains ethanol concentration while
the liquid loses ethanol concentration. One tray is
roughly equal to one distillation in a pot similar to
that discussed earlier. Technically, the vapour con-
denses in the liquid of the tray above it, and gives off
its heat of vaporization. This heat of vaporization in
turn revaporizes a corresponding volume of vapour
which is richer in ethanol.
The still is conceptually divided into two sections,
the stripping section and the rectifying section. The
stripping section is the part of the still that is on and
below the feed-tray level. This section is referred to as
4488 III / WHISKY: DISTILLATION
Figure 8 Typical plate flow detail.
the stripping section, because here the residual alco-
hol is essentially stripped from the feed stream so that
the still bottoms have a negligible presence of ethanol.
The still must be designed not only to produce a spirit
of the desired proof, but also to limit base losses.
Typically, the stripping section has about 16}20
plates. The plates in this section must be designed to
minimize the likelihood of fouling due to the grain
residue being present here. Almost exclusively, sieve
trays are used for this purpose because they have
larger, less complex vapour openings and wider toler-
ances to help prevent plugging with grain particles.
The space between the trays must also be sufRciently
wide to prevent foam and other entrained liquid on
one tray from inSuencing the tray above it.
It is possible for grain particles in the feed to be
entrained in and carried upwards by the vapour pass-
ing through the feed tray. This can happen on any
tray in the stripping section, but it is most critical on
the feed tray. Various approaches have been utilized
to minimize this problem; almost all are mechanical
alterations to the still itself. The most common de-
entrainment device is the use of one additional sieve
tray immediately above the feed tray.
The rectifying section of the still is the part that is
above the feed tray. In this section, the alcohol is
concentrated to the desired product proof. Typically,
the rectifying section has between two and Rve plates.
The plates in this section are designed to cause more
efRcient commingling of the vapour with the liquid as
the vapour passes through the plate. Since solids are
no longer an issue, the contacting mechanisms can be
more intricate and tolerances closer in this section.
The plates in this section may also be more closely
spaced because foaming and entrainment are much
less of a problem here than in the stripping section.
The Beer Heater
As a rule of thumb, the conditions of the feed stream
to the still should match, as closely as possible, the
conditions on the tray to which the feed is introduced.
It has already been noted that the ethanol concentra-
tion of the feed stream is generally between 5 and 9%
by weight. The concentration can be closely predicted
from heat and material balance calculations where no
empirical data exist for a given feed stream. It has
also been noted that the feed tray is generally near the
18th plate. The feed tray location can also be pre-
dicted from detailed will design calculations. The
only problem that remains, then, is the feed temper-
ature. When fermentation is complete, the beer
temperature is generally about 343C. The feed tray
liquid temperature should be about 933C. Since the
vapour from the still generally has to be cooled and
condensed, it provides a convenient source of heat to
pre-heat the beer. Usually, the beer feed is passed
through a shell and tube-type heat exchanger with
large diameter tubes to help alleviate plugging. The
vapours from the still are on the shell side of the
exchanger.
The condensate from the beer heater is generally
returned to the still as reSux. ReSux is the liquid
returned to the top of the still. It alters the number of
trays required to perform the desired degree of separ-
ation as well as the tower cross-sectional area and the
heating and cooling loads required for vaporization
and condensation. ReSux is generally referred to as
a ratio of the liquid returned to the still versus prod-
uct collected. As the reSux ratio goes up, the number
of trays required to perform the separation goes
down, but the requisite heating and cooling loads go
up. At inRnite reSux, the minimum number of trays is
achieved, but the maximum heating and cooling
loads are required. At minimum reSux, an inRnite
number of trays is required for the separation, but
minimum energy requirements are achieved. An opti-
mum reSux ratio can be calculated and the beer
heater can be designed to provide that reSux.
A general rule of thumb for still design would
require that the reSux be introduced to the still on the
top plate because of its temperature and composition.
However, many distillers have made the decision to
enter the reSux lower down the column near the beer
feed plate for quality reasons. Additionally, the Rnal
distillation proof in whisky production is only partly
determined by economic considerations. Depending

on the type of whisky being produced, there are
generally governmentally prescribed maximum
ethanol concentrations which may be permitted dur-
ing the distillation process. In the case of bourbon, the
US Bureau of Alcohol, Tobacco, and Firearms pre-
scribes that the distillate may be taken from the still at
no higher than 160 proof (80% ethanol by volume).
Of utmost importance, however, are the organoleptic
considerations which go into the production par-
ameters for the whisky. Product taken off at a
lower proof retains more of the grain character, while
product taken off at a higher proof tends to have less
of the grain Savour constituents.
The Doubler
Many distillers utilize a doubler in their whisky distil-
lation process. The doubler is basically a pot still, like
the one discussed earlier. The doubler acts as one
additional distillation stage. It is used in practice for
Rnal proof adjustment and for product quality en-
hancement. There are two fundamentally different
ways of operating the doubler. The Rrst is called true
double distillation. In true double distillation, the still
vapours generally pass through the beer heater Rrst,
and then one or more condensers, so that the product
is completely condensed back to a liquid form. This
liquid is then charged to the doubler where it is heated
with steam coils and re-vaporized. The vapour from
the doubler is then condensed again and taken off as
product.
The doubler can also be operated as a thumper. In
this case, the doubler is Rtted with a large sparger.
The doubler is charged with liquid to a level just
above the sparger. The liquid is typically de-
mineralized water or the low proof tails cut from
a previous distillation.The vapour from the still Rrst
passes through the beer heater then through the spar-
ger in the doubler where it bubbles through the liquid.
As the vapour passes through the liquid in the doub-
ler, it Sash condenses and gives off its heat of vapor-
ization which, in turn, revaporizes a corresponding
volume of vapour which is richer in ethanol. The
thumper gets its name from the sound made as the
vapour condenses while passing through the liquid.
Finally, the ethanol-enriched vapour passes through
one or more condensers and is taken off as product.
Reboilers
Most beer stills are heated by direct steam injection
from a low pressure steam sparger located in the base
of the still. A reboiler is a type of heat exchanger
which permits the use of higher pressure steam than
a steam sparger will allow. There are several advant-
III / WHISKY: DISTILLATION 4489
ages to using a reboiler. First, it acts as one theoretical
plate in the distillation column. Second, it saves on
the amount of waste to be disposed of from the still
bottoms because it adds no water to the system.
However, reboilers are not generally used in
whisky production because they have a great tend-
ency to scorch the grain in the bottoms and, hence,
degrade the product quality. In other stills with no
grain residue, reboilers have been used quite success-
fully.
Process Control
Control of the continuous beer still is generally ac-
complished by means of three interrelated control
loops. These loops regulate the level of the liquid in
the bottom of the still, the Sow of steam into the
bottom of the still and the Sow of the beer feed near
the top of the still.
Typically, the liquid level in the bottom of the still
is sensed by a level transmitter which, in turn, regu-
lates a control valve on the discharge of a continuous-
ly running base level pump. Alternatively, in certain
conRgurations, the base level can be regulated very
simply by means of a Soat valve set at a certain level.
This requires that the discharge be capable of gravity
Sow away from the still bottom. Additionally, newer
technology has made it possible to dispense with the
control valve on the base level pump. The level trans-
mitter can provide a signal to a frequency inverter
which controls the frequency of the electrical current
running the pump. This frequency shift will cause the
pump to speed up or slow down in relation to the
signal from the level transmitter.
In a similar manner, the steam Sow to the still is
generally held at a constant base pressure or a con-
stant Sow rate. Base pressure control is the most
common means of steam control. A pressure trans-
mitter in the base of the still above the liquid level
provides a control signal to a control valve which, in
turn, regulates the Sow of low pressure steam into
the steam sparger in the bottom of the still. If steam
Sow control is desired, an oriRce plate or vortex Sow
meter is inserted into the steam line. The Sow-sensing
device provides the control signal to regulate the
control valve. In the past, some distillers have used
the still top temperature as a means of regulating the
steam Sow, while holding the beer feed constant.
While this means is effective, it tends to be less re-
liable due to the relatively large amount of process
response lag time.
The beer feed is generally regulated by means of
sensing the still top temperature, which is directly
related to the proof of the distillate. A temperature
transmitter generally provides a control signal to
4490 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
a process control valve in the beer feed line, which is
fed by a constantly running feed pump. More recent
technology has made it possible to control the
proof more directly by using the temperature-correc-
ted density function of a mass Sow meter, which
can be correlated to the proof of the discharge
from the still. The only downside to the use of
a mass Sow meter is the process lag time that results
from having to measure the proof of the distillate
after condensation. Additionally, the control valve
can be eliminated from this loop by using a frequency
inverter, as described above. Some distillers employ
a more sophisticated means of controlling the beer
feed to the still by use of a cascaded control loop.
Typically, a magnetic Sow meter is used to measure
the Sow of beer to the still and control the operation
of the control valve. The still top temperature trans-
mitter provides a signal which is used to manipulate
the control settings for this Sow control loop.
In addition to the above controls, one or more
condensers must also be controlled. Generally, a con-
trol valve on the inlet cooling water line is used to
control this process. The control signal typically com-
es from a temperature transmitter which can either be
located on the discharge water line or the discharge
product line. Due to the relatively quick Sow rate of
the cooling water with respect to the product Sow
rate, process control response is generally much bet-
ter if the temperature transmitter is located on the
cooling water discharge line.
Finally, if the product is double-distilled in a true
doubler, one additional control loop is required. The
steam Sow to the steam coils inside the doubler
must be regulated. Almost without exception, this
loop consists of a steam control valve and a temper-
ature transmitter on the vapour discharge from the
doubler. In a manner similar to the still top control,
WINE: GAS AND LIQUID
CHROMATOGRAPHY
J. Guasch and O. Busto, Universitat Rovira i Virgili,
Tarragona, Spain
Copyright ^ 2000 Academic Press
Introduction
From the chemical point of view, wines are aqueous
alcoholic solutions containing more than 1000 com-
ponents that can be present at high concentrations
(g L\1), but also at trace levels (ng L\1). Some of
new technology has made it possible to control the
discharge proof more directly using a mass Sow meter.
Conclusion
A sign at the Stitzel-Weller distillery in Louisville,
Kentucky sums up the traditional view of the impact
of science on the beverage alcohol industry:
No Chemist Allowed
Nature and the oldtime know-how of the master distiller
get the job done here. Because traditional Kentucky
whisky is a natural product, we disdain synthetics, scien-
tist, and their accompanying apparatus. This is a distil-
lery, not a whisky factory.
Pappy Van Winkle
Tradition handed down through the generations is
the predominant means of whisky production. There
are numerous stories of a distiller who had to replace
his still because it had worn out. When the new still
was being installed, the distiller would make sure that
it was identical to the one it was replacing, right
down to the dent in the side of the still, which was
generally reapplied by the master distiller himself.
As a result, technological change is slow to be
adopted in an industry where any change in the pro-
cess may result in a changed taste. Technology is
gaining a foothold in the area of process control,
where new and better Rnal control elements, trans-
mitters and control systems are always being applied.
This traditional approach has also resulted in an
almost complete lack of published literature on the
topic of whisky distillation, which at best is viewed by
the industry as only part science and part art.
See Colour Plate 127.
See also: III/Wine: Gas and Liquid Chromatography.
these components determine the organoleptic proper-
ties of wines, while others are signi"cant for classify-
ing their origin and/or for checking whether some
adulteration has taken place. Concentration levels of
these compounds vary according to the variety of
vine, the climatic conditions under which the grapes
were grown, and the conditions under which vini"ca-
tion and ageing processes have been developed.
The quality of wines is established by sensory anal-
ysis, which is clearly correlated to their chemical
4490 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
a process control valve in the beer feed line, which is
fed by a constantly running feed pump. More recent
technology has made it possible to control the
proof more directly by using the temperature-correc-
ted density function of a mass Sow meter, which
can be correlated to the proof of the discharge
from the still. The only downside to the use of
a mass Sow meter is the process lag time that results
from having to measure the proof of the distillate
after condensation. Additionally, the control valve
can be eliminated from this loop by using a frequency
inverter, as described above. Some distillers employ
a more sophisticated means of controlling the beer
feed to the still by use of a cascaded control loop.
Typically, a magnetic Sow meter is used to measure
the Sow of beer to the still and control the operation
of the control valve. The still top temperature trans-
mitter provides a signal which is used to manipulate
the control settings for this Sow control loop.
In addition to the above controls, one or more
condensers must also be controlled. Generally, a con-
trol valve on the inlet cooling water line is used to
control this process. The control signal typically com-
es from a temperature transmitter which can either be
located on the discharge water line or the discharge
product line. Due to the relatively quick Sow rate of
the cooling water with respect to the product Sow
rate, process control response is generally much bet-
ter if the temperature transmitter is located on the
cooling water discharge line.
Finally, if the product is double-distilled in a true
doubler, one additional control loop is required. The
steam Sow to the steam coils inside the doubler
must be regulated. Almost without exception, this
loop consists of a steam control valve and a temper-
ature transmitter on the vapour discharge from the
doubler. In a manner similar to the still top control,
WINE: GAS AND LIQUID
CHROMATOGRAPHY
J. Guasch and O. Busto, Universitat Rovira i Virgili,
Tarragona, Spain
Copyright ^ 2000 Academic Press
Introduction
From the chemical point of view, wines are aqueous
alcoholic solutions containing more than 1000 com-
ponents that can be present at high concentrations
(g L\1), but also at trace levels (ng L\1). Some of
new technology has made it possible to control the
discharge proof more directly using a mass Sow meter.
Conclusion
A sign at the Stitzel-Weller distillery in Louisville,
Kentucky sums up the traditional view of the impact
of science on the beverage alcohol industry:
No Chemist Allowed
Nature and the oldtime know-how of the master distiller
get the job done here. Because traditional Kentucky
whisky is a natural product, we disdain synthetics, scien-
tist, and their accompanying apparatus. This is a distil-
lery, not a whisky factory.
Pappy Van Winkle
Tradition handed down through the generations is
the predominant means of whisky production. There
are numerous stories of a distiller who had to replace
his still because it had worn out. When the new still
was being installed, the distiller would make sure that
it was identical to the one it was replacing, right
down to the dent in the side of the still, which was
generally reapplied by the master distiller himself.
As a result, technological change is slow to be
adopted in an industry where any change in the pro-
cess may result in a changed taste. Technology is
gaining a foothold in the area of process control,
where new and better Rnal control elements, trans-
mitters and control systems are always being applied.
This traditional approach has also resulted in an
almost complete lack of published literature on the
topic of whisky distillation, which at best is viewed by
the industry as only part science and part art.
See Colour Plate 127.
See also: III/Wine: Gas and Liquid Chromatography.
these components determine the organoleptic proper-
ties of wines, while others are signi"cant for classify-
ing their origin and/or for checking whether some
adulteration has taken place. Concentration levels of
these compounds vary according to the variety of
vine, the climatic conditions under which the grapes
were grown, and the conditions under which vini"ca-
tion and ageing processes have been developed.
The quality of wines is established by sensory anal-
ysis, which is clearly correlated to their chemical
 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
composition. To assure and control this quality, some
essential parameters and characteristic compounds
are determined by physical and chemical analyses,
which are established principally by the OfTce Inter-
national de la Vigne et du Vin (OIV). Other constitu-
ents whose determination is included in these
methods are those associated with the toxicity of
wines or which are allowed to be present to some
maximum permissible levels.
Until recently, the techniques used in the analysis of
this broad variety of properties and compounds have
been based on classical methods (mainly gravimetric,
titrimetric and colorimetric), which allow an ad-
equate control of the viniRcation in the wineries.
Although chromatographic techniques are not widely
used in the OIV methods, the complexity of wine
composition has pointed to the use of chromato-
graphy in many oenological laboratories, and the
improvement in the analysis of wines is undeniably
bound up with the development of chromatography.
Almost all the chemical compounds present in
wines can be analysed by chromatography, either by
direct injection or by prior derivatization. For vol-
atile, thermally stable compounds, gas chromato-
graphy (GC) is the most used technique, while for the
analysis of nonvolatile and thermally unstable com-
pounds, high performance liquid chromatography
(HPLC) is preferred. These techniques are considered
in more detail below.
Gas Chromatography
Gas chromatography has been responsible for the
most important advances in the knowledge of the
volatile fraction of wines. Although the non-
volatile fraction can also be analysed by GC after
derivatization of the analytes, HPLC methods are
simpler and therefore they are preferred for these
compounds.
The main application of GC to wine analysis is the
study, characterization and determination of the
aroma of wines, which originates from their volatile
components. Aroma compounds are usually classiRed
according to their chemical functionality, the most
important being esters, alcohols, acids, lactones, car-
bonyl compounds, volatile phenols and sulfur- and
nitrogen-containing volatiles. With the exception of
ethanol and glycerol, the concentrations of the indi-
vidual aroma compounds range from 100 mg L\1 to
0.1 ng L\1. The human sensory organs are extremely
sensitive to certain aroma substances, which can
show sensory thresholds much lower than their con-
centrations in wine. The analysis of wine aroma must,
therefore, be optimized in order to determine all these
compounds. To perform this kind of analysis, two
4491
different steps must be considered: sample prepara-
tion and gas chromatographic separation.
Sample Preparation
The sample pretreatments are conditioned by the
wine matrix and the character and the concentration
of the analytes to be determined. Wines can be di-
rectly injected, but it is more common to inject the
extract obtained after the application of pretreatment
techniques.
Direct injection is applied to the analysis of the
volatiles whose concentration is close to the mg L\1
level, so they are easily detected by GC detectors.
When injecting wine in this way, the nonvolatile
fraction remains in the injector and may be thermally
degraded, giving rise to potential interfering substan-
ces (artefacts) and unstable baselines. Distillation of
the volatile fraction or Rltration, after addition of
a water-miscible solvent to reduce the polarity of the
wine matrix, helps to minimize this problem.
Injection after clean-up and concentration treat-
ments is usually applied to the analysis of trace
compounds (
g L\1 to ng L\1) to enhance their detec-
tability. At the same time, interfering substances are
removed during the isolation procedure. The main
problems with these techniques are the occasional
quantitative and qualitative changes of the analytes,
the formation of artefacts by chemical reactions or
thermal decomposition, and the introduction of im-
purities. For these reasons, the suitability of the isola-
tion and concentration methods for a particular
analyte has to be carefully evaluated.
Distillation is normally used to isolate the wine
volatiles from the nonvolatiles. It can be carried out
at atmospheric or reduced pressure in different distil-
lation modes (direct, steam and fractional). By work-
ing at reduced pressure and low temperature chemical
reactions or thermal decomposition can be mini-
mized. The most important disadvantage is that the
isolates obtained are diluted and it is necessary to
combine the distillation with other methods that con-
centrate the volatile fraction.
Solvent extraction is used to simultaneously isolate
and concentrate the volatiles. It is carried out in batch
mode (simple or multiple) or continuous mode (by
using continuous liquid}liquid extractors). The
choice of the solvent is conditioned by the high con-
centration of ethanol (10}15%) in wines. Owing to
their low boiling points and very low polarities, pen-
tane, dichloromethane and their azeotropic mixtures
are commonly used because they discriminate against
ethanol. Other solvents used are diethyl ether and ethyl
acetate. Fluorocarbons were widely used because
of their extraction efRciency and very low boiling
points, but nowadays they are environmentally
4492 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
unacceptable. One of the disadvantages of solvent
extraction is the use of large volumes of solvents
(normally not free of contaminants) and the large
amount of time spent in the extraction. Whenever
possible, the use of minimum solvent/sample volume
ratios enhances the concentration and minimizes con-
tamination problems. At the same time, the efRciency
of the extraction can be raised appreciably by salting-
out the solution with sodium chloride.
The low boiling points of these solvents allow the
concentration of the extracted volatiles by distilling
off the solvent. By using a Kuderna}Danish concen-
trator the loss of volatiles is minimized. Furthermore,
a gas stream is used to remove the solvent excess from
the extract. This procedure is very effective, but may
lead to the introduction of contaminants from the gas
and to losses of the most volatile compounds.
Simultaneous distillation}extraction, using the ap-
paratus originally described by Likens and Nikerson,
has not been commonly applied to wine analysis.
Supercritical Suid extraction is not common, but it is
becoming increasingly accepted.
Solid-phase extraction (SPE) has also been used
for the isolation of wine aroma compounds. The
most common adsorbents are charcoal, silica gel
and porous polymers (Chromosorb2+, Porapak2+,
Amberlite XAD2+ and Tenax2+). The volatile
compounds retained are usually eluted and/or frac-
tionated by pentane, diethyl ether, dichlorometane,
ethyl acetate or their mixtures. This technique is
preferred for the analysis of a speciRc group of
volatiles, since the adsorbent used is normally
selective. Large volumes of adsorbents and solvents
are used in order to assure the whole recovery
of analytes, so dilute solutions are obtained. A
Rnal step including solvent evaporation is therefore
needed.
Headspace techniques are widely used in wine
aroma determinations because they enable the direct
analysis of the headspace gas above the heated sam-
ples, where the compounds responsible for the aroma
detected by the human nose are transferred. The
static headspace technique is suitable for the analysis
of the aromatic compounds of highest concentration,
but for the analysis of trace levels it is necessary to use
dynamic headspace (purge and trap) techniques. The
retention of the volatiles is usually achieved by using
either cryo or sorbent traps. Sorbent traps are nor-
mally preferred because the retention of water and
ethanol is minimized, using the same adsorbents men-
tioned for SPE. The trapped volatiles are recovered by
extraction with small volumes of solvent or by ther-
mal desorption, which can be carried out in the
chromatographic injector, enabling the overall
sample to be analysed in a single step. On the
other hand, the volatiles obtained by solvent extrac-
tion are more dilute, but the sample can be fractioned
and therefore injected in several chromatographic
runs.
Solid-phase microextraction (SPME) is a single-
step solvent-free extraction technique that combines
the advantages of both SPE and headspace tech-
niques. It has been increasingly applied to the isola-
tion of Savour compounds. The adsorbent (usually
polydimethylsiloxane-, divinylbenzene- or polyac-
rylate-coated fused silica Rbres) is Rxed in the needle
of a specially designed chromatographic syringe and
exposed either to the liquid sample or to the head-
space above it. After exposure of the Rbre to the
sample, absorbed analytes are recovered from the
Rbre by thermal desorption in a conventional GC
injection port.
Chromatographic Separation
The chromatographic separation is normally carried
out in a gas chromatograph equipped with a split/
splitless injector and a Same ionization detector
(FID). On-column injectors with retention gaps are
very useful for the analysis of traces because they
enable the injection of large volumes of wine extracts
and the concentration of the volatile fraction at the
head of the chromatographic column. Programmed
temperature vaporizer (PTV) injection is also suitable
for the direct desorption of volatile compounds trap-
ped on injector glass liners Rlled with adsorbents.
The detection of the analytes is usually carried out
with a FID or a mass spectrometer detector (MSD).
Other detectors are used only to detect more speciRc
compounds. The Same photometric detector (FPD)
and, more recently, the sulfur chemiluminiscence
detector (SCD), are widely used for detection of sul-
fur-containing compounds, mainly thiols, sulRdes,
disulRdes and heterocyclic compounds. The electron-
capture detector (ECD) is used to detect halogenated
substances, such as chlorophenols and chloroanisoles,
which are associated with cork taint off-Savours. The
ECD and the nitrogen phosphorus detector (NPD)
are widely used for the analysis of pesticide residues
and some speciRc additives.
To characterize the wine Savour, gas chromatogra-
phy}olfactometry (GCO) has been coupled with dif-
ferent methods that determine the relative aroma
intensity. The smell of the different components of
the wine aroma is assessed by snifRng the efSuent of
the chromatographic column in parallel with FID
detection.
All kinds of chromatographic columns can be used
to separate the volatile analytes of wines, but fused
silica capillary columns with different stationary
phases are the most common. Polyethylene glycol
 III / WINE: GAS AND LIQUID CHROMATOGRAPHY 4493

phases are preferred for the evaluation of the global
wine aroma, while less polar phases (such as methyl-
siloxane and methylphenylsiloxane phases) are
needed for assessing the identiRcation of individual
compounds. Chiral phases are used for the separation
of the enantiomers of volatile compounds, which
exhibit very different sensory properties. Multi-
dimensional gas chromatography is used for the anal-
ysis of volatiles that are not well separated with
a single column.
Selected Applications
Many of the sources listed in the Further Reading
section deal with the analysis of wines by GC. The
following methods are usually performed in oenologi-
cal laboratories for routine control and research
studies.
According to the OIV method, the determination
of methanol and ethyl acetate in wines is carried out
by GC-FID. Wine distillates are injected in the split
injection mode on a polyethylene glycol column un-
der isothermal conditions. This method enables the
simultaneous determination of other compounds
present in the distillate, such as acetaldehyde, 1-pro-
panol, 2-methylpropanol, 1-butanol, 2- and 3-methyl-
butanol, 1-pentanol, 1-hexanol, 2-phenylethanol,
ethyl lactate, ethyl succinate, 3-methylbutyl acetate,
acetic acid, some polyalcohols and so on. In routine
analysis, this procedure is usually simpliRed when
wines are directly injected. Figure 1 shows an
example of the chromatogram obtained under these
conditions.
The main aroma compounds of wine distillates are
analysed by direct split injection of samples with
a temperature programme that optimizes the separ-
ation of the different substances. The compounds
determined in wine distillates by this OIV method are
methanol, acetaldehyde, acetals, higher alcohols,
ethyl esters of fatty acids, acetates of the main alco-
hols and volatile fatty acids. There are other aroma
compounds that can be detected by splitless injection
of the extract obtained with a batch extraction using
ether/hexane (1 : 1) and magnetic stirring. Figure 2
shows an example of the chromatogram obtained
when a methylene chloride wine extract is injected.
In research work, the isolation of the global aroma
of wine is usually performed by continuous solvent ex-
traction using different ratios of pentane}dichloro-
methane and different times of extraction. The ex-
tract is dried over anhydrous sodium sulfate and
concentrated either in a Kuderna}Danish device or
with a gas stream. The concentrate is injected into the
GC-FID, working in splitless mode and with a pro-
grammed column temperature. Purge and trap
methods are suitable alternatives to this procedure
and, more recently, SPME has also found some ap-
plications in the determination of the ethyl esters of
spirit beverages and in the analysis of the main aroma
of fruit juices.
Currently, many other volatile compounds are in-
vestigated by GC for their sensory contribution. The
most important are terpenes, lactones (solerone,
sotolone and oak lactones), carbonyl compounds
(hexenals,
-damascenone and - and
-ionone), vol-
atile phenols (alkylphenols and alkylguamK acols), thiols,
sulRdes, disulRdes, pyrazines and vitispiranes. The
particular methods of analysis for these compounds
are fully described in publications listed in Further

Figure 1 Chromatogram from direct injection of a wine sample (GC-MS). Key: 1, carbon dioxide; 2, acetaldehyde; 3, ethyl acetate;
4, 1-propanol; 5, 2-methyl-1-propanol; 6, 2-methyl-1-butanol; 7, 3-methyl-1-butanol; 8, acetone; 9, ethyl lactate; 10, acetic acid;
11, (D)-2,3-butanediol; 12, meso-2,3-butanediol; 13, 1,2-propanediol; 14, 3-ethoxy-1-propanol; 15, 2-phenylethanol.
4494 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
Figure 2 Chromatogram of the injection of a concentrated wine extract obtained by methylene chloride extraction (GC-MS).
Reading. Two examples of chromatograms obtained
from the analysis of sulfur compounds (Figure 3) and
pyrazines (Figure 4) are shown.
High Performance Liquid
Chromatography (HPLC)
The Rrst studies on the application of HPLC to the
analysis of wines appeared at the end of the 1970s for
the analysis of polyphenols. Since then, HPLC has
been applied to the separation, characterization and
determination of a large number of wine compounds
Figure 3 Sulfur compounds found in the headspace of
a cryogenically trapped wine (GC-FPD). Key: 1, hydrogen sulfide;
2, methanethiol; 3, carbon disulfide; 4, ethanethiol; 5, dimethyl
sulfide; 6, methyl ethyl sulfide (internal standard); 7, diethyl sul-
fide; 8, methyl propyl sulfide; 9, ethanol; 10, tiophene (internal
standard); 11, methyl thioacetate; 12, dimethyl disulfide; 13, ethyl
thioacetate; 14, ethyl methyl disulfide; 15, diethyl disulfide. (Re-
produced with permission from Mestres M, Busto O and Guasch
J (1997) Chromatographic analysis of volatile sulfur compounds
in wines using the static headspace technique with flame photo-
metric detection. Journal of Chromatography A 773: 261}269.
Copyright 1997, Elsevier Science.)
or groups of compounds. The literature concerning
the application of HPLC in wine and must analysis is
very extensive, but it is important to emphasize the
particular interest of this technique in the study of
polyphenols, amino acids, biogenic amines, organic
acids and sugars.
The use of HPLC is rarely recommended in the
ofRcial methods. However, carboxylic acids, sacchar-
ose, hydroxymethylfurfural and some additives such
as sweeteners can be determined by ofRcial HPLC
methods.
As mentioned above, one of the main prob-
lems when dealing with wines is the complexity
of the matrix. Although HPLC offers the possibility
to choose columns, solvents, detectors and derivatiz-
ing reagents, many of the chromatographic proced-
ures developed for HPLC determinations in wines
involve some kind of sample pretreatment. These
procedures generally make use of either ion exchange
Figure 4 Pyrazines found in the headspace of a wine after
SPME (GC-NPD). Key: 1, 3-isopropyl-2-methoxypyrazine; 2, 3-
ethyl-2-methoxypyrazine; 3, 3-sec-butyl-2-methoxypyrazine; 4, 3-
isobutyl-2-methoxypyrazine; H, 3-isopropyl-2-ethoxy-pyrazine (in-
ternal standard).
 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
chromatography and/or solvent of solid-phase ex-
traction.
Sample Preparation
Filtration of samples through a 0.45
m membrane is
always recommended before injecting wines into the
HPLC system. This process can be before or at the
same time as more complex pretreatments.
One of the methods for avoiding the presence of
interfering substances is eluting wine through a low
pressure liquid chromatographic column. Polyvinyl-
pyrrolidone (PVP), polyvinylpolypyrrolidone (PVPP)
and polyamide adsorbents are used to eliminate poly-
phenolic substances and silica is used to retain pro-
teins. Ion exchangers are commonly used either to
clean up samples or to isolate ionized amines and
organic acids from wines.
The most common method used to pretreat sam-
ples is solvent extraction with ether or ethyl acetate,
although many researchers use more selective sol-
vents.
Solvent extraction has generally been replaced by
SPE and SPME. Although there are some applications
of carbon and ion exchange cartridges, octadecyl-
silane (C18) is the most commonly used adsorbent.
Many of the analytes whose determination in wines is
of interest (such as polyphenols, amino and aroma
compounds) can be selectively retained or eluted with
slight modiRcations of matrix conditions (such as pH
or addition of ion pair reagents) or by transforming
the analytes by derivatization.
Chromatographic Separation
Reversed stationary phase (RP) are the most popular
in the HPLC analysis of wines, although it is fully
recognized that they are not capable of separating all
kinds of analytes in wine. Apart from some special
applications, silica is utilized almost exclusively as the
support material and C18 as the bonded phase.
Since wines are constituted of analytes spanning
a wide range of polarities, linear solvent strength
gradients are preferred for analysis. Mobile phases
normally consist of binary mixtures of either meth-
anol or acetonitrile and slightly acidiRed water.
The variable UV-visible (UV-vis) wavelength de-
tector is the most popular, although Suorescence and
refractive index detectors are also common. Photo-
diode array detection has also found some applica-
tion.
Selected Applications
In contrast to GC, the analysis of wines by HPLC has
focused on determining compounds with similar
chemical functionality.
4495
Figure 5 Chromatogram of free amino acids in white wine after
derivatization with PITC and DAD detection. (Reproduced with
permission from Calull M, FaH bregas J, MarceH RM and Borrull
F (1991) Determination of free amino acids by precolumn derivat-
ization with phenylisothiocynate. Application to wine samples.
Chromatographia 31: 272}276.)
Ion exchange chromatography has become the
most popular technique for the determination of
amino acids due to the use of autoanalysers. After
chromatographic separation, the analytes are de-
rivatized with ninhydrin, Suorescamine or o-phthal-
dialdehyde and detected by spectrophotometry.
Amino acids can also be determined by RP-HPLC,
which is faster than ion exchange chromatography.
The stationary phases are based on amino and, espe-
cially, C18 chemical groups. Although isocratic elu-
tion is used in some applications, gradient elution is
preferred because it enables the simultaneous deter-
mination of amino acids of different polarities. Mo-
bile phases are normally of binary composition
(methanol or acetonitrile and an aqueous buffer solu-
tion). As in ion chromatography methods, the amino
acids are derivatized, but this time dansyl chloride or
phenylisothiocyanate (Figure 5) are used for UV-vis
detection and o-phthalaldehyde for Suorimetric de-
tection.
Amines are also determined by HPLC, either by
direct injection or, more commonly, by derivati-
zation. When they are directly injected, they are sep-
arated by ion pair chromatography on a C18 column
and detected by conductimetry or spectrophoto-
metry. The main limitation of these procedures is that
mobile phases shorten the life of the column, so
procedures involving the separation of derivatized
amines are preferred. Ninhydrin is one of the reagents
commonly used in post-column derivatization. The
separation is done either by RP-HPLC or by ion
exchange chromatography. However, these methods
are very time-consuming and so pre-column derivat-
izations are preferred. The derivatizing reagents used
in this case are the same as when dealing with
4496 III / WINE: GAS AND LIQUID CHROMATOGRAPHY
Figure 6 OPA-derivatives of biogenic amines in red wine after
SPE and fluorescence detection. Key: 1, ethanolamine; 2, hista-
mine; 4, ethylamine; 5, tyramine; 6, isopropylamine; 8, tryp-
tamine; 10, phenethylamine; 11, putrescine; 14, cadaverine; ,
peak corresponding to the excess of OPA; x, unknown. (Repro-
duced with permission from Busto O, Guasch J and Borrull
F (1995) Improvement of a solid-phase extraction method for
determining biogenic amines in wines. Journal of Chromatogra-
phy A 718: 309}317. Copyright 1995, Elsevier Science.)
amino acids, o-phthalaldehyde being the most used
(Figure 6). The derivatives are separated by RP-HPLC
and detected by spectrophotometry or Suorimetry.
Although carboxylic acids can be determined by
GC, after suitable derivatization, the OIV proposes
the use of HPLC for determining carboxylic acids in
wines, as an alternative to the usual enzymatic pro-
cedures. RP-HPLC is used either by direct injection of
the wine or by derivatization of the acids before
separation. Direct injection of carboxylic acids is
a simple method, but the use of mobile phases at low
acidic pH (to avoid acid ionization) considerably re-
duces the life of the analytical column, hence derivat-
ization is recommended. Furthermore, when acids are
transformed into their corresponding esters, detection
is more sensitive. The different derivatization
methods reported so far for the determination of
organic acids with spectrophotometric detection and
which are worthy of special mention are those which
use organic compounds containing the chromophore
groups phenacyl, naphthacyl and p-nitrobenzyl. The
organic groups containing courmarin and anthracene
groups, on the other hand, are used for Suorimetric
detection. Mobile phases are of binary solvents (nor-
mally methanol and water) and the elution is done
isocratically or with a linear gradient.
Anion exchange chromatography is an alternative
to RP-HPLC, using mobile phases made of organic
Figure 7 Chromatogram of a wine obtained by high resolution
ion exclusion chromatography and refractometric detection. Key:
1, citric acid; 2, tartaric acid; 3, glucose; 4, malic acid; 5, fructose;
6, acetic acid; 7, glycerol; 8, lactic acid; 9, methanol; 10, ethanol.
solvents buffered at a pH close to 8 and with conduc-
timetric or refractometry detection. Although good
results are obtained by this method, ion exclusion
chromatography using strong cation exchange phases
has become the best technique. There are stationary
phases speciRcally developed for determining car-
boxylic acids in fermented products, which permit the
simultaneous determination of sugars, ethanol, meth-
anol and glycerol in a single run. Mobile phases
consist of slightly acidic water solutions and the
detectors used are either UV or refractive index.
Figure 7 shows an example of the chromatogram that
is obtained when wine is directly injected under these
conditions.
Some papers have reported the separation of car-
boxylic acids by NP-HPLC and ion pair chromato-
graphy, but the results are not comparable to those
obtained from the methods described above.
Although the ofRcial methods of analysis of sugars
are based on enzymatic techniques, sugars can also be
determined by HPLC. The preferred methods are
based on the use of speciRc ion exclusion polymeric
columns because they enable the simultaneous deter-
mination of carbohydrates and other analytes, as al-
ready mentioned. The mobile phase used is dilute
sulfuric acid and the detection is carried out by spec-
trophotometry or refractometry. RP-HPLC of sugar
benzoylated derivatives followed by spectro-
photometric detection has also been used to deter-
mine carbohydrates in wines (Figure 7).
According to the OIV methods, saccharose is ana-
lysed by HPLC. In this case, the column used is based
on 3-aminopropylsiloxane-bonded phases and the
mobile phase is acetonitrile and water with refractive
index detection.
There are classical methods for estimating the total
phenol content of wines, but HPLC is necessary
for the determination of individual polyphenolic
 III / WINE: GAS AND LIQUID CHROMATOGRAPHY 4497

Figure 8 HPLC chromatograms of a must monitored (A) at 280 nm for all phenolic compounds, and (B) at 520 nm to selectively
detect anthocyanins. Key: 1, gallic acid; 2, cis-caffeoyltartaric; 3, trans-caffeoyltartaric; 4, S-glutathionylcaftaric; 5, cis-coumaroyltar-
taric; 6, trans-coumaroyltartaric; 7, procyanidin B1; 8, catechin; 9, procyanidin B2; 10, delphinidin-3-glucoside; 11, epicatechin; 12,
cyanidin-3-glucoside; 13, petunidin-3-glucoside; 14, peonidin-3-glucoside; 15, malvidin-3-glucoside; 16, cyanidin-3-glucoside acetate;
17, rutin; 18, petunidin-3-glucoside acetate; 19, peonidin-3-glucoside acetate; 20, malvidin-3-glucoside acetate; 21, peonidin-3-
glucoside acetate (p-coumarate); 22, malvidin-3-glucoside acetate (p-coumarate). (Reproduced with permission from Lamuela RM and
Waterhouse AL (1994) A direct HPLC separation of wine phenolics. American Journal of Enology and Viticulture 45: 1}5.)

compounds. Chromatographic procedures are condi-
tioned by the lack of suitable standards and the com-
plexity of chromatograms. Thus, the determination is
tackled from the point of view of the two different
families of phenolic compounds: Savonoids (an-
thocyanins, Savanols and procyanidins), and non-
Savonoids (hydroxycinnamic and hydroxybenzoic
derivatives). Nevertheless, only the pretreatment of
samples is different depending on the fraction that
has to be isolated. HPLC on reversed-phase columns
is almost universally used for anthocyanin separation.
The most common used support is C18. Extremely
acid solvents are required to suppress ionization
of the analytes. Solvents such as methanol/
water and acetonitrile/water (in varying propor-
tions), acidiRed at low pH with phosphoric, per-
chloric or formic acid, have been used with different
solvent programmes. When dealing with Savonols
and procyanidins, extraction and puriRcation of
wines prior to HPLC is needed. HPLC analysis of
Savonols is achieved on C18 columns with binary sol-
vent systems consisting of acetonitrile and acetic acid
in water and using gradient elution programmes.
Procyanidins are chromatographed on C18, C8 or cyano
columns and dilute acid is normally required as a com-
ponent of the solvent to obtain satisfactory peak shapes.
Hydroxycinnamic acids are also analysed by HPLC, by
using C18 columns and methanol/water eluents slightly
acidiRed with acetic acid. In all cases, detection is car-
ried out spectrophotometrically (Figure 8).
4498 III / ZEOLITES: ION EXCHANGERS
Other substances which are present in wine but
which are not determined as frequently as the ones
discussed above can also be determined by HPLC.
These include additives such as sorbic, salicylic, ben-
zoic and ascorbic acids, which can be determined,
according to OIV methods, by RP-HPLC coupled
with either spectrophotometric or refractive index
detectors.
Future Trends
Both GC and HPLC techniques are widely used in
wine analysis. Although the methodologies are nor-
mally based on traditional separations, multidimen-
sional chromatographic methods (with or without
chiral phases) are increasingly being introduced, fre-
quently coupled online with other analytical devices.
More recently, capillary electrophoresis and super-
critical Suid chromatography have also been used for
wine determinations, but they are still in an early
stage of application.
At present, most of the chemical compounds pres-
ent in wine can be determined by means of a great
variety of chromatographic methods described in the
literature. Future trends, however, will focus more on
internal method validation rather than on develop-
ment of new methodologies. Future ofRcial methods of
analysis will then include the minimum requirements
(accuracy, precision, limit of detection, robustness,
and so on) that an analytical method must fulRl in
order to guarantee the validity of the results obtained.
See Colour Plate 128.
See also: II / Chromatography. Extraction. II / Chrom-
atography: Gas: Headspace Gas Chromatography.
III / Amines: Gas Chromatography. Amino Acids: Gas
XENOBIOTICS: MAGNETIC AFFINITY
SEPARATIONS
See III / BIOLOGICALLY ACTIVE COMPOUNDS AND XENOBIOTICS:
MAGNETIC AFFINITY SEPARATIONS
ZEOLITES: ION EXCHANGERS
C. D. Williams, University of Wolverhampton,
Wolverhampton, UK
Copyright ^ 2000 Academic Press
Chromatography; Liquid Chromatography. Phenols: Gas
Chromatography; Liquid Chromatography; Solid-Phase
Extraction.
Further Reading
Acree TE and Teranishi R (eds) (1993) Flavor Science.
Sensible Principles and Techniques, ACS Professional
Reference Book. Washington, DC: American Chemical
Society.
Doneche B (ed.) (1993) Les Adquisitions Re& centes en
Chromatographie du Vin. Paris: Lavoisier.
Gordon MH (ed.) (1990) Principles and Applications of
Gas Chromatography in Food Analysis. Chichester: Ellis
Horwood.
Horwitz W (ed.) (1990) OfTcial Methods of the Associ-
ation OfTcial of Analytical Chemistry, 15th edn. Arling-
ton, VA: AOAC.
Linskens HF and Jackson JF (eds) (1988) Wine Analysis.
Berlin: Springer.
Maarse H (ed.) (1991) Volatile Compounds in Foods and
Beverages. New York: Dekker.
Maarse H and Beltz R (eds) (1985) Isolation, Separation
and IdentiTcation of Volatile Compounds in Aroma
Research. Dordrecht: D. Reidel.
Morton ID and MacLeod AJ (eds) (1982) Food Flavours.
Amsterdam: Elsevier.
Nollet LML (ed.) (1992) Food Analysis by HPLC. New
York: Dekker.
NykaK nen L and Suomalainen H (1983) Aroma of Beer,
Wine and Distilled Alcoholic Beverages. Dordrecht: D.
Reidel.
OIV (1990) Recueil des Me& thodes Internationales dAna-
lyse des Vins et des MouL ts. Paris: OfRce International de
la Vigne et du Vin.
OIV (1994) Recueil des Me& thodes Internationales dAna-
lyse des Boissons Spiritueuses, des Alcohols et de la
Fraction Aromatique des Boissons. Paris: OfRce Interna-
tional de la Vigne et du Vin.
The ion exchange properties of zeolites have been
known since 1858, when Eichhorn studied the use of
chabazite as an ion exchanger. In the 1920s and
1930s several ion exchange studies were reported.
 III / ZEOLITES: ION EXCHANGERS 4499

Table 1 Phosphate based molecular sieve characteristics

Number Structure type Porosity Pore size (
m) Saturation H2O capacity (cm3 g\1)

Structure fully determined

5 Novel Large 0.80 0.31
11 Novel Medium 0.60 0.16
14 Novel Small 0.40 0.19
15 Leucophosphite
16 Zuvnite Very small 0.30 0.30
17 Erionite Small 0.43 0.28
20 Sodalite Very small 0.30 0.24
25 Novel Very small 0.30 0.17
46 Novel
Structures inferred from X-ray powder patterns
37 Faujasite Large 0.80 0.35
34 Chabazite Small 0.43 0.30
35 Levynite Small 0.43 0.30
42 [A] Small 0.43 0.30
43 Gismondine Small 0.43 0.34
44 Chabazite Small 0.43 0.3}0.34
47 Chabazite Small 0.43 0.3}0.34
Unknown structures
36 Novel Large 0.80 0.31
40 Novel Large 0.70 0.33
31 Novel Medium 0.65 0.17
41 Novel Medium 0.60 0.22
18 Novel Small 0.43 0.35
26 Novel Small 0.43 0.23
33 Novel Small 0.40 0.23
39 Novel Small 0.40 0.23
28 Novel Very small 0.30 0.21

During the 1960s many groups studied the ion ex- of other substituted aluminophosphates that do have
change behaviour of the new synthetic zeolites then ion exchange properties (Table 2).
being produced. Due to the enormous commercial Since the early 1980s several new zeotypes, based
potential of zeolites, many research groups world- on oxoanion frameworks, have been developed. The
wide began serious efforts to sythensize new micro- major group of materials of interest as far as ion
porous zeolites and zeotype materials. The Rrst major exchange properties are concerned are the layered
breakthrough was made by workers at Union Car- group IV acid salts. These include phosphates, ar-
bide, who in 1982 produced the aluminophosphate senates, molybdates, tungstates, antimonates, sili-
molecular sieves (Table 1). Although these materials cates and silicophosphates. Most of these materials
are electrically neutral and have no instrinsic ion act as cation exchangers. Early attempts at synthesis
exchange properties, they did lead to the development mimicked zeolite preparations using reactive
amorphous gels crystallized at temperatures between
120 and 2003C. This crystallization produced a var-
Table 2 Phosphate based molecular sieves with ion exchange
character
iety of materials, which have been classiRed by their
structure type: -layered exchangers, -layered ex-
Number Structure types changers, Rbrous exchangers, 3-D net exchangers and
unsolved structure exchangers. Table 3 lists some of
40 Novel the more important -layered ion exchangers.
41 Novel
34, 44 Chabazite
The structure of ZrP was determined by ClearReld
35 Levyne in 1969. The inorganic layers are formed by a plane
37 Faujasite of octahedral Zr atoms that are linked together alter-
42 A natively above and below via phosphite groups.
17 Erionite Three oxygen atoms of the phosphite group are coor-
20 Sodalite
5, 11, 16, 31 AlPO4
dinated in this way and the fourth bears a hydrogen
atom (Figure 1).
4500 III / ZEOLITES: ION EXCHANGERS
Table 3 Important  layered ion exchangers
Compound Formula
Titanium phosphate Ti(HPO4)2 ) H2O
Zirconium phosphate Zr(HPO4)2 ) H2O
Hafnium phosphate Hf(HPO4)2 ) H2O
Germanium phosphate Ge(HPO4)2 ) H2O
Tin(IV) phosphate Sn(HPO4)2 ) H2O
Lead(IV) phosphate Pb(HPO4)2 ) H2O
Titanium arsenate Ti(HAsO4)2 ) H2O
Zirconium arsenate Zr(HAsO4)2 ) H2O
Tin(IV) arsenate Sn(HAsO4)2 ) H2O
The ZrP exchangers have been characterized by
carrying out potentiometric titrations against MCl
and MOH solution mixtures. X-ray analysis of the
solid phases shows that ZrP is initially converted to
ZrMH(PO4)2 ) nH2O then on further exchange is
converted to Zr(MPO4)2 ) nH2O. At any point during
the ion exchange process these two phases coexist
together with the solution phase. The interlayer dis-
tance is large enough to accommodate unhydrated
Li#, Na# and K#; however Rb# and Cs# are too
large to enter without lattice expansion. The energy
to expand the lattice is supplied by a base, neutraliz-
ing the lattice protons and allowing larger cations to
enter.
The -layered compounds are far less common than
the  compounds. Both ZrPZr(HPO4)2 ) 2H2O and
TiPTi(HPO4)2 are known, but both suffer from hy-
dration at high exchange levels. Both materials have
large interlayer distances and as a consequence can
accept large cations such as Cs#.
Pastor et al. have reviewed the synthesis, character-
ization and ion exchange, ion transport, sorptive and
catalytic properties of inorganically pillared layered
metal(IV) phosphates, typiRed by Zr(BPO4)2 ) H2O.
Porous nanostructures are generally prepared from
metal(IV) phosphates either by ion exchange of poly-
nuclear species or by intercalation from solutions of
Figure 1 Structure of Zr-P.
Interlayer distance (As ) Ion exchange capacity (mmol g\1)
7.56 7.76
7.56 6.64
7.56 4.17
7.6 7.08
7.76 6.08
7.8 4.79
7.77 5.78
7.78 5.14
7.8 4.80
condensed species obtained by the hydrolysis of or-
ganometallic precursors using sol-gel methods. Ther-
mal treatment is used to eliminate organic moieties,
condense hydroxyl groups, eliminate water and con-
solidate the structure by grafting the pillar to the
layer. The different strategies devised to overcome the
problem presented by the high layer charge density of
- and -structured phosphates in obtaining porous
solids are described, including exfoliation and local
surface growth of pillaring ions, and modiRcation of
the zirconium phosphate matrix ix to reduce the
cation exchange capacity. Structural and textural
characteristics of Al, Cr, mixed Al-Cr, Fe-Cr, Ga-Al
and of Si-pillared phosphates obtained from X-ray
analysis by Rne structure (XAFS), X-ray photo-
electron spectroscopy (XPS), and magic angle spin-
ning nuclear magnetic resonance (MAS-NMR) are
presented, and the perspectives of nanocomposite pil-
lared layered solids in general are discussed in the
current context of mesoporous solids synthesized
using templates.
The Rbrous materials are exempliRed by cerium
and thorium(IV) phosphates. Their Rbrous nature
allows then to be fabricated into papers that allow
fast separation of cations. The precise structure of
these phosphates is unclear but is probably M(HPO4)2 )
H2O, where M"Ce or Th.

The three-dimensional materials have the general
formula NM2(IV)(XO4)3, where M(IV) is Ti, Zr, Th
or Ge; X is P or As and N is a univalent cation. The
structures consist of XO4 tetrahedra and M(IV)O6
octahedral linked by corner sharing to form 3-D net-
works; this linking forms cavities, occupied by N#. If
phosphate is progressively replaced by silicate the
cavities open up, allowing free movement of the
N# cations and leading to the cation exchange prop-
erties. Numerous speciRc examples of these materials
can be found in the literature, particularly the gallium
phosphate-derived materials.
A more recent series of exchangers are those of the
titanosilicate type, which have zeolite type pores/cav-
ities. The materials have a formula Na2Ti2O3SiO4 )
2H2O and are synthesized from an alkaline medium
under similar conditions to those used to crystallize
zeolites. The structure has been solved using Rietveld
reRnement and shows titanium atoms in clusters of
four, octahedrally coordinated by oxygen atoms. The
silicate groups link the titanium clusters into a square
which then shares corners with other titanium cluster
squares to form a 3-D network. Half of the sodiums
are linked into the framework while the other half are
labile and available for ion exchange.
In 1991 a zincosilicate containing three-, four- and
Rve-member rings connected together to form a por-
ous eight- and intersecting nine-member ring channel
was reported. Initial studies indicate that the labile
monovalent cations can be exchanged.
A great deal of synthetic work has been directed at
replacing the aluminium in various zeolites with other
metals/nonmetals, including the crystallization of
ferrisilicates, borosilicates, gallosilicates, vanadosili-
cates and titanosilicates. In 1992 the synthesis of
a zincophosphate anionic eight-ring three-dimen-
sional framework was reported. During the synthesis
the anionic framework was stabilized by cationic,
protonated diazabicyclo[2.2.2.]octane or dabco
[(H2N2C6H12)2#] molecules and water. These mater-
ials, although chemically similar to ClearRelds
layered phosphates/phosphites, beneRt from having
a stable open 3-D structure. No ion exchange data
has been given but thermal analysis shows that the
framework is stable even after the organic dabco has
been removed. This calcined material has potential as
a cation exchanger. In 1991 the synthesis of sodium
zirconium phosphate with a zeolite-type framework
was reported. The synthesis followed typical
aluminophosphate preparations using triethylamine
as a template. The synthesis was carried out in an
acidic medium, resulting in the template becoming
protonated. Once crystalline, the sample had the tem-
plate removed by calcination and the adsorption
properties of the new material were studied. The
III / ZEOLITES: ION EXCHANGERS 4501
material remained microporous after calcination;
however, no ion exchange studies were carried out.
Initial studies suggest that this material would act as
a cation exchanger.
However, although an enormous number of new
materials have been synthesized since 1990, there are
few reports on the ion exchange characteristics of
these materials.
Future Developments
Over the past Rve years increasing emphasis has been
placed on the investigation into microporous mater-
ials based on oxoanion networks other than the
aluminosilates (zeolites). The vast array of micropor-
ous materials with potential ion exchange properties
is enormous. The number of reported nonzeolite mo-
lecular sieves now tops 130. The range of materials
includes gallosilicates, borosilicates, ferrosilicates,
germanium aluminates, titaniosilicates, silico
alumino phosphate (SAPO) and metal alumino phos-
phate (MeAPO) molecular sieves. Most of these new
materials have not yet been characterized for their ion
exchange properties. The potential of these materials
is as yet unrealized but, with increasing environ-
mental demands, it is only a matter of time before
these materials are explored.
See also: II/Ion Exchange: Historical Development;
Inorganic Ion Exchangers; Novel Layered Materials:
Phosphates; Novel Layered Materials: Non-Phosphates;
Theory of Ion Exchange.
Further Reading
Annen MJ, Davis ME, Higgins JB and Schlenker JL (1991)
The physicochemical properties of VPI-7: a micro-
porous zinco-silicate with three membered rings. Mater-
ials Research Society Symposium Proceedings 233:
245}253.
Breck DW (1974) Zeolite Molecular Sieves. New York:
Wiley.
Dongare MK, Singh P and Suryavanshi PM (1992) Hy-
drothermal synthesis and characterisation of crystalline
sodium zirconium phosphates. Materials Research Bull-
etin 27: 637}645.
Dyer A, Hudson MJ and Williams PA (eds) (1997) Pro-
gress in Ion Exchange, Advances and Applications.
Cambridge: Royal Society of Chemistry.
Harrison WTA, Martin TE, Thurman EG and Stucky GD
(1992) Tetrahedral atom zincophosphate structures:
synthesis and structural characterisation of two novel
anionic eight ring frameworks containing cationic 1,4
diazabicyclo[2.2.2.]octane guests. Journal of Materials
Chemistry 2: 175}181.
Notari B (1993) Titanium silicates. Catalysis Today 18:
163}172.
4502 III / ZEOLITES: ION EXCHANGERS

Pastor PO, Torres PM, Castellon ER et al. (1996)
Chemistry of Materials 8: 1758}1769.
Poojary DM, Cahill RA and ClearReld A (1994) Synthesis,
crystal structure, and ion exchange properties of
a novel porous titano-silicate. Chemical Materials 6:
2364}2368.
Ratnasami P and Kumar R (1991) Ferri-silicate analogues
of zeolites. Catalysis Today 9.
Szostak R (1989) Molecular Sieves: Principles of Synthesis
and IdentiTcation. New York: Von Nostrand Rheinold.
Szostak R (1992) Handbook of Molecular Sieves.
New York: Von Nostrand Rheinold.
Wilson ST, Lok BM, Messina CA and Flanigen EM (1984)
Synthesis of AlPO4 Molecular Sieves. Proceedings of the
6th International Zeolite Conference (Reno Confer-
ence). Guildford: Butterworth ScientiRc, pp. 97}109.

ZINC ORES: FLOTATION

See III / LEAD AND ZINC ORES: FLOTATION

ZONE REFINING COUNTERCURRENT

CHROMATOGRAPHY
See III / PH-ZONE REFINING COUNTERCURRENT CHROMATOGRAPHY
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS 4533

ESSENTIAL GUIDES FOR

ISOLATION/PURIFICATION OF CELLS
J. Bauer, Max-Planck-Institut fu( r Biochemie, Methods for Cell Separation without
Martinsried, Germany Biological Modi\cation
Copyright ^ 2000 Academic Press The above deRnition rules out some technologies,
frequently used to prepare homogeneous cell popula-
In cell-separation technology the term component of
tions, because they include biological modiRcations
a mixture corresponds to a group of cells, which is
of cells. For example, the enrichment of cell types of
usually called a cell population and shares a number
interest by establishing cell lines or cell clones will not
of common features. How many common features
be considered as a subject in this chapter. Cell lines or
a group of cells has to share in order to be called a cell
cell clones may be very useful sources of important
population depends on the interest of the separator.
genes and gene products. However, their cells are
For example, a T-cell population may be a group of
transformed in unnaturally fast growing states, in
mononuclear white blood cells bearing CD3 antigens,
order to separate them from unwanted accompanying
while a helper cell population usually comprises
cells. Also cell separation/puriRcation techniques us-
mononuclear white blood cells bearing CD3 and CD4
ing different capabilities of various kinds of cells to
antigens.
adhere to surfaces of, for example, culture dishes or
Cells metabolize as long as they live independently,
Rbres or to bind antibodies or macromolecules label-
whether they remain in an actual state of activation
led by Suorescence dyes or magnetic beads, will not
or differentiation or they proceed to another one.
be described, because cell interactions with foreign
This means that a whole cell must not change its
components or antibody binding sites very often in-
appearance or functions, but some cell components
duce biological modiRcations. Of course, cell-puriR-
are chemically modiRed either anabolically or
cation methods like those mentioned above are very
catabolically. So for discussing cell separation the
useful in research and biotechnology. The reader may
term chemical modiRcation should be converted to
Rnd more informations regarding these techniques in
biological modiRcation and in this chapter the ex-
the Further Reading.
pression without biological modiRcation will be de-
This chapter focuses on application of counter-
Rned as puriRcation of cells without changes or sig-
current centrifugal elutriation (CCE) and free-Sow
nals for changes of cellular states of activation and/or
electrophoresis (FFE). These methods use differences
differentiation.
of physical cell parameters such as speciRc cell den-
No technology has been developed so far which
sity, cell size or negative surface charge density but do
allows picking of cell populations directly out of
not include steps of cell labelling or cell transforma-
pieces of plant or animal tissues. So a mixture which
tion. They have the advantage that cells can be puri-
will be separated is normally a suspension of single
Red within a short time while they are kept suspended
cells prepared from parts of plants, from organs or
in biocompatible Suids or even culture media. Cell
body Suids of animals and humans or from two- or
contacts to foreign surfaces and/or biologically active
three-dimensional in vitro cell cultures. These cell
substances are thereby minimized and signals of ac-
sources already contain preselected groups of cells,
tivation and differentiation are delivered to cells dur-
the so-called organ or Suid (e.g. blood) speciRc cells.
ing the isolation procedure to a minimal extent. Both
Still, a series of populations differing in important
methods may help to obtain sufRcient numbers of
features are present in most plant or animal body
identical cells with a high degree of purity and vitality
compartments. In these instances, it may be of inter-
for studying the biological role, which a deRned cell
est to separate cells for studying their biology or
population may play within an organism or for
for using some of their capabilities in medicine or
transplantation of cells with states of activation and
biotechnology.
differentiation suitable to Rt in the new organism of
Thus the following reSections on essential
a recipient.
guides for separation/puriRcation of cells are based
on separations deRned as processes of any scale
by which cell populations of single-cell suspensions
Single-Cell Suspensions
are separated from each other without biological Up to the present, cell separation by physical methods
modiRcation. has required single-cell suspensions. Some cell
4534 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS
compartments such as peripheral blood, ascites,
lymph or other body Suids already contain single
cells. Cells of organs such as bone marrow, spleen or
thymus can easily be removed, for example with the
help of needles. The dissociation of single cells from
two- or three-dimensional tissue cultures and from
solid body tissues needs more rigorous methods.
These cells not only adhere to each other, but are also
more or less Rrmly attached to the extracelluar
matrix, a complex network of collagen, proteog-
lycans and cross-linking proteins such as laminin and
Rbronectin. Mechanical dissociation by scraping cell
monolayers from their surfaces or by forcing tissue
pieces or cell aggregates through oriRces or syringes
or pipettes very often damages the cells and results in
a poor yield. An enzymatic treatment or pretreatment
of cell cultures or organs is thus often applied in order
to digest the extracelluar matrix and/or to weaken the
cell}cell attachment sites. The selection of the en-
zymes, their concentration and their time and temper-
ature of application depend on the type of organ and
its originating organism. Enzymes frequently used for
animal cell preparation are collagenase, trypsin,
pronase, dispase, papain, chymotrypsin, hy-
aluronidase, lysozyme and DNase, while cellulase is
a typical enzyme for plant tissue dissociation. Some-
times the action of the enzymes is supported by the
presence of EDTA (ethylenediaminetetraacetic acid),
which destroys binding sites mediated by Ca2# ions.
Details regarding techniques of preparing single cells
may be found in books quoted in the Further Read-
ing. In general, the enzymatic treatment has to be
optimized for each cell-separation process, because
the enzymes may not only attack cell membrane com-
ponents which keep the cells within the tissue but
may also destroy important cell-membrane functions.
If neither mechanical nor enzymatic methods lead
to satisfactory results, an alternative way may be to
incubate pieces of tissue on surfaces which challenge
the cells to move out of the tissue and to form mono-
layers surrounding the tissue. For example, cells of
human prostate tumour sections, which cannot be
dissociated in viable single cells by a number of mech-
anical and enzymatic techniques, migrate out of the
tissue and form a monolayer, when incubated in
culture dishes for a few weeks. After removal of the
tissue, the cells can easily be scraped off the plastic
dish surface.
Pre-Separation
As soon as single cells are available, countercurrent
centrifugal elutriation or cell electrophoresis may be
applied. However, some cell-separation tasks need
pre-enrichment of the cells of interest. Especially
where an investigator is interested in a peripheral
blood leukocyte population such as a lymphocyte,
granulocyte, monocyte or even reticulocyte popula-
tion, the erythrocytes comprising more than 99% of
the blood cells have to be removed before one of these
white blood cell populations may be puriRed. In these
instances, it has proved useful to perform a Rrst step
of density-gradient centrifugation, which does not
need pre-labelling of cells. The method allows the
separation of mononuclear leukocytes consisting
mainly of lymphocytes and monocytes from
granulocytes and erythrocytes. The separation prin-
ciple is based on different speciRc densities of the
various cell populations. In practice, a tube is Rlled
with a biocompatible isotonic medium with a speciRc
density adjusted between the speciRc densities of the
cells to be separated and the cell mixture is layered on
the top of this medium. Then the whole sample is
exposed to a few hundred g by centrifugation. The
forces cause mononuclear leukocytes with a density
lower than the separation medium to remain at the
top, while those with higher density sediment to the
bottom. The speciRc density of the medium is ad-
justed by silica colloids, which are coated with an
inert material and have low osmolality. Although
modern commercially available density-gradient sep-
aration media are very inert and direct damage of the
cells is seldom observed, the silica colloids are
pinocytosed by some cells.
If this is a problem, prolonged centrifugation of
whole blood may be an alternative route. During such
a centrifugation procedure, a layer of white blood
cells is formed above the erythrocytes. This layer,
called a buffy coat, contains mononuclear as well as
polymorphonucelated leukocytes and lies directly on
the erythrocytes. The white cells may be collected.
Although co-collection of a considerable number of
red cells is usually unavoidable, a degree of white
blood cell pre-enrichment can be achieved which
allows reasonable further separation by, e.g. CCE.
Countercurrent Centrifugal Elutriation
(CCE)
The method of CCE and the equipment required for
cell separation according to cell size have already
been described so they are summarized only brief
here. Cells loaded into the elutriation chamber are
subjected to centrifugal sedimentation forces gener-
ated by rotation in an outward direction and to
counterSow Suid forces pumped into the separation
chamber in an inward direction. As long as sedi-
mentation forces are balanced by the opposite Suid
forces, different cell populations take different
chamber positions depending mainly on their sizes
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS
Figure 1 Table-top elutriator (middle) together with an infusion
pump driving the counterflow (left) and a volume analyser (right).
and to a lesser extent on their speciRc densities. If the
counterSow rate is increased by speeding up a pump
and/or the sedimentation forces are decreased by re-
ducing centrifugation velocity, the various cell popu-
lations are washed out sequentially with increasing
size ranges.
Commercially, small elutriation chambers with
5 mL separation volume and large ones with 40 mL
are available together with suitable centrifuges and
rotors from Beckman Instruments (Palo Alto, USA).
They can accommodate up to 109 and 1010 cells,
respectively. A laboratory device has also been con-
structed; it consists of a table-top centrifuge with a
small rotor which has a separation chamber with a
volume of 0.5 mL to which 106 tissue cells or 2;107
mononuclear leukocytes may be loaded (Figure 1).
In order to fractionate cells with different sizes into
different fractions, counterSow rates and rotor speeds
have to be adjusted depending on rotor size, chamber
volume and the size of the cells to be separated. The
result of each separation should be controlled by
recording volume distributions of the cells of each
fraction with the help of cell size analysers. Beckman
Table 1 Examples of counterflow and rotor speed adjustments using a Beckman elutriator equipped with JE-6 rotor (small separation
chamber)
Cell mixtures: sheep erythrocytes/reticulocytes
Pre-enrichment: buffy coat
Cell size range: 28}42
m3
Rotor speed: 3000 rpm
Counterflow: 9}24 mL min\1
Desired cells: reticulocytes
Eluted at: 24 mL min\1
Use: analysis of volume regulation
For details, see: Lauf PK and Bauer J (1987) Biochemical and Biophysical Research Communications 144: 849}855 and Bauer J and
Hannig K (1984) Electrophoresis 5: 269}274.
4535
rotors are frequently operated at speeds ranging from
1000 to 4000 rpm. Dependent on the actual rotor
speed, the counterSow through small chambers may
be started at rates between 8 and 20 mL min\1 and
increased for fractional elution stepwise up to
100 mL min\1 (Table 1); counterSow rates through
large separation chambers may start at 50 mL min\1.
The table-top centrifuge is operated at counterSow
rates between 1 and 6 mL min\1, while the rotation
speeds are between 500 and 2200 rpm for tissue cell
separation and between 1500 and 2800 rpm for
leukocyte separation (Table 2). Using any of the in-
struments, separation times are short and the cells
may be kept suspended in culture medium. Thus
unwanted exposure of the cells to stimulatory envi-
ronments are minimized so that characteristics of
separated populations will rather closely reSect the
status of the original cells before fractionation.
Because of these advantages, CCE has proved most
useful, if applied for the separation tasks listed below:
E For cell cycle analyses, the cellular DNA content is
normally determined. However, cells have to be
killed in order to make their DNA accessible for
intercalating Suorescence molecules. If living cells
in different steps of the cell cycle need to be separ-
ated, an increase of cell size during passage through
the cell cycle may be used as a separation para-
meter. With the help of CCE the small cells, which
are in G1 phase can be separated from S-phase cells
which have intermediary size and from the G2/M-
phase cells which have the largest size of the cell
population. Many Sow cytometric analyses of the
DNA content of separated cells have already pro-
ved that CCE enables the separation of cells of cell
lines in fractions, which have up to 100% G1 phase
cells, up to 80% S-phase cells and up to 80%
G2/M phase cells, respectively.
E A number of different kinds of cells such as mono-
nuclear phagocytes recognize very non-speciRcally
human mononuclear leukocytes cultured human lymphocytes/
macrophages
density-gradient centrifugation none
180}400
m3 180}2000
m3
2460 rpm 2460 rpm
16.5}40 mL min\1 16.5}81 mL min\1
lymphocytes/monocytes macrophages
22 mL min\1/40 mL min\1 80 mL min\1
further enrichment, immunological surface charge analysis
tests
4536 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS
Table 2 Examples of counterflow and rotor speed adjustments using the self-made table-top elutriator
Cell mixture: cultured human mononuclear leukocytes
Pre-enrichment: none
Cell size range: 180}2000
m3
Rotor speed: 2800, 500 rpm
Counterflow: 2.5, 4, 6 mL min\1
Desired cells: antibody-producing cells
Eluted at: 6 mL min\1/500 rpm
Use: antibody secretion
For details, see: Bauer J and Hannig K (1988) Journal of immunological Methods 112: 213}218 and Bauer J, Grimm D, Hofstaedter
F and Wieland W (1992) Biotechnological Progress 8: 494}500.
foreign molecules and particles entering an organ-
ism. So despite many alternative methods such as
antibody-dependent sorting or panning, CCE,
which does not involve cell adhesion to matrices or
to antibodies, is often preferred, to separate mono-
cytes from peripheral blood or bone marrow and to
purify macrophages from alveolar tissues or Kup-
ffer cells from liver and to enrich mast cells, if
contacts to stimulatory surfaces and substances
must be avoided.
E Problems still exist in the detailed study of the
biological and physiological features of healthy
and malignant animal tissue cells and plant proto-
plasts. These cells have not yet been characterized,
as well as, for example, lymphoid cells. Antibodies
against the surface epitopes of such cells are not
isolated in great abundance, so fractionation of
single-cell populations, obtained from tissues of
various organisms, by CCE, is a competitive way to
provide important homogeneous cell populations
for biological, toxicological and pharmacological
studies.
E CD34-positive hematopoietic stem cells are very
helpful to restore hematopoiesis of patients, who
have to undergo whole-body radiation or rigorous
chemotherapy. In the past, CD34-positive cells
were separated either by panning, immunomag-
netic sorting or Suorescence-activated cell sorting.
All these techniques include expensive time-
consuming steps of labelling cells by antibodies and
generate problems of removing the antibodies/
ligands from the surface of the puriRed cells. CCE
thus appears to be an alternative method for
CD34-positive stem cell puriRcation as the stem
cells have a similar volume as mononuclear
leukocytes. However, resolution improvements
still seem to be necessary.
Free-Flow Cell Electrophoresis (FFE)
Another method for purifying cell populations with-
out antibody tagging or cell adherence is free-Sow
human erythrocytes/granulocytes cultured human tissue cells
buffy coat none
90}400
m3 1000}3000
m3
2800, 1500 rpm 2200}500 rpm
4}6 mL min\1 4}6 mL min\1
granulocytes hyperdiploid cells
6 mL min\1/1500 rpm 6 mL min\1/500 rpm
analysis analysis
electrophoresis. Its basic principle has already been
described and is repeated brieSy here. A laminar
buffer stream Sows between two narrowly spaced
parallel glass plates forming a separation chamber.
Near one end of the chamber, a cell suspension is
injected as a narrow band into the Suid Sow which
carries the cells through an electric Reld applied per-
pendicularly to the carrier Suid Sow. Cells exposed to
the electric Reld migrate laterally towards the posit-
ively charged electrode with velocities depending on
their negative surface charge densities. Thus cells
with different negative surface charge densities mi-
grate at different speeds, arrive at different points
along the opposite edge line and can be collected for
preparative isolation.
This principle is called free-Sow zone electrophor-
esis (FFZE) and is still the only electrophoresis mode
applicable to cell separation, although it has poorer
resolution than other electrophoresis modes such as
isoelectric focusing (IEF) and isotachophoresis (ITP),
because it is a non-focusing process. In addition, most
whole cells do not tolerate a Suid pH below 6.9 and
above 7.5 and need media which allow reasonable
electrophoretic mobilities, but are simultaneously
biocompatible. So for quite a long time, cell elec-
trophoresis was rarely applied, particularly as re-
solution was often not high enough to purify cell
populations with different mean electrophoretic
mobilities but overlapping distribution curves and
this second drawback negatively inSuenced cell vital-
ity. Cells had to be suspended in media lacking NaCl
or other physiologically important ions, because too
many ions in the chamber medium caused problems
of performance such as overheating of the medium
and short electromigration distances, as long as only
conventional devices with homogeneous chamber
media were available.
Recently, a new type of FFE was developed which
opened new possibilities of electrophoretic cell
separation. It is called Octopus and is commercially
available from the Dr. Weber GmbH, Kirchheim,
Germany (Figure 2). It is quite suitable to perform
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS
Figure 2 An Octopus free-flow electrophoresis apparatus with
the electrophoresis chamber in vertical position (left) and imple-
ments such as a power supply, a pump and the control unit (right).
(A generous gift of the Dr. Weber GmbH, D-85551 Kirchheim,
Germany. More information about the machine may be found at
http://members.aol.com/ffeweber/default.htm.)
preparative cell electrophoresis but can easily be ad-
justed to IEF and ITP of sub-cellular particles or
molecular substances. Its electrophoresis chamber has
a length of 500 mm and a width of 100 mm and can
be Rxed in a vertical or a horizontal position, as long
as specimen sedimentation does not play a role. The
thickness is variable, between 0.4 mm and 0.2 mm, so
that heat-removal efRciency may be enhanced, if ions
are required in cell suspension media and the applica-
tion of high electric Relds is necessary. An optical
particle detection system allows control of process
stability.
The major advantage of the new system is that
various media may Sow through the chamber adjac-
ent to each other and the sample may be introduced at
the optimal site (Figure 3). This means for cell elec-
trophoresis, that one central cell suspension medium,
which may contain up to 50 mmol L\1 NaCl is
pumped between two margin media with elevated
quantities of ions Sowing at both edges (Table 3).
They cover the electrode membranes, protect the sep-
aration medium from detrimental inSuences of the
electrodes, prevent diminution of Na# and Cl\ ion
concentrations within the central chamber area and
conduct the electric current to this area of cell trans-
port with minimal voltage drop.
Like CCE, FFE is most advantageous if antibodies
coupled to Suorescent dyes or magnetic beads are not
4537
available or must not be applied. So the method is
quite useful, when cells are separated by CCE because
of the reasons explained above and the resulting frac-
tions still contain cells which belong to different
populations, but have equal size, while their elec-
trophoretic mobilities are different. For example, cell
fractions are routinely obtained, which contain more
than 90% monocytes, if pre-enriched mononuclear
leukocytes are elutriated. In such fractions, up
to 0.2% antibody-producing cells with equal size
as monocytes but different electrophoretic mobilities
(EPM) are often co-collected. The antibody-produ-
cing cells can be further enriched by FFE. Similarly,
T-cell fractions obtained by CCE contaminated by
accessory cells, of equal size have been submitted to
a following step of FFE puriRcation. T-cells of indi-
vidual blood donors were obtained, which did not
respond to concanavalin A unless accessory cells were
re-added.
A cell feature, which cannot be deRned by antibod-
ies is the negative surface charge density. Its biolo-
gical role is still very poorly understood. Observa-
tions made during recent cell electrophoretic studies
appear currently like very scattered mosaic stones
which do not allow the whole picture to be revealed.
For example, erythrocytes change their EPM in pa-
tients suffering various kinds of diseases, monocytes
change their EPM when maturing to non-activated
macrophages, B-cells change their EPM when devel-
oping to antibody-producing cells in vivo but not in
vitro, and mice with different erythrocyte EPM have
different sensitivities to malaria infection (see Further
Reading). These accumulating data suggest that fur-
ther efforts in studying the biological relevance of
the negative surface charge density by FFE will be
worthwhile.
Since electrophoresis media with 20}50 mmol L\1
NaCl can be used for cell separation, tissue cells can
be processed without clotting. Now it is possible to
electrophorese cell suspensions obtained from tissues
directly or indirectly after a few passages of culture.
The separations performed so far have revealed quite
interesting new tissue cell sub-populations. Hence,
future application of FFE to fractionation of viable
tissue cells appears promising.
Conclusions
Essential guides for separation/puriRcation of cells
have been described in this chapter following a deRni-
tion of cell separations as processes of any scale by
which cell populations are separated from each other
without biological modiRcation, i.e. without changes
of their actual states of activation and differentiation.
As explained above and shown in Figure 4 single cells
4538 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF CELLS

Figure 3 Scheme of free flow electrophoresis chambers working with homogeneous (left) and segmented (right) carrier fluids.

suspended in suitable media after preparation from
human or animal body Suids, from human, animal or
plant tissues or from in vitro cultures are prerequisites
of such processes. If a cell suspension with a reason-
able number of desired cells is available, methods
such as countercurrent centrifugal elutriation and
free-Sow electrophoresis may be applied, either each
of them alone or in combination. As both cell-separ-
ation methods are rapid and work while cells are kept
in suspensions with minimal contact with foreign
surfaces but do not require labelling of cell surfaces
by antibodies or other macromolecules, a fair chance
can be expected to obtain homogeneous cell popula-
tions retaining their original states of activation and

Table 3 Examples of buffer systems for homogeneous and segmented FFE chamber fluids

Cell-suspension medium Margin buffers Electrode buffer(s)

Homogeneous
27 mmol L\1 triethanolamine 342 mmol L\1 triethanolamine
4 mmol L\1 potassium acetate 40 mmol L\1 potassium acetate
27 mmol L\1 sucrose pH 7.2 adjusted by acetic acid
1 mmol L\1 glucose
216 mmol L\1 glycine
pH 7.2 adjusted by acetic acid

Segmented
central: 10 mmol L\1 triethanolamine anodal: 50 mmol L\1 triethanolamine anodal: 200 mmol L\1 sodium acetate
2 mmol L\1 sodium acetate 250 mmol L\1 Na2SO3
50 mmol L\1 NaCl pH 7.2 adjusted by acetic acid
2 mmol L\1 glucose
180 mmol L\1 sucrose cathodal: 50 mmol L\1 triethanolamine cathodal: 100 mmol L\1 HCl
pH 7.2 adjusted by acetic acid 250 mmol L\1 NaCl 100 mmol L\1 NaCl
75 mmol L\1 sucrose 200 mmol L\1 imidazole
pH 7.2 adjusted by acetic acid
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
Figure 4 Flow diagram showing a survey of the processes of
cell purification described in this article.
differentiation, even if appropriate antibodies are not
available.
Cells puriRed without biological modiRcations may
be especially useful if it is of interest to study their
original in vivo status or to use them for transplanta-
tion purposes and if size or surface charge-related
phenomena are to be investigated. As knowledge of
possible cellular characteristics and components is
continuously accumulating, questions on their actual
ESSENTIAL GUIDES FOR ISOLATION/
PURIFICATION OF DRUG METABOLITES
I. P. Nnane and A. J. Hutt, Kings College London, UK
L. A. Damani, Chinese University of Hong Kong,
Hong Kong
This article is reproduced from Encyclopedia of Analyti-
cal Science, Copyright ^ 1995 Academic Press
Metabolite Isolation and Identi\cation
Following the administration of drugs to either ani-
mals or man, very few of the drugs are excreted
4539
expression under normal and pathological conditions
will frequently arise. For studying such questions,
homogeneous cell populations retaining their original
in vivo status may become so important that tech-
niques and instruments required for their puriRcation
will be further improved.
See also: Cells and Cell Organelles: Field Flow Frac-
tionation.
Further Reading
Bauer J (1987) Electrophoretic separation of cells. Journal
of Chromatography 418: 359}383.
Bauer J (1994) Cell Electrophoresis. Boca Raton: CRC
Press.
Bauer J (1998) Advances in cell separation: recent develop-
ments in counterSow centrifugal elutriation and con-
tinuous Sow cell separation. Carrier free electrophor-
esis. Electrophoresis 19: Special issue.
Bauer J (1999) Journal of Chromatography 722: 55}69.
Coleman R, Wilton JC, Stone V and Chipman JK (1995)
General Pharmacology 26: 1445}1453.
Dixon RA and Gonzales RA (1994) Plant Cell Culture:
A Practical Approach. Oxford: Oxford University
Press.
Merrill GF (1998) In: Mather JP and Barnes D (eds)
Methods in Cell Biology, vol. 57, pp. 229}249. San
Diego: Academic Press.
Pretlow TG and Pretlow TP (1982) Cell Separation:
Methods and Applications. New York: Academic Press.
Shapiro HM (1995) Practical Flow Cytometry, 3rd edn.
New York: Wiley-Liss.
Specto DL, Goldmann RD and Leinwand LA (1998) Cells.
A Laboratory Manual, vol 1: Culture and Biochemical
Analysis of Cells. New York: Cold Spring Harbor
Laboratory Press.
unchanged. The majority undergo biotransforma-
tions by interaction with a complex series of enzymes.
This process, known as drug metabolism, is not re-
stricted to drugs but occurs with all chemicals that are
taken in by living systems, including food additives,
pesticides, carcinogens, etc. These chemicals are
termed exogenous compounds, as opposed to endo-
genous, or naturally present, compounds.
Metabolic studies have made, and continue
to make, fundamental contributions to the drug
4540 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES

discovery process and also to the elucidation of mech-

anisms of both drug action and toxicity. During the
early stages of drug development, an evaluation of the
metabolic dispositional proRle of a compound may
yield valuable information and signiRcantly contrib-
ute to the drug candidate selection procedure. In
addition, drug metabolism has a central role in the
safety evaluation of novel drug substances, and the
regulatory authority guidelines for toxicity testing all
make reference to metabolic and pharmacokinetic
data.
The reactions of drug metabolism may be divided
into two groups, the phase I or functionalization
reactions and the phase II or conjugation reactions.
The phase I reactions involve either the introduction
or unmasking of a functional group, e.g. hydroxyl,
carboxyl or amino group, within a molecule by the
processes of oxidation, reduction or hydrolysis. The
groups introduced generally result in an increase in Figure 1 Relationships between phase I and phase II meta-
the polarity, and therefore the aqueous solubility, bolic transformations.
of the metabolite compared with the parent com-
pound. Depending on the reaction type, the change
in physicochemical properties may be relatively ly complex matrix containing numerous potentially
minor, e.g. dealkylation of a tertiary to a secondary interfering endogenous materials. The isolation and
amine, or substantial, e.g. hydrolysis of an ester or characterization of metabolic products therefore
amide. requires a range of analytical methodologies, pri-
The phase II reactions are biosynthetic and involve marily in the areas of separation and spectroscopic
the addition of an endogenous molecule to the drug, techniques.
or a phase I metabolite of the drug, by reaction with
a suitable functional group, e.g. carboxyl, hydroxyl,
amino, etc. The products of these reactions are
Sample Types
generally polar, hydrophilic molecules that are The range of techniques used to investigate the meta-
ionized under physiological conditions, and hence bolism of drugs is relatively wide (Table 1) and the
the excretion of the foreign compound into urine bioanalyst may therefore be presented with samples
or bile is facilitated. Examples of reaction types in- which vary markedly in terms of both their nature
clude conjugation with glucuronic acid, sulfate, and origin. In in vivo metabolic studies, following the
glutathione and amino acids, all of which result in an administration of a drug to either animals or humans,
increase in the polarity of the product compared to the major sample types examined include blood,
the drug. Some conjugation reactions, namely plasma, urine, faeces and less commonly bile and
methylation and acetylation, may result in an in- milk. In in vitro methodology, sample types may
crease in the lipid solubility of the metabolite com-
pared that of the drug; however, this depends very
much on the nature of the substrate. The two phases Table 1 Biological techniques used in drug metabolism
of drug metabolism are intimately linked, as shown in Methodology Example
Figure 1.
As a result of the metabolic transformations out- Administration of the drug Human or standard laboratory
lined here, and analytical sample of biological origin to whole animals species, e.g. rat, dog, etc.
may contain several substances which vary markedly Isolated perfused organs Liver, kidney, lung, intestine
Tissue slices Liver, kidney
in their physicochemical properties. For example Isolated cells Hepatocytes, renal cells, lung
metabolic products may be acidic, basic, neutral or cells, enterocytes, blood cells
zwitterionic and relatively hydrophilic or hydropho- Subcellular fractions Whole tissue homogenates,
bic. The examination of such samples therefore pres- postmitochondrial supernatant,
ents the bioanalyst with a considerable challenge as microsomal fractions, cytosol
Purified enzymes Cytochrome P-450 and flavin
the sample will contain relatively small quantities of monooxygenases
structurally related materials dispersed in an extreme-
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
range from relatively clean perfusion Suids to com-
plex tissue homogenates. Thus the bioanalyst may be
presented with liquid, semisolid and solid samples for
evaluation, each of which presents different
problems. Plasma, for example, contains relatively
high concentrations of proteins which may interfere
with the chromatographic separation of metabolites
or damage chromatographic stationary phases. The
samples therefore require deproteination prior to
analysis. Samples which are solids, or semisolids,
may affect the separation characteristics of solid-
phase extraction cartridges and it is frequently the
case that such samples are homogenized prior to
analysis.
Preliminary Sample Pretreatment
Because of the nature of the sample types and the
potential range of physiochemical properties of the
analytes, the samples encountered in metabolic stud-
ies generally require extensive pretreatment prior to
instrumental analysis. It is essential that any manip-
ulations carried out on the sample do not result in
ex vivo changes to the analytes. A general approach
to sample treatment is presented in Figure 2. Prelimi-
nary sample preparation plays an important role in
the speciRcity of an analytical procedure. The initial
step involves sample clean-up to remove potentially
interfering substances, fractionation of the metabolic
products according to their physicochemical proper-
ties, concentration of the sample for analysis and
possible hydrolysis of conjugated metabolites. Hav-
ing obtained a primary extract, the analytes are fur-
ther puriRed, generally by a chromatographic proced-
ure, prior to characterization by conventional spec-
troscopic techniques.
Figure 2 General approach for the isolation and characteriza-
tion of metabolic products.
4541
Hydrolysis of Conjugates
One of the most frequently encountered phase II
pathways is conjugation with glucuronic acid. Sev-
eral functionalities undergo this reaction, yielding
a variety of bond types between the sugar moiety and
the aglycone, e.g. carboxyl groups yield acyl
glucuronides, phenolic and alcoholic hydroxyl
groups yield ether linkages, thiols and amino func-
tions yield S- and N-glucuronides respectively, and
examples of carbon glucuronides are also known.
The stability of these various linkages varies consider-
ably, as does their susceptibility to hydrolytic treat-
ments, the common methods of hydrolysis being
mild alkali, acid or treatment with the enzyme

-glucuronidase.
A number of problems may occur during sample
pretreatment; acyl glucuronides for example are rela-
tively labile under mild alkaline conditions and may
undergo hydrolysis in samples which are not stored
with care. An additional problem with compounds of
this type is that they also undergo facile intramolecu-
lar rearrangement at mild alkaline pH giving rise to
mixtures of the corresponding 2-, 3- and 4-O-acyl
esters of glucuronic acid. Such glucuronic acid esters
are resistant to hydrolysis by
-glucuronidase but
may be hydrolysed by treatment with mild alkali.
Thus the amount of aglycone liberated by treatment
of the conjugate with the enzyme may be lower than
that found following treatment with alkali. Ethereal
glucuronides are stable to treatment with mild alkali,
but may be hydrolysed with acid or
-glucuronidase.
The stability of both N- and S-glucuronides to either
chemical or enzymatic treatment is highly dependent
on the nature of the aglycone and the bond type.
Carbon-linked glucuronide conjugates are resistant to

-glucuronidase.
The liberation of an aglycone upon incubation
of a conjugate with
-glucuronidase may only be
taken as presumptive evidence that the conjugate
is a glucuronide if adequate controls have been
carried out, e.g. inhibition of hydrolysis by the speci-
Rc
-glucuronidase inhibitor, saccharo-1,4-lactone,
identiRcation of the carbohydrate moiety by
chromatography and detection with naphthoresor-
cinol.
Sulfation is a relatively common conjugation reac-
tion for phenolic hydroxyls, alcohols and some amino
compounds. Sulfate conjugates may be hydrolysed by
aryl sulfatases. However, the commercially available
preparations may be contaminated with
-
glucuronidase, which should be inhibited by the addi-
tion of saccharo 1,4-lactone. Acid hydrolysis of
solvolysis may also be employed but the reactivity
of the conjugates may vary considerably.
4542 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES

Figure 3 General scheme for the fractionation of drug metabolites using solvent extraction.

In contrast, amino acid conjugates of carboxylic presented in Table 2. Thus a particular drug or its
acids are relatively stable and may be isolated and metabolites can be selectively isolated from a biolo-
characterized by conventional methodology. gical matrix by a consideration of the physicochemi-
cal properties of the material, careful choice of sol-
vent and adjustment of the pH of the aqueous me-
Methods of Isolation + Extraction dium. Ideally the solvent should completely extract
Techniques the drug and its metabolites in a single extraction
while keeping the amount of coextracted endogenous
Extraction techniques may be used for the prelimi-
compounds to a minimum. However, the efRci-
nary puriRcation and fractionation of metabolic
ency of the extraction procedure must be investi-
products from the biological matrix. Further puriRca-
gated and a double or triple extraction coupled with a
tion of the individual metabolites may be achieved
salting out procedure may be necessary.
using chromatographic techniques.
Solvents for extraction should be of high purity
grade and it is frequently important that they are
Liquid+Liquid Extraction
distilled prior to use. This distillation will ensure that
As pointed out previously, drug metabolites vary the solvents are free of trace quantities of solutes
greatly in terms of their physicochemical properties. which on sample concentration may be present in
Selective isolation of material may be achieved by
extraction from the aqueous-based biological sam-
Table 2 Examples of commonly used solvents for extraction of
ples, after appropriate adjustment of pH, using an biological fluids
immiscible solvent. The choice of solvent is critical
and it may be possible to fractionate the analytes by Solvent Dielectric Boiling point
sequential extraction using solvents of different constant (3C)
polarities, with or without adjustment of sample pH.
n-Hexane 1.9 68.7
Having obtained an organic extract of the com- n-Heptane 2.0 98.4
pounds of interest the sample may be cleaned up Carbon tetrachloride 2.2 76.8
further by back-extraction into an aqueous phase Toluene 2.4 111.0
with appropriate pH adjustment. A general scheme Diethyl ether 4.3 34.5
for the extraction of a drug and metabolites is pre- Chloroform 4.8 61.2
Ethyl acetate 6.0 77.1
sented in Figure 3 and examples of commonly used Dichloromethane 9.1 40.2
solvents. Used either alone or in combination, are
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
concentrations greater than that of the analytes and
may therefore interfere with the subsequent analysis.
Phthalates for example are universal contaminants
and are frequently observed in the mass spectra of
metabolic products. Solvents such as diethyl ether
frequently contain aldehydic impurities which may
react with the analyte, e.g. (!)-ephedrine reacts with
the acetaldehyde, propionaldehyde or formaldehyde
present in ether to yield a series of oxazolidine deriva-
tives during isolation. Chlorinated solvents should be
used with caution in the presence of basic com-
pounds; several analgesic agents, e.g. dextromethor-
phan, pethidine and methadone, have been shown to
undergo alkylation during the concentration of di-
chloromethane extracts of the drugs. Traces of per-
oxides rapidly form in diethyl ether and may oxidize
drugs, or their metabolites. The products of such
reactions may then be erroneously identiRed as
metabolites. Additional problems may also arise dur-
ing concentration of extracts because of the degrada-
tion of thermolabile compounds and the loss of vol-
atile compounds. Evaporation under reduced pres-
sure and/or freeze-drying are useful alternative
approaches.
Ion Pair Extraction
Highly polar metabolites cannot usually be extracted
efRciently using solvents. In this case the metab-
olites, in their ionized state, can be paired with
a counterion of opposite charge. Isolation of several
biogenic amines, together with some of their metab-
olites, has been achieved by this approach. Cat-
echolamines and their derivatives, for example, have
been extracted from biological samples using di(2-
ethylhexyl)phosphoric acid as a counterion.
Liquid+Solid Extraction
The isolation of drugs and their metabolites by ad-
sorption methods offers an alternative approach
to the traditional solvent extraction. Liquid}solid ex-
traction, also known as solid-phase extraction (SPE)
employs a wide range of materials including coated
charcoal, silica, alumina and ion exchange resins. The
biological Suid is passed through a column packed
with the adsorbent and the materials of interest are
separated from the components of the matrix by
elution with an appropriate solvent. The success of
the technique depends on the afRnity of the
analytes for the adsorbent and the strength of the
eluting solvent.
In recent years the development of chemically
bonded silica stationary phases for liquid chromatog-
raphy (LC) has resulted in cartridge forms of these
materials for liquid}solid extraction and SPE. The use
4543
of these packings for extraction is based on the prin-
ciples of LC; the packing materials however are of
larger particle size (50
m) than those used in LC.
Thus analytes with a greater afRnity for the sta-
tionary phase are retained and highly polar endogen-
ous components of the matrix may be eluted with
polar solvents. The retained materials may then be
selectively eluted from the packing depending on their
physicochemical properties, the nature of the adsor-
bent and the elution solvent. The range of phases for
SPE is fairly extensive and includes C2, C8 and
C18 alkyl, phenyl, cyclohexyl, amino, diol and cyano,
in addition to a variety of ion exchange phases. This
range allows considerable versatility in terms of both
selectivity and speciRcity for analyte isolation. SPE is
superior in many respects to solvent extraction; it is
highly reproducible, efRcient and easier to auto-
mate, it generates less waste solvent and the only
major drawback is the relatively high cost of the
cartridges.
Methods of Isolation +
Chromatographic Techniques
Chromatographic techniques are extensively used in
bioanalysis for the separation, isolation and puriRca-
tion of drugs and their metabolites from biological
Suids. Such techniques can provide useful preliminary
information concerning the physicochemical proper-
ties of the metabolites in relation to those of the drug.
Although they give little information concerning
speciRc chemical structures, comparison of the
chromatographic properties of an analyte with those
of an authentic reference compound may provide
sufRcient information to establish the identity of
a particular metabolic product.
Thin-Layer Chromatography
Thin-layer chromatography (TLC) is relatively cheap,
easy to use, rapid and robust. These features account
for the widespread use of the technique in metabolic
studies, particularly for the isolation and puriRcation
of analytes prior to their characterization by spectro-
scopic techniques. A wide variety of stationary phases
is available; these include silica gel, alumina and
a number of bonded hydrocarbon phases (e.g. C2, C8,
C18) for reversed-phase and ion exchange separations.
Such phases are coated onto plastic, aluminium foil
or glass supports. A recent innovation in TLC station-
ary phase technology involves the introduction of
high performance thin-layer chromatographic
(HPTLC) plates which are coated with a layer
(200
m) of 5-
m particle size silica which offers
improved performance in terms of resolution and
4544 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
speed of chromatographic development. A number of
chiral TLC phases have also been introduced. How-
ever the application of these phases in bioanalytical
studies has been limited.
The chromatographic phases can be prepared with
Suorescent indicators which facilitate the detection of
ultraviolet absorbing analytes. The visualization of
analytes may also be achieved by the use of a wide
range of chromogenic spray reagents. The colour re-
actions observed with appropriate reagents may be of
assistance in the determination of class of metabolite,
e.g. glycine conjugates yield characteristic red-orange
colours on treatment with p-dimethylaminobenzal-
dehyde, naphthoresorcinol is used for the detection of
glucuronides, potassium dichromate}silver nitrate for
sulphur(II) and ninhydrin for glutathione conjugates.
The main application of TLC in metabolic studies
is in the isolation of metabolites, for subsequent iden-
tiRcation by spectroscopic techniques. However, it
can also be used quantitatively if radiolabelled
drugs are used, or by the application of scanning
densitometry.
Gas Chromatography
Gas chromatography (GC) is the technique of choice
for the separation and determination of volatile,
thermally stable and relatively low relative-molecu-
lar}mass drugs and metabolites. GC separation may
be carried out using either packed or capillary
columns. GC columns are usually made of glass,
stainless-steel, copper, aluminium or PTFE (polytet-
raSuoroethylene). However in metabolic studies glass
columns are preferred to minimize the potential ther-
mal breakdown of metabolites during analysis. For
example, the decomposition of primary and second-
ary hydroxylamines, formed on metabolic N-oxida-
tion of the corresponding amines, takes place readily
on the heated surfaces of metal columns.
A wide range of stationary phases is available for
GC separation and a variety of solid support mater-
ials for packed columns is encountered. The amount
of loading of the stationary phase on the support may
also vary considerably, thus the number of possible
combinations of phase and support are essentially
unlimited. Liquid phases based on polymers of
poly(ethylene glycol) and dimethylsilicone have been
widely used in bioanalysis. The poly(ethylene glycols)
are polar and Carbowax 20M is frequently used, both
with and without potassium hydroxide, as a station-
ary phase. Carbowax 20M is a particularly useful
phase for analysis of low relative-molecular-mass
amines and their metabolites, e.g. amphetamine and
related compounds. Problems arise with compounds
of greater molecular mass because of the maximum
operating temperature of 2003C for Carbowax 20M.
In contrast the silicon phases may be used up to c.
3503C, and common phases encountered in meta-
bolic studies include OV1 and OV101 polymers and
the more polar phenylmethylsilicones OV17 and
OV25. The most commonly used support materials in
bioanalysis include Chromosorb G and W, Gas
Chrom Q, Haloport F and Chromosorb 750. These
support materials are not entirely inert and are fre-
quently washed with acid or silanized prior to being
coated with the stationary phase. Glass beads may
also be used as support material for GC because of
their inert nature and they may also be coated with
low loadings of stationary phase. N-Hydroxyam-
phetamine, a thermolabile metabolite of ampheta-
mine, was Rrst chromatographed successfully without
prior derivatization using a column containing Car-
bowax 20M (0.2%) coated onto a glass bead support.
Several GC detector systems have been described.
However only four types are commonly used in bio-
analysis, the Same ionization detector, the electron
capture detector, the nitrogen}phosphorus detector,
and the mass spectrometer. Of these the mass spec-
trometer is the most speciRc and can provide informa-
tion on the structural features of compounds. The
Same ionization detector is the most widely used in
metabolic studies, and sample quantities as low as
1 ng can be determined depending on detector design.
Electron capture detectors respond to halogenated
compounds, and compounds containing nitro groups
or conjugated carbonyls, etc., and thus the suitability
of this detector for metabolic studies depends on the
structure of the analyte. This selectivity, together
with sensitivity (detection of 1 pg of material is poss-
ible) increases the utility of this system. Derivatiz-
ation of compounds that do not contain halogens
with appropriate reagents frequently yields volatile
products which are ideal for analysis by GC using this
detector system, e.g. the analysis of debrisoquine and
its 4-hydroxy metabolite following derivatization
with hexaSuoroacetylacetone to yield the corre-
sponding bis(triSuoromethyl)pyrimidine derivatives.
Nitrogen}phosphorus detectors are extremely sensi-
tive and are 20 000}40 000 times more sensitive to
nitrogen than to carbon. In bioanalysis such detectors
are useful particularly for heterocyclic compounds,
which are less likely to lose nitrogen via metabolic
deamination than are acyclic compounds.
The products of metabolism are generally more
polar than the drug, have less volatility, long GC
retention times and produce tailing peaks. Thus
metabolites are frequently derivatized prior to analy-
sis to increase volatility, modify chromatographic
properties and increase detector response. The most
common techniques are: silylation, the replacement
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
of active hydrogen by trimethylsilyl groups (e.g. OH,
SH, NH2); alkylation using diazomethane, dimethyl-
formamide, dialkyl acetals, or boron triSuoride and
an alcohol; and acylation using perSuoroacyl re-
agents to yield triSuoroacetyl, heptaSuorobutyryl,
etc. derivatives.
Liquid Chromatography
LC is an extremely versatile technique and has a num-
ber of advantages over GC in terms of bioanalysis.
For example, highly polar compounds which are dif-
Rcult to extract from aqueous solutions, e.g.
glucuronide and sulfate conjugates and quaternary
ammonium derivatives, may be analysed directly
without the necessity for extraction; the technique is
normally carried out at room temperature and there-
fore thermolability of analytes is not a major consid-
eration; and the technique is nondestructive so that
the eluent containing the analytes may be collected
and used for additional ofSine characterization.
A variety of different stationary phases is
available for LC, but the reversed-phase packings
(e.g. C2, C8, C18 , phenyl) are the most commonly used
in metabolic studies. The mobile phases utilized with
such columns are based on aqueous solvents contain-
ing variable quantities of organic modiRers. Such
systems allow analyte sample preparation to be sim-
pliRed, reducing the preanalysis manipulation steps.
Provided the chromatographic system has sufR-
cient resolving power, it may be possible to inject
directly either a dilute sample or a plasma sample
following precipitation of plasma proteins with an
organic solvent compatible with the LC mobile phase.
This approach has the obvious advantages of reduc-
tion of tedious sample manipulation steps and also in
the reduction of human exposure to samples of clini-
cal origin. A disadvantage of the direct injection ap-
proach is that protein precipitation may occur on
injection of plasma samples into the chromatograph.
However, the use of precolumns can protect the ana-
lytical column and the instrument.
The most frequently used detection systems in LC
analysis are either Rxed- or variable-wavelength
ultraviolet (UV) detectors, Suorescence detectors or
electrochemical detectors. While analyte derivatiz-
ation is not as commonly used with LC, as with GC,
derivatives may be used to enhance the detector sensi-
tivity, particularly if Suorescence is used. Multiple-
wavelength UV detectors and the diode array detector
are particularly useful in metabolic studies. The use of
such detectors in analysis provides a three-dimen-
sional chromatographic retention and the entire spec-
trum of a sample may be determined in a single
chromatographic run. The main applications of these
4545
detector systems in bioanalysis are to ensure
chromatographic peak purity and to provide initial
spectroscopic data for metabolite identiRcation. If the
compound under investigation is radiolabelled then
a suitable LC radiodetector may also be used to
facilitate the detection of metabolites.
Capillary Electrophoresis
A technique that is likely to make an impact on
bioanalysis in the near future is capillary electrophor-
esis (CE). This technique, and its variants, offers
a number of advantages over conventional chromato-
graphic techniques, e.g. high column efRciencies
and short analysis times, particularly where samples
are complex mixtures. At present CE instruments are
limited in terms of sample size, but this may be an
advantage when working with biological Suids as it
minimizes potential contaminants. An additional ad-
vantage of the technique is that by manipulation of
the analytical conditions the nature of the compo-
nents entering the capillary may be controlled and
interference effects reduced. A variety of detector
systems has been described for CE, including Suores-
cence, electrochemical and mass spectrometry sys-
tems, but the majority of commercially available in-
struments incorporate a sensitive UV detector system.
Bioanalytical applications of CE have been described
for the determination of cefpiramide and anticancer
agents in human plasma.
Methods of Identi\cation
The elucidation of metabolite structure is dependent
on the use of spectroscopic techniques, either directly
linked to chromatographic systems (the so-called
hyphenated: techniques, GC-MS, LC-MS), or used
ofSine following the isolation and puriRcation of
analytes. The main techniques used in bioanalysis are
ultraviolet (UV), infrared (IR) and nuclear magnetic
resonance (NMR) spectroscopy and mass spectro-
metry (MS). However, the technique or combination
of techniques Rnally adopted will be dependent on the
complexity of the problem and the amount of pure
isolated material available.
Ultraviolet Spectroscopy
This technique is widely used for the quantitative
analysis of drugs and metabolites in biological sam-
ples. However, the amount of structural information
which can be obtained from a UV spectrum of
a metabolite is limited. If the site of metabolism in
a molecule is at, or adjacent to, a chromophore then
the UV spectrum is likely to yield useful structural
information, e.g. oxidation of an aromatic ring to
4546 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF DRUG METABOLITES
yield a phenol, the spectrum of which can be in-
Suenced by alteration of pH, reduction of an aro-
matic nitro group to yield an amine, and reactions
which result in the introduction of conjugated double
bonds, e.g. aromatization.
Infrared Spectroscopy
IR spectra are highly diagnostic and the examination
of an IR spectrum provides a simple, rapid and often
reliable method of assigning metabolite structure. Un-
til relatively recently the sensitivity of IR spectrom-
eters restricted their application in metabolic studies
but the development of Fourier transform infrared
(FTIR) and the application of these instruments as
detector systems for both GC and supercritical Suid
chromatography has increased the potential of the
technique in bioanalysis signiRcantly.
Nuclear Magnetic Resonance Spectroscopy
NMR spectroscopy has been routinely used in meta-
bolic studies as a means of structure elucidation for
a number of years. The major limitation of the tech-
nique has been sensitivity. However instrumental de-
velopments, with improvements in resolution, ana-
lytical power and sensitivity, have changed the way
the technique is used in bioanalysis.
There are now a number of reports of direct NMR
examination of biological samples with minimal or
no preliminary sample clean-up processes and biolo-
gical Suids have been placed directly in the NMR
sample tube. It has also been possible, using NMR
techniques, to examine metabolism and distribution
in cells and organs.
Proton NMR is the most widely used technique in
drug metabolism because of its high sensitivity and
the large number of observable protons in most drugs
and their metabolites. As bioanalytical samples are
initially obtained in aqueous solution the intense
water signal present must be either eliminated, sup-
pressed or edited out of the spectrum. Signals from
endogenous materials may also obscure signals of
interest because of overlap resulting from the narrow
chemical shift range in proton NMR. Direct proton
NMR has been used to examine the urinary disposi-
tion of paracetamol following the oral administration
of the drug to humans. Using this approach it was
possible to examine the urinary metabolite proRle
following both therapeutic doses and overdoses of the
drug.
Other nuclei of interest in bioanalysis include 19F,
31
P, 15N and 13C, although sensitivity may be a prob-
lem with some of these nuclei (e.g. 15N). An approach
which may yield useful information with these nuclei
is to use compounds appropriately labelled with
stable isotopes. This approach may be particularly
useful if the label is introduced at a site which under-
goes transformation. For example the fate of [car-
boxyl-13C]phenylacetic acid has been examined fol-
lowing administration to a horse. This compound
undergoes both amino acid and glucuronic acid con-
jugation and the major metabolites could readily be
distinguished following direct NMR examination of
urine samples and observation of the 13C-carbonyl
resonances. NMR has also been used directly linked
with HPLC for metabolite work.
Mass Spectrometry
As a result of its extreme sensitivity and ability to
provide diagnostic information, MS is the standard
technique for the identiRcation of metabolic prod-
ucts. All the major MS techniques have been utilized
in metabolic investigations and thus it is relatively
easy to Rnd publications detailing applications of
electron impact, chemical ionization, Reld desorption
and fast atom bombardment in drug metabolism. The
major advantage of MS is the ability to use the tech-
nique in combination with either GC or LC. Such
hyphenated techniques, particularly GC-MS, have
been extensively used in metabolic studies and the
development of thermospray ionization for LC-MS
has enabled spectra of nonvolatile hydrophilic
analytes, e.g. metabolic conjugates, to be determined
directly. As a result of the variety of ionization modes
available, and the ability to link the technique to GC
or LC, there are few bioanalytical problems to which
MS cannot make a valuable contribution.
There are essentially three main applications of
MS, particularly the hyphenated techniques, in meta-
bolic studies.
1. The characterization and structure elucidation of
metabolic products.
2. Quantitative analysis using the mass spectrometer
as a sensitive chromatographic detector for se-
lected single or multiple ion monitoring, using for
example compounds labelled with stable isotopes
as internal standards. An alternative approach in-
volves the administration of a labelled compound
to a patient who is at steady state, and use of the
nonlabelled material to examine the phar-
macokinetics of what is effectively a single
drug dose under these conditions.
3. Mechanistic investigations, e.g. the source of oxy-
gen in a metabolite may be determined by carrying
out appropriate experiments using 18O2 or H218O;
the use of compounds labelled with stable iso-
topes, e.g. replacement of hydrogen with
deuterium to determine kinetic isotope effects
on metabolism.
APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS
MS has made a number of important contributions
to drug metabolism and the further development of
the technique, together with advances in instrumenta-
tion, will enhance its application in this area.
See also: II/Chromatography: Gas: Column Techno-
logy; Detectors: Mass Spectrometry. Chromatography:
Liquid: Detectors: Ultraviolet and visible Detection; Nu-
clear Magnetic Resonance Detectors. Chromatography:
Thin-Layer (planar): Layers. Electrophoresis: Capillary
Electrophoresis; Capillary Electrophoresis Detection;
Capillary Electrophoresis-Mass Spectrometry; Capil-
lary Electrophoresis-Nuclear Magnetic Resonance.
Extraction: Analytical Extractions; Solid-Phase Extrac-
tion; Solvent Based Separation; III/Drugs of Abuse:
Solid-Phase Extraction. Drugs and Metabolites: Liquid
ESSENTIAL GUIDES FOR ISOLATION/
PURIFICATION OF ENZYMES AND PROTEINS
S. Doonan, University of East London, UK
Copyright ^ 2000 Academic Press
Nature of the Problem
The puriRcation of proteins presents a unique chal-
lenge in the Reld of separation science. Typically, the
particular protein to be isolated will constitute 1% or
less (sometimes much less) of the material in the
original extract and all of the contaminants will have
basically the same chemical characteristics, i.e. they
are all proteins. There is the added complication that
for most purposes it is necessary to retain the bio-
logical activity of the protein, and the inherent insta-
bility of protein structures restricts the range of tem-
peratures and solvent compositions that can be used
during puriRcation.
Tools for its Solution
Clearly, methods for the separation of proteins must
be based on those characteristics in which they dif-
fer from one another. The most important of these are
listed in Table 1 along with the separation techniques
that exploit those differences. These various
properties are not of equal generality or of equal
utility for puriRcation purposes.
By far the most widely used technique for protein
isolation is ion exchange chromatography. The gener-
ality of the method arises from the fact that proteins
contain ionizable amino acids and hence carry a net
4547
Chromatography-Mass Spectrometry; Liquid Chro-
matography-Nuclear Magnetic Resonance-Mass Spectro-
metry.
Further Reading
Gibson GG, ed. (1993) Progress in Drug Metabolism, vol.
13. London: Taylor & Francis.
Moffat AC, Jackson JV, Moss MS and Windopp B, eds
(1986) Clarkes Isolation and identiTcation of Drugs in
Pharmaceuticals, Body Fluids and Post-Mortem Mater-
ial. London: The Pharmaceutical Press.
Reid E and Wilson ID (1992) Methodological Surveys in
Biochemistry and Analysis, vol. 22. London: Royal So-
ciety of Chemistry.
charge at all pH values except the unique pH (the
isoelectric point) at which the positive and negative
charges are equal. Moreover, two proteins that have
the same charge at a particular pH are likely to
differ in charge at some other pH. Ion exchange
chromatography is technically simple and can be ad-
opted for use over a very large range of scales.
Chromatofocusing and isoelectric focusing are
methods that depend on the differences in
isoelectric points between proteins but are less widely
used for preparative work than is ion exchange
chromatography because of increased cost, restric-
tions of scale and technical difRculty.
Electrophoresis is a special case. Electrophoretic
methods are of central importance in analytical pro-
tein chemistry but, until recently, have not proved
Table 1 Properties of proteins that can be exploited for purifica-
tion and associated experimental methods
Electrical charge Ion exchange chromatography
Chromatofocusing
Isoelectric focusing
(Electrophoresis)
Hydrophobic surface regions Hydrophobic chromatography
General surface properties Salt fractionation
Size Size-exclusion chromatography
Membrane filtration
Specific binding site Affinity chromatography
Dye-binding chromatography
Surface carbohydrate Lectin chromatography
Metal-binding site Metal chelate chromatography
Antigenic determinants Immunoaffinity chromatography
4548 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS
useful for puriRcation purposes. The reason for the
change has been the development of ultra high sensi-
tivity techniques for structural analysis that has
blurred the distinction between analytical and prep-
arative methods. Hence the inclusion of electrophor-
esis in Table 1 as a preparative method, although it
serves that purpose for a limited range of applications
(see later).
Although the surfaces of most soluble proteins are
predominantly polar, many of them have patches of
hydrophobic amino acids that, under appropriate
conditions (usually at high salt concentrations), can
bind to hydrophobic matrices. This provides
a method for separation provided that elution from
the matrix can be achieved under conditions that do
not lead to loss of biological activity.
Proteins differ from one another in their
solubilities in salt solutions. Clearly, in a complex
mixture of proteins the solubilities of the components
will overlap and hence fractional precipitation with
salt, usually ammonium sulfate, provides only a crude
separation method. However, it is widely used as
a Rrst step in puriRcation procedures, particularly
when working on a large scale. Occasionally, frac-
tional precipitation with an organic solvent (ethanol,
acetone) is used but there is a possibility of protein
denaturation at high solvent concentrations.
Proteins also differ from one another in size
and this can be exploited in size-exclusion
chromatography. This is inherently a method of low
resolution and can rarely achieve more than separ-
ation of mixtures of proteins into broad size classes.
However, a very important application of size-exclu-
sion chromatography is for changing the composition
(e.g. the pH) of the solvent between steps in a puriR-
cation procedure. Dialysis can also be used for this
purpose but is much slower. The same restriction of
low resolution applies to separations using membrane
Rltration, but this technique is of enormous utility at
various stages in a puriRcation schedule for reducing
the volume of protein solutions: the single major
contaminant in a protein solution is water.
Whereas the methods above depend on differ-
ences in structures of proteins there is also a set of
procedures that depend essentially on differences
in biological activity. In the vast majority of cases,
biological activity of a protein depends on it recogniz-
ing and binding to a ligand. For example, enzymes
bind to substrates and inhibitors, hormones bind to
receptors, antibodies bind to antigens and so on. This
speciRc biological activity can be exploited by con-
struction of a matrix to which the appropriate ligand
is (usually covalently) attached. Passage of a protein
mixture through the resulting afRnity matrix
should result in binding of one or a small number of
proteins that recognize the ligand. Subsequent elution
can be achieved by passage of a solution of the ligand,
or a suitable analogue, through the column. This
method has seen widespread application in the puriR-
cation of enzymes and is in principle capable of very
high selectivity because of the speciRcity of en-
zyme/substrate or enzyme/inhibitor binding. The sel-
ectivity achieved is, however, often limited by the fact
that the ligand may be charged and hence gives rise to
ion exchange effects, or it may be hydrophobic
and give rise to nonspeciRc hydrophobic interactions.
Despite this, afRnity chromatography is a very
powerful method and its use is restricted more by the
fact that it is often necessary to design and synthesize
the afRnity matrix oneself rather than by inherent
limitations.
Dye binding chromatography is a variant of af-
Rnity chromatography and relies on the fact that
a variety of chlorotriazine textile dyes interact moder-
ately speciRcally with enzymes that have nucleotide
(ATP, NAD(H), coenzyme A (CoA)) binding sites.
The ability of dye-containing matrices to recognize
nucleotide-dependent enzymes is not a purely af-
Rnity effect } indeed the structural similarity
between the dyes used and the cofactors is not obvi-
ous } and includes elements of ion exchange and
hydrophobic effects. Nevertheless, these methods
often work remarkably well for isolation of groups of
nucleotide-dependent enzymes or even of individual
members when biospeciRc elution methods are used.
Lectin chromatography and metal chelate
chromatography are available when the protein of
interest has either surface carbohydrate or a metal-
binding site, respectively. The former method de-
pends on the fact that various plants produce proteins
(lectins) that bind speciRcally to particular classes of
carbohydrate. If the lectin is coupled to an appropri-
ate support then the product matrix will speciRcally
bind glycolproteins from a mixture of proteins. Elu-
tion can be effected by passage of a solution of
the appropriate monosaccharide through the column.
In metal chelate chromatography the matrix has
a chelating agent covalently attached and loaded with
an appropriate metal ion. On passage of a mixture of
proteins through the column those with a binding site
for the metal will be retained and subsequently can be
eluted by passage of a solution of metal ions through
the column.
ImmunoafRnity chromatography is in principle the
most speciRc method available for protein isolation.
It involves raising an antibody to the target protein,
attaching the antibody to a supporting
material and then using this as an afRnity matrix.
The extreme speciRcity of antigen}antibody interac-
tions should ensure high selectivity in binding the
APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS 4549

target protein. However, there are two problems.

Clearly, the protein has to have been isolated pre-
viously in order for an antibody to be produced.
A highly puriRed protein will be required to raise
polyclonal antibodies. Alternatively, a partly puriRed
antigen can be used to produce monoclonal antibod-
ies but this adds an extensive new dimension to
a puriRcation procedure. The major restriction on the
application of the method, however, is the difR-
culty of elution of proteins from the immunoaf-
Rnity matrix once bound. The tightness of binding
often requires extreme conditions for efRcient
elution (very high or low pH, presence of chaotropic
agents) such that many protein molecules become
denatured during the elution process.

Putting them Together

Faced with the variety of methods available for the
separation of proteins the question arises as to which
of them to use and in which order for development of
a puriRcation schedule for a particular protein. The
answer to this depends on a whole host of issues such
as:
E how much protein is required?
E what sources of the protein are available?
E has the gene for the protein been cloned?
E is the protein required to be completely pure?
E is it necessary to retain biological activity?
E has it been done before?
If the answer to the last question is positive, the
obvious approach is to try to reproduce the reported
puriRcation procedure. It may not work exactly as
described } small variations in procedures between
laboratories can give rise to signiRcant differ-
ences in the behaviour of proteins during puriRcation
} but it should be relatively easy to adjust conditions
to get it right. Development of a new protocol is
time-consuming and not usually worthwhile unless it
is to be used repeatedly and an existing method ap-
pears to be unnecessarily cumbersome; even then the
published method should provide a valuable guide on
how to make improvements.
What follows are descriptions of the sorts of sched-
ules of methods that might be used in a variety of
situations.
Large-Scale Isolation of an Active Protein
Large-scale here is taken to mean a laboratory-scale
puriRcation of a few tens of milligrams of protein.
Industrial scale puriRcation might well follow the
same general pattern but there would be engineer-
ing problems associated with scaling up that will
not be dealt with here. In the case of a protein to
Figure 1 Flow chart for the purification of an enzyme.
be used for therapeutic purposes there would also
be speciRc requirements imposed by regulatory
agencies that are beyond the scope of the present
discussion.
The Sow chart in Figure 1 outlines steps in the
protocol for puriRcation of an enzyme developed in
the authors laboratory. The starting material was
5 kg of pig liver. If the source of the enzyme or other
protein is not of importance for the purpose of the
investigation then the best choice is to use an animal
tissue that can be obtained in quantity from a com-
mercial abattoir. Animal tissues are generally easy to
homogenize in a domestic food blender. Other sour-
ces such as fungi, bacteria and plants present dif-
Rculties in disruption of the tissue and are best avoid-
ed unless the source is constrained by the problem in
hand.
Ten litres of buffer was used for homogeniz-
ation and, after removal of debris, the volume of
protein solution was 8.5 L. This volume of solution is
difRcult to handle and hence fractional precipita-
tion with ammonium sulfate was used both to obtain
an initial crude puriRcation and, more importantly,
4550 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS
reduce the volume. After centrifugation of the active
fraction, resuspension of the pellet and dialysis the
volume was reduced to 500 mL.
The next step was ion exchange chromatography.
There were choices to be made of whether to use
a cation or anion exchanger, and as to which of
the various available supporting materials (cellu-
lose, Sepharose, Superose) was to be preferred.
In the present case, a cellulose-based matrix was
chosen. This was essentially because the amount
of protein in the sample (about 100 g) made it
necessary to use a large amount of exchanger and
correspondingly a large column. Cellulose-based
exchangers are much cheaper than other varieties
and, in addition, large columns of cellulose ex-
changers have better Sow characteristics than do
those of other materials. The choice of car-
boxymethyl (CM) cellulose rather than the anion
exchanger diethylaminoethyl (DEAE) cellulose was
dictated by previous experience of the behaviour of
the two materials for the separation of crude protein
mixtures.
The protein of interest was retained by the CM-
cellulose and was eluted by application of a gradient
of increasing sodium chloride. This is to be preferred
over the other possibility of using conditions where
the protein is not retained on the column since a
higher degree of puriRcation is likely to be achieved
on gradient elution. After combination and con-
centration of the active fractions the volume of
the sample had been reduced to about 50 mL and
the amount of protein to about 1 g. These amounts
were suitable for the application of a variety of
small-scale but more highly resolving techniques.
For example, had the enzyme of interest been a glyco-
protein then lectin afRnity chromatography
would have been a good choice. Similarly, hydropho-
bic chromatography could have been used. An ad-
vantage of using the latter technique would have been
that it would not have been necessary to remove the
sodium chloride from the sample after gradient elu-
tion from CM-cellulose given that in hydrophobic
chromatography the sample is applied in a solution of
high salt content to promote interaction with the
matrix.
In practise it was relatively easy in the present case
to develop an afRnity matrix for the enzyme
based an analogue of the substrate. It was worthwhile
to do this because it was intended to repeat the
puriRcation frequently so that the time involved in
preparing the afRnity matrix was subsequently
recovered. If a puriRcation is essentially one-off then
this is unlikely to be the case.
After afRnity chromatography the product was
examined by electrophoresis and found to contain
two minor contaminants, both more basic than the
target enzyme. Hence, a Rnal step using an anion
exchanger under conditions where the protein of in-
terest was absorbed but the contaminants were not,
or were bound more weakly, suggested itself. The
exchanger chosen was DEAE-Sepharose, which has
a greater resolving power than cellulose-based
products.
The Rnal product of the puriRcation procedure was
28 mg of protein that was homogeneous, as judged by
the usual criterion of producing a single band after
sodium dodecyl sulfate}polyacrylamide gel elec-
trophoresis (SDS-PAGE). The enzyme had been
puriRed 6000-fold compared with the original
homogenate and the yield was about 50%.
Small-Scale Isolation of an Active Protein
Isolation of a few milligrams of active protein follows
the same general principles as outlined above but can
often be achieved in a smaller number of steps. For
example it is not necessary to carry out fractional
precipitation with salt because the small volume of
protein solution will allow direct use of ion exchange
chromatography as the Rrst puriRcation step. In addi-
tion, because of the small scale of the procedure, the
high resolving power of ion exchange chromatogra-
phy or of hydrophobic chromatography in fast pro-
tein liquid chromatography (FPLC) mode can be ex-
ploited, which may allow a reduction in the number
of chromatiographic steps required. FPLC differs
from conventional column chromatography in that it
employs very Rne particle size matrices that offer
greater resolving power for protein mixtures. Fully
automated equipment is also available that allows for
greater reproducibility between runs than do conven-
tional methods. Capacity is, however, limited.
Proteins from Sub-Cellular Organelles
Many proteins in higher organisms exist in discrete
subcellular organelles such as mitochondria, chloro-
plasts and lysosomes. If the target protein is one of
these it may be advantageous to make a preparation
of the organelle and isolate the protein from that
rather than from a total issue homogenate. Isolation
of subcellular organelles is usually carried out by
preparative differential centrifugation. Proteins
can subsequently be extracted from the organelles
and puriRed by standard techniques.
The advantage of this approach is obvious. Given
that the organelles contain a more restricted range of
proteins than does the parent cell then the puriRca-
tion procedure is likely to be much simpler. The
downside is that the isolation of subcellular structures
is time-consuming and, except on a relatively small
APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS 4551

scale, there may not be a net saving of time in been achieved in modifying genes by the attach-
adopting this approach. ment of an export signal so that the host organ-
ism excretes the protein product into the culture
Membrane Proteins medium.
Other approaches to facilitating puriRcation of
Integral membrane proteins present special problems
cloned gene products involve the construction of
because of their location within membranes and be-
fusion proteins. One example of this is where a tail of
cause they are not soluble in aqueous buffer
basic residues (lysine or arginine) is engineered onto
solutions. The Rrst step will be to obtain a prepara-
the protein. This tail will make the protein very basic
tion of the membrane of interest, usually by dif-
and hence increase its afRnity for ion exchangers
ferential centrifugation. Next, the protein has to be
such as CM-Sephadex. If, after elution from the ex-
extracted from the membrane preparation, most
changer, further puriRcation is required then the tail
commonly by using solutions of detergents such as
can be removed (by digestion with carboxypep-
Triton X-100, Lubrol PX, digitonin, sodium cholate,
tidase B) followed by further chromatography under
etc. This is a crucial step and the best detergent to use
the same conditions. The decreased basicity conse-
to obtain optimum release of the protein from the
quent on removal of the tail will ensure that the
membrane fragments can be determined only by trial
protein now behaves differently compared with
and error.
any contaminants whose properties will not have
Once a soluble extract of the protein has been ob-
been modiRed.
tained its puriRcation can be achieved using the usual
Other approaches involve engineering afRnity
chromatographic techniques except that, because of
labels onto the protein. For example, fusion products
solubility problems, it will be necessary to maintain
between a target protein and maltose-binding protein
a standing concentration of detergent in the buf-
from Escherichia coli can be very readily puriRed by
fers. This frequently adversely affects the perfor-
amylose afRnity chromatography. Similarly, anti-
mance of ion exchange materials and more success in
bodies to certain small peptide sequences, referred to
isolation of membrane proteins has been achieved by
as Sags, have been raised so that fusion proteins
exploiting their binding properties, that is, by using
bearing these Sag sequences can be readily puriRed by
various forms of afRnity chromatography.
immunoafRnity chromatography.
A Rnal problem, once the protein has been puriRed,
Obviously, in any particular case the question
will usually be to remove the detergent from the
needs to be asked as to whether the time and cost
preparation or to change the detergent type. This can
involved in genetic engineering of the desired protein
be achieved by a variety of methods, including equi-
product is justiRed in terms of the time saved in
librium dialysis, gel Rltration and a variety of
subsequent puriRcation. The answer is likely to be
chromatographic methods.
positive only if the puriRcation is to be repeated
Peripheral membrane proteins, that is, those that
frequently.
are only loosely associated with the membrane, do
not usually present special problems. They can be
released from membrane preparations by salt extrac- Special Cases
tion or by changes in pH, are usually soluble in
The procedures described above should be used when
aqueous buffers, and are amenable to the usual
it is important to retain the biological activity of the
puriRcation methods.
protein of interest. Essentially, this means using ex-
perimental conditions under which the native three-
Products from Cloned Genes
dimensional structure of the protein is preserved.
As a result of the rapid developments in genetic tech- There are some situations where this is not necessary
nology in recent years it is now relatively easy to and all that is important is that the primary struc-
clone the gene for any protein of interest and express ture of the protein remains unchanged. An impor-
it in a suitable bacterial host. This does not change tant example of this is where the protein is required
the methods that are available for puriRcation but for amino acid sequence analysis. In this case addi-
it does allows for simpliRcation of the puriRcation tional techniques can be used for puriRcation. For
procedure. An obvious example is that expression example reverse-phase HPLC using hydrocarbon
of the gene can be manipulated so that its protein (C4}C18) stationary phases provides for high-re-
product represents a very high percentage of the solution separation of proteins but elution often
protein in the host cell. Values of up to 50% have requires the use of organic solvents such as aceto-
been achieved, which obviously simpliRes the sub- nitrile, which frequently leads to denaturation. The
sequent puriRcation. Similarly, some success has method is, however, extremely powerful for Rnal
4552 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF ENZYMES AND PROTEINS
separation of partly puriRed proteins for sequence
analysis.
For a variety of applications, including N-terminal
sequence analysis using modern high-sensitivity
techniques, only very small amounts of protein (a few
micrograms) are required. For these applications the
resolving power of SDS-PAGE can be exploited to
separate even relatively crude mixtures. The protein
of interest is then removed from the acrylamide gel,
for example by using an appropriate blotting tech-
nique, and the blot subjected to analysis.
More recently this approach has been extended to
the identiRcation of proteins in cell homogenates. The
total cell extract is separated by two-dimensional
electrophoresis, most commonly using isoelectric fo-
cusing in the Rrst dimension and SDS-PAGE in the
second dimension. Individual spots are excised from
the gel, the protein subjected to digestion with tryp-
sin, and the trypsin fragments analysed by mass spec-
trometry. The set of peptide masses obtained is then
scanned against a data bank of the masses of tryptic
peptides from all known proteins. In most cases this
allows unique identiRcation of the protein in the gel
spot, provided that its sequence is known either from
direct analysis or by translation of a DNA sequence.
Detection and Quanti\cation
It is clearly of central importance in any puriRcation
procedure that a method is available for detecting the
presence of the protein of interest in the fractions
from the various separation steps. In the case of
enzymes this is easy because they possess catalytic
activity that can be measured by some appropriate
analytical technique. Other proteins might require
the use of a bioassay, or an immunoassay, or perhaps
the identiRcation of the protein as a particular band
produced on analytical electrophoresis.
What might not be so obvious is the importance of
quantiRcation of the recovery of the protein at each
stage of the puriRcation procedure } that is, of keep-
ing an inventory. Unless this is done it is very easy to
end up with a disappearing yield of the protein of
interest and not to know at which step or steps it
disappeared. At each step it is important to measure
the total protein content and the amount of the
protein of interest. This allows not only the recovery
but also the degree of enrichment of the protein to be
determined. Any step for which either of these is low
should be abandoned.
In the case of enzymes, keeping this inventory is
straightforward; it is simply necessary to measure the
catalytic activity of a known volume of the fractions.
In other cases it is much more difRcult. Bioassays
can be very time consuming. Immunoassays are not
usually too difRcult, but in this case it is necessary
to bear in mind that immunological reactivity of
a protein may be retained even though biological
activity has been lost. In the case of a protein with no
known biological activity, or where the activity is
very difRcult to measure, then recovery can be
assessed from the measurements of the intensity of
the appropriate band produced by analytical elec-
trophoresis. Whatever the difRculties, however,
keeping a score card is essential if a successful puriR-
cation protocol is to be developed.
See also: I/Affinity Separation. Centrifugation. II/Affin-
ity Separation: Hydrophobic Interaction, Chromato-
graphy; Immobilised Boronates and Lectins; Immuno-
affinity Chromatography. Chromatography: Protein
Separation; Size Exclusion Chromatography of Polymers.
Chromatography: Liquid: Mechanisms: Size Exclusion
Chromatography. Electrophoresis: Isoelectric Focusing;
Two-dimensional Electrophoresis. Membrane Separ-
ations: Membrane Bioseparations. III / Proteins: Centri-
fugation; Electrophoresis; Field Flow Fractionation; High-
Speed Countercurrent Chromatography; Ion Exchange.
Further Reading
Deutscher MP (ed.) (1990) Guide to Protein PuriTcation.
San Diego: Academic Press.
Doonan S (ed.) (1996) Protein PuriTcation Protocols.
Totowa, Humana.
Harris ELV and Angal S (eds) (1989) Protein PuriTcation
Methods. Oxford: IRL Press.
Harris ELV and Angal S (eds) (1990) Protein PuriTcation
Applications. Oxford: IRL Press.
Kenney A and Fowell S (eds) (1992) Practical Protein
Chromatography. Totowa: Humana.
Walker JM (ed.) (1998) Protein Protocols on CD-Rom.
Totowa: Humana.
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS
ESSENTIAL GUIDES FOR
ISOLATION/PURIFICATION OF
IMMUNOGLOBULINS
A. Layer, P. Schneider, J.-D. Tissot and
M. A. Duchosal, Service Re& gional Vaudois de
Transfusion Sanguine, Lausanne, Switzerland
Copyright ^ 2000 Academic Press
Introduction
The immunoglobulins (Igs), proteins produced by
B lymphocytes, have been extensively studied both at
molecular and genetic levels. They consist of two
identical heavy chains and two identical light chains
having therefore the same isotype and the same type,
respectively. Igs are puriRed for three main purposes.
(i) as therapeutic injections to patients; (ii) for use as
a tool in research or clinical diagnosis; and (iii) for
their biochemical analysis (speciRcity, isotype or
clonal diversity). Most of these applications require
that the binding activity of Igs be retained throughout
all the puriRcation procedures.
PuriRcation of Igs can be performed according to
their physicochemical properties, their biological
activities or a combination of both. The technique
used will depend on the desired degree of purity
and the amount and nature of the starting mater-
ial. The methods that have been described are gener-
ally directly applicable to crude materials such as
serum, ascitic Suid or cell culture supernatant. Two-
dimensional polyacrylamide gel electrophoresis (2D-
PAGE) affords an efRcient way of evaluating
the degree of purity reached in afRnity puriRca-
tions. Several aspects of 2D-PAGE analysis are
described in detail in two other articles Electro-
phoresis/Two-dimensional PAGE and &Clinical Ap-
plications of Electrophoresis/Electrophoresis in this
Encyclopedia.
As a general rule, and independently of the tech-
nique used, the starting material should always be
devoid of any insoluble substances and the puriRca-
tion be preceded by centrifugation or Rltration. Vis-
cous Suids, such as serum, may be diluted before use,
especially for chromatographic procedures. The solu-
tions should contain a bacteriostatic agent, such as
0.02% sodium azide (NaN ), and be kept on ice. Ig

solutions should be handled gently, avoiding bubbl-
ing or frothing, because such manipulations may be
4553
accompanied by denaturing effects, and may
lead to protein precipitation.
Puri\cation by Precipitation
Solubility of the proteins, particularly Igs, in water
relies mainly on the ability to make hydrogen bonds
between polar or ionic groups with water molecules
(hydrophilic interactions), and on the capacity to
maintain hydrophobic groups that cannot interact
with water molecules buried inside the proteins. In
addition, the solubility of Igs is temperature-depen-
dent. Any external factor capable of modifying hy-
drogen bonds or decreasing the medium hydrophilic-
ity will decrease the solubility of the proteins and may
eventually lead to their precipitation. Each protein
has its own physicochemical characteristic, including
solubility. For this reason, several differential
precipitation procedures can be developed to isolate
Igs from various Suids. These procedures are present-
ed below.
Differential Ethanol Precipitation
The Rrst fractionation of plasma proteins for thera-
peutic use was described in 1949 by E.J. Cohn. The
basic procedure, with few modiRcations, is still
widely used in industrial fractionation centres.
Basically, ethanol is added progressively to the
medium to a Rnal concentration varying from 8 to
40%. Subsequently, the temperature is decreased
to !33C and then to !53C. Finally, the pH
is decreased from 7.3 to 4.8. These steps yield pre-
cipitation fractions, called Cohns fractions I}V.
Fraction II contains the
-globulins or Igs. The
treatment of this fraction with caprylic acid (see be-
low) allows the preparation of Igs that are enriched in
IgA and IgM. This approach is used when large
amounts of Igs are needed, i.e. for therapeutic pur-
poses (from up to 5000 L plasma) and will not be
detailed further.
Ammonium Sulfate Precipitation
Small and highly charged ions, such as ammonium
ions, replace bound water molecules when present at
a sufRcient concentration. This decreases protein
4554 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS
solubility and, when the ammonium sulfate concen-
tration is increased stepwise, a sequential precipita-
tion of proteins may be obtained.
Practically, a saturated solution of ammonium sul-
fate (761 g L\1; 4.1 mol L\1) is slowly added to
a stirred (in order to avoid local over-increase in
concentration) solution of Igs, until the Rnal desired
concentration, usually expressed as a percentage of
ammonium sulfate saturation, is reached. Although
some interspecies variability is observed for Ig solu-
bility, a 50% ammonium sulfate saturation is usually
appropriate for most Ig precipitation procedures. Pre-
precipitation at 40% ammonium sulfate saturation
may be useful to remove large protein aggregates and
proteins that may precipitate at low ammonium ion
concentration. The precipitation is allowed to occur
at 43C for 6}12 h. The precipitate is recovered by
centrifugation at 2000}5000 g for 20}30 min. In gen-
eral, the pellet is then gently solubilized in a minimal
volume of a physiological buffer, and dialysed to
remove the residual ammonium sulfate. Alterna-
tively, Ig purity can be increased by washing the pellet
with a 50% saturated solution and solubilization in
PBS, followed by another round of ammonium sul-
fate Ig precipitation. When starting with serum or cell
supernatant containing serum, this method allows the
removal of most albumin and haemoglobin, but the
precipitated fraction still contains several serum pro-
teins in addition to Igs. In this case this approach
must be coupled with another puriRcation technique
if pure material is needed. When starting with
a serum-free cell culture supernatant, this method is
convenient to isolate, and to concentrate, monoclonal
Igs in one step. In addition, the method is easy, cheap
and can be applied to large volumes.
Caprylic Acid Precipitation
The solubility of proteins is altered by the presence of
some short chain fatty acids, such as octanoic acid
(caprylic acid) at mildly acidic pH. Basically, caprylic
acid increases medium hydrophobicity. In practice
the pH of the starting solution must be adjusted to
4 by the addition of about 2 vol of a 60 mmol L\1
sodium acetate buffer. Then, 0.04}0.07 vol of
caprylic acid (depending on the starting material as
well as on the animal species of the Igs) is added drop
by drop while stirring, and the solution is incubated
at room temperature for 30 min. Under these condi-
tions, most of the serum proteins are precipitated,
with the exception of IgG, which is recovered in the
supernatant after centrifugation at 5000 g for 10 min.
This method bears similarities with that of am-
monium sulfate precipitation. In particular, it needs
to be coupled with another one to yield highly puri-
Red Ig fractions.
Chromatographic Methods
In chromatographic procedures, compounds in solu-
tion are separated by allowing them to Sow through
a selective medium poured in a column. Differen-
tial interactions between molecules and matrix are
responsible for them migrating at various speeds, or
even completely immobilizing them. Separated mol-
ecules are recovered in the efSuent of the column.
Several commercially available preparations allow
separation of proteins according to their various
physicochemical properties. Detailed information
about the use of such media is furnished by the
manufacturers or can be found in the literature for
particular applications.
Ion Exchange Chromatography
An ion exchanger consists of a positively or negative-
ly charged group covalently bound to an insoluble
matrix. Charged molecules with complementary po-
larity to that of the immobilized groups bind to the
matrix through electrostatic interactions, whereas
uncharged or similarly charged molecules pass freely
through the matrix. Since the net charge of a pro-
tein depends on the pH, the starting experimental
conditions must be carefully chosen. The bound pro-
teins may be desorbed either by change in pH, or by
change in the ionic strength. The former modiRes the
charge of the protein, whereas salts compete with the
binding of the protein to the resin. Addition of NaCl
is most frequently used for elution. As the strength of
the protein}matrix interaction depends on the net
protein charge, a sequential elution can be performed
by gradually increasing salt concentration. Because
Igs have a more basic isoelectric point than most
other serum proteins, ion exchange chromatography
can be used for their puriRcation. Practically, the
matrix should be extensively washed with
0.5 mol L\1 HCl or 0.5 mol L\1 NaOH before use,
and then equilibrated with the binding buffer. In
addition, the sample must be dialysed against the
binding buffer before being loaded on the resin.
At pH 6.5 (5 mmol L\1 phosphate), Igs will not
bear negative charges, and therefore will not bind to
a positively charged matrix such as diethylaminoethyl
(DEAE) matrix, which is not the case for other serum
proteins. A bulk Ig fraction can therefore be re-
covered in the Sow-through. In contrast, Igs will bear
a negative net charge at pH 8.5 (10 mmol L\1 Tris),
and thus will bind to a positively charged matrix
(DEAE). Sequential elution of proteins bound to the
matrix can be performed by increasing the concentra-
tion of NaCl from 0.05 to 1 mol L\1. Igs are among
the Rrst serum proteins to be eluted, at salt concentra-
tions usually below 0.5 mol L\1. Ig isotypes can be
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS
differentially puriRed using this method. Rela-
tively pure IgG is usually recovered in the Rrst eluted
fraction; IgM, the last eluted Ig isotype, may also be
recovered in quite a pure form, whereas IgD and IgA,
with intermediate elution properties, are only poorly
resolved with this method. Ion exchange chromatog-
raphy can yield sufRciently pure antibodies if the
starting material is a cell culture supernatant or an
ascitic Suid, but it must be coupled to an additional
puriRcation step when samples such as serum are
used. The method is also cheap and is convenient for
large initial volumes.
Hydroxyapatite Chromatography
Immobilized hydroxyapatite (calcium phosphate hy-
droxide) is used for another kind of adsorption
chromatography. At pH 6.8, Igs bind to the matrix,
and are eluted when a linear gradient of phosphate
buffer from 120 to 300 mmol L\1 is applied.
When highly puriRed Ig fractions are needed, this
technique must be coupled to another one, again
depending on the starting material.
Gel Filtration Chromatography
A gel Rltration matrix consists of beads containing
pores of various sizes. As the sample Sows through
the matrix, the largest molecules are excluded from
the beads. They stay only in the mobile phase, and
move fast. The smaller molecules, depending princi-
pally on their sizes but also on their shapes, dif-
fuse more or less inside the pores, and move more
slowly within the column. Thus, this system allows
the separation of the proteins according to their
sizes and shapes. Various matrices with particular
structures, pore sizes and excluding limits (the Mr at
which the proteins are no more able to enter into the
beads) are available commercially. Gel Rltration can
be used within broad pH ranges, with or without
detergents such as 1% sodium dodecyl sulfate (SDS),
or dissociating agents such as urea or guanidine. Igs
puriRcation is usually performed without sophisti-
cated conditions, and allows the separation of IgM
molecules that are considerably larger than IgG as
well as many other serum proteins. Using an exclu-
sion limit of 300}500 kDa, and a column volume of
at least 20 times larger than that of the starting
solution, IgM is easily recovered in the excluded
peak. Due to the time needed to allow a complete
passage, the use of a fraction collector is highly rec-
ommended. Fractions corresponding to 1}3 initial
volumes should be collected. Gel Rltration must usu-
ally be coupled with other methods to yield sufR-
cient Ig purity, and is mainly limited to the puriRca-
tion of IgM.
4555
Precipitation of Immune Complexes
The Precipitin Reaction
When antigens are added in adequate proportions to
a mixture of antibodies, speciRc Igs and antigens will
form a lattice which is susceptible to precipitate (the
precipitin reaction). This macromolecular complex
can then be easily recovered by centrifugation. The
required amount of antigens to be added must be
determined by establishing a precipitation curve. This
implies that one should also have a procedure allowing
the precipitate to be quantiRed. The use of radiolabel-
led antigens is particularly suitable for this when
restricted amounts of antigens are available. Comp-
lement activation, which occurs in normal fresh serum,
is able to inhibit the precipitation strongly. Therefore,
it is necessary to make either a pre-puriRcation of the
Igs or to inhibit the complement cascade by the addi-
tion of ethylenediaminetetraacetic acid before per-
forming a precipitin reaction. The precipitated lattice
is subsequently re-solubilized either by incubation
with serum or with an excess of free antigen, or by
digestion with papain, which generates Fc fragments.
Polyethylene Glycol Precipitation
Soluble (nonlattice) immune complexes, either pres-
ent naturally, or generated by adding a corresponding
antigen, precipitate from serum in the presence of
3}4% polyethylene glycol (PEG; Mr 6000) after
2}12 h incubation at 43C. This method has been
widely used to isolate circulating immune complexes
in various pathological situations. Other high mo-
lecular weight proteins, as well as aggregated Igs, can
also precipitate using above-mentioned conditions.
Therefore, additional steps of puriRcation, using pro-
tein G or A (see below), are generally warranted
before sufRciently pure material can be used.
Solubilization of most PEG-precipitated immune
complexes can easily be performed using most buf-
fers that do not contain PEG. In contrast, dissociation
of immune complexes requires quite harsh condi-
tions, that are frequently not compatible with tech-
niques allowing the further puriRcation of free Igs
devoid of antigens. Immunoprecipitation of speciRc
antibodies is therefore mainly limited to analytical
purposes that do not need biological activity. Ig light
and heavy chains from immune complexes are easily
solubilized in buffers containing SDS, and can be
analysed by SDS-PAGE or 2D-PAGE.
Af\nity Chromatography
In afRnity chromatography, samples containing
Igs are incubated in the presence of a matrix consist-
4556 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS

ing of an Ig-binding molecule covalently coupled to

a bead support. Unbound molecules are removed by
washing, and speciRcally bound Igs are then eluted
using an appropriate buffer. The method is high-
ly speciRc and high Ig purity is usually reached in
a single step. Ig-binding molecules belong to three
major groups: (i) bacterial protein A or protein G; (ii)
speciRc antigens; and (iii) monospeciRc antibodies
directed to epitopes on Igs (such as goat anti-human
Ig antibody). Various bead supports and coupling
procedures have been studied. The use of commer-
cially available activated beads has now allowed
preparation of afRnity media within most labor-
atories. In particular cyanogen bromide (CNBr)-
activated Sepharose beads can easily be used for most
applications and detailed instructions are furnished
by the manufacturers. After coupling, the resin
should be extensively washed to remove all un-
coupled Ig-binding molecule, and equilibrated in
binding buffer. The binding capacity of each
resin should be determined experimentally before use.
If sample binding is indifferently performed in
a container (batch procedure) Rxed on a rotating
wheel or through a column, washing and elution are
best performed through a column. The use of a peri-
staltic pump is highly recommended to ensure a Rxed
Sow.
Washing and elution steps are best followed online
with a UV detector set to monitor at 280 nm. Alterna-
tively, fractions may be collected and tested individ-
Figure 1 (A) 2D pattern of IgG purified over protein G-
ually using either UV absorption or more speciRc Sepharose. Human serum was incubated with commercially
procedures. When necessary (see elution conditions available protein G-Sepharose (Pharmacia), and IgG eluted as
below), the eluate has to be collected in a neutralizing indicated in the text.
heavy chains migrate within pIs ranging
solution, and the resin immediately re-equilibrated in from 6 to more than 10, with a size of about 50 kDa. Light chains
the binding buffer. Unused resins should be display pIs ranging from 5 to 10 and size between 21 and 26 kDa.

, IgG heavy chain; , , light chains. (B) 2D pattern of IgA affinity
stored at 43C in the presence of a bacteriostatic agent, purified over a homemade anti- chain-Sepharose resin. The
such as 0.02% sodium azide or 20% ethanol. affinity resin was prepared from commercially available CNBr-
activated Sepharose (Pharmacia), according to the manufac-
Af\nity Chromatography using Immobilized turers recommendations, and commercially available goat anti-
Protein G and A human  chain antibodies.  chains migrate with a pI ranging from
4.9 to 6.1 and a size of about 58 kDa. , IgA heavy chain; , , light
Protein A and protein G are present within the bacter- chains.
ial cell walls of Staphylococcus aureus and of group
G streptococci, respectively. Both proteins have high
afRnity for the Fc region of IgG, but bind dif- binding to samples, the resin is washed with about
ferentially IgG subclasses from various species. 20 vol of the binding buffer until online UV
Whereas protein G bind all human and mouse IgG absorption of the Sow-through medium gives an op-
subclasses, protein A presents only low binding capa- tical density back to zero. Elution of IgG is best
city for human IgG3 and mouse IgG1. Ready-to-use performed by addition of 100 mmol L\1 glycine pH
matrix-immobilized protein G or A is commercially 2}3 in tubes containing a suitable amount of a neu-
available, and detailed information about the binding tralizing buffer such as 1 mol L\1 Tris, pH 8 or
properties of these two proteins can be found in the 1 mol L\1 phosphate, pH 7. As illustrated in Fig-
literature or furnished by the manufacturers. Binding ure 1, IgG is highly puriRed after a single-step proced-
buffers usually contain 100 mmol L\1 Tris or ure over protein G column, with no other heavy chain
10 mmol L\1 phosphate with 0.15 mol L\1 NaCl, at class detectable. The strength of the protein A(G)}Ig
pH 8 for protein A and pH 7 for protein G. After Ig interaction, which is mainly based on hydrophobic
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS 4557

interactions, can be increased by raising salt concen-

trations of the binding and washing buffers to
3.3 and 3 mol L\1, respectively. This high salt con-
centration method was initially described for the
puriRcation of mouse IgG1 on protein A, but is now
obsolete, due to the availability of protein G. The use
of an excess of protein A allows IgG1, IgG2 and IgG4
to be depleted from samples and IgG3 to be puriRed
using protein G in a second step. A sequential elution
of all four IgG subclasses from protein A using a pH
gradient has been described, but the resolution is
quite low and subclass-speciRc puriRcation of IgG1 4
\
is now best performed using immobilized mono-
speciRc anti-Ig raised against either
1,
2,
3 or

4 heavy chains of the IgG molecule.
The elution yield from protein G or A using stan-
dard procedures is never 100% (see below) and resid-
ual IgG can be eluted during a second run procedure
and may contaminate the new IgG being puriRed with
IgG from the previous run. In order to avoid any Ig
contamination from a previous experiment, it is
therefore highly advisable to use either the same
batch of protein G or A for the same initial IgG
preparation or to use a new batch of resin for each
procedure.
Af\nity Chromatography using Immobilized
anti-Igs
In this method, the immobilized binding molecules
are Igs (mouse, rabbit, goat, sheep) directed against Ig
heavy and/or light chains. Using antibodies of various Figure 2 (A) 2D pattern of a mixture of affinity purified IgM and
speciRcities, it is possible to isolate either total Igs IgA; (B) 2D pattern of a mixture of affinity purified IgM and IgG.
(using anti- and - chains), particular Ig isotopes Immunoglobulins were purified over homemade anti- chain-
Sepharose resin, anti -chain-Sepharose resin and anti
-chain-
(using anti-, -
, -, - or - chain) or IgG subclasses
Sepharose resin and mixed as indicated before electrophoresis.
(using anti-
1, -
2, -
3 or -
4 chains). The interaction Resins were prepared according to the manufacturers instruc-
between immobilized and targeted immunoglobulins tions, from CNBr-activated Sepharose and commercially avail-
is just a particular type of antibody}antigen interac- able goat anit-human ,  and
chains. Immunoglobulins were
tion. Binding and elution are therefore basically per- prepared from serum as indicated in the text.  chains migrate
with pIs between 5.6 and 6.4, and size of 72 kDa. , IgM heavy
formed using the same conditions as those used for
chain; , IgA heavy chain;
, IgG heavy chain; , , light chains.
immobilized antigen supports (see below). Figure 1B
shows that IgA puriRed from a human serum sample
over an anti- chain resin does not display any other class. Disadvantages are that the methodology re-
heavy but  chain isotope. The resolution obtained by quires larger amounts of resin and several immobi-
2D-PAGE in separating various Ig heavy chains is lized antibodies and that, when biological Suids are
illustrated in Figure 2. processed, the Rnal preparation still contains proteins
IgG subclasses can be puriRed using two dif- other than IgG.
ferent procedures named positive and negative isola-
tions. In positive isolation, the desired subclass is
Af\nity Chromatography using Immobilized
immobilized on a resin, washed and recovered by
Antigens
elution, whereas in the negative isolation, all un-
wanted subclasses are bound on resin and the desired A commonly used method for purifying and recover-
subclass is recovered in the Sow-through. The ad- ing antigen-speciRc, and antigen-free, Igs from a poly-
vantage of this latter approach is that the Rnal prep- clonal mixture of antibodies involves the use of
aration is not exposed to strong nonphysiological matrix-bound antigens that bind speciRc, antibodies
conditions which may denature the puriRed IgG sub- at physiological pH and salt concentration
4558 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS
Figure 3 2D patterns of affinity purified anti-tetanus toxoid anti-
bodies. Immunoglobulins were purified from two severe combined
immunodeficient (SCID) mice (A and B) previously injected with
human lymphocytes and boosted with tetenus toxoid. Homemade
tetanus toxoid-Sepharose resin was prepared according to manu-
facturers recommendations, from commercially available (Phar-
macia) CNBr-activated Sepharose and a home preparation of
tetanus toxoid. Igs were SDS-eluted, as indicated in the text.
,
IgG heavy chain; , IgM heavy chain; , , light chains, a, albumin.
(10 mmol L\1 Tris or phosphate buffer saline
(PBS) 0.15 mol L\1 NaCl, pH 7.5). The unbound
material is washed away with about 20 resin-volumes
of binding buffer. Elution of bound antibodies is
commonly performed by successive washes with
100 mmol L\1 glycine, pH 2}3 and 100 mmol L\1
triethylamine, pH 11}12. The eluted fractions are
subsequently recovered into tubes containing neu-
tralizing buffer. Elution solutions such as
5 mol L\1 LiCl/PBS, 3.5 mol L\1 MgCl2/PBS, 1%
SDS, 2}8 mol L\1 urea, 3 mol L\1 thiocyanate, 10%
dioxane, or 50% ethylene glycol can also be used.
The puriRcation of anti-tetanus toxoid Igs over
a toxoid-coated resin is presented in Figure 3. Both
2D-PAGE light chain patterns shown depict a
limited number of easily distinguishable spots,
typical of oligoclonal Igs. These patterns are clearly
different from that of total Igs (and puriRed total
IgG, not shown) from the same serum, indicating
that a subpopulation of Igs was puriRed. Whereas the
anti-tetanus toxoid antibodies shown in Figure 3A
consist only of IgG, some IgM anti-tetanus toxoid
antibodies were also present in the case shown in
Figure 3B.
Af\nity Chromatography Using Jacalin or
Complement
Jacalin (a carbohydrate-binding molecule) allows the
separation of both subclasses of pre-puriRed IgA
(jacalin binds IgA1 but not IgA2). Complement C1q
will bind antigen-complexed Igs. Anti-complement
Igs will bind immune complexes bound to compo-
nents of the complement system.
Recovery from Af\nity Resins
As already mentioned, elution from protein G or
A-Sepharose may be incomplete. In our hands, puriR-
cation of 6}12 mg batches of IgG from various sour-
ces resulted in a recovery of about 50%, using acidic
elution. Similar recovery yields have also been re-
ported by others, using similar elution. PuriRcation of
anti-tetanus toxoid antibodies on tetanus toxoid-
Sepharose resulted likewise in a 50% loss of antibody
activity. We further investigated antibody recovery
yield using afRnity-puriRed radiolabelled anti-
bodies. When puriRcation was scaled down to 10
g
Ig (an amount that allows enzyme-linked immunosor-
bent assay or electrophoresis techniques), recovery of
bound material from protein G or tetanus toxoid-
Sepharose was only about 10%. The percentage of
lost Igs was roughly inversely proportional to the
initial Ig amount. Some loss is acceptable when
purifying large batches of monoclonal antibodies. On
the other hand, when purifying antibodies for analyti-
cal purposes, one should keep in mind that antibody
losses may skew the Rnal results; the composition
(isotype, subclass and diversity) of the eluted fraction
may indeed no longer reSect the composition of Igs
that were initially loaded on the resin. The problem of
low recovery could be solved by heating Ig-loaded
protein G- or tetanus toxoid-Sepharose in the pres-
ence of SDS and dithioerythritol; more than 97% of
bound Igs could be recovered by this way. Of course,
such treatment does limit further analysis to methods
that do not require biological activity of Igs, such as
electrophoresis, since Igs are denatured under such
conditions.
Conclusion
Many different methods have been described
over the years to purify Igs, and the most important
have been brieSy presented in this article. Of course,
we have made a choice between the many methods
available, and the list is not exhaustive. Numerous
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF IMMUNOGLOBULINS 4559

Table 1 Summary of the major approaches for purifying immunoglobulins

Starting material Methods Purposes

Plasma, asciitis Affinity chromatography on protein G or A Isolation of pure IgG

Plasma, asciitis, pre-purified immunoglobulin
fractions
Plasma, asciitis, pre-purified immunoglobulin
fractions
Plasma, asciitis, pre-purified immunoglobulin
fractions
Plasma, asciitis, pre-purified immunoglobulin
fractions
Plasma, asciitis, pre-purified immunoglobulin
fractions
Affinity chromatography on purified
antigens
Affinity chromatography using mono-
specific antibodies (anti-, -
, -, - or -)
(NH4)2SO4/DEAE Sepharose
(NH4)2SO4-Hydroxyapatite
Gel filtration/DEAE Sepharose
Purification of monospecific antibodies
Isolation of immunoglobulins of a single
isotype
Preparation of large amounts of relatively
pure immunoglobulin fractions
Preparation of large amounts of relatively
pure immunoglobulin fractions
Preparation of relatively pure IgM frac-
tions

variations and/or combinations of methods may be
used to satisfy a particular need, depending on the
starting material, as well as for the purpose of the
puriRcation. However, for most current applications,
afRnity puriRcation procedures appear to be the
most elegant and selective methods. The binding ca-
pacities of afRnity resins are usually high (up to
20 mg of immunoglobulins per mL resin), and their
reusability allows the puriRcation of quite large
amounts of pure immunoglobulins in relatively short
times.
Table 1 summarizes the most efRcient methods
of purifying Igs.
Further Reading
AkerstroK m B and BjoK rck L (1986) A physicochemical study
of protein G: a molecule with unique immunoglobulin
G-binding properties. Journal of Biological Chemistry
261: 10240}10247.
AkerstroK m B, Brodin T, Reis K and BjoK rck L (1985) Protein
G: a powerful tool for binding and detection of mon-
oclonal and polyclonal antibodies. Journal of Immunol-
ogy 135: 2589}2592.
Bukovsky J and Kennett RH (1987) Simple and rapid puriR-
cation of monoclonal antibodies from cell culture super-
natants and ascites Suids by hydroxyapatite chromatog-
raphy on analytical and preparative scales. Hybridoma
6: 219}228.
Burnouf T (1994) New trends in plasma fractionation and
plasma products (review). Vox Sanguinis 67 (suppl 3):
251}253.
Crowley-Nowick PA, Campbell E, Schrohenloher RE et al.
(1996) Polyethylene glycol precipitates of serum con-
tains large proportion of uncomplexed immunog-
lobulins and C3. Immunological Investigations 25:
91}101.
Harlow E and Lane D (1988) Antibodies: A Laboratory
Manual, pp. 283}318. Cold Spring: Cold Spring Harbor
Laboratories.
Labrou N and Clonis YD (1994) The afRnity techno-
logy in downstream processing. Journal of Biotechnol-
ogy 36: 95}119.
Langone JJ (1982) Applications of immobilized protein
A in immunochemical techniques. Journal of Immunolo-
gical Methods 55: 277}296.
Langone JJ (1982) Protein A of Staphylococcus aureus and
related immunoglobulin receptors produced by strepto-
cocci and pneumococci. Advances in Immunology 32:
157}252.
Page M, Baines MG and Thorpe R (1994) Preparation of
puriRed immunoglobulin G (IgG). Methods in Molecu-
lar Biology 32: 407}432.
Perosa F, Carbone R, Ferrone S and Dammacco F (1990)
PuriRcation of human immunoglobulins by sequential
precipitation with caprylic acid and ammonium sulfate.
Journal of Immunological Methods 128: 9}16.
Rojas G, Jimenez JM and Gutierrez JM (1994)
Caprylic acid fractionation of hyperimmune
horse plasma: description of a simple procedure for
antivenom production. Toxicon 32: 351}363.
Scholz GH, Vieweg S, Leistner S et al. (1998) A
simpliRed procedure for the isolation of immuno-
globulins from human serum using a novel type
of thiophilic gel at law salt concentration.
Journal of Immunological Methods 219: 109}118.
4560 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
ESSENTIAL GUIDES FOR ISOLATION/
PURIFICATION OF NUCLEIC ACIDS
G. A. Monteiro, J. M. S. Cabral and
D. M. F. Prazeres, Centro de Engenharia
Biolo& gica e Quimica, Instituto Superior Te& cnico,
Lisboa, Portugal
Copyright ^ 2000 Academic Press
Introduction
Isolation/puriRcation of nucleic acids (NAs } dsDNA,
ssDNA, tRNA, mRNA, etc.) is a common and crucial
step in most molecular biology, biomolecule engin-
eering, cancer research, recombinant DNA, forensic
analysis, gene therapy, DNA vaccines and diagnostics
(e.g. DNA chips) applications. Most often, NAs are
isolated and puriRed merely as a way of obtaining
quantitative and qualitative information about a cer-
tain sample (e.g. paternity testing, screening for vi-
ruses in clinical samples, etc.). Several puriRcation
methods have been developed by laboratories and
companies involved in the above-mentioned areas.
These protocols and processes are based on the same
general principles, and follow three main stages: ob-
taining the NA source (cell growth, tissue isolation,
Figure 1 Strategies for the isolation/purification of nucleic acids.
chemical or enzymatic reaction), primary isolation
and puriRcation. Different techniques exist that
can be used alone or in combination within each step
of the isolation/puriRcation scheme (Figure 1).
When choosing an isolation/puriRcation protocol
or process, NA researchers and manufacturers should
take into account several aspects. Especially impor-
tant are the nature of the target NA, the Rnal applica-
tion, cost and availability of the technique (Table 1).
The Rnal application will determine which speciRca-
tions (yield, purity, safety, efRcacy, identity) the
Rnal NA preparation should follow.
Nucleic Acids
The success of NA puriRcation relies on a minimal
understanding of the molecular composition and
structure of the target molecule. NAs are linear poly-
mers of nucleotides (adenylate, guanylate, cytidylate
and uridylate in RNA and deoxyadenylate,
deoxyguanylate, deoxycytidylate and thymidylate in
DNA) linked by phosphodiester bonds. The presence
of negatively charged phosphate groups in the
backbone of the molecule confers a polyanionic
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS 4561

Table 1 Relevant aspects to the choice of a nucleic acid isolation/purification protocol or process

Target nucleic acid

DNA or RNA
Size (from a few base pairs to 100 kb or more)
Base composition (e.g. poly(A)# segments)
Structure (single/double strand, secondary and tertiary structure, supercoiling)

Application
Material for research (cloning, sequencing, in vitro translation, etc.)
Material for therapeutic use (gene therapy, gene marking, DNA vaccines, antisense)
Material for identification and quantification (diagnostics, forensics, medicine)

Specifications
Purity
Yield
Potency
Safety
Identity

Source
Cells
Prokaryotic (bacteria)
Eukaryotic (plant, animal, fungi including yeast)
Viruses (M13, phage
, human immunodeficiency virus, hepatitis)
Chemical and enzymatic mixture
Oligonucleotides from solid-phase synthesis
Restriction digests
Polymerase chain reaction mixtures
Labelling and modifying reaction mixtures

Type of processing
Single sample
High-throughput
Parallel, n samples with the same target NA
Parallel, n samples with different target NA
Sequential, n samples with same target NA
Sequential, n samples with different target NA

Cost
Per number of isolation (diagnostics)
Per amount of target DNA (large scale manufacture)

Other
Time
Scaleability of process
Environmental issues
Safety of protocol
Robustness
Automation

nature to NAs. DNA molecules are often double-
helix structures formed by two strands winding
around each other and around a common axis. These
structures are stabilized by hydrogen bonds and
mainly by stacking forces. The helix axis can also be
coiled, forming a higher order structure, named
supercoiling. RNA can take on the same conRgura-
tions as DNA, having secondary and tertiary struc-
tures; it can be single-stranded (more often) or
double-stranded, linear (more often) or circular, and
it can also form hybrid helices with DNA. Double-
stranded NAs melt to single strands fairly sharply as
condition (temperature, OH\, denaturants like for-
maldehyde or formamide) becomes more denaturing.
Renaturation is achieved when the reversible de-
naturation condition is removed. However, if the
denaturation condition is quickly removed, the pro-
cess is irreversible. The stability of double strands
increases with the mole fraction of GC pairs and
decreases as the pH is varied towards either side of
neutrality. RNA is particularly sensitive to alkaline
conditions.
4562 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
Primary Isolation
Cell Disruption
Cells from the source organism are Rrst recovered
from the starting material, whether it is a cell culture,
tissue or clinical sample, and resuspended in an ap-
propriate buffer. The isolation of the target NA
then starts with the disruption or lysis of the cells.
There are several cell disruption methods that can be
used alone or in combination (Table 2).
The choice of disruption method should consider
factors such as the effect on the Rnal NA, the cell
type (animal or plant culture or tissues; fungi or
Gram-negative or -positive bacteria spheres or Rla-
ments; viruses), the scale (laboratory- or industrial-
scale), sample volume and number of samples, asso-
ciated costs, duration and Rnal application.
Usually, cell disruption is combined with a chem-
ical or an enzymatic step for the inactivation of in-
tracellular nucleases, which can degrade the target
NA if a controlled method is not properly adopted.
The physical and chemical conditions present during
cell lysis are a key step in NA isolation. Temperature
should be kept around 43C to avoid nuclease activ-
ities that degrade the NAs. When an enzymatic step is
included (performed at around 25}353C), for in-
stance by adding proteinase K or RNase to hydrolyse
proteins or contaminating RNAs, caution should be
taken with the presence of endogenous DNases that
are active in the same temperature range.
The release of NAs during lysis signiRcantly in-
creases the viscosity of the solution, making mixing
a difRcult task. The lysates show non-Newtonian
properties, exhibiting a rheological behaviour that
makes Sow and handling of the material very dif-
Rcult. This issue is particularly troublesome on a large
scale. The mixing at this stage should be very gentle
to avoid shearing of the target NA. In certain proto-
Table 2 Summary of cell disruption methods
Physical Chemical Biological
Pressure Detergents Enzymes
Ultrasoud SDS RNase
Blades Triton Lysozyme
Grinding Acid Proteinase K
FreezeIthaw Alkali Viruses
Osmotic shock Organic solvents
Dehydration Phenol
Chaotropic salts
Urea
Guanidinium hydrochloride
Guanidinium thiocyanate
Caesium trifluoroacetate
Thiol reduction
cols and processes it is important to maintain the
contaminating NAs with the highest molecular
weight possible, in order to facilitate their removal in
the subsequent steps. Glucose or sucrose is often
included in the lysis buffer or after disruption, in
order to protect NAs against shearing. Sensitivity to
shear increases with molecular weight, and a single-
stranded NA is more sensitive than a double-stranded
one with the same length. Special care should thus be
taken when handling (pipetting, pouring, mixing) NA
solutions of high molecular weight.
The lysis buffer usually contains a chelating agent
such as ethylenediaminetetraacetic acid (EDTA). The
removal of divalent cations (mainly Ca2# and Mg2#)
from biological membranes and cell walls (when pres-
ent) destabilizes their structure, facilitating lysis, and
reduces the activity of endogenous Mg2#-dependent
nucleases, preventing NA degradation. Protein-de-
naturing agents (e.g. detergents, phenol, guanidinium
hydrochloride) are commonly added. The detergents
denature proteins, including the ones associated with
NAs, and solubilize lipids from membranes, facilitat-
ing cell disruption. Finally, the lysis buffer pH
must be tightly controlled and local pH extremes
avoided, since low pH values promote the hydrolysis
of DNA, and alkaline pH favours the cleavage of
RNA and the irreversible denaturation of chromo-
somal DNA. Moreover, very low pH values lead to
depyrimidation and depurination.
RNA is notoriously susceptible to degradation and
special care is required in its puriRcation. The pres-
ence of endogenous and/or exogenous RNases is
a major concern, since RNases can recover activity
even after harsh denaturation treatments (e.g.
boiling). High concentrations of strong chaotropic
agents (e.g. guanidinium hydrochloride, guanidinium
thiocyanate, caesium triSuoroacetate) is often used to
inactivate RNases irreversibly and simultaneously
promote the disruption of cellular membranes.
Clari\cation of Lysates
After cell lysis and inactivation of endogenous nu-
cleases, cellular debris, proteins and other precipi-
tated molecules must be removed. This is usually
achieved by means of centrifugation or Rltration
steps. During this operation, certain amounts of NAs
can be lost in the liquid entrained in the solid phase.
Extensive removal of water from the solid phase can
be performed to increase the yield, but this operation
usually also increases the amount of impurities. Min-
imum shear should be exerted in order to preserve the
integrity of the target NA. The solution obtained after
clariRcation is more or less rich in several impurities.
Small molecules such as nucleotides, nucleosides,
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
amino acids, sugars and inorganic ions are easily
removed, but DNA, RNA, polysaccharides and pro-
teins are difRcult to remove because these macro-
molecules have some common physical and chemical
characteristics. The impurity proRle and amount in
a clariRed lysate depend not only on the cell type, but
also on factors such as phase growth and growth
conditions (carbon and nitrogen sources, media rich-
ness, dissolved CO2 and O2, pH, etc), as well as on the
method used for disruption (Table 2).
Puri\cation
After the primary isolation steps, most impurities in
solution are comprised of NAs from the source organ-
ism, denatured forms of the target NA, proteins and
endotoxins (Gram-negative bacteria). The similarities
between these molecules and their wide molecular
weight range make puriRcation difRcult. Several
methods are available to purify NAs, depending on
the amount, yield and purity needed, and also on the
availability of methods and associated costs.
Gradient Centrifugation
A classical method for separating DNA or RNA from
each other, and from proteins and polysaccharides,
uses equilibrium-density gradient (or isopycnic) cen-
trifugation, which is based on differences in par-
ticle density. The density gradients can be self-form-
ing or pre-formed continuous gradients, prepared
with caesium salts (e.g. CsCl, Cs2SO4) or iodinated
compounds (e.g. metrizoate, metrizamide, nycodenz).
Equilibrium-density gradient centrifugation is one of
the most efRcient methods available for the puri-
Rcation of NAs. The major drawbacks are its depend-
ence on high cost reagents and equipment (the ultra-
centrifuge) and the fact that it is a time-consuming
operation (typically at least 48 h of centrifugation).
When using this method, it should be kept in mind
that centrifugal conditions may affect the native
characteristics of the sample. For instance, the pres-
ence of endogenous and/or exogenous nucleases dur-
ing a long-term process can lead to NA degradation.
In addition, the chemical constituents of the solutions
can affect the integrity of NAs. Physical degrada-
tion as a result of shear stress should be avoided, too.
The position of NAs in the gradient is usually
located by measuring the absorbance at 260 nm.
However, in a CsCl isopycnic separation, the DNA
molecules band at approximately the same density.
An alternative approach in this case is to band DNA
in CsCl gradients in the presence of ethidium bromide
or propidium iodide, which intercalate differen-
tially between the bases of DNA, resulting in dif-
4563
ferent molecule buoyant densities. This technique
allows the separation of supercoiled plasmids, nicked
or relaxed plasmid forms, genomic DNA, RNA and
proteins.
Liquid^Liquid Extraction
Solvent extraction is often used to remove proteins
and lipids from NAs. The pH values of the extractant
organic phase and of the starting aqueous phase are
very important because partition coefRcients of
NAs are pH-dependent. A classical system for solvent
extraction uses a sequential extraction with phenol :
chloroform (1 : 1). Although phenol is an efRcient
denaturant of proteins, it does not completely inacti-
vate RNases, and it solubilizes RNA with long
poly(A)# tails. These problems can be partially over-
come by introducing isoamyl alcohol in the phenol/
chloroform mixture.
In spite of the efRciency of solvent extraction,
residual contaminants still remain in solution. Fur-
thermore, solvents like phenol and chloroform are
extremely toxic for both the operator and the envi-
ronment. Even small amounts in the Rnal NA prep-
aration are potentially toxic to live recipient cells and
can interfere in downstream experiments. This makes
solvent extraction an unacceptable method when
preparing NA for therapeutic use.
Extraction of NAs has also been performed with
aqueous two-phase systems. This technique relies on
the fact that concentrated aqueous solutions of poly-
saccharides, such as dextran and polyethylene glycol
(PEG), are immiscible. Many biological components
(polymers, cells, cell organelles) including NAs show
different solubility in the two phases formed, and
will therefore separate by partitioning. By manipula-
ting the system conditions (e.g. buffer, polymer
and salt concentration) it may be possible to separate
DNA from RNA, native from denatured DNA and
single from double or triple strands. The technique,
however, has not been studied in depth, and therefore
is not commonly used.
Precipitation
Precipitation with ethanol or isopropanol is com-
monly used to concentrate NAs. Typically, 2 vol of
ethanol or 0.7 (v/v) isopropanol is added to the NA
solution. The process is more efRcient if per-
formed in the presence of moderate concentrations of
monovalent cations and at low temperatures (below
43C). The concentration and type of salt (cation:
ammonium, lithium, sodium, potassium; anion; acet-
ate, chloride) used in precipitation should be opti-
mized for the target NA. The duration and speed of
the centrifugation step used to recover the precipitate
4564 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
material are also important, and should be adjusted
according to the concentration of NA (longer and
faster centrifugation for lower concentrations). If the
amount of target NA is low, an inert carrier such as
glycogen can be added to the mixture to increase
precipitation efRciency. After draining liquid
from the precipitated material, an appropriate buf-
fer is added to redissolve the NAs.
Another method for precipitation of NA uses
PEG/salt (NaCl, MgCl2) systems. The method is
based on the fact that the size of the DNA molecule
precipitated by PEG is dependent on the concentra-
tion and molecular weight of PEG. It can thus be used
either to fractionate DNA according to molecular
mass, or simply to precipitate the total DNA content.
The method is rapid and inexpensive but consistent
yields are difRcult to obtain.
The precipitation methodology, although generally
used to concentrate NAs, is also effective as
a puriRcation step. Selective precipitation could be
achieved, for instance, with high concentrations of
different salts or changes in pH to precipitate
proteins or nontarget DNA or RNA. For example,
ammonium sulfate is often used to precipitate con-
taminating proteins. Lithium chloride precipitation
has also been used in the preparation of large RNA. It
takes advantage of the fact that small RNAs (tRNAs
and 5S RNA) are soluble in solutions of high ionic
strength, whereas large RNAs (rRNAs and mRNAs)
are insoluble and precipitate out. After centrifu-
gation, the high molecular weight RNA is redissolved
in water or buffer.
Gel Electrophoresis
Gel electrophoresis on agarose or polyacrylamide gels
is a powerful technique commonly used to separate,
identify and purify NAs. Under the effect of an
electrical Reld, NA molecules migrate through the gel
matrix at a rate that is inversely proportional to their
size. The resolving power of gel electrophoresis is
extremely high, enabling the separation of molecules
that differ in size by as little as 1 bp. The position
of the individual molecules in the gel can be deter-
mined by staining with a Suorescent intercalating dye
such as ethidium bromide. By using a number of
techniques, such as electroelution or low-melting
agarose gels, it is possible to recover DNA of high
purity from the gels. However, the scale associated
with electrophoresis is generally small.
Chromatography
Chromatography has been increasingly used to purify
NAs. The technique is powerful, rapid and simple to
perform, while avoiding the use of toxic compounds
common to competing technologies like ultracen-
trifugation and phenol/chloroform extraction.
Disposable column cartridges packed with par-
ticles of a stationary phase, and operating in the
gravity Sow format, often constitute an essential part
of commercial isolation kits for lab-scale applica-
tions. In many cases, centrifugation is combined with
disposable spin columns to enhance and speed the
isolation. High-throughput and automated formats
using these column cartridges are increasingly avail-
able that enable the processing of several samples at
the same time, minimize hands-on preparation time,
and reduce the risk of sample mix-ups. PuriRcation
using high pressure liquid chromatography (HPLC)
has also been performed as a preparative step to
isolate small quantities of NAs. In large scale applica-
tions, column chromatography is a central process
step. For large scale applications, economic con-
straints demand regeneration and re-use of the expen-
sive chromatographic media. Furthermore, if mater-
ial is being produced for clinical use, validation of the
cleaning is indispensable.
Different types of chromatography, such as gel
Rltration, ion exchange, hydrophobic interaction, re-
versed-phase, adsorption and afRnity, have been
used for the puriRcation of NAs (Table 3).
Except in the case of gel Rltration, the mode of
operation of the chromatographic columns, whether
small or large, is similar. Two possibilities exist:
1. Interaction of the target NA with the chromato-
graphic support: column feed, binding of the tar-
get NA to the stationary phase, removal of impu-
rities by washing and selective elution and, Rnally,
elution of the target NA (Figure 2A).
2. No interaction of the target NA with the
chromatographic support: column feed, collection
of the target nucleic in the Sow-through, binding
of impurities to the stationary phase and, Rnally,
column disposal or removal of the impurities by
selective elution (Figure 2B).
A chromatographic column can also be used in
conjunction with an enzyme process in order to im-
prove the selectivity of the method. For instance,
DNase (RNase) treatment of bound NAs will remove
DNA (RNA), while leaving RNA (DNA) behind.
PuriRcation by anion exchange takes advantage of
the interaction between negatively charged phosphate
groups in the DNA or RNA backbone and positively
charged ligands on the surface of the particles that
constitute the stationary phase. A salt gradient is
usually used to displace the different NA species,
which in principle should elute in order of increasing
overall net charge, which in turn is a function of chain
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
Table 3 Purification of nucleic acids by chromatography
Chromatography Basis of separation
Anion exchange Charge, charge density
Hydrophobic interaction Hydrophobic interaction
Reversed-phase Hydrophobic interaction
Adsorption Selective adsorption
Affinity Structure recognition
Gel filtration Size, shape
length. When column cartridges are used, the salt
gradient is always a step for convenience, while in
process applications linear gradients can improve sel-
ectivity.
Anion exchange chromatography can also be used
as a nonsize-based NA puriRcation tool. In some
cases, base sequence and composition affect the
elution pattern of NAs in anion exchangers. The
shape and size of the molecules may also play an
important role. This is the case, for example, in the
puriRcation of plasmid variants. In some anion ex-
changers, the more compact supercoiled plasmid
forms, which have a higher charge density, elute later
than the open circular forms, which have a lower
overall charge density.
Since many of the NAs being isolated are normally
very large molecules, their binding to most of the
existent anion exchangers is likely to occur only at the
surface. This constitutes an important capacity lim-
itation, especially in large scale applications. Dif-
ferent types of stationary phases have been used for
anion exchange chromatography of NAs. Typical
examples include weak ligands such as diethyl-
aminoethyl and dimethylamino coupled to silica,
polymeric or composite (inorganic#polymeric) ma-
trices and strong ligands such as quaternary amines
coupled to polymeric matrices.
Hydrophobic interaction chromatography has not
been described for the puriRcation of NAs, but recent
results indicate that matrices derivatized with mildly
hydrophobic residues are able to separate double-
stranded DNA from RNA and single-stranded DNA
under nondenaturing conditions.
4565
Examples and applications
Plasmid purification
Plasmid copy number analysis
Fractionation of restriction fragments
Separation of polymerase chain reaction products
Separation of dsDNA from ssDNA and RNA
Fractionation of supercoiled and relaxed plasmids
Separation of dsDNA from ssDNA and RNA
Separation of supercoiled and relaxed plasmid
Purification of oligonucleotides (chemical synthesis)
Capture of nucleic acids by silica (glass powder, diatomaceous earth)
Separation of dsDNA from ssDNA by hydroxyapatite
Separation of RNA from DNA with boronic acids
Triple helix formation
Hybridization of poly(A)# tails to oligo(dT) ligands
Plasmid purification with acridine dye ligands
Buffer exchange
Salt and oligonucleotide removal
Fractionation of supercoiled and relaxed plasmids
Removal of endotoxins
In reversed-phase chromatography, NAs are also
retained by the hydrophobic interaction of the bases
with the chromatographic resin, but here, the density
of ligands is much higher. Separation of single- and
double-stranded DNA has been accomplished in
C18 columns and differentiation between DNA
and RNA due to the effect of ribose and de-
oxyribose has also been reported. In reversed-phase
chromatography, the binding is stronger than in
hydrophobic interaction chromatography due to
the higher ligand density. This requires elution
to be carried out under more severe conditions,
for instance, by using eluents with organic solvents,
which can have a deleterious effect on the struc-
ture of the target molecule. For this reason, reversed-
phase chromatography is more suited for the puriRca-
tion of smaller NAs that are less prone to denatura-
tion. This is the case with synthetic oligonucleotides,
used as primers or as blocking agents in antisense
technologies, that are synthesized by solid-phase
chemistry and puriRed by reversed-phase chromato-
graphy.
In adsorption chromatography, a stationary phase
is used that selectively binds NAs in the presence of
chaotropic salts, which remove water from hydrated
molecules in solution, ensuring separation from
complex biological mixtures. For instance, NAs
adsorb to silica (glass) in the presence of sodium
iodide or guanidinium hydrochloride, and to hy-
droxyapatite in the presence of urea. Polysaccharides
and proteins do not adsorb and are removed by wash-
ing. Elution is performed using a low salt buffer.
Hydroxyapatite further displays an ability to separate
4566 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS

Figure 2 Mode of operation of chromatographic columns in nucleic acid purification: (A) interaction of the target nucleic acid with the
chromatographic support; (B) no interaction of the target nucleic acid with the chromatographic support.

single-stranded NAs that bind less tightly from
double-stranded NAs.
A method for the selective adsorption of RNA in
the presence of DNA employs immobilized boronic
acid derivatives which, above their pKa values, form
cyclic complexes with vicinal diols. Since the
deoxyribose sugars lack the vicinal diol of ribose
sugars in RNA, DNA is poorly adsorbed and can be
washed away easily.
AfRnity chromatography is based on the recog-
nition of a particular structure in the target NA mol-
ecule by an immobilized ligand. In triple-helix af-
Rnity chromatography, the formation of triple helices
between oligonucleotides linked to a chromato-
graphic matrix and duplex sequences present on
dsDNA is explored. The afRnity of poly(A)# tails
of mRNA to oligo(dT) probes has also been explored
as a way of capturing and purifying mRNA. The
combination of afRnity ligands with magnetic
particles is another recent development that avoids
the use of packed columns. It allows the binding of
the target molecule directly from solution, with the
particle complex being recovered with a magnet. The
high selectivity of afRnity chromatography makes
it a powerful tool for the one-step puriRcation of
NAs. However, since each ligand targets a speciRc
base sequence, the versatility of the technique is low
and the associated cost high.
Size exclusion or gel Rltration chromatography has
been used to fractionate NA molecules on the basis of
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF NUCLEIC ACIDS
their relative size. By selecting a gel with an appropri-
ate selectivity for the size range in question it
is possible, for example, to isolate supercoiled plas-
mid DNA from microbial contaminants such as
genomic DNA, RNA, proteins and endotoxins.
Another feature of size exclusion chromatography,
often explored when purifying NAs, is the possi-
bility of exchanging buffers and removing
salts, nucleotides, excess primers and other small
molecules. This type of column is commercially avail-
able in the cartridge format for the clean-up of NA
solutions.
Quality Control of Final Nucleic Acid
The Rnal NA quality criteria will vary with the NA
inherent complexity, its intended use and the method
and complexity of manufacture. The researcher
and/or producer will select several quality control
tests that depend on their own application. NAs can
be quantitatively and qualitatively analysed by
a number of chemical, biochemical and physical as-
says. Some quality controls of NA preparations are
summarized in Table 4.
A method used widely to estimate the amount of
RNA or DNA is spectrophotometric analysis. The
absorbance at 260 nm is relatively accurate (depend-
ing on base composition) and reproducible when ap-
plied to puriRed samples without signiRcant amounts
of contaminants (e.g. proteins, other NAs, phenol),
and to moderately diluted or concentrated prepara-
Table 4 Quality control tests for DNA or RNA preparations
Test Method
Appearance Visual inspection
Identity Restriction enzyme analysis
Gel and capillary electrophoresis
Sequencing
Concentration Spectrophotometry A260 nm
HPLC
Gel and capillary electrophoresis
Other nucleic acids Gel and capillary electrophoresis
DNA or RNA hybridization
HPLC
Proteins Colorimetric assay (e.g. BCA)
Sodium dodecyl sulfate-polyacrylamide
gel electrophoresis
Immunoassays
Polysaccharides Specific assays (e.g. HPLC, enzymatic
Lipopolysaccharides assays, LAL)
Nucleases Nuclease-specific assays
Sterility Cytopathic effects
Reverse transcriptase assay
Electron microscopy
Bioburden assay
4567
tions. Spectrophotometric scans between 220 and
320 nm are also used to detect salt and organic con-
taminations. The purity of NAs based on the absorb-
ance ratio 260 nm/280 nm is commonly used.
A 260 nm/280 nm ratio between 1.8 and 2.0 is usu-
ally considered to be a good estimation of purity. This
method was initially developed to quantify the con-
taminating NAs in protein preparations. For this rea-
son, it often fails when used for NA purity assess-
ment. Therefore, when a high level of purity is criti-
cal, care should be taken while using this method, and
it should be complemented by other purity analysis.
HPLC, capillary electrophoresis or methods using
speciRc Suorescent dyes, quantitative DNA or RNA
hybridization and quantitative polymerase chain re-
action are preferable.
Conclusions
Isolation/puriRcation of nucleic acids is becoming
more and more important with the emergence of
areas such as genomics, gene therapy and DNA vacci-
nation and the growing importance of clinical diag-
nostics and forensics. The general strategies and tech-
nologies commonly used in the puriRcation and isola-
tion of NAs have been reviewed in this article.
Critical issues and bottlenecks that still hamper the
puriRcation of nucleic acids have also been high-
lighted. Although the future will certainly bring
more efRcient isolation/puriRcation methodolo-
gies, designed for high-throughput and automated
preparation/analysis, the core technologies and strat-
egies described here will most likely retain their
importance.
Further Reading
Harwood AJ (1996) Basic DNA and RNA Protocols.
Methods in Molecular Biology, vol. 58. New Jersey:
Humana Press.
Muller W (1985) Partitioning of nucleic acids. In: Walter
H, Brooks DE and Fisher D (eds) Partitioning in Aque-
ous Two-Phase Systems: Theory, Methods, Uses and
Application to Biotechnology, pp. 227}266. Orlando,
FL: Academic Press.
Prazeres DMF, Ferreira GNM et al. (1999) Large scale
production of pharmaceutical grade plasmid DNA for
gene therapy: problems and bottlenecks. Trends Bio-
technology 17: 169}174.
Rickwood D (1984) Centrifugation, A Practical Approach,
2nd edn. Oxford: IRL Press.
Sambrook J, Fritsch EF and Maniatis T (1989) Molecular
Cloning, 2nd edn., Cold Spring Harbor, FL: CSH Labor-
atory Press.
Sinden RR (1994) DNA Structure and Function. San Diego,
CA: Academic Press.
4568 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES

Sofer G and Hagel L (1997) Handbook of Process
Chromatography: A Guide to Optimization, Scale-up,
and Validation. San Diego, CA: Academic Press.
Thompson JA (1986) A review of high-performance liquid
chromatography in nucleic acids research III. Isolation,
puriRcation, and analysis of supercoiled plasmid DNA.
BioChromatography, 1: 68}80.
US FDA (1991) Points to Consider in Human Somatic
Cell Therapy and Gene Therapy. Rockville, MD:
FDA.

ESSENTIAL GUIDES FOR ISOLATION/

PURIFICATION OF POLYSACCHARIDES
R.-C. Sun and J. Tomkinson,
University College of Wales, Bangor, UK
Copyright ^ 2000 Academic Press
Introduction
The isolation of polysaccharides from biological
sources represents an important source of these valu-
able materials. Biomass such as cereal straws and
grasses, are an enormous underutilized energy re-
source as raw materials in the production of paper,
panel products, chemicals and other industrial prod-
ucts. On a dry-weight basis, the straws and grasses
contain 65}85% of polysaccharides, with hemicel-
luloses ranked second to cellulose in abundance;
however, it must be noted that the chemical con-
tent of the hemicellulose, with respect to saccharide
ratios changes with plant growth and maturity
(Table 1).
At the present time, there is widespread interest in
the use of hemicelluloses, particularly arabinoxylan-
rich hemicelluloses as precursors in food gums. More
recently, water-soluble xylans from corn cobs have
shown biological activity as immuno-modulating
compounds. Other potential industrial applications
of hemicelluloses are to be found in the Relds of
adhesives, thickeners in foods, stabilizers, biodegrad-
able Rlm formers and emulsiRers. They can also be
easily converted to primary chemicals such as xylose,
xylitol, furfural, hydroxymethylfurfural and levulinic
acid.
Hemicelluloses, however, are the most complex
components in the cell wall of straws and grasses.
They form hydrogen bonds with cellulose, covalent
bonds (mainly -benzyl ether linkages) with lignins
and ester linkages with acetyl units and hydroxy-
cinnamic acids. To investigate the potential utiliz-
ations of polysaccharides from straws and grasses,
a thorough study of the isolation procedures is
necessary. Details of the method are reviewed as
follows.
Cell Wall Preparation
The straw or grass is cut into 1}2-cm lengths, air-
dried, and ground to pass through a 0.5}0.8-mm
screen. The ground sample is then further dried in
a cabinet oven with air circulation at 50}603C for
12}16 h. Dried material is dewaxed by reSuxing with
toluene}EtOH (2 : 1, v/v) or defatted with chloro-
form}methanol (2 : 1, v/v) for 6 h in a Soxhlet appar-
atus. The dewaxed sample is treated with -amylase
to degrade the starch or extracted with phenol}acetic

Table 1 Chemical composition of agricultural residues (per cent dry matter)

Species Water-solubles Cellulose Hemicelluloses Lignin Extract Ash

Wheat straw 4.7 38.6 32.6 14.1 1.7 5.9

Rice straw 6.1 36.5 27.7 12.3 3.8 13.3
Rye straw 4.1 37.9 32.8 17.6 2.0 3.0
Barley straw 6.8 34.8 27.9 14.6 1.9 5.7
Oat straw 4.6 38.5 31.7 16.8 2.2 6.1
Maize stems 5.6 38.5 28.0 15.0 3.6 4.2
Corn cobs 4.2 43.2 31.8 14.6 3.9 2.2
Esparto 6.1 35.8 28.7 17.8 3.4 6.5
Sugar beet pulp (pectin 27.1) 5.9 18.4 14.8 5.9 1.4 3.7
Bagasse 4.0 39.2 28.7 19.4 1.6 5.1
Oil palm fibre 5.0 40.2 32.1 18.7 0.5 3.4
Abaca fibre 3.7 60.4 20.8 12.4 0.8 2.5
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES 4569

oven at 603C for 16 h. Water-soluble polysaccharides

are isolated by precipitation of the concentrated
supernatant with 4 volumes of 95% ethanol and
recovered by centrifugation. The resultant solid is
subsequently puriRed by extensive washing with 75%
ethanol and then freeze-dried. As shown in Table 1,
water-soluble polysaccharides obtained from various
straws and grasses account for 3.7}6.8% of the dry
matter and have much lower weight-average molecu-
lar weights (MM w"&8000 g mol\1) when compared
to those of hemicellulosic fractions. Neutral sugar
analysis shows that arabinose and galactose are the
major sugar constituents, whereas xylose and glucose
are present in only small amounts in this fraction.
Detailed studies of water-soluble polysaccharides
from rapeseed meal, sorghum stalk, and dehulled
legume seeds and their hulls have been reported.

Pectic Polysaccharides
The pectic polysaccharides are restricted to the pri-
mary cell wall and middle lamella of higher plant
tissues and growth zones. They are most abundant in
Figure 1 Scheme for procedures in cell wall preparation from
straw or grass. soft tissues, such as rinds of citrus fruit (&30%),
sugar beet pulp (&25%), and apple peels (&15%),
but are present in only small proportions in woody
acid}water (2 : 1 : 1) to remove proteins. The pro-
teins can also be extracted from the residue with
sodium dodecyl sulfate solution containing 10 mM
1,4-dithiothreitol at room temperature for 3 h. To
degrade the proteins, proteolysis is started by addi-
tion of proteases at 373C for 4}6 h in 0.1 M sodium
phosphate buffer, pH 7.5, containing 0.02}
0.05% sodium azide as bactericide. The cell walls are
recovered by Rltration, extensively washed with
water, and then treated with 80% ethanol to release
the ethanol}water-soluble components. These consist
mainly of free phenols and ethanol-soluble lignins
together with small amounts of low-molecular-
weight polysaccharides (&10% w/w of ethanol ex-
tract). After Rltration, the cell wall preparations are
dried by solvent exchange through ethanol and di-
ethyl ether. Figure 1 summarizes the procedures for
cell wall preparation.

Fractional Extraction of Cell Wall

Polysaccharides
Water-Soluble Polysaccharides
The scheme for fractional extraction of cell wall poly-
saccharides is illustrated in Figure 2. The prepared
cell walls are stirred with distilled water at 75}803C
for 2 h. The residue is Rltered off on a nylon Figure 2 Scheme for fractional extraction of polysaccharides
cloth, washed with ethanol and ether, and dried in an from straw or grass.
4570 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES
tissue such as straw and grass (0.5}1.5% dry matter).
The term pectins or pectic substances is associated ides (1.1% of dry matter) can be obtained from wheat
with acidic polysaccharides consisting of a backbone
of mainly (1P4)-
-bound D-galacturonic acid resi-
dues interrupted by the insertion of
-linked L-rham-
nose residues. Other constituent sugars such as L-
arabinose, D-galactose, D-xylose, L-fucose, and traces
of 2-O-methyl-D-xylose and 2-O-methyl-L-fucose are
attached in the side chains. The pectins can be extrac-
ted with dilute acid, e.g. HCl solution at pH 1.5
or with chelating agents such as 0.2% aqueous
ammonium oxalate, disodium ethylenediaminetet-
raacetic acid (EDTA), and sodium hexametaphos-
phate (SHP) solution at pH 3}4 for 4 h at 853C. After
extraction, the Rltrates are adjusted to pH 4.0 with
1 M NaOH. The pectic polysaccharides are isolated
by precipitation of the Rltrates in 4 volumes of 95%
ethanol and recovered by centrifugation. Crude pec-
tins are extensively washed with 70% acidiRed
ethanol and fractionated by ion exchange chromato-
graphy on a column (550 mm;15 mm) of DEAE
Sepharose Fast Flow (Pharmacia, Sweden), initially
equilibrated in 0.005 M NaAc}buffer pH 5.0 or
on a QAE Sephadex A-25 column (80;1.5 cm i.d.;
Pharmacia, Sweden) equilibrated with 10 mM
imidazole-HCl buffer (pH 7.0). Other columns
used to fractionate the pectic polysaccharides include
a DEAE-Sepharose CL6B column (23;5 cm; Phar-
macia, Sweden) equilibrated with 0.005 M sodium
succinate buffer at pH 4.8, and a DEAE-
Sephadex A-50 column (20;2 cm; Pharmacia,
Sweden) previously equilibrated with 10 mM potassi-
um phosphate buffer (pH 6.0). The neutral and
acidic pectic polymers are hydrolysed with 2 M triSu-
oroacetic acid at 1203C for 2 h in sealed ampoules or
with pectinase. In comparison, treatment by pec-
tinase has a signiRcant effect on the hydrolysis of
the pectic polymers, while the reverse trend is ob-
served during the treatment of the neutral pectic poly-
saccharides by acid hydrolysis, since galacturonic
acid residues in polysaccharides are known to be
resistant to acid hydrolysis. For example, endo-1,4-
-
polygalacturonase, pectin and pectate lyases can sub-
stantially cleave the
-(1P4)-GalpA glycosidic
bonds between contiguous galacturonic residues. Pre-
vious studies have shown that extraction of the cell
wall preparation of sugar beet pulp with water fol-
lowed by treatment with HCl at pH 1.5, 0.2% am-
monium oxalate, 0.2% EDTA, and 0.2% SHP at pH
3.3 for 4 h at 853C yielded 32.1%, 26.1%, 29.3%,
and 30.0% (percent dry cell wall preparation) of
pectic polysaccharides, respectively. An optimum ex-
traction procedure was found to be treatment with
dilute HCl at pH 1.5 for 4 h at 853C, in which the
pectin contained 84.6% anhydrogalacturonic acid
and only 2.7% ash. Xylose-rich pectic polysacchar-
straw by extraction with dilute HCl at pH 1.6 for 4 h
at 853C, which contain 44.8% galacturonic acid (re-
leased by pectolyase), and 28.4% neutral sugars (re-
leased by acid hydrolysis).
Hemicelluloses
The cell wall components that are readily hydrolysed
by hot dilute mineral acids, or dissolved by hot dilute
alkalis or cold 5% sodium hydroxide solutions have
been termed hemicelluloses. Hemicelluloses belong
to a group of heterogeneous polysaccharides which
are formed through biosynthetic routes different
to the glucose}UDP route of cellulose (a homo-
polysaccharide). Hemicelluloses of Gramineae such
as cereal straws have a backbone of (1P4)-linked
-D-xylpyranosyl units. The chain may be linear, but
is often branched and usually has other glycosidically
bound sugar units. Some xylan chains have D-gluco-
pyranosyluronic acid units attached, but the most
important acidic hemicelluloses are O-acetyl-4-O-
methyl-D-glucuronoxylans and L-arabino(4-O-
methyl-D-glucurono)xylans. In dicots, where the
hemicelluloses are mostly xyloglucan, hydroxypro-
line-rich glycoproteins also comprise a substantial
amount of the cell wall and cross-link the carbohy-
drate polymers to form a rigid matrix. In cereal
straws of grasses, these proteins have been replaced
by esteriRed and etheriRed phenolic compounds. The
xylans present in the hemicellulosic backbone can
be substituted by arabinose, galactose, glucuronic
acid, and methyl glucuronic acid. The arabinoxylan
can be isolated directly from fully ligniRed straws
or grasses by extraction with aqueous potassium
hydroxide or sodium hydroxide. Usually 80}95% of
the total xylan present in straw and grass contains
a relatively high percentage of associated lignin
(5}10%). This high lignin percentage considerably
darkens the polysaccharide limiting their industrial
application.
Xylans undergo only partial hydrolysis in alkaline
solution at room temperature under an atmosphere of
nitrogen. The product obtained } except for the fact
that acetyl-, feruloyl-, and p-coumaroyl appendices
have been partially or completely saponiRed } is quite
similar to the native polysaccharide. The hemicel-
luloses which have a light brown colour contain
a relatively small amount of bound lignin (1}2%),
and can be quantitatively isolated from the holocel-
lulose by extraction with aqueous alkali. DeligniR-
cation of the depectinated cell wall preparations
obtained from straws or grasses has been performed
with sodium chlorite at 753C for 2 h in acidic solution
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES
(pH 4.2}4.7) adjusted by 10% acetic acid. The
acetylated xylans are soluble in water and in solvents
such as dimethylsulfoxide (DMSO), formamide, and
N,N-dimethylformamide. Although only a part of the
xylan can be extracted, the advantage is that no
chemical changes take place. Aqueous solutions of
potassium and sodium hydroxide are mostly used as
the alkaline solvent of choice for the extraction of
hemicelluloses. The preferred hydroxide is potassi-
um, mainly because the subsequent potassium acetate
formed during the neutralization is more soluble in
the alcohol used for precipitation than is sodium
acetate. Comparison of the extraction ability of 1 M
solutions of potassium, sodium, and lithium hydrox-
ide on wheat straw, shows an approximately equal
effect in the rate and yield of solubilization of
hemicellulose. However, it has been found that so-
dium hydroxide and lithium hydroxide are more
powerful than potassium hydroxide in removing
hemicelluloses, especially mannans from wood sam-
ples. The yield of hemicelluloses obtained using cal-
cium hydroxide and liquid ammonia was markedly
lower than for the respective alkali metal hydroxide
solutions. Liquid ammonia, in general, can be used
for pre-swelling prior to an alkaline extraction.
Moreover, the yield of hemicelluloses strongly de-
pends on a number of important factors, e.g. type of
alkali, concentration, temperature and time of extrac-
tion. Addition of sodium borate to the alkali facili-
tates the dissolution of galactoglucomannans and
glucomannans. However, any ester groups present
are simultaneously saponiRed during the alkali ex-
tractions.
Hemicelluloses obtained by alkali can be subfrac-
tionated into hemicellulose A, B, and C. Hemicel-
lulose A, the more linear and less acidic fraction, is
isolated from the supernatant by acidifying to pH 5.0
with acetic acid followed by centrifugation. Hemicel-
lulose B, the more acidic or branched fraction, is
obtained from the mother liquor by precipitation
with 4 volumes of 95% ethanol, then Rltered and
Table 2 The yield and neutral sugar composition as well as content of uronic acids of hemicellulosic subfractions of DMSO-solubles,
A, B, and C extracted sequentially by DMSO at 803C for 2 h and 10% KOH}2% H3BO3 at 253C for 16 h from wheat straw holocellulose
Hemicellulosic Yield (%)* Neutral sugar composition (%)**
subfractions
Ara Xyl
DMSO-solubles 4.8 8.5 68.1
A 7.2 5.2 86.3
B 18.5 11.0 70.0
C 2.7 12.2 80.3
*Percent dry matter (w/w).
**Percent hemicellulosic subfractions (w/w).
***ND"not detectable.
4571
washed with 70% ethanol. The resultant solid is
redissolved in water (after the residual salts were
dialysed against water until free from salts), and
the hemicellulose B recovered by evaporation under
reduced pressure at 453C or by lyophilization.
The fraction that remains soluble in aqueous ethanol
is named hemicellulosic fraction C and is isolated
by dialysis with water and ethanol until free from
salts (Figure 2). The yield and neutral sugar com-
position as well as content of uronic acids of hemi-
cellulosic subfractions of DMSO-solubles, A, B,
and C extracted sequentially by DMSO at 803C
for 2 h and 10% KOH}2% H3BO3 at 253C for
16 h from wheat straw holocellulose, are given in
Table 2.
Precipitation of the polysaccharide fractions by ad-
dition of miscible organic solvents to aqueous solu-
tions is one of the main methods of recovery and
puriRcation. Ethanol is the solvent most commonly
used, but methanol, acetone, and other organic sol-
vents have also been applied for fractionation of
hemicelluloses. In ethanol}water (80 : 20, v/v) the
major portion of the polysaccharides are precipitated
and only a small amount of short-chain material is
left in solution. In addition to the neutral organic
solvents, some more speciRc precipitation agents are
also known, e.g. barium hydroxide for glucomannans
and cetyltrimethylammonium bromide or hydroxide
for glucuronoxylans. Fehlings solution or other cop-
per salts can be used for precipitation of both
glucomannans and glucuronoxylans. Hemicellulosic
complexes precipitated by iodine in calcium chloride
appear to be relatively unsubstituted by non-xylose
residues, whereas the material remaining in solu-
tion is more highly substituted by such residues.
The methods used to fractionate the hemicelluloses
of straws and grasses are similar to those used
to fractionate hemicelluloses from woods. The
hemicelluloses may be fractionated as their acetates
by precipitation from solution by ammonium sul-
fate, or by chromatography on DEAE-cellulose. The
Uronic acids Acetyl content
(%)** (%)**
Man Gal Glc
Trace 3.8 6.5 5.2 7.3
0.4 2.4 2.4 2.6 ND***
1.2 5.0 5.3 6.8 ND
0.5 3.5 2.0 0.6 ND
4572 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES

precipitated hemicellulose preparations can if desired
be further puriRed by column chromatography.
Gel permeation chromatography is one of the most
useful tools for determining the average molecular
weights of the isolated hemicellulosic fractions. Ion
exchangers based on cellulose, dextran or agarose
such as diethylaminoethyl cellulose in different
ionic forms can also be used to separate hemicel-
luloses from each other. Chromatography (in its vari-
ous forms) is routinely used for the characterization
of the acidic hydrolysis products of isolated hemicel-
luloses. Dialysis of aqueous solutions often removes
inorganic salts and other low-molecular-weight im-
purities prior to chromatography. Alternatively, salt
may be removed by electrodialysis, by treatment of
solutions with ion-exchange resins, or by gel Rltra-
tion, using Sephadex, a cross-linked dextran column.
More recently, it has been reported that alkaline
peroxide is an effective agent for both deligniR-
cation and solubilization of hemicelluloses from
straws and grasses. This was Rrst proposed in 1984 in
studies on the alkaline peroxide deligniRcation of
agricultural residues to enhance enzymatic sacchariR-
cation. Hydrogen peroxide is widely used in the pulp
and paper industry to bleach lignin-rich pulps. It has
been generally accepted that the bleaching action of
hydrogen peroxide is attributable to the hydroperox-
ide ion (HOO\), formed in alkaline media, which is
the principal active species in hydrogen peroxide
bleaching systems. This anion is a strong nucleophile
that preferentially attacks ethylenic and carbonyl
groups present in lignin. On the other hand, hydrogen
peroxide is unstable in alkaline conditions and readily
decomposes, particularly in the presence of certain
transition metals such as manganese, iron, and cop-
per. This metal-catalysed decomposition of hydrogen
peroxide is undesirable in the bleaching operation.
However, this decomposition generates more active
radicals, such as hydroxyl radicals (HOz) and
z
superoxide anion radicals (O\ 2 ), participating in
degradation reaction of lignin and solubilization of
hemicelluloses, which therefore, results in signiRcant
solubility of the lignin and hemicelluloses. The ad-
vantages of hemicellulose extraction with alkaline
peroxides are low investment cost, accompanying
strong bleaching effect, lower biological and
chemical oxygen demand (BOD and COD) efSu-
ents as well as the recovery of the solubilized macro-
molecular hemicelluloses with a minimal degrada-
tion. Results show that more than 80% of the original
hemicelluloses and over 90% of the original lignin is
solubilized during the treatment of cereal straws such
as wheat, barley, rice, oat, and rye straw, and maize
stems with 2% H2O2 at 483C for 16 h at pH
12.0}12.5 (Table 3). These hemicellulose prepara-
tions are white in colour and contain very small
amounts of associated lignin (3}5%).
To gain maximum dissolution of the hemicel-
luloses, it is not necessary to continuously regulate
the reaction pH, even though over the course of the
treatment (483C, 16 h) the reaction pH rises from
12.0 to 12.5 and from 12.9 to 13.1, respectively. As
the reaction pH becomes more alkaline, increasing
amounts of hemicelluloses are solubilized, and the
yield of the residue decreases. Incremental increase of
the initial reaction pH from 11.5 to 12.5 results in an
increase of hemicellulose dissolution of about 20%.
During the initial stages of stirring, oxygen evolution
is active, and substantial frothing occurs, requiring
extractions to be conducted in vessels with volumes
two to three times those of the extraction mixtures.
After treatment, the cellulose-rich insoluble residue is
collected by Rltration, washed with distilled water
until the pH of the Rltrate is neutral, and then dried at
603C. The supernatant Suid is adjusted to pH 5.5
with 10% HCl and then concentrated. The sol-
ubilized hemicelluloses are precipitated by pouring
the concentrated supernatant into 4 volumes of

Table 3 The yield, neutral sugar composition, and content of uronic acid and lignin of hemicelluloses obtained by treatment of wheat,
rice, rye, barley, and oat straw, and maize stems respectively with 2% H2O2 at 483C for 16 h at pH 12.2

Cereal Yield (%)* Neutral sugar composition (%)** Uronic acids Lignin content
straw/stems (%)** (%)**
Rha Fuc Ara Xyl Man Glc Gal

Wheat 29.6 0.8 ND*** 13.8 60.5 0.4 9.8 4.5 4.9 5.1
Rice 22.3 0.6 0.3 14.9 56.3 ND 22.3 4.8 4.3 4.7
Rye 26.6 0.5 ND 11.2 65.3 0.3 6.1 3.3 8.1 4.8
Oat 25.6 0.4 ND 10.8 68.3 0.3 6.4 3.6 5.5 4.3
Barley 23.3 0.4 ND 10.6 66.1 0.5 7.6 3.7 5.8 4.5
Maize 22.7 0.5 0.3 15.2 62.7 0.8 6.3 4.7 5.3 3.7

*Percent dry matter (w/w).

**Percent hemicelluloses (w/w).
***ND"not detectable.
 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES
Figure 3 FT-IR spectra of 2% aqueous hydrogen peroxide-
soluble hemicellulosic preparations extracted with 2% H2O2 at
503C for 12 h at pH 11.5 from wheat straw (spectrum a), rye
straw (spectrum b), maize stem (spectrum c), and rice straw
(spectrum d).
ethanol, from which they settle out as a white Soccu-
lent precipitate which is freeze-dried. The Fourier
transform infrared (FT-IR) spectra of 2% aqueous
hydrogen peroxide-soluble hemicellulosic prepara-
tions extracted with 2% H2O2 at 503C for 12 h at pH
11.5 are shown in Figure 3: wheat straw (spectrum
a), rye straw (spectrum b), maize stem (spectrum c),
and rice straw (spectrum d). Figure 4 illustrates the
13
C-NMR spectrum of the hemicelluloses extracted
with 2% H2O2 at 453C for 12 h at pH 12.0 from
maize stems. Obviously, both FT-IR spectra and 13C-
NMR spectrum of the hemicelluloses appear to be
those of the typical hemicelluloses such as xylan from
cereal straws and grasses. These observations reveal
Figure 4 13C-NMR spectrum of the hemicelluloses extracted
with 2% H2O2 at 453C for 12 h at pH 12.0 from maize stems.
4573
that the alkaline peroxide treatments under the condi-
tions given do not affect the overall structure of
the macromolecular hemicelluloses. This effec-
tive and convenient method of alkaline peroxide
treatment may be used for most isolation of hemicel-
luloses from straws and grasses.
Steam treatment enables the lignocellulosic mater-
ials present in straws and grasses to be separated into
hemicelluloses, lignin, and cellulose in reasonable
yields and purity. Steam-explosion treatments or
other steam treatments have the advantage of using
a widely-available solvent without signiRcant cost or
environmental impact. It is generally agreed that the
steaming process is basically an acid-catalysed
autohydrolysis due to the small amounts of acetic
acid liberated early in the process through cleavage of
the hemicellulosic acetyl groups. A portion of the
starting material, mainly hemicellulose, is converted
into water-soluble products through this acid-
catalysed hydrolysis process. The cellulose is not sig-
niRcantly solubilized but undergoes a change in its
crystallinity or is partially depolymerized. During this
process, the lignocellulosic substrate is stream treated
at temperatures ranging from 140 to 2403C for
1}10 min followed by a rapid pressure release (i.e.
explosion) through a discharge valve. The hemicel-
luloses solubilized during the steam or steam-ex-
plosion process, are isolated after extraction in water.
They comprise mainly acetyl- and 4-O-methyl-
glucuronosyl-substituted xylans, and contain
15}25% and 5}10% bound lignin, respectively. The
associated lignins can be removed by treatment of the
crude hemicelluloses with 1 M NaOH at 203C for
2}4 h. All the hemicellulosic preparations have
a lower degree of polymerization (40}65), with
MM w ranging between 5000 and 10 000 g mol\1.
Cellulose
Cellulose is the main constituent of agricultural resi-
dues. Approximately 35}45% of the dry substance in
most straw and grass species is cellulose, located
predominantly in the secondary cell wall. The term

-cellulose is given to the residue remaining after
deligniRcation by sodium chlorite in acidic solution
and separation of hemicelluloses by extraction of the
holocellulose with 24% KOH (or 17.5% NaOH) at
253C for 2 h or 10% KOH (or 7.5% NaOH) at 253C
for 16 h. This term was originally coined for wood
cellulose which is insoluble in strong sodium hydrox-
ide solution. The portion which is soluble in the
alkaline medium but precipitated from the neu-
tralized solution was called -cellulose. -Cellulose
is the name for the portion which remains soluble
even in the neutralized solution. This method was
4574 APPENDIX 1 / ESSENTIAL GUIDES FOR ISOLATION/PURIFICATION OF POLYSACCHARIDES
modiRed in various ways and is now established as
the standard method for the determination of
-, -,
and -cellulose from straws and grasses. The cellu-
lose-rich residues remaining after alkaline peroxide
treatment under the conditions given earlier contain
80}90% cellulose and 10}20% hemicelluloses as
well as approximately 5% bound lignin. During
steam treatment, cellulose undergoes a change in its
crystallinity and can also partially depolymerize, de-
pending on the treatment conditions. The cellulose-
rich Rbres generated by the steam treatment/ex-
plosion process are generally more accessible to
chemicals and enzymes under derivatization condi-
tions.
Further Developments
The current views on the fractional isolation of poly-
saccharides from cereal straws and grasses are based
on the culmination of information gained over the
past thirty years. In spite of the many studies on the
polysaccharides of straws and grasses, little is known
about (a) isolation of pectic polysaccharides, (b) iso-
lation of the polysaccharide fraction present in delig-
niRcation liquors, and (c) isolation of non-glucosyl
residues and residual lignins in
-cellulose. Proced-
ures designed to extract pectic substances may extract
material that might otherwise be described as partly
hemicellulosic, and the converse may also happen.
Many procedures used in the isolation of polysac-
charides from straws and grasses would lead to the
loss of any such polysaccharides present. During de-
ligniRcation of wheat straw with sodium chlorite in
acidiRed solution, 1.2}2.4% of the total hemicel-
luloses passes into solution. It is also more difR-
cult to isolate pure cellulose without degradation
since the hemicelluloses and lignin are strongly asso-
ciated with cellulose; many of the solubilized hemicel-
luloses are irreversibly absorbed on the cellulose. In
addition, drying straw and grasses in an oven at 603C
may cause autohydrolysis or other changes in poly-
saccharide morphology. Furthermore, the various
procedures used prior to the exhaustive extraction of
hemicelluloses may affect quantitative or struc-
tural conclusions about the hemicelluloses as they
occur in nature. It is also of interest to note that the
hemicellulosic materials extracted by alkali are, only
in part, precipitated in ethnol after neutralization of
the extract. Under these conditions, a small part of
the hemicellulosic materials remains in solution, and
is commonly not recovered. All of these points re-
quire further investigation to obtain a more reliable
standard method for fractional isolation of polysac-
charides from cereal straws and grasses.
Further Reading
Aspinall GO (1970) Polysaccharides. Oxford: Pergamon
Press.
Browning BL (1967) Methods of Wood Chemistry, Volume
II. New York: John Wiley.
Coughlan MP and Hazlewood GP (1993) Hemicellulose
and Hemicellulases. London: Portland Press.
Fengel D and Wegener G (1989) Wood Chemistry, Ultra-
structure, Reactions. Berlin: Walter de Gruyter.
Gould JM (1984) Alkaline peroxide deligniRcation of agri-
cultural residues to enhance enzymatic sacchariRcation.
Biotechnology and Bioengineer 26: 46}52.
Lawther JM, Sun RC and Banks WB (1995) Extraction,
fractionation, and characterization of structural poly-
saccharides from wheat straw. Journal of Agricultural
and Food Chemistry 43: 667}675.
Montane D, Farriol X, Salvado J, Jollez P and Chornet
E (1998) Application of stem explosion to the fractiona-
tion and rapid vapour-phase alkaline pulping of wheat
straw. Biomass and Bioenergy 14: 261}276.
SjoK stroK m E (1981) Wood Chemistry } Fundamentals and
Applications. New York: Academic Press.
Steaniforth AR (1979) Cereal Straw. Oxford: Clarendon
Press.
Sun RC, Fang JM, Mott L and Bolton J (1999) Fractional
isolation and characterization of polysaccharides from
oil palm trunk and empty fruit bunch Rbres. Holzfor-
schung 53: 253}260.
Theander O and A> man P (1978) Chemical composition of
some Swedish cereal straws. Swedish Journal of Agricul-
tural Research 8: 189}194.
Water RH (1991) The Chemistry and Technology of Pec-
tin. San Diego: Academic Press.
Wilkie KCB (1979) The hemicelluloses of grasses and cer-
eals. Advances in Carbohydrate Chemistry and Bio-
chemistry 50: 215}264.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN AFFINITY
CHROMATOGRAPHY
B. J. Sines and W. H. Velander, Virginia Polytechnic
Institute and State University, VA, USA
Copyright ^ 2000 Academic Press
Introduction
Highly selective afRnity-based separations have evol-
ved considerably over the past two decades to im-
prove characteristics related to target speciRcity, dy-
namic adsorptive capacity and chemical robustness of
the afRnity matrix. These separation matrices are
used as screening devices for molecular interactions
as well as for the puriRcation of complex mixtures at
the analytical, preparative and large scale. AfRnity
adsorptive phases can be synthesized by the attach-
ment of structures called ligands to immiscible poly-
meric Suids, solvated gels and porous solids. Ligands
are selected for their afRnity to a soluble or colloidal
target species. Advances in technology have been as-
sociated with the synthetic and biosynthetic design of
afRnity ligands and matrices. These advances have
been made through a better understanding of phe-
nomena associated with the transport of the target
species to the ligand and the nature of molecular
interactions between target and ligand. Specialized
processing equipment for large and small scale ap-
plications has also improved afRnity separations.
Matrix Materials and Processing
Geometries
Separations achieved by the selective capture of a tar-
get from a Suid phase by an afRnity ligand conRned to
an adsorptive phase are most often done in aqueous
systems. Therefore, the adsorptive phase for a hy-
drophilic target species is often a hydrophilic mater-
ial. More hydrophobic targets require matrices which
offer a more hydrophobic environment. In some
cases, the adsorptive phase can be a separate Suid
phase or soluble component having afRnity ligands
which will capture and concentrate the target species
from complex aqueous mixtures.
Mattiasson demonstrated that heterobifunctional
ligands could be used for selective isolation of pro-
teins by afRnity precipitation. Heterobifunctional
ligands can bind the target and are also covalently
coupled to a polymer which can be used to induce
precipitation of a target/ligand complex. Alterna-
4575
tively, multiple but identical binding sites can be
occupied on the target by homobifunctional ligands.
This results in a cross-linked network between targets
which can be precipitated. Polymers used for afRnity
precipitation applications include chitosan, alginate,
polyethyleneimine, and Eudragit S-100 (a copolymer
of methyl methacrylate and methacrylic acid). By
manipulating such parameters as the pH or temper-
ature, the polymer/ligand/target complex can be re-
versibly rendered soluble and insoluble. Lactate dehy-
drogenase and alcohol dehydrogenase have been pre-
cipitated using Cibracron Blue 3GA coupled with
Eudragit2+ S-100.
A signiRcant advance in the Reld of bioseparations
is the coupling of afRnity ligands to aqueous two-
phase systems of water-soluble polymers. AfRnity
partitioning is the selective extraction of target pro-
teins from crude mixtures using afRnity ligands im-
mobilized on to either of two incompatible, water-
soluble polymer phases. Mattiasson used triazine
dyes coupled to immiscible polymer Suids to purify
proteins by afRnity partitioning. Reactive triazine
dyes have been widely exploited using this separation
technique. In comparison to classical chromato-
graphic systems, afRnity equilibrium is attained more
rapidly in aqueous two-phase partitioning. In addi-
tion, high afRnity binding capacities and protein re-
coveries have been achieved. Alternatively to two-
phase aqueous polymer partitioning, McCreath and
Chase immobilized human immunoglobulin G on to
perSuorocarbon emulsions for the selective adsorp-
tion of Staphylococcus aureus cells containing
membrane-bound protein A. These liquid emulsion
droplets were comprised of a perSuorocarbon oil
cross-linked with poly(vinyl alcohol).
Most of the applications for afRnity-based separ-
ations are developed for chromatographic processing
of aqueous systems. Thus, the ligand is immobilized
on a porous bed of solids packed in a column through
which liquid can be pumped. Most of the matrices are
made of hydrogels composed of cellulose, agarose
and dextran. A chief advantage of hydrogel matrices
is the combination of an easily derivatized, hy-
drophilic environment that provides molecular
accessibility that can be extended to 106 Da. Other
semi-synthetic and synthetic polymeric matrices, such
as copolymers of vinyldimethyl azlactone and
methylene-bis-acrylamide, can also provide similar
4576 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
intraparticle transport features. Coleman immobi-
lized protein A at high density on to azlactone-func-
tionalized polymeric matrices. These afRnity matrices
demonstrated high throughput capacity combined
with low operating pressures. As an alternative to the
use of pellicular matrices or matrices comprised of
reduced bead particle diameter, Rodriguez and Liapis
described the mechanism of perfusion chromatogra-
phy in which intraparticle mass transfer resistance is
reduced by increasing the particle permeability.
Velander optimized the molecular accessibility and
mechanical stability of uncross-linked cellulose ad-
sorbents by using large diameter beads ('0.3 mm)
which operate at low pressures even at high Sow
rates. These hydrogel matrices had a solids content of
only 3% or less and were shown to provide fast
intraparticle transport, apparently through a mixed
mode of both diffusion and convection. Larsson dem-
onstrated the use of superporous agarose matrices.
The agarose beads contained a typical internal porous
network in addition to larger pores. These larger
pores constituted a signiRcant portion of the bead, up
to one-third to one-tenth of the bead particle dia-
meter. Mass transfer characteristics were improved
since the larger pores allowed a considerable fraction
of the bulk chromatographic Sow to penetrate and
Sow through the individual bead particles. AfRnity
matrices were prepared using these superporous
agarose beads containing immobilized NAD# ana-
logue for the isolation of bovine lactate dehydrogen-
ase and protein A for the isolation of rabbit im-
munoglobulin G. These afRnity matrices operated
under much higher processing Sow rates compared to
conventional homogeneous bead columns. The pro-
tein A and NAD# analogue matrices had processing
throughputs Rve and three times higher, respectively,
in comparison to processing throughputs demon-
strated by conventional afRnity matrices. Thus, a sig-
niRcant improvement to afRnity technology is the
design of matrices having greater intraparticle acces-
sibility and transport of the target species.
Advances in transport phenomena needed for tar-
get contacting with the afRnity adsorbent have result-
ed in matrix designs having large or small particulate
as well as membrane geometries. Smaller particles,
having mean diameters of about 0.1 mm or less, yield
a greater surface area to volume for improved target
mixture contacting, but higher drops when operated
in a packed bed mode. Thus, specialized afRnity sep-
arations using high pressure liquid chromatography
have been developed for small and large scale. How-
ever, these media require pre-clariRcation of the feed
stream before chromatographic processing to prevent
column fouling. Alternatively, small diameter
particles having a higher density than water can be
Suidized to achieve both Rltration and afRnity adsorp-
tion in a single step using expanded bed chromatog-
raphy. Expanded bed adsorption chromatography
has been developed to provide single stage operation
for the isolation of target proteins from crude mix-
tures such as milk, hybridoma cell culture Suid and
fermentation broth. For example, humanized lgG4
antibody was isolated from myeloma cell culture Suid
by expanded bed adsorption using recombinant pro-
tein A immobilized on to Pharmacia Streamline2+
media.
Planar membranes are often used for afRnity separ-
ations in analytical assays but some cross-Sow,
stacked sheet geometries have been developed for
large scale applications. AfRnity membranes for large
scale work can also be cast in tubular hollow Rbres.
One example of planar membranes was made of a gel
formed from two dispersed liquid phases. The chief
advantages of afRnity membrane systems are the low
pressure operation and fast intraparticle transport
due to the short diffusion distances in thin mem-
branes. Thus, the kinetics of ligand/target interac-
tions can become rate limiting to afRnity separations
using thin membranes. Etzel showed that limitations
in membrane performance can arise due to variations
in porosity and thickness which can result in diffuse
breakthrough loading proRles. The stacking of planar
membranes can sharpen afRnity breakthrough pro-
Rles at higher loading Sow rates.
Af\nity Ligand Selection and
Design
AfRnity ligands have evolved from antibodies, enzy-
matic substrates, co-factors, nucleic acids, coen-
zymes, hormones, immunoglobulin, lectins, effectors
and inhibitors to a great diversity of small, low mo-
lecular weight peptides, polypeptides and other or-
ganic structures. These newer classes of ligands can
be made using biosynthetic and wholly synthetic
methods. Common to all selection strategies is the
need to begin with a diversity of structures from
which to discover candidate afRnity ligands. The
chance discovery of a ligand with desirable properties
has been enhances through the use of phage display
and synthetic combinatorial libraries which contain
many orders of randomized structural permutations.
The structural permutations within a library rely on
the polymeric assembly of bi- or multifunctional
monomeric molecules into compounds which are typ-
ically greater than 103 Da in molecular mass. In the
case of biosynthetic libraries, the assembled struc-
tures are necessarily polypeptides and the subunits
are amino acids. In the case of wholly synthetic struc-
tures, a variety of bi- and multifunctional organic
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
subunits that provide high selectivity and yield reac-
tions have been sequentially assembled using soluble
or solid-phase linked chemistries. For example, vari-
ous perturbations of bifunctional molecules having
reactive oxazoline chemistry have been used to create
assembled combinatorial chemical libraries having
103 or more structures. In addition, some wholly
synthetic libraries have used amino acids and novel
branched chain peptide linkages as core structural
platforms.
In 1985, Smith demonstrated the use of phage
technology to display candidate polypeptide ligands
from a bacterial virus particle which also contains the
DNA encoding the polypeptide sequence. The poly-
peptide is displayed from the virus particle surface in
a manner that also enables afRnity interactions with
a targeted species. Thus, candidate ligands can be
screened by the speciRc adsorption of phage to target
species which have been immobilized either to
a microtitre assay plate or to a chromatographic
matrix. NonspeciRcally adsorbed material is then
washed away and the speciRcally adsorbed phage
particles are eluted. The eluted phage particles are
cultured using microbiological plate-streaking tech-
niques. Candidate phages from the Rrst afRnity
screening are passed through a second afRnity screen-
ing and reculturing. Phages obtained from the two
afRnity and culture selection processes are subjected
to DNA sequencing. The DNA sequence of the dis-
played polypeptide is readily determined because of
speciRc endonuclease restriction sites engineered into
the phage genome. The polypeptide ligand DNA se-
quence then is readily used to create ligands in mass
quantity through recombinant fermentation techno-
logy. About 108 polypeptides can be created within
a single phage display library where about
101 to 102 candidate ligands typically result after a
sequence of two afRnity and culture screenings. Thus,
the chief advantage of the biosynthetic phage display
libraries is the large number of library members and
facile screening due to the coupled nature of the
encoding DNA and displayed polypeptide ligand of
the phage particle. The chief disadvantage is the in-
herent limitation to structures which are polypep-
tides. However, as a further improvement of phage
library technology, Ladner demonstrated that the use
of random permutations about a core polypeptide
ligand sequence can greatly enhance the afRnity of
initial ligand candidates. The chemical robustness of
polypeptides is typically limited to moderate pH and
aqueous environments.
Examples of the use of phage-display libraries in
the development of afRnity ligands are found
throughout the recent literature. Ladner generated
a series of libraries comprised of variants of the Rrst
4577
Kunitz domain of human lipoprotein-associated co-
agulation inhibitor (tissue-factor pathway inhibitor-
I). A typical human Kunitz domain was chosen as the
parental protein since immunogenicity would be min-
imal due to its human origin and lack of glycosylation
which would facilitate the use of phage-display tech-
nology. The library was screened against human plas-
min, which is a serine protease that participates in the
Rbrinolytic process. This study synthesized a protease
inhibitor exhibiting a high afRnity and speciRcity for
plasmin. Small (&58 amino acid residues), stable
Kunitz domains, which lack glycosylation, and con-
taining nearly human sequences were selected and
determined to have high afRnity and speciRcity to-
wards plasmin. The same phage-display library and
screening methodology has been used to select high
afRnity and high speciRcity ligands for human plasma
kallikrein and human thrombin. Markland demon-
strated that certain constraints may be imposed upon
primary and tertiary structures to increase speciRcity
relative to more simple linear peptide ligands directed
towards the same target. For example, a phage-dis-
play library was constructed that displayed an 18
amino acid residue peptide containing two Rxed cys-
teine residues to allow disulRde bond formation. In
addition, several variable residue positions between
and adjacent to the two cysteine residues were in-
cluded in the peptide sequence. This library was
screened against streptavidin and an anti-
-en-
dorphin monoclonal antibody. The screening yielded
phage displaying disulRde-constrained microproteins.
The selected phage clones required a disulRde bond
for the high afRnity binding to both target proteins.
Other core peptide motifs have resulted in phage-
display libraries displaying protease inhibitory do-
mains derived from wild-type bovine pancreatic tryp-
sin inhibitor. In one case, Ladner selected a ligand
which was an inhibitor of human neutrophil elastase
that had a 3.6 million-fold higher afRnity than that
for the parental protein and was reported to exceed
the highest afRnity cited for any human neutrophil
elastase inhibitor by 50-fold.
Wholly synthetic combinatorial libraries typically
use a robotic assembly of structures. The robotic
equipment used to generate orders of structural per-
mutations essentially consists of a miniaturized chem-
ical factory. These multifunction automated work
stations perform sequences of sample mixing,
thermostatic reactions, volume reduction and puriR-
cation of the reaction masses. The identity of indi-
vidual reaction masses is catalogued. IdentiRcation of
the newly created chemical structures is deduced
from automated combinations of liquid chromato-
graphy and mass spectrometry. Libraries having 103
to 104 distinct members have been synthesized and
4578 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
characterized in this way. The screening of wholly
synthetic libraries is laborious due to the need to
putatively identify ligand/target complexes and sub-
sequently to test the ligand structure in an immobi-
lized state. In summary, a chief disadvantage of these
libraries is the many order fewer library members that
can be generated using the current robotic methods.
Another disadvantage is cumbersome information ac-
quisition and management associated with both the
characterization of the ligand structure and target
afRnity. Unlike phage display, the primary molecular
structure is not encoded and coupled to the wholly
synthetic ligand as enabled by DNA of the phage
particle. However, the diversity of chemically robust
structures that can be produced from wholly syn-
thetic molecules is an important advantage. In addi-
tion, rational design models can greatly decrease the
number of permutations necessary from the initial
discovery of ligands having lower afRnity to the de-
sign of high afRnity ligand candidates based upon
initially discovered, ligand structural motifs. If com-
bined with rapid and sophisticated assay screening
methods, combinatorial methodologies provide an
efRcient process for the development of novel drug
leads and afRnity ligands. Future improvements in the
efRciency of screening combinatorial ligands will like-
ly be made using automated chemical sensor techno-
logy.
AfRnity peptide ligands can be made using solid-
phase chemistry. Houghten showed that the solid-
phase-linked chemistries have advantages of separ-
ation/wash cycling between reaction sequences as
well as being able to provide information about the
base structure of the candidate ligands. Independent
studies by Pingali and Fassina have demonstrated the
use of afRnity peptide ligands generated from solid-
phase peptide synthesis. Pingali produced an afRnity
peptide that bound human Rbrinogen and a peptide
of chromochrome c containing haem that captured
human serum albumin (HSA). The anti-Rbrinogen
tetrapeptide was made using standard FMOC (N-
Suorenylmethoxycarbonyl) chemistry. The afRnity
matrix made from this peptide was highly speciRc as
puriRed Rbrinogen was obtained directly from crude
human plasma. Frontal analysis determined that the
dynamic binding capacity of the antiRbrinogen col-
umn was 10.2 mg Rbrinogen per millilitre of gel.
Similarly, by frontal analysis, an HSA-afRnity peptide
column was found to have a dynamic binding capa-
city of 19 mg HSA per millilitre of gel. Fassina used
a core motif consisting of a small branched-chain
peptide to discover a protein A mimetic afRnity
ligand. This peptide recognized the Fc fragment of
immunoglobulin G from rabbit, goat, sheep and
mouse and provided a one-step isolation method for
highly puriRed immunoglobulin G directly from
crude sera. Frontal analysis determined that the dy-
namic binding capacity was 2 mg rabbit immunog-
lobulin per millilitre of gel. The peptide also exhibited
stability towards sanitization reagents such as
0.1 mol L\1 sodium hydroxide and ethanol. Both of
the above-mentioned studies show that peptide-based
afRnity columns derived from standard solid-phase
peptide synthesis can provide a viable alternative to
the use of immunoafRnity chromatography.
Molecular modelling is used to deduce likely inter-
actions between the target and afRnity ligand, and is
frequently used as a basis for further rational design
of both biosynthetic and wholly synthetic combina-
torial libraries. Once certain nominal afRnity motifs
have been identiRed from initial ligand screening,
rational design has been used to enhance afRnity by
creating permutations which build upon structural
motifs having lower afRnity for the target species. In
addition, recent studies have been conducted which
attempt to develop peptidomimetic structural com-
pounds, such as aminimides, which have core struc-
tures similar to that of peptides, but with improved
characteristics. Aminimides are much more chemic-
ally stable, have enhanced solubility characteristics
and are resistant to enzymatic degradation. Ringe
formulated a method for structure-based design in-
volving aminimides in which complete binding surfa-
ces of the targets are mapped. Combinatorial chem-
istry is then used to identify and improve structural
speciRcations in lead candidates targeted at the corre-
sponding binding sites. Ringe developed a pep-
tidomimetic aminide inhibitor or porcine pancreatic
elastase based upon crystal structures of an aminim-
ide analogue.
AfRnity dyes are another class of wholly synthetic
afRnity ligands which are not derived from combina-
torial libraries. Reactive synthetic textile dyes are
resistant to chemical and biological degradation and
can bind selectively and reversibly to a wide range of
enzymes and proteins. These ligands evolved from
ligands which were Rrst classiRed as enzyme co-factor
mimetics. Many of these co-factor mimetics were
members of reactive dyes used for staining textiles
and paper. Since the mid-1950s, these reactive textile
dyes have been used in the afRnity puriRcation of
proteins. The use of commercial textile dyes as
ligands in the large scale processing of diagnostic,
therapeutic and genetically engineered proteins is
documented throughout the literature. Since that
time, many variations of original dye ligand struc-
tures, such as reactive Cibacron Blue, have been syn-
thesized. These new structure}function motifs are no
longer considered a simple co-factor mimetic rela-
tionship as was recognized between nicotinamide
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
and Cibacron Blue. Cibracron Blue F3G-A, the
most thoroughly studied dye ligand, binds with a var-
iety of proteins such as adenine coenzyme-dependent
oxidoreductases, phosophokinases, hydrolases, trans-
ferases, nucleases, polymerases, synthetases, lyases,
decarboxylases, in addition to glycolytic enzymes and
plasma proteins. Lowe has shown that the triazine
dye structure binds to clefts of proteins which have no
known co-factor binding domains, but have much
more subtle, yet speciRc interactions with triazine dye
derivatives. Molecular modelling has also improved
the rational afRnity design of triazine derivatives
through the addition of various chemical moieties to
the core triazine structure. Lowe demonstrated the
use of computer-aided molecular modelling and de-
sign in the development of structurally modiRed bio-
mimetic dyes based on Cibracron Blue F3G-A and
Procion Blue MX-R. A dye ligand was engineered
speciRcally for the capture of horse liver alcohol de-
hydrogenase.
There are several advantages to the puriRcation of
pharmaceutical products using synthetic afRnity
ligands. A major advantage is that they are not de-
rived from biological sources. Therefore they impart
less of a product contamination risk. Hence, regula-
tory issues concerning the presence of unknown con-
tamination and infectious agents in the Rnal product
are circumvented. Other major advantages of these
ligands include: lower manufacturing cost; they can
be readily immobilized under a wide variety of coup-
ling chemistries to an extensive range of commercial-
ly available supports; they can be modiRed to enhance
speciRcity or stability; and they are more stable and
less susceptible towards denaturation. In addition,
ligands derived from combinatorial libraries may
contain a diversity of novel molecular structures such
as organically derived or non-natural amino acid resi-
dues such as L-amino acids (D-optical isomers). Well-
characterized afRnity ligands may ultimately dictate
decreased Rnal product costs associated with less ex-
pensive process costs (i.e. more robust afRnity ma-
trices and less expensive ligands) and higher through-
put due to a decrease in number of required
chromatographic steps (i.e. decreased buffer usage).
AfRnity separation of pathogens may be better, en-
abled by the increased accessibility of small ligands to
subtle, yet conserved domains of viruses.
Installation of Af\nity Ligands
The covalent installation of the afRnity ligand into
a chromatographic or membrane-based matrix can
profoundly affect the ligand/target binding efRciency.
Matrix coupling chemistries usually covalently attach
ligands through highly reactive groups such as -
4579
amino groups. Common activation methods for poly-
saccharide matrices are cyanogen bromide, divinyl-
sulfone, epoxy, organic sulfonyl chlorides, carbonyl
diimidazole, and N-hydroxysuccinimide. Polyacryl-
amide matrices are commonly activated by using
glutaraldehyde or hydrazine. Isothiocyanate and -
glycidoxypropylsilane activation methods are com-
monly used for silica-based matrices.
Ligands which contain amino acids can be coupled
through the -amino group of lysine, carboxyl groups
of aspartate and glutamate, and the phenolic group of
tyrosine. Chemistries which randomly couple these
residues also result in random orientation and spac-
ing of the immobilized ligand. Some of the most
thorough studies of ligand coupling chemistries have
been done with proteins, especially antibodies. For
example, Velander demonstrated that antibodies used
as afRnity ligands exhibit best performance character-
istics when the binding conformation of the antibody
is protected during covalent coupling to the adsor-
bent phase. A conformationally related effect is the
orientation of the immobilized ligand. The use of
orientated immobilization methods for antibodies is
also applicable to synthetic and biosynthetic afRnity
ligands that have asymmetric structure due to the
presence of both target binding and nonbinding do-
mains. The nonbinding domain is best used for
covalent coupling. Coupling of ligands to the support
matrix through reactive moieties present within the
binding domain can be detrimental from both ori-
entation and non-native conformational effects act-
ing upon the ligand. However, masking or shielding
of the binding domain prior to immobilization can be
employed to circumvent these effects. Velander used
synthetic antigens consisting of water-soluble adducts
of poly(2-methyloxazoline) polymers and a synthetic
peptide epitope for the masking of monoclonal anti-
body during immobilization. The mask was then re-
moved from the immobilized antibody. A loss of as
much as 50% of the theoretical binding capacity of
an immobilized antibody was attributed to orienta-
tion effects based upon anti-Fc and anti-Fab probing
of immunosorbents made using masked and unmas-
ked antibodies.
Activated matrices which preferentially couple
through speciRc functional groups on ligands can aid
in the site-directed, orientated immobilization of af-
Rnity ligands. For example, Domen developed several
activated matrices in which each couple antibodies
through different functional groups. Iodoacetyl
groups on SulfoLink2+ gel are designed to couple
through sulfhydryl groups found predominantly in
the hinge region of antibodies. CarboLink2+ ac-
tivated gels couple through aldehyde groups. The
aldehyde groups on antibodies can be formed by the
4580 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN AFFINITY CHROMATOGRAPHY
oxidation of carbohydrate found primarily in the Fc
region of antibodies. The site-directed method of
coupling for a bivalent antibody through the oxidized
carbohydrate groups using hydrazide chemistry re-
sulted in the theoretical maximum of two antigen
molecules for every antibody immobilized. However,
Velander and Orthner found some antibodies immo-
bilized on to matrices using hydrazide chemistry,
which couple through the carbohydrate groups, pro-
duced no signiRcant differences in immunosorbent
antigen binding efRciency in comparison to im-
munosorbents prepared using random coupling
chemistries such as cyanogen bromide. These anti-
bodies were found to have carbohydrate in the bind-
ing domain.
The covalent attachment of the ligand through
spacer arms becomes necessary when the nonbinding
domain of the ligand does not offer sufRcient molecu-
lar spacing between the molecular structure of the
adsorbent and the binding domain to enable unen-
cumbered binding of the target. The use of spacer
arms is well documented throughout the literature.
For example, Cuatrecasas demonstrated that exten-
sion of the ligand from the matrix through a hydro-
carbon spacer arm can considerably increase
ligand/target binding efRciency. Cuatrecasas pre-
pared speciRc staphylococcal nuclease afRnity ma-
trices by immobilizing the competitive inhibitor,
pdTp-aminophenyl, using spacer arms of varying
length on to agarose and polyacrylamide matrices.
AfRnity matrices prepared with the ligand attached
directly to cyanogen-bromide-activated Sepharose2+
yielded a binding capacity of 2 mg nuclease per mil-
lilitre of gel whereas afRnity matrices, prepared using
a ligand attached to the support through a 3,3-dia-
minodipropylamine spacer arm, yielded a binding
capacity of 10 mg nuclease per millilitre of gel. In
addition to ligand proximity to the support surface,
the proximal spatial positioning of adjacent ligands
within the support inSuences binding efRciency as
well.
A high local ligand density can also encumber tar-
get binding between proximally immobilized ligands.
Velander also demonstrated that antibodies immobi-
lized with a low, local density gave as much as 50%
of theoretical capacity based on a 2:1 target/anti-
body, molar stoichiometry. Thus, the majority of
activity loss associated with the installation of anti-
bodies into matrices can be attributed to local density
and orientation effects. As mentioned above, intra-
matrix transport phenomena can also affect afRnity
sorbent performance, particularly for large target
molecules. Thus, ligands are best installed into do-
mains with rapid target accessibility. However, ori-
entation, spacer arm and local spatial ligand density
effects must be evaluated at any location within the
adsorptive matrix as part of the afRnity sorbent op-
timization process.
Future Developments
A diversity of new wholly synthetic afRnity ligands
will be synthesized by microSuidic devices which will
essentially be micro-chemical factories on a chip.
Engineering afRnity matrices for optimum perfor-
mance will also include evaluating different immobil-
ization environments to enhance ligand/target inter-
actions. Since many different target binding environ-
ments may need to be screened for a given ligand,
miniaturized matrices installed on sensors will likely
replace the inefRcient batch and chromatographic
analysis of optimal immobilization environments.
Older pharmaceutical processes, such as blood
plasma fractionation, will eventually be supplanted
by afRnity separation processes.
See also: II / Affinity Separation: Affinity Membranes;
Affinity Partitioning in Aqueous Two-Phase Systems; Con-
valent Chromatography; Dye Ligands; Hydrophobic Inter-
action Chromatography; Immobilised Boronates and Lec-
tins; Immobilised Metal Ion Chromatography; Immunoaf-
finity Chromatography; Imprint Polymers; Rational Design,
Synthesis and Evaluation: Affinity Ligands; Theory and
Development of Affinity Chromatography; Chromatogra-
phy: Protein Separation. Appendix: 1/Essential Guides
for Isolation / Purification of Enzymes and Proteins.
Essential Guides for Isolation/Purification of Im-
munoglobulins.
Further Reading
Baumbach GA and Hammond DJ (1992) Protein puriRca-
tion using afRnity ligands deduced from peptide libra-
ries. BioPharmacology 5: 24}29.
Cannon LE, Ladner RC and McCoy D (1996) Phage-dis-
play technology. IVD Technology 2 (6).
Coleman PL, Walker MM, Milbrath DS, Stauffer DM,
Rasmussen JK, Krepski LR and Heilmann SM (1990)
Immobilization of protein A at high density on azlac-
tone-functional polymeric beads and their use in afRnity
chromatography. Journal of Chromatography 512:
345}363.
Domen PL, Nevens JR, Mallia AK, Hermanson GT and
Klenk DC (1990) Site-directed immobilization of pro-
teins. Journal of Chromatography 510: 293}302.
Fassina G, Verdoliva A, Odierna MR, Ruvo M and Cassani
G (1997) Protein A mimetic peptide ligand for afRnity
puriRcation of antibodies. Journal of Molecular Recog-
nition 9: 564}569.
Garg N, Yu I and Mattiasson B (1996) Dye-afRnity tech-
niques for bioprocessing: recent developments. Journal
of Molecular Recognition 9: 259}274.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS
Gupta MN, Kaul D, Guoqiang D, Dissing U and Mattias-
son B (1996) AfRnity precipitation of proteins. Journal
of Molecular Recognition 9: 356}359.
Hogan Jr, JC (1996) Directed combinatorial chemistry.
Nature 384: 17}19.
Kaster JA, de Oliveira W, Glasser W and Velander WH
(1993) Optimization of pressure-Sow limits, strength,
intraparticle transport and dynamic capacity by hydro-
gel solids content and bead size in cellulose immunosor-
bents. Journal of Chromatography A 648: 79}90.
Lowe CR, Burton SJ, Burton NP, Alderton WK, Pitts JM
and Thomas JA (1992) Designer dyes: biomimetic
ligands for the puriRcation of pharmaceutical proteins
by afRnity chromatography. Tibtech 10: 442}448.
McCreath GE and Chase HA (1996) Applications of per-
Suorocarbon afRnity emulsions for the rapid isolation of
Staphylococcus aureus. Biotechnology Progress 12:
77}83.
Markland W, Ley AC and Ladner RC (1996) Iterative
optimization of high-afRnity protease inhibitors using
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN CAPILLARY
ELECTROPHORESIS
S. K. Poole, Parke-Davis Pharmaceutical
Research, Division of Warner-Lambert Company,
Ann Arbor, MI, USA
C. F. Poole, Wayne State University,
Detroit, MI, USA
Copyright ^ 2000 Academic Press
Introduction
Capillary electrophoresis encompasses a number of
related separation approaches, some of which are
adapted to the requirements of speciRc applications
(Figure 1). They share in common the use of electro-
lyte solutions as mobile phase, the use of capillary
tubes as the separation column, and the use of an
electric Reld to induce sample and mobile phase trans-
port. This allows a similar instrument platform to
service all capillary electrophoretic separation tech-
niques with only minor modiRcations for speciRc
applications. Detection is usually by UV-visible ab-
sorption through the fused silica capillary wall, or
occasionally by Suorescence, electrochemical or mass
spectrometric detection. Contemporary instruments
are also highly automated for ease of use and im-
proved control of critical experimental variables.
ClassiRcation of capillary electrophoretic tech-
niques according to their usual applications is given
in Table 1. These techniques can be considered as
4581
phage-display. 2. Plasma kallikrein, and thrombin. Bio-
chemistry 35: 8045}8057.
Orthner CL, Highsmith FA, Tharakan J, Madurawe RD,
Morcol T and Velander WH (1991) Comparison of the
performance of immunosorbents prepared by site-di-
rected or random coupling of monoclonal antibodies.
Journal of Chromatography 558: 55}70.
Pingali A, McGuinness B, Keshishian H, Fei-Wu J, Varady
L and Regnier F (1996) Peptides as afRnity surfaces for
protein puriRcation. Journal of Molecular Recognition
9: 426}432.
Roberts BL, Markland W, Ley AC, Kent RB, White DW,
Guterman SK and Ladner RC (1992) Directed evolution
of a protein: Selection of potent neutrophil elastase in-
hibitors displayed on M13 fusion phage. Proceedings of
the National Academy of Sciences USA 89: 2429}2433.
Velander WH, Subramanian A, Madurawe RD and Or-
thner CL (1991) The use of Fab-masking antigens to
enhance the activity of immobilized antibodies. Biotech-
nology Bioengineering 39: 1013}1023.
general, sample-type speciRc, in an early development
phase, or of minor importance. Such a broad range of
descriptive terms requires further elaboration to indi-
cate how we propose to treat these techniques in this
article. Capillary zone electrophoresis (CZE), or sim-
ply capillary electrophoresis, and micellar elec-
trokinetic chromatography (MEKC) are widely used
and complementary techniques for the separation of
ionic and neutral molecules. They are the most
important and general in terms of the number of
applications and frequency of use. Capillary electro-
chromatography (CEC) is a relatively new and prom-
ising technique with a range of applications similar to
liquid chromatography. Since electro-driven Sow has
been shown to provide both theoretical and practical
advantages over pneumatic-driven Sow, it has the
potential to become a major separation technique. At
present, too little is known about the technique to
provide a deRnitive guide to method development,
especially as in the future it is likely that new column
materials will be developed speciRcally for capillary
electrochromatography with properties different
to those currently used. Capillary gel electrophoresis
(CGE) is an important technique for the separation of
biopolymers but is little used outside of laboratories
that perform this type of analysis. Capillary isoelec-
tric focusing (CIEF) is a specialized technique within
the Reld of macromolecule zwitterion separations,
4582 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS

Figure 1 Classification of capillary electrophoretic separation methods based on buffer type and mechanism. CZE"capillary zone
electrophoresis; CGE"capillary gel electrophoresis; MEKC"micellar electrokinetic chromatography; CEC"capillary electro-
chromatography; CIEF"capillary isoelectric focusing; and CITP"capillary isotachophoresis.

largely proteins, requiring special buffers to generate
a continuous pH gradient. Capillary isotachophoresis
(CITP) is not widely used for separations, it can be
rather difRcult and tedious to optimize, and
yields an integral signal that is different to other
separation techniques. Many samples that can be
separated by capillary isotachophoresis can also be
separated by other electrophoretic techniques more
familiar to separation chemists. It is Rnding increas-
ing use to preconcentrate ions for separation by capil-
lary zone electrophoresis. With this framework in
mind we propose to provide general guidelines for
method development in capillary zone electrophor-
esis, micellar electrokinetic chromatography, and gel
electrophoresis with only comments and brief instruc-
tions applicable to the other capillary electrophoretic
techniques.
Sample Suitability
Table 1 provides a general guide to method selection
by analogy to established applications. For bio-
polymers capillary electrophoretic techniques often
select themselves, for other compounds the capillary
electrophoretic techniques have to be considered in
terms of suitability drawn against other existing
chromatographic methods. Reasonable solubility in
aqueous solution is required for most separation
modes. Non-aqueous capillary electrophoresis is little
developed (although promising) and techniques such
as micellar electrokinetic chromatography can separ-
ate hydrophobic compounds but provide little selec-
tivity. Gas chromatography is usually a better choice
for the separation of volatile hydrophobic com-
pounds. High pressure liquid chromatography is
often a better choice when low level detection, struc-
tural elucidation by mass spectrometry or prepara-
tive-scale separations are required. The concentration
sensitivity of the capillary electrophoretic techniques
using UV-visible absorption detection is limited by
the small cross column pathlength and small injection
volumes to solutions containing at least
1}10
g mL\1 and for ease of operation 0.1 mg mL\1
or above is preferred. Various stacking and precon-
centration techniques may improve detection limits
but these require additional effort and time for
optimization that may not be justiRable if another
technique is suitable for the separation. Within these
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS 4583

Table 1 Common separation methods using capillary electrophoretic techniques

Technique Separation mechanism Applications

Zone electrophoresis Differences in charge-to-size ratios Inorganic and organic ions

Ionizable compounds
Zwitterions
Biopolymers

Micellar electrokinetic chromatography Distribution of neutral and partially ionized Water-soluble neutral compounds
compounds between charged micelles and Weak acids and bases
electrolyte solution

Gel electrophoresis Differences in size and charge (but not size-to- DNA fragments
charge ratio) by migration through a gel matrix SDS proteins
or entangled polymer network with a range of Macromolecules
pore sizes

Electrochromatography Distribution between a solid stationary phase and Neutral compounds

mobile electrolyte solution Weak acids and bases
Ions

Isoelectric focusing Differences in isoelectric points in a continuous Proteins

pH gradient Zwitterionic compounds

Isotachophoresis Differences in electrophoretic mobility of ions Preconcentration of ions

sandwiched between two buffers containing
ions of greater (leading) and lower (trailing)
mobility

restrictions it is obvious that many sample types and
problems can be handled by capillary electrophoretic
techniques accounting for its expanding use in ana-
lytical chemistry.
Selecting System Variables
Virtually all separations are carried out in fused silica
capillary columns 50}100
m internal diameter and
up to 1-m long. Large-bore capillaries provide greater
loading capacity and a higher detector response be-
cause of the longer pathlength (on-column detection)
but generate larger currents and are less efRcient
at heat dissipation. Small-diameter columns show
increased adsorption character due to their larger
inner surface area-to-volume ratio but provide more
efRcient heat dissipation. If detection limits are not
a problem, then a small inner diameter column
should be used. The choice of capillary length is
a compromise between speed (short columns) and
separation capacity (long columns). Unless the separ-
ation is unusually complicated capillaries should
be short (25}50 cm). When a new capillary is put
into use or is suspected of being contaminated, a
conditioning procedure is required. Washing with
a solution of sodium hydroxide, water, and buf-
fer as indicated in Table 2 is normally sufRcient.
Capillaries with an interior coating are used to alter
electroosmotic Sow or to minimize analyte ad-
sorption by the capillary wall, particularly for
macromolecules. Electroosmotic Sow is optimized
to obtain useful separations in MEKC and CEC, is
often used to improve separations and total sample
detection for ions of opposite charge in CZE, but
is usually undesirable in CGE, CIEF and CITP.
So it is in the later techniques that capillaries with
chemically bonded or physically adsorbed coatings
are used.
Separations are usually performed with a voltage of
10}30 kV. High voltages provide faster separations
with higher efRciency provided that the heat gen-
erated is effectively dissipated. A plot of current
against applied voltage can be used to optimize oper-
ating conditions. The fastest and most efRcient
separations are obtained at the upper end of the linear
portion of the plot. A positive deviation in the plot
indicates that the heat removal capacity of the system
is being exceeded. Capillary electrophoretic separ-
ations are usually performed at or close to room
temperature (253C). Temperature control, however,
is important and separation capillaries are thermo-
stated in an air or liquid bath. Thermostating is used
to remove heat and to establish a constant temper-
ature. Poor thermostating results in lower efR-
ciency and poor reproducibility of migration times.
Temperature is a useful operating variable, which can
be used to modify migration times and selectivity, but
is generally considered only suitable for Rne tuning
4584 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS

Table 2 A guide for selecting initial conditions in capillary electrophoretic separations

Parameter Setting

Column Initial experiments use a fused silica capillary 30}50-cm long and 50- or 75-
m internal diameter. Short
columns are appropriate for trial experiments. The complexity of the sample dictates the length. For 2}10
analytes use 35}40 cm; 11}50 analytes 50}60 cm; 50}80 analytes 70}80 cm; and '80 analytes
90}100 cm. Smaller diameter columns (25 or 50
m) provide higher efficiency but lower sample loading
capacity.
Initial conditioning Rinse with 0.1 M sodium hydroxide for 30 min. Flush with water for 15 min followed by the separation buffer
for 15 min.
Voltage Usual range is 10}30 kV. High voltages provide faster separations and greater separation efficiency. The
method employed to dissipate heat, the column internal diameter, and buffer type and concentration all
affect this decision. Use the highest voltage that does not exceed 100
A current as a rough guide.
Otherwise plot current against voltage (2.5-kV increments) and operate at a voltage towards the upper
portion of the linear plot.
Temperature Initial experiments use 20}253C. Selectivity and separation speed varies with temperature, which is
optimized to fine-tune a separation (vary from 20 to 603C in 53C increments).
Injection Hydrodynamic (e.g. 3 s at 0.5 p.s.i.) or electrokinetic (2}5 nL)
Detection Absorption maximum of the analyte of interest, for which the weakest signal is expected because of low
concentration or low absorbance. If analyte detection properties are unknown try 200}230 nm.

nearly acceptable separations. Subambient temper- Capillary Zone Electrophoresis

atures are not commonly used, as they are less
convenient and result in poorer kinetic separation
properties.
In general, the sample should be prepared such that
the analytes of interest are present in a suitable solu-
tion, free from interferences, and at an appropriate
concentration for detection. The ionic strength of the
sample should be no greater than that of the buf-
fer, with a more or less similar pH to the buffer,
and free of matrix problems associated with column
wall adsorbing materials and particle matter. For the
best peak shapes and resolution the concentration of
the injected sample should be about 100 times lower
than the concentration of the buffer. Syringe
Rlters for particle removal and ion exchange mem-
brane Rltration devices to reduce excessive concentra-
tions of common matrix ions are available. Proteins
and similar macromolecules, if not of interest to the
analysis, should be precipitated prior to separation to
minimize column fouling. Analytes of low water solu-
bility may have to be dissolved in a water-miscible
organic solvent or mixture of organic solvent and
separation buffer. For other samples it is com-
mon practice to dissolve the sample in the run buf-
fer, a diluted solution of the run buffer, or water.
Samples are introduced into the separation capillary
by hydrodynamic or electrokinetic injection. Both
methods provide reproducible injection volumes but
sampling bias is associated with electrokinetic injec-
tion, which injects increasing amounts of sample
components in proportion to their mobility. Hy-
drodynamic injection is not suitable for CEC and
CGE because of the high Sow resistance of packed
columns.
Once the system variables are set within reasonable
ranges the parameters that have most effect on
migration times and selectivity are the composition,
concentration and pH of the run buffer and the
presence of additives, if used, to provide additional
selectivity optimization. For a good separation by
CZE four features are important: (i) the individual
mobilities of the analytes must be different; (ii)
the background electrolyte must be homogeneous
and the Reld strength uniform along the column; (iii)
neither analytes nor matrix components must interact
with the column wall; and (iv) the conductivity of
the buffer must substantially exceed the total con-
ductivity of the sample components. Suitable com-
mon buffer recipes for a wide pH range are given
in Table 3. Additional buffers with their pKa and
anion mobility values are given in Table 4.
Ionic strength and pH greatly affect selectivity
and separation time and should be course tuned in
initial screening experiments. Low pH is favourable
for separating anions (all anions are less mobile) and
a high pH is preferred for cation separations. The
practical pH range is limited roughly to between
2 and 12. If the pKa of the sample components is
known or can be reasonably estimated, pH optimiza-
tion should start with a pH+pKa. Weak acids and
bases change from the neutral form to the fully
ionized form over about 4 pH units. In the neutral
form their electrophoretic mobility is zero and they
all migrate at a Rxed velocity due to the electroos-
motic Sow in common with all neutral species. When
totally ionized the ion moves with a constant elec-
trophoretic velocity and may be separated from other
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS 4585

Table 3 Recipes for preparing some common electrophoretic buffers (100 mL of 60 mM buffer)

pH Buffer system Acid Base

Phosphate 85% Phosphoric acid Potassium dihyrogenphosphate

2 395.3 mg 349.9 mg
2.5 205.3 mg 574.3 mg
3.0 81.4 mg 720.5 mg
Acetate 1.0 M Acetic acid Sodium acetate
3.5 5.67 mL 26.6 mg
4.0 5.08 mL 75.8 mg
4.5 3.81 mL 174.6 mg
5.0 2.13 mL 317.6 mg
5.5 0.89 mL 419.1 mg
Phosphate Sodium dihyrogenphosphate (1H2O) Disodium hydrogenphosphate (2H2O)
6.0 779.2 mg 61.9 mg
6.5 692.8 mg 174.3 mg
7.0 512.2 mg 407.2 mg
7.5 280.7 mg 705.9 mg
8.0 115.5 mg 919.0 mg
Borate Boric acid Disodium tetraborate (10H2O)
8.0 320.9 mg 77.3 mg
8.5 232.7 mg 213.2 mg
9.0 59.3 mg 480.6 mg
Borate Disodium tetraborate (10H2O) 0.1 M Sodium hydroxide
9.5 371.0 mg 41.77 mL
10.0 371.0 mg 52.72 mL

ions based on differences in their charge-to-size
ratio. When partially ionized the ions migrate with an
effective mobility that changes between the two
extreme values in a sigmoid fashion as the pH is
varied (Figure 2). Ions may be separated in their fully
ionized form or partial ionized form as a matter of
circumstance; that is, at those conditions that maxi-
mizes the difference in charge-to-size ratios. Be-
cause changes in mobility tend to be large for par-
tially ionized solutes small pH changes (0.1}0.5 pH
units, or smaller for complex mixtures) are used to
optimize the separation.
If the pKa values for a sample are unknown, con-
duct initial separations in a series of buffers at or
near pH 2.5, 4.0, 5.5, 7.0, 8.5 and 10 (see Table 3 for
appropriate buffers). To obtain reproducible re-
sults over the pH range 4 to 7, careful column condi-
tioning is important. From the plot of the effec-
tive mobility against pH identify the most promising
pH range for the separation. Optimization then pro-
ceeds in smaller changes in pH units as before.
To optimize the buffer concentration initial
experiments are performed with a concentration of
30}100 mM for 50-
m internal diameter columns
and 20}50 mM with 75-
m internal diameter col-
umns. Lower ionic strength buffers are used to
obtain faster separations, when selectively is high,
and to separate simple mixtures containing a few
analytes. Higher ionic strength buffers are used
for the separation of complex mixtures or to separate
analytes with small differences in their elec-
trophoretic mobility. If stacking is used to enhance
analyte detectability then the difference in ionic
strength between the buffer (high ionic strength)
and the sample should be maximized. From Table 4
inorganic buffers are likely to provide better
peak shapes for high mobility ions and Good-type
(zwitterionic) buffers for low mobility ions.
Zwitterionic buffers are useful for many applica-
tions where a high concentration and buffering
capacity is desirable because of their low speciRc
conductivity, which allows more favourable kinetic
separation conditions to be employed.
For difRcult separations the selectivity can be
further modiRed by employing secondary chemical
equilibria and solvation effects by adding appro-
priate reagents or solvents to the electrolyte system
(Table 5). Increasing the ionic strength of the electro-
lyte by adding salts such as potassium sulfate modiRes
the charge and/or conformation of proteins and re-
duces wall interactions. Metal cations such as Cu2#,
Zn2#, Ca2# coordinate to proteins and peptides
modifying the net charge. Also, alkanesulfonic acids
bind selectively to proteins and peptides through
hydrophobic interactions modifying the surface
charge as well as reducing wall interactions. Slow
4586 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS

Table 4 Suitable buffers for capillary electrophoresis. Mobility values are at zero ionic strength and 253C (in 10\9 m2 V\1 s\1)

Buffer pKa Mobility

Phosphoric acid 2.12 (pK1) !35.10

7.21 (pK2) !58.30
12.32 (pK3) !71.50
Malonic acid 2.90 (pK1)
5.70 (pK2)
Citric acid 3.13 (pK1) !28.70
4.76 (pK2) !54.30
6.40 (pK3) !70.80
Lactic acid 3.85 !35.80
Hydroxyisobutyric acid 3.97 !33.50
Glutamic acid 4.38 !28.90
Acetic acid 4.76 !42.40
MES [2-(N-morpholine)ethanesulfonic acid] 6.13 !26.80
MOPS [3-(N-morpholine)propanesulfonic acid] 7.20 !24.40
MOPSO [2-hydroxy-4-morpholinepropanesulfonic acid] 6.79 !23.80
ACES [N-2-acetamido-2-aminoethanesulfonic acid] 6.84 !31.30
Imidazole 7.17 52.00
BES [2-(bis
2-hydroxyethylamino)ethanesulfonic acid] 7.16 !24.00
HEPES [N-2-hydroxyethylpiperazine-N -2-ethanesulfonic acid] 7.51 !21.80
TRICINE [N-
tris(hydroxymethyl)methylglycine] 8.15
TRIS [tris(hydroxymethyl)aminoethane] 8.08 29.50
TAPS [3-
tris(hydroxymethyl)methylaminopropanesulfonic acid] 8.30 !25.00
BICINE [N,N-bis(2-hydroxyethyl)glycine] 8.35
Glycylglycine 8.40
Ammonia 9.26
Ethanolamine 9.50 44.3
CHES [2-(cyclohexylamino)ethanesulfonic acid] 9.50
Triethylamine 9.87
CAPS [3-(Cyclohexylamino)propanesulfonic acid] 10.40
Diethylammonium 11.40 37.9

adsorption/desorption interactions with the column
wall cause peak broadening and tailing and irrevers-
ible adsorption leads to modiRcation of the capillary
wall. These problems are caused by electrostatic or
hydrophobic interactions between macromolecules
Figure 2 Separation of two hypothetical weak acids as a func-
tion of pH by capillary zone electrophoresis.
(usually) and the column wall. Solutions to this prob-
lem include using extreme pH buffers, high ionic
strength electrolytes, and by using dynamic or chem-
ically bonded wall-coated capillaries. There are no
universal solutions and effective methods have to
be tailored to the properties of the analyte. Buf-
fer additives are usually used at concentrations of
5}60 mM except for modiRcation of the ionic
strength of the electrolyte where much higher concen-
trations are often required (e.g. 50}250 mM). Urea,
which forms hydrogen-bond complexes with pro-
teins and peptides, but is nonionic, is often used at
concentrations of 7 M. The separation of metal
cations (alkaline earths, transition metals and lan-
thanides) is difRcult because of their similar ionic
conductance. In this case complexing agents, such as
-hydroxyisobutyric acid or citrate are required.
Since many cations lack a chromophore complexa-
tion is an effective method of introducing a chro-
mophore for convenient detection. There is now
considerable literature on the separation of anions by
capillary electrophoresis. For fast separations it is
necessary to reverse the direction of the electro-
osmotic Sow by adding cationic surfactants below
their critical micelle concentration to the buffer
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS 4587

Table 5 Secondary equilibria used to optimize selectivity in capillary electrophoresis

Additives Function

General considerations
Inorganic salts
Crown ethers
Organic solvents
Urea
Metal ions
Alkanesulfonic acids
Cellulose polymers
Cationic surfactants
Organic acids
Minimize wall interactions, induce protein conformation changes
Modify mobility by selective formation of inclusion complexes
Modify electroosmotic flow, increase solubility of organic ions, modify ion solvation, reduce wall
interactions
Modifies the mobility of proteins by hydrogen-bond complexation
Modify mobility of anions and electroosmotic flow
Modify mobility by ion pair formation, wall adsorption leads to changes in surface properties
Mask active sites on the capillary wall, modify electroosmotic flow
Use to reverse the polarity of the fused silica capillary wall
Modify mobility by ion pair formation

Ion complexation
Chelate formation (metals) Polycarboxylic acids (lactate, tartrate, hydroxyisobutyric acid)
Ethylene-1,2-diaminetetraacetic acid
Dihydroxyazobenzene-5, 5-disulfonate
Ion pairing Ionic surfactants ((critical micelle concentration)
Cetyltrimethylammonium bromide, tetradecyltrimethylammonium bromide
Polyvalent metal cations (Ca2#, Al3#, etc.)
CHES and other alkanesulfonic acids, perchlorate
Ion inclusion Crown ethers (15-crown-6, 18-crown-6, etc.)

Solvent effects
Organic solvents Acetonitrile, methanol, 2-propanol, tetrahydrofuran, etc.
Electrolyte Ionic strength, concentration of the probe (co-ion)

system. The electroosmotic Sow and electrophoretic
migration now occur in the same direction. For
difRcult to separate anions normal (counterSow)
operation may be the better option at the expense of
longer separation times. To reduce peak broadening
the mobility of the sample anions should be matched
to those of the background electrolyte. For UV-visible
detection indirect detection is frequently employed.
This requires the addition of a probe (co-ion) of high
molar absorption, in low concentration, with the
same charge as the analytes. Examples include chro-
mate (most popular), benzoate, salicylate, phthalate,
etc.
Micellar Electrokinetic
Chromatography
The addition of an ionic surfactant above its critical
micelle concentration to the buffer provides an
additional separation mechanism based on distribu-
tion of the analytes between the micelles and electro-
lyte. The velocity with which the micelles migrate to
the detector is usually different to the velocity of
the bulk electrolyte allowing separations based purely
on differences in the analyte distribution constants
for neutral compounds. For ions differences in both
distribution constants and electrophoretic mobility
are important. An acceptable separation also requires
favourable kinetic properties (efRciency), provis-
ion of an adequate migration window (peak capacity)
and a reasonable total separation time. Normally, the
experimental conditions are set to establish an accept-
able separation time and migration window under
conditions where the efRciency is not compro-
mised and the outcome of the experiment controlled
by selectivity optimization. Selectivity is inSuenced
largely by the identity of the surfactant and the addi-
tion of complexing agents and/or organic solvents to
the buffer.
Some common surfactants and their relative solva-
tion properties are summarized in Table 6. Method
development usually begins with sodium dodecyl sul-
fate because of its favourable kinetic and chromato-
graphic properties. (Table 7). Other surfactants are
selected based on their complementary properties to
sodium dodecyl sulfate using the system constants of
the solvation parameter model as a guide (Table 6).
For example, sodium cholate (representative of the
bile salts) is a stronger hydrogen-bond base and
weaker hydrogen-bond acid than sodium dodecyl sul-
fate. By similar reasoning a working list of surfactants
for selectivity optimization would include sodium
dodecyl sulfate, sodium cholate, lithium perSuorooc-
tanesulfonate, sodium N-dodeconyl-N-methyltaurine
and tetradecyltrimethylammonium bromide. Table 6
also provides a framework to identify new surfactants
4588 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS

Table 6 Characteristic properties of common surfactants for micellar electrokinetic chromatography

Surfactant Critical Aggregation Solvation parameter model system constantsH

micelle number
concentration m r s a b
(mM)

Sodium dodecyl sulfate 8.2 62 2.99 0.46 !0.44 !0.30 !1.88

Tris(hydroxymethyl)aminoethane dodecyl sulfate 2.56 0.57 !0.66 !0.33 !1.56
Sodium dodecyl sulfonate 9.8 54 2.51 0.51 !0.70 !0.14 !1.51
Sodium cholate 13}15 2}4 2.45 0.63 !0.47 0 !2.29
Sodium taurocholate 2.8 4 2.43 0.60 !0.34 0 !2.06
Sodium deoxycholate 4}6 4 2.67 0.66 !0.47 0 !2.47
Sodium taurodeoxycholate 2}4 8 2.62 0.67 !0.45 0 !2.17
Sodium N-dodecanoyl-N-methyltaurine 8.7 3.07 0.72 !0.50 0.22 !2.58
Lithium perfluorooctanesulfonate 2.30 !0.52 0.34 !0.82 !0.53
Tetradecyltrimethylammonium bromide 4.4 64 2.82 0.36 !0.29 0.90 !2.67
Hexadecyltrimethylammonium bromide 0.026 169 3.40 0.61 !0.55 0.58 !3.08
MicroemulsionHH 3.05 0.28 !0.69 !0.06 !2.81

HThe m system constant is a measure of the difference in cohesive energy and dispersion interactions for the micelles and electrolyte;
the r system constant the difference in interactions with lone pair electrons; the s system constant the difference in interactions of
a dipole type; the a and b system constants the difference in hydrogen-bond base and hydrogen-bond acid interactions, respectively.
The sign of the constant indicates whether the interaction favours distribution to the micelles (positive) or electrolyte system
(negative). HHMicroemulsion consisting of 1.4%wt. sodium dodecyl sulfate, 6.49% wt. butan-1-ol and 0.82%wt. heptane.

with complementary properties to those available at When selectivity optimization using different
present and to avoid unnecessary experiments with surfactant types is exhausted further optimization is
surfactants with different structures but nearly achieved by the use of additives (see Table 7). For this
identical selectivity properties. purpose the common approaches are the use of mixed

Table 7 Starting conditions for method development in micellar electrokinetic chromatography

Parameter Setting

Sample 1}2 mg mL\1 dissolved in methanol or water

Column Fused silica capillary 30}50-cm long with an internal diameter of 50
m
Initial conditioning Flush with 0.1 M sodium hydroxide for 3 min and rinse with the run buffer for 5 min. These conditions
will have to be varied depending on the previous use (if any) of the column. It is preferable to
reserve individual capillaries for each surfactant.
Buffer 20 mM sodium phosphate}sodium tetraborate pH 8 buffer (or see Table 3 for suitable single buffers)
containing 50 mM sodium dodecyl sulfate
Voltage 20}25 kV
Temperature 253C
Injection 50 mbar 1}2 s (hydrodynamic)
Detection 210 nm (or absorption maximum for analyte with lowest absorbance)

Course tuning selectivity

Surfactant Choose surfactants of different selectivity (see Table 6)
Sodium cholate (72 mM)
Sodium N-dodecanoyl-N-methyltaurine (50 mM)
Tetradecyltrimethylammonium bromide (50 mM) with reverse polarity
Other suitable surfactants
pH Optimize migration window and separation time (lower pH to extend and raise pH to lower) for neutral
compounds. Weak acids and bases may show significant changes in electrophoretic behaviour
Additives Mixed surfactants formed with neutral and ionic surfactants. For example, Brij 35 (polyoxyethylene[23]
dodecyl ether) 1}25 mM
Organic solvents methanol, 2-propanol, acetonitrile, tetrahydrofuran 1}25% (v/v)
Higher molecular mass solvents of low water solubility 1}5% (v/v)
Complexing additives such as -, -, -cyclodextrins, hydroxypropyl--cyclodextrin and
heptakis-(2,3,6-tri-O-methyl)--cyclodextrin (5}20 mM)

Fine tuning selectivity

Modify system properties such as column length, temperature, voltage, buffer type and ionic strength.
Surfactant concentration changes the phase ratio but has little effect on selectivity
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS 4589

success. Neutral surfactants such as Brij 35 are often

Figure 3 Change in the system constants obtained from the
solvation parameter model as a function of the composition of the
mixed micelles formed with the neutral surfactant Brij 35
(1}50 mM) and 50 mM sodium dodecyl sulfate. See Table 6.
(Reproduced with permission from Poole SK and Poole CF (1997)
Variation of selectivity with composition for a mixed-micellar
buffer in micellar electrokinetic chromatography. Journal of High
Resolution Chromatography 20: 174}178.)
surfactant micelles, organic solvents and inclusion
complexing agents. A large number of mixed micelles
can be employed without any certain prospects of
chosen Rrst to adjust selectivity and/or the size of the
migration window. Figure 3 shows an example of the
use of Brij 35 to change the selectivity of sodium
dodecyl sulfate micelles. The solvation properties of
the mixed micelles are not changed radically, even at
high concentrations of the neutral surfactant, in
agreement with predictions made by the interphase
retention model. The main change is the gradual
decrease in the hydrogen-bond acidity of the mixed
micelles, which should provide a useful change of
selectivity for the separation of hydrogen-bond bases.
Selectivity modiRcation by addition of organic sol-
vent to the buffer is by no means as useful as in
reversed-phase liquid chromatography. At low con-
centrations modiRer effects are small and not
strongly dependent on solvent identity, and at higher
concentrations they lead to deleterious effects on
system efRciency and the separation time. By con-
trast, the use of complexing additives, such as urea
and cyclodextrins has to be considered one of the
success stories of MEKC for achieving the separation
of isomers, enantiomers, and other difRcult to
separate compounds capable of forming suitable in-
clusion complexes. Figure 4 provides an example of
the separation of pharmaceutically important estro-
gens that were only adequately separated in the sys-
tem containing the complexing additive. The incor-
poration of low molecular mass organic solvents and
cyclodextrins in the micelles is very low. Their main
effect on the distribution properties of the system

Figure 4 Separation of estrogens by MEKC using a 20 mM sodium phosphate-borate pH 8 buffer containing 50 mM sodium dodecyl
sulfate (A) and the same buffer containing 20 mM -cyclodextrin (B). Separation conditions: capillary 48.5 cm (effective length 40 cm),
internal diameter 50
m, temperature 253C, and field strength 20 kV. Compounds: 1"estriol; 2"17-estradiol; 3"17-estradiol;
4"17-dihydroequilenin; 5"17-dihydroequilenin; 6"17-dihydroequilenin; 7"17-dihydroequilin; 8"estrone; 9"equilenin;
and 10"equilin. (Modified from Poole SK and Poole CF (1996) Separation of pharmaceutically important estrogens by micellar
electrokinetic chromatography. Journal of Chromatography A 749: 247}225, with permission from Elsevier Science.)
4590 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS
is due to changes in the relative solubility of the
analytes in the electrolyte.
Capillary Gel Electrophoresis
Capillary gel electrophoresis is used for the separ-
ation of macromolecules such as proteins and nucleic
acids, whose mass-to-charge ratios do not vary much
with size. Separation requires a sieving medium made
up of a crosslinked gel or an entangled polymer net-
work. The capillaries are often wall-coated or chem-
ically bonded to minimize electroosmotic Sow that
tends to destabilize the columns. Columns Rlled with
rigid crosslinked gels, usually polyacrylamide, are
characterized by the total amount of monomer and
crosslinking agent (%T) and the ratio of crosslinking
agent to total amount of monomer and crosslinking
agent (%C) used to prepare the column. Larger pore
size gels (lower %T) are used for separating DNA
sequencing reaction products whereas the narrow-
pore media are best for proteins and small oligonuc-
leotides. Entangled polymer networks of linear
polyacrylamide, methylcellulose or dextran have the
advantage that they can be forced into the capillary as
a solution and replaced when needed. Unlike gels,
columns are easily prepared in the laboratory and
tend to the be more robust. Electrokinetic injection is
used for sample introduction. The buffer pH is
selected such that the analytes of interest are ionized.
TRIS/borate and TRIS/phosphate buffers in the
pH range 7.5 to 8.5 (50}200 mM) are fairly general
conditions. Sometimes urea (7}8 M) or ethylene
glycol (1.5}3.0 M) is added to the buffer as
a nonionic denaturing or solubilizing agent and
EDTA (about 2 mM) to protect against cation inter-
ferences. When SDS-proteins are separated sodium
dodecyl sulfate (0.1% w/v) is added to the run buf-
fer. For many practical applications of capillary gel
electrophoresis the column materials and reagents
required can be purchased in kit form.
Capillary Isoelectric Focusing and
Isotachophoresis
Capillary isoelectric focusing is used to separate poly-
peptides based on differences in their isoelectric
points (pI) in wall-coated fused silica capillaries to
eliminate electroosmotic Sow and nonspeciRc ad-
sorption of the sample with the capillary wall. The
capillary is Rlled with the sample and a mixture of
ampholytes capable of producing a pH gradient that
covers the pI values of the proteins. Ampholytes are
a mixture of hundreds to thousands of amphoteric
compounds, generated by the random addition of
acrylic acid to a mixture of linear and branched
oligoamines, providing pI values are fairly well dis-
tributed along the pH scale from 3 to 10. In practice
about 94% of proteins can be separated in the
pH range 3}8.5. This allows a single capillary to
be used for hundreds of separations by minimizing
alteration to the capillary wall coating. When a volt-
age is applied (e.g. 15 kV for 4 min) the sample
components focus into narrow zones according
to their pI values. The zones are then mobilized
by hydraulic, electroosmotic or ion addition (by
adding 80 mM sodium chloride to either the source
or destination vial and applying an electric Reld)
to move them past the detector. The destination
vial contains a buffer (catholyte) at a pH higher
than the pI of the most basic ampholyte (40 mM
sodium hydroxide) and the source vial contains a buf-
fer (anolyte) at a pH lower than the pI of the most
acidic ampholyte (20 mM phosphoric acid). To
avoid protein precipitation in the focused zones a
surfactant or urea can be added to the buffer,
the sample diluted, or gel-Rlled capillaries can be
used.
In capillary isotachophoresis sample ions are separ-
ated by differences in their mobility in a hetero-
geneous buffer system created by sandwiching
the sample between a leading and terminating
buffer with different and speciRed compositions.
It is general practice to separate mixtures in the
constant current mode using chemically bonded or
dynamically coated capillaries to eliminate electroos-
motic Sow. As well as fused silica capillaries of stan-
dard dimensions wide-bore TeSon (0.5}0.8 mm)
tubes have been used in purpose-built apparatus for
isotachophoresis. Before commencing the separation
both the capillary and destination buffer vial is
Rlled with the leading electrolyte (assuming sup-
pression of the electroosmotic Sow). The leading elec-
trolyte ion must have a higher mobility than any of
the analytes to be separated and the counterion must
be able to set the pH for the separation by ensuring
sufRcient (but generally not complete) dissocia-
tion of weak acids and bases in their own zones.
Either sample cations or anions can be determined in
the separation but not both simultaneously. The ter-
minating electrolyte is placed in the source vial and
should have a lower mobility than any of the analyte
ions. Recommendations for buffer selection and
operating conditions are summarized in Table 8. If
solubility is a problem nonionic or zwitterionic sur-
factants or urea can be added to both the leading
electrolyte and the sample. When fused silica capillar-
ies are used hydroxypropylmethylcellulose, polyethy-
lene glycol or polyvinyl alcohol can be added to the
buffers to suppress electroosmotic Sow through
dynamic coating of the column wall. Detection of
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN CAPILLARY ELECTROPHORESIS 4591

Table 8 Composition of some common capillary isotachophoresis buffersH

Property pH

2.0 3.3 4.5 6.0 8.8

Separation Cations Anions Cations Anions Anions

Leading ion 10 mM HCl 10 mM HCl 10 mM KOAc 10 mM HCl 10 mM HCl
Leading counterion -Alanine HOAc Histidine Ammediol
Leading additive 0.2% HPMC 0.2% HPMC 0.2% HPMC
Terminating ion 10 mM TRIS 10 mM caproic acid 10 mM HOAc 10 mM MES 10 mM -Alanine
Terminating counterion HCl TRIS Ba(OH)2
Terminating pH 8.5 6.0 9.0

Recommendations
Cations Anions
Leading ion K#, NH# 4 , Na
#
Cl\
(20}30 mM)
Terminating ion H#, or weak base OH\, or weak acid
(mobility'H#) (mobility'OH\)
Terminating counterion Weak acid, Weak base,
pK"pHL$0.5 pK"pHL$0.5
Typical counterions pHL pHL
Formate 3.2}4.2 -Alanine 3.1}4.1
Acetate 4.2}5.2 Histidine 5.5}6.5
MES 5.7}6.7 Imidazole 6.6}7.6
Glycine 9.1}10.1 TRIS 7.6}8.6
Ethanolamine 9.0}10.0

See Table 4 for buffer abbreviations; Ammediol"2-amino-2-methyl-1,3-propanediol; HPMC"hydroxypropylmethylcellulose; and

OAc"acetate.

the separated zones is usually by conductivity or
UV-visible absorption. The method has high peak
capacity since separated zone boundaries are sharp
and close to each other to maintain continuity of the
current. When the experimental conditions are cor-
rect a steady state is reached in which all zones are
migrating at the same speed and the detector output is
a series of steps, the length of which corresponds to
the concentration of the ion. At Rrst sight the data
presentation may be confusing and this combined
with the complex method development has sup-
pressed interest in capillary isotachophoresis in
favour of other chromatographic methods. The com-
pelling advantage of isotachophoresis is its ability to
trace enriched dilute samples, by 100-fold or more,
and as a preconcentration or preseparation technique
for capillary zone electrophoresis it is enjoying some-
thing of a renaissance.
Conclusions
The capillary electrophoretic methods are sufR-
ciently established to ensure their continued laborat-
ory use but not so mature that signiRcant develop-
ments are unexpected in the near future. These devel-
opments are likely to be application driven and will
impact on the method development process. New
systems for separation of biopolymers using gels and
entangled polymers, a wider range of surfactants for
selectivity optimization in micellar electrokinetic
chromatography, and tailor-made sorbents for selec-
tivity optimization and control of electroosmotic Sow
in electrochromatography are just some expected
improvements. Better models for predicting sample
migration should aid computer-aided method devel-
opment strategies and experimental design ap-
proaches for multiparameter optimization of com-
plex mixtures should grow in popularity.
Further Reading
Baker DR (1995) Capillary Electrophoresis. New York:
Wiley-Interscience.
Bossi A, Olivieri E, Castelletti L, GelR C et al. (1999)
General experimental aspects of the use of isoelectric
buffers in capillary electrophoresis. Journal of Chrom-
atography A 853: 71}82.
Doble P and Haddad PR (1999) Indirect photometric detec-
tion of anions in capillary electrophoresis. Journal of
Chromatography A 834: 189}212.
Jimidar M, Yang Q, Smeyers-Verbeke J and Massart DL
(1996) Method development and optimization for small
ion capillary electrophoresis. Trends in Analytical
Chemistry 15: 91}102.
Kaniansky D, Nasar M, Marak J and Bodor R. (1999)
Capillary electrophoresis of inorganic ions. Journal of
Chromatography A 834: 133}178.
4592 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION
Krivankova L and Bocek P (1997) Synergism of capillary
isotachophoresis and capillary zone electrophoresis.
Journal of Chromatography B 689: 13}34.
McLaughlin GM, Weston A and Hauffe KD (1996)
Capillary electrophoresis methods development and sen-
sitivity enhancement strategies for the separation of in-
dustrial and environmental chemicals. Journal of
Chromatography A 744: 123}134.
Muijselaar PG, Otusuka K and Terabe S (1997) Micelles as
pseudo-stationary phases in micellar electrokinetic chro-
matography. Journal of Chromatography A 780: 41}61.
Poole CF and Poole SK (1997) Interphase model for
retention and selectivity in micellar electrokinetic
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN EXTRACTION
J. R. Dean, University of Northumbria at Newcastle,
Newcastle upon Tyne, UK
Copyright ^ 2000 Academic Press
Introduction
Samples for extraction can be broadly categorized as
solid, liquid or gaseous matrices. It is obvious that the
different methods of extraction of analytes from
Figure 1 Sample preparation protocol.
chromatography. Journal of Chromatography A 792:
89}104.
Reijenga JC, Verheggen TPEM, Martens JHPA and
Everaerts FM (1996) Buffer capacity, ionic strength
and heat dissipation in capillary electrophoresis. Journal
of Chromatography A 744: 147}153.
Rodriguez-Diaz R, Zhu M and Wehr T (1997) Strategies to
improve performance of capillary isoelectric focusing.
Journal of Chromatography A 772: 145}160.
Watzig H, Matthias D and Kunkel A (1998) Strategies for
capillary electrophoresis: method development and vali-
dation for pharmaceutical and biological applications.
Electrophoresis 19: 2695}2752.
these matrices will also vary. This guide provides an
overview of the different approaches for extrac-
tion of analytes from these different matrices.
It is important to consider that extraction is only
one part of the sample preparation protocol. Other
steps are highlighted in Figure 1. A typical solid
sample is most likely to be heterogeneous. This is
a problem in the analysis, if appropriate steps have
not been taken to remove a representative sample
using a statistical approach. Failure to do so can make
any subsequent extraction and analysis results mean-
ingless.
Also of relevance to any subsequent extraction and
analysis is whether the sample has been stored (and
preserved, if necessary) in the appropriate manner to
prevent losses of the analyte due to degradation
and/or adsorption. It is necessary to consider, in the
Figure 2 Extraction of analytes from solid matrices.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4593

Figure 5 Shake-flask extraction.

Figure 3 Soxhlet extraction.

Solid Matrices
Extraction of analytes from solid matrices can be
case of a soil sample, whether it should be dried
classiRed according to the scheme shown in Figure 2.
(volatile analytes may be lost) or extracted in the
The main extraction techniques are Soxhlet extrac-
unadulterated state. If possible, drying is favoured, as
tion, soxtec extraction, supercritical Suid extraction
the subsequent matrix can be ground and sieved to
increase its surface area (smaller particle size).

Figure 4 Ultrasonic extraction. Figure 6 Supercritical fluid extraction.

4594 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION
Figure 7 Microwave-assisted extraction.
(SFE), pressurized microwave-assisted extraction
(pMAE), atmospheric microwave-assisted extrac-
tion (aMAE), pressurized Suid extraction (PFE) or
accelerated solvent extraction (ASE), ultrasonic ex-
Figure 8 Pressurized fluid extraction (or accelerated solvent
extraction).
Figure 9 Matrix solid-phase dispersion.
traction, shake-Sask extraction and matrix solid
phase dispersion (MSPD). Method development
approaches for each extraction technique are shown
in Figures 3}10.
Liquid Matrices
Liquid extraction approaches are essentially centred
around methods of preconcentration. Typically, this
involves the use of sorbent and/or an organic solvent.
The choice of solvent/organic solvent depending
upon the nature of the analyte, e.g. polar/nonpolar.
The main extraction approaches are liquid}liquid
Figure 10 Thermal desorption.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION
Figure 11 Separating funnel liquid}liquid extraction.
extraction (LLE), solid-phase extraction (SPE) and
solid-phase microextraction (SPME). A guide to
method development for each extraction technique is
shown in Figures 11}14.
Figure 12 Solid-phase extraction.
4595
Figure 13 Solid-phase microextraction.
Gaseous/Atmospheric Samples
Gaseous samples are normally analysed by gas
chromatography (GC). The volatile nature of the
analytes means that some form of trapping is re-
quired. Typical approaches for analyte trapping are
shown in Table 1.
Solvent Selection
Extraction of an analyte is dependent upon overcom-
ing any interactions between the matrix with the
extraction technique. These interactions, for organic
molecules, are predominantly based on weak forces
of attraction between the analyte and the matrix, e.g.
4596 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Table 1 Common approaches for gaseous samples

Technique Comments

Solid phase trapping
Liquid trapping
Headspace sampling
Purge and trap
Solid-phase microextraction
Gaseous sample passed through a sorbent, e.g. Tenax, activated charcoal, etc. Trapped analytes
are eluted with a suitable solvent.
Gaseous sample is bubbled through a suitable trapping solvent. To improve trapping efficiency it is
important to minimize the flow rate and/or lower the temperature. The use of multiple traps or
impingers may be necessary.
Solid or liquid sample placed in a sealed glass vial until equilibrium is reached. Volatile analytes
sampled from the headspace using a gas-tight syringe or solid-phase microextraction.
See Figure 14.
See Figure 14 and Headspace sampling, above.

Table 2 Calculation of individual group contributions for a solvent (methanol) and the analyte, DDT

Molecule Group Group contribution to Group contribution to Group contribution to Molar volume (V )
dispersion (Fd) polarity (Fp) hydrogen bonding (Uh) cm3 mol\1
J 1/2 cm3/2 mol\1 J1/2 cm2 mol\1 J mol\1

Methanol CH3 420 0 0 33.5

OH 210 500 20 000 10.0
Total 630 500 20 000 43.5
DDT 2;-Ph- 2540 220 0 104.8
2;Cl-CH" 900 1100 800 48
3;Cl 1350 1650 1200 72
1;CH 80 0 0 !1.0
'C( !70 0 0 !19.2
Total 4800 2970 2000 204.6

Table 3 Total Hildebrand solubility parameter and its individual components

Solvent/analyte Dispersion coef[cient, Polarity, Hydrogen bonding, Total Hildebrand

d (MPa1/2)
p (MPa1/2)
h (MPa1/2) solubility parameter,

t (MPa1/2)

Methanol 14.48 11.49 21.44 28.31

Acetonitrile 14.78 19.13 6.59 25.06
Acetone 14.52 9.90 5.07 18.29
Dichloromethane 18.25 8.58 3.53 20.48
iso-Hexane 14.27 0.00 0.00 14.27
DDT 23.46 9.75 3.13 25.60

Van der Waals, hydrogen bonding, etc. While the
choice of extraction technique is important, often for
economic and environmental concerns, its phys-
ical/chemical properties are largely inSuenced by the
choice of solvent (in most cases). This is not to say
that the effects of heat, pressure, agitation and
sorbent are negligible, but that these on their own are
largely unimportant without the presence of an or-
ganic solvent and that the choice of solvent is critical.
Apart from general rule of thumb guidelines for sol-
vent selection, i.e. like extracts such as a nonpolar
analyte can be extracted by a nonpolar solvent, little at-
tempt has been made to offer a scientiRc approach.
The solvent prediction scheme used is based on the
Hildebrand solubility parameter (
t). The solubility
parameter is a measure of the internal energy of
cohesion in the solvent/solute. Solvents with similar
solubility parameter form mixtures, hence an analyte
and a solvent that have similar solubility parameters,
should also form mixtures.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4597

Table 4 Pressurized fluid extraction of DDT from contaminated

soil followed by GC-MSD quantitation, n"6a

Solvent Mean g g\1 SD

Methanol 89 10.1
Acetone 163 7.4
Dichloromethane 220 13.9
Acetonitrile 65 2.9
Iso-Hexane 120 4.4

a
Extraction conditions: sample size 2 g; temperature, 1003C;
pressure 2000 psi; static extraction time 10 min; one static/flush
cycle.

three components:
h, hydrogen bonding ability con-

tribution;
d, dispersion co-efRcient contribution;
and,
p, polarity contribution. They are linked by the
following equation:

2t"
2h#
2p#
2d [2]
Figure 14 Purge and trap.
The individual components of
t can be determined
using a group contribution additive method. The data

t is deRned as the square root of the cohesive available allows each groups contribution to polar-
energy density or: ity, dispersion and hydrogen bonding (Fp, Fd, and Uh,
respectively) to be calculated using the following

t"(Ev/V)1/2 [1] equations
p,
h, and
d:

where
t"total Hildebrand solubility parameter,
d"(zzFd)/V [3]

Ev"energy of vaporization at a given temperature
and V"molar volume of the molecule.
p"(zzFp)/V [4]
Hansen (1967) took this work further and assumed
that the total cohesive energy is a linear addition of
p"(zzF2p)1/2/V [5]

h"((zzUh)/V)1/2 [6]

For molecules with more than one polar group pres-

Figure 15 Comparison of calculated solvent and analyte frac-
tional parameters.
ent, then eqn [5] must be used instead of eqn [4] to
take into account the interactions between the polar
groups.
An example calculation of the individual compo-
nents of the solubility parameter for a solvent (meth-
anol) and an analyte (1,1,1-trichloro-2-2-bis(p-
chlorophenyl)ethane (DDT)) are shown in Table 2.
The individual Hansen parameters for a range of
solvents and an analyte (DDT) are shown in Table 3.
As an example, the calculated total Hildebrand solu-
bility parameter,
t, for methanol (28.3 MPa1/2)
compared favourably with the literature value of
29.6 MPa1/2.
In order to normalize the data, fractional para-
meters of the Hildebrand solubility parameter can
be calculated and plotted on a triangular graph in
order to give a visual representation of the extent of
contribution from the three components (polarity,
4598 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Box 1 Soxhlet extraction of polycyclic aromatic hydrocarbons from contaminated soil.

Extraction conditions
Sample size: 10 g plus 10 g anhydrous sodium sulfate
Solvent: 150 mL dichloromethane
Extraction time: 24 h
Comments: sample heated using an isomantle. Typically, refluxing of solvent occurs at the rate of 4 h\1.
Extracts were concentrated to 10 mL using rotary evaporator and then diluted twofold before addition of the
internal standards.

Analysis of extracts by GC
Separation and identification of individual PAHs was done on a HP 5890 series II#GC fitted with a HP 5972A mass spectrometer.
A 30 m;0.25 mm i.d.;0.25
m film thickness DB-5 capillary column was used with temperature programming from an initial
temperature held at 853C for 2 min before commencing a 63C min\1 to 3003C, with a final time of 7 min. The split/splitless injector
was held at 3003C and operated in splitless mode with the split value closed for 1 min following sample injection. The split flow was
set at 40 mL min\1, and the mass spectrometer transfer line was maintained at 2703C. Electron impact ionization at 70 eV with an
electron multiplier voltage set at 1500 V was used while operating in single-ion monitoring (SIM) mode.

Typical results: Saim N, Dean JR, Abdullah MP and Zakaria Z (1997) Journal of Chromatography 791A: 361.

dispersion and hydrogen bonding). A plot of selected Selected Examples of Extraction of

solvents and DDT is shown in Figure 15. Using this
plot, it can be seen that dichloromethane (DCM) is
predicted to be the optimum solvent for extraction of
DDT. Table 4 shows results for the extraction of
DDT contaminated soil for selected solvents using
accelerated solvent extraction (ASE). It is clearly
shown that DCM gives the highest recovery of DDT.
Similarly, it is also predicted and conRrmed that both
isohexane and acetone would remove signiRcantly
more of the DDT than methanol and acetonitrile.
Work is on-going to identify whether the model can
be applied to other systems.
Analytes from Environmental Matrices
In order to provide speciRc details on particular ex-
traction techniques selected examples are provided
from the authors own laboratory. In particular, the
following techniques are covered: Box 1, Soxhlet
extraction of polyaromatic hydrocarbons (PAHs)
from contaminated soil; Box 2, shake Sask extrac-
tion of four phenols from soil; Box 3, SFE of OCPs
from soil and Celite; Box 4, pressurized microwave-
assisted extraction of PAHs from soil; Box 5, atmo-
spheric microwave-assisted extraction of PAHs from
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4599

Box 2 Shake flask extraction of phenols from soil

Extraction conditions
Sample size: 1 g
Solvent: 50 mL methanol}water (60}40% v/v)
Extraction time: 30 min
Comments: Sample and solvent placed in a 100 mL screw-capped bottle and extracted on a rotating disc Warburg
mixer. Resultant sample/solvent was filtered under vacuum. Sample extracted filtered through a 0.45
m membrane
Acrodisk prior to analysis.

Analysis by HPLC
Separation and quantification was achieved using a 25 cm;4.6 mm i.d. ODS2 column with UV detection at 275 nm. The
mobile phase was operated under isocratic conditions acetronitrile}H2O}acetic acid (40#59#1) at a flow rate of
1 mL min\1. A 20
L Rheodyne injection loop was used to introduce samples and standards on to the column (303C).

Typical results: Hancock P and Dean JR (1997) Analytical Communications 34: 377.

soil; Box 6, pressurized Suid extraction of DDT,
1,1-dichloro-2,2-bis(p-chlorophenyl)ethane (DDD)
and 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene
(DDE) from soil; Box 7, liquid}liquid extraction of
PAHs from water; Box 8, SPE of phenols from water;
Box 9, solid-phase microextraction of benzene,
toluene, ethyl benzene and xylene (BTEX) from
water; and, Box 10, purge and trap of BTEX from
water.
Further details on the theoretical and technical
aspects of these and other extraction techniques can
be found in the relevant entries in the Encyclopedia.
4600 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Box 3 Supercritical fluid extraction of organochlorine pesticides from soil and Celite

Extraction conditions
Sample size: 1 g
SFE conditions: pressure, 250 kg cm\2; temperature, 503C; static
extraction time, 15 min followed by 40 min dynamic extraction time;
and a flow rate of liquid CO2, 2 mL min\1.
Comments: Extracts collected in a vial containing 3}4 mL DCM. Es-
caping CO2 and analytes vented through a C18 SPE cartridge which
was back-flushed with 1}2 mL methanol after each extraction.

Analysis by GC
Separation and identification of individual OCPs was done on a HP 5890 series II#GC fitted with a HP 5972A mass
spectrometer. A 30 m;0.25 mm i.d.;0.25
m film thickness DB-5 capillary column was used with temperature programming
from an initial temperature held at 853C for 0.75 min before commencing a 163C min\1 to 2853C, with a final time of 2 min. The
split/splitless injector was held at 2803C and operated in splitless mode with the split valve closed for 1 min following sample
injection. The split flow was set at 40 mL min\1, and the mass spectrometer transfer line was maintained at 2903C. Electron
impact ionization at 70 eV with an electron multiplier voltage set at 1500 V was used while operating in single-ion monitoring
(SIM) mode.

Typical results: Dean JR, Barnabas IJ and Owen SP 1996 Analyst 121: 465.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4601

Box 4 Pressurized microwave-assisted extraction of polycyclic aromatic hydrocarbons (PAHs) from soil.

Extraction conditions
Sample size: 2 g
Solvent: 40 mL acetone
pMAE conditions: power, 30% (for a 950 W system); temperature, 1203C;
extraction time, 20 min.
Comments: After extraction, extraction vessels allowed to cool.
Contents of vessels were then filtered through a GF/A glass microbore filter.
Extracts were concentrated to 5 mL using a rotary evaporator before addition
of internal standards.

Analysis by GC
Separation and identification of individual PAHs was done on a Carlo Erba HRGC 5300 Mega Series with on-column injection
and flame ionization detection. A 30 m;0.32 mm i.d.;0.1
m film thickness DB-5 HT capillary column was used with
temperature programming from an initial temperature held at 503C for 2 min before commencing a 153C min\1 to 903C;
hold for 2 min; increase at 63C min\1 to 3003C with a final hold time of 8 min. The detector temperature was set at 2903C.

Typical results: Saim N, Dean JR, Abdullah MP and Zakaria Z (1997) Journal of Chromatography 791A: 361, with
permission from Elsevier Science.
4602 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Box 5 Atmospheric microwave-assisted extraction of polycyclic aromatic hydrocarbons (PAHs) from soil.

Extraction conditions
Sample size: 2 g
Solvent: 70 mL DCM
pMAE conditions: power, 99% (for a 300 W system); extraction
time, 20 min.
Comments: Contents of extraction vessel was then filtered through
a GF/A glass microbore filter. Extracts were concentrated to 5 mL
using a rotary evaporator before addition of internal standards.

Analysis by GC
Separation and identification of individual PAHs was done on a Carlo Erba HRGC 5300 Mega Series with on-column
injection and flame ionization detection. A 30 m;0.32 mm i.d.;0.1
m film thickness DB-5 HT capillary column was
used with temperature programming from an initial temperature held at 503C for 2 min before commencing a 153C min\1
to 903C; hold for 2 min; increase at 63C min\1 to 3003C with a final hold time of 8 min. The detector temperature was set
at 2903C.

Typical results: Saim N, Dean JR, Abdullah MP and Zakaria Z (1997) Journal of Chromatography 791A: 361, with
permission from Elsevier Science.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4603

Box 6 Pressurized fluid extraction of DDT, DDD and DDE from soil.

Extraction conditions
Sample size: 2 g
PFE conditions: pressure, 2000 psi; temperature, 1003C; static
extraction time, 10 min; and three static/flush cycles.
Comments: Sample placed in stainless steel extraction cell
on top of a filter to prevent cell frit blockage. Hydromatix was
used to fill the headspace to reduce solvent consumption.

Analysis by GC
Separation and identification of DDT, DDD and DDE was done on a HP 5890 series II#GC fitted with a HP 5972A
mass spectrometer. A 30 m;0.25 mm i.d.;0.25
m film thickness DB-5ms capillary column was used with temperature
programming from an initial temperature held at 1203C for 2 min before commencing at 53C min\1 to 2903C, with a final time
of 2 min. The split/splitless injector was held at 2803C and operated in splitless mode. The mass spectrometer transfer line
was maintained at 2803C. Electron impact ionization at 70 eV with an electron multiplier voltage set at 1500 V was used while
operating in single-ion monitoring (SIM) mode.

Typical results: Fitzpatrick LJ and Dean JR (2000) Journal of Chromatography, in press.

4604 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Box 7 Liquid}liquid extraction of polyaromatic hydrocarbons (PAHs) from water

Extraction conditions
Sample volume: 25 mL
LLE conditions: sample extracted with 2;3 mL of DCM plus 1 g salt (NaCl). Each extract was shaken for
5 min each.
Comments: Combined extracts placed in a volumetric flask, internal standard added, prior to analysis.

Analysis by GC
Separation and identification of individual PAHs was done on a HP 5890 series II GC fitted with a HP 5971A mass spectro-
meter. A 30 m;0.25 mm i.d.;0.25
m film thickness HP-5ms capillary column was used with temperature programming from
an initial temperature held at 903C for 2 min before commencing a 73C min\1 to 2853C, with a final time of 20 min. The split/
splitless injector was held at 2803C and operated in splitless mode with the split valve closed for 1 min following sample
injection. The split flow was set at 40 mL min\1, and the mass spectrometer transfer line was maintained at 2803C. Electron
impact ionization at 70 eV with an electron multiplier voltage set at 1500 V was used while operating in single-ion monitoring
(SIM) mode.

Typical results: Arenaz-Laborda MP (1998) MSc dissertation, University of Northumbria at Newcastle, UK.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4605

Box 8 Solid phase extraction of phenols from water.

Extraction conditions
Sample volume: 25 mL
SPE sorbent: PS-DVB, 230 mg
SPE conditions: conditioning, 5 mL of acetonitrile followed by 5 mL of water;
sample loading; interference elution, 2 mL of water; and analyte elution,
4 mL of acetonitrile.
Comments: sample extract made up to 10 mL with water.

Analysis by HPLC
Separation and quantitation was achieved using a 25 cm;4.6 mm id ODS2 column with UV detection at 275 nm. The mobile
phase was operated under isocratic conditions acetonitrile}H2O}acetic acid (40#59#1) at a flow rate of 1 mL min\1.
A 100
L Rheodyne injection loop was used to introduce samples and standards on to the column (353C).

Typical results: Madier C (1997) BSc project, UNN, Newcastle upon Tyne, UK.

Analysis of phenol, 4-nitrophenol and 2-methylphenol.

Calibration range: 0}400 ng mL\1
Correlation coefficients: 0.9993}0.9979.
4606 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION

Box 9 Solid phase microextraction of BTEX from water.

Extraction conditions
Sample volume: 10 mL
Fibre: 100
m polydimethylsiloxane
Conditions: SPME: fibre inserted into either the sample or headspace above the sample
(with/without stirring; with/without salt) for varying amounts of time.

Analysis by GC
Separation and identification of BTEX was done on a Carlo Erba HRGC 5300 Mega Series with split/splitless injection and flame
ionization detection. A 30 m;0.25 mm i.d.;0.1
m film thickness DB-5 capillary column was used with temperature
programming from an initial temperature held at 503C for 3 min before commencing a 163C min\1 to 1203C with a final hold time of
7 min. The detector temperature was set at 2503C.

Typical results: Ahmed HK (1996) MSc dissertation, University of Northumbria at Newcastle, UK.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN EXTRACTION 4607

Box 10 Purge and trap (P&T) of BTEX from water.

Extraction conditions
Sample volume: 2}10 mL
P&T conditions: Sample sparged for 2}5 min using N2.
BTEXs trapped on Tenax trap maintained at 203C for 1}5 min.
Analytes desorbed by rapid heating to 2603C for 1 min.
Comments: GC column initially maintained at 503C to concentrate
analytes.

Analysis by GC
Separation and identification of BTEX was done on a Carlo Erba HRGC 5300 Mega Series with split/splitless injection and
flame ionization detection. A 30 m;0.25 mm i.d.;0.1
m film thickness DB-5 capillary column was used with temperature
programming from an initial temperature held at 503C for 3 min before commencing a 163C min\1 to 1203C with a final hold
time of 7 min. The detector temperature was set at 2503C.

Typical results: Ahmed HK (1996), MSc dissertation, University of Northumbria at Newcastle, UK.

See also: I/Extraction; Chromatography: Thin-Layer
(Planar): Theory of Thin-Layer (Planar) Chromatography.
Extraction: Analytical Extractions; Analytical Inorganic
Extractions; Microwave-Assisted Extraction; Solid-Phase
Extraction; Solid-Phase Microextraction; Solvent Based
Separation; Steam Distillation; Supercritical Fluid Extrac-
tion; Ultrasound Extractions. III/Airborne Samples:
Solid-Phase Extraction. Bioanalytical Applications:
Solid-Phase Extraction. Drugs of Abuse: Solid-Phase
Extraction. Environmental Applications: Solid-Phase
Microextraction; Soxhlet Extraction; Supercritical Fluid
Extraction. Herbicides: Solid-Phase Extraction. Im-
mobilised Boronic Acids: Extraction. Immunoaffinity
Extraction. Molecular Imprints for Solid-Phase Extrac-
tion. Multiresidue Methods: Extraction. On-line
Sample Preparation: Supercritical Fluid Extraction.
Pesticides: Extraction from Water. Phenols: Solid-Phase
Extraction. Pressurized Fluid Extraction: Non-Environ-
mental Applications. Solid-Phase Extraction with
Discs. Sorbent Selection for Solid-Phase Extraction.
Appendix: 2/Essential Guides to Method Develop-
ment in Solid-Phase Extraction.
Further Reading
Barton AFM (1983) The Handbook of Solubility Para-
meters and other Cohesion Parameters. Boca Raton:
CRC Press Inc.
4608 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
Dean JR (1998) Extraction Methods for Environmental
Analysis. Chichester. John Wiley and Sons.
Handley AJ (ed.) (1999) Extraction Methods in Organic
Analysis. ShefReld: ShefReld Academic Press.
Hansen CM (1967) Journal of Paint Technology 39: 104.
Pawliszyn J (1997) Solid Phase Microextraction: Theory
and Practice. New York: Wiley-VCH.
Pawliszyn J (1999) Applications of Solid Phase Microextrac-
tion. Cambridge: Royal Society of Chemistry, Cambridge.
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN FLOTATION
E. Woodburn, UMIST, Manchester, UK
Copyright ^ 2000 Academic Press
General
This article is designed to develop methods for an
interested non-specialist, by showing how they can be
used as a basis for a Chemical Engineering Unit Op-
erations course.
Flotation is practised extensively in industry. The
technique requires a detailed knowledge in physical
metallurgy, the physical chemistry of surfaces, a com-
petence both in mathematics and practical hydro-
dynamics.
The operation is based simply on the attachment of
an air bubble to either a small or low-density particle,
or to a liquid droplet.
Method 1: Selective Separation
Mineral Sotation has by far the greatest usage, pro-
cessing 20 billion tons per year; however the process
of delinking newsprint is currently at about 25 mil-
lion tons per year and is expected to grow signiR-
cantly in the next decade. In these operations the
selective attachment of a bubble to the valuable or an
unwanted component of a particle is required. In
de-inking, this refers to the removal of ink particles
from cellulosic Rbres. For mineral processing, a high-
er degree of selectivity is required, to recover a valu-
able particle from a suspension of waste particles.
This operation is very seldom used on its own but is
part of a Sowsheet in which, after pretreatment which
includes size reduction, a solid suspension in water is
fed to the Sotation circuit.
In the circuit, cells may be arranged in sequence
with each successive cell treating the concentrate
Ramsey ED (1998) Analytical Supercritical Fluid Ex-
traction Techniques. London: Kluwer Academic Pub-
lishers.
Thurman EM and Mills MS (1998) Solid Phase Ex-
traction: Principles and Practice. New York: Wiley-
Interscience.
van Krevelen DW and Hoftzyer PJ (1976) Properties of
Polymers; Their Estimation and Correlation with
Chemical Structure. Amsterdam: Elsevier.
from the previous one to improve its purity; this is,
called roughing. The Rnal concentrate from the
rougher bank is fed to a bank of cleaning cells.
The reject stream from the last of the cleaning cells
is itself recycled to improve the Rnal recovery and
is called scavenging. The concentrate from the
Rnal scavenger stream is recirculated to the feed of the
Rrst of the rougher cells. The waste from the Rnal
scavenging cell is discharged as the overall plant
waste. This may be recycled, or treated to minimize
its environmental impact. The Rnal cleaner concen-
trate is essentially the plant product, although it may
also have to be processed possibly by recleaning and
drying.
In waste paper, de-inking the ink-rich stream tail-
ings appears in what in mineral processing is the
concentrate and the de-inked paper in what is usually
the mineral processing tailings.
Method 2: Non-Selective Separations
The other class of operations require only the non-
selective attachment of air bubbles to a particle/drop-
let, producing an aggregate of high buoyancy, so that
the attached material can be withdrawn from the top
of the Sotation vessel. Processes of this type include
the off-shore recovery of crude oil which may be
5}50% oil by volume, containing dispersed oil in the
form of 10}50
m oil droplets in water. After pro-
cessing, virtually all the oil is recovered containing
only 0}5% water. Other processing operations of this
class include water treatment, in which the rate of
setting of the Socculants on their own is very slow
while the buoyancy of the air bubble/Socculant is
high. Also the separation of rejected plastics from
general wastes is economically attractive, with poly-
ethylene terephthalate (PET), polyethylene (PE),
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
polyvinyl chloride (PVC) and polypropylene (PP) be-
ing recoverable.
The effectiveness of any Sotation separation is
described by the fractional recovery of the desired
material, R, in the concentrate stream and its purity,
the Grade G. The practical application of this is
a principle more honoured in the breach than the
observance, and is an area where signiRcant economic
improvements are possible.
Method 3: Measures of Separation
Ef\ciency Potentially Achievable
In mineral processing, it is convenient to represent
both quantities on a plot 04R, G41. The Grade
G is deRned as 1!(1!xC)/(1!xF) where xF and
xC are the mass fractions of the desired component in
the feed and concentrate streams respectively, and the
fractional recovery of the desired component
R"(CxC)/(FxF). A perfect separation is therefore one
in which R"G"1.
Actual simulations lie on the upper boundary of the
Grade}Recovery plot and describe the best R at a de-
Rned Grade. The area on the plot whose upper
boundary represents optimum operation is called the
attainable region. The function of the research
worker ultimately is to devise techniques whereby the
attainable region may be expanded. The Grade}
Recovery plot should also be used by operators to
monitor and control plant performance. The applica-
tions of automatic feed back control is attractive, but
the requirements of online instrumentation, such as
image processing and chemical analysis detectors, are
still in the development stage. This is particularly so
with the control actions which are necessary to be
able to compensate for deviations from optimum op-
erations. These include changing in air Sow rates and
the addition of chemicals. It is in this latter area that
there is a large measure of uncertainty, which should
be addressed as it again offers the possibility of
signiRcant economic improvements. The treated tail-
ings which are usually in great bulk still have to be
disposed of, at a signiRcant cost, sometimes requiring
slime dams to be built or as landRll. This is an envir-
onmental factor which may be very costly, both in the
operation of existing plants, and in assessing the vi-
ability of new projects. Recycling the waste is envir-
onmentally acceptable and may also be proRtable.
Method 4: Macro Dry Separations
It is signiRcant that siliceous materials can be Soated
as they may well be the source of gangue contamina-
tion in metallic ore Sotation. These gangues are in
general disadvantageous to the separation and may
4609
have to be depressed. However the recycling of simple
oxide materials such as corundum, haematite and
goethite may well be valuable although Soated at
lower rates.
Ore-dressing } although this is not strictly a Sota-
tion operation } effectiveness determines the lim-
iting separation achievable in the Sotation circuit. In
the Rrst ore-dressing step, separation is based on size
reduction and primary mechanical classiRcation.
Once again, the details of the circuit depend on the
nature and throughput of the raw material to be
processed. The general principles involve crushing
followed by dry separation in spirals (which are in
fact of a helical design). Crushing and dry separations
such as screening are relatively cheap and their use
should be maximized to achieve the cheapest possible
separation of high value materials from those which
are exclusively gangue. Of these dry separation
methods, the spirals depend on the difference in
density between the waste rock and the valuable
material. Vibratory screens may follow or be oper-
ated in parallel with the spirals. These are only useful
if the Rnes are largely gangue. The valuable-rich
stream from the dry separations is then fed to rod and
wet or dry ball mills, the product of which goes to
hydrocyclones whose underSow is a solid suspension
in water whose solids lie in the size range 50}500
m.
This is a size range at which the subsequent Sotation
operations will function satisfactorily. The Rne prod-
uct from the mills is aimed at producing two separate
powder streams in which there is a sharp change in
the valuable material content; this process is referred
as the liberation of the valuable material. The power
costs of milling are extremely high. The overSow
from the cyclones are called slimes and go to settling
tanks from which the Rnal solids and liquid wastes
may be recycled or discharged. These streams are the
primary source of environmental pollution and are
vulnerable to objections which may require new
treatment techniques.
Method 5: The Characterization of
the Solid Material
In Sotation, the complete characterization of the solid
material to be Soated, which varies considerably for
different materials, is fundamental to the separ-
ation. In paper de-inking the type of ink used and its
method of attachment to the waste Rbre, determines
the nature of the process required. This again is an
area in which the technology is still developing. In
mineral separations for example, the chemical type of
both the valuable material and the gangue have to be
identiRed, and it is also crucial to be able to identify
and determine the distribution of individual minerals,
4610 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
throughout the solid matrix of the primary ore.
A mineral may be present as individual grains whose
boundaries are a source of mechanical weakness in
the solid. This facilitates breakage at the grains fol-
lowing impact and also most interestingly, breakage
at the grains following differential thermal ef-
fects generated in a microwave Reld. Alternatively,
the mineral may be distributed uniformly throughout
the solid matrix. The initial characterization is usu-
ally done by microscope examination which, in the
hands of an experienced operator, is extremely in-
formative, but which usually has to be supplemented
by X-ray Suorescence (XRF) analysis.
Method 6: Wet Processing ^
Hydrodynamics of Cell Design
The practice of mineral beneRciation by Sotation is
based on the production of an aqueous suspension of
particles in the micron-size range and a dispersion of
bubbles in a millimetre-size range. It is the objective
within this suspension to achieve particle}bubble col-
lision which will be followed by a selective attach-
ment of the particles. In terms of the method of
interception/attachment there are two basic types of
cell.
Mechanical Cells
In mechanical cells there is a relatively small inner
turbulent region enclosed by a larger diameter quiesc-
ent zone. The turbulent region is generated by an
impeller which in addition has to disperse air into the
solid suspension, and then pump the aerated suspen-
sion into the quiescent zone. There are a plethora of
designs available which have to be carefully evalu-
ated. The design variations should be related to the
achievement of a desired interception/attachment ef-
Rciency as a function of particle size. These cells are
the most commonly encountered in industrial prac-
tice principally because they can have very high ca-
pacities, with single cells of up to 300 m3.
In the inner region the suspension is exposed to
a high level of turbulence, in which particle}bubble
interception depends on the different paths be-
tween the water eddies and the particle trajectories.
Although there are no theoretical calculations to pro-
vide a basis for these effects, it is assumed that
the interception efRciency is determined by the
local turbulent intensity and is increased for large
particle sizes with a large differential density
between the particle and water. Unfortunately, there
is a probability that large particles which have been
attached to the bubbles, following the turbulent inter-
ception may subsequently become detached from the
bubble surface. The detachment will be associated
with the overall strength of the attachment forces; as
these forces are surface effects and the speciRc
surface is low, their detachment probability is high.
For the small particles with a high speciRc surface
with an overall high attachment force, the detach-
ment probability is low. It is however, also apparent
that these smaller lighter particles, as they follow the
eddies closely, will have lower interception ef-
Rciencies. The net degree of attachment of particles to
bubbles will depend on the design of the agitators
(impellers).
Cylindrical Vertical Column Cells
The second cell type takes the form of a cylindrical
vertical column in which the suspension is dilute,
typically containing 5% solids. In these cells the at-
tachment of a desired particle is more speciRc than in
the mechanical cells. These cells may be between 3 m
in diameter and 15 m high. The cells are described by
the collection and froth zones. There is also an inter-
mediate zone between the top of the collection zone
and the froth. This zone functions very similarly to
the top of the collection zone. The solid suspension
from the milling circuit is added to the top of the
collection zone and the air as bubbles through a spar-
ger at the bottom. The bubbles and particles move
countercurrently through the relatively quiescent col-
lection zone. The hydrodynamics of the collection
zone are relatively easy to describe. Originally the
probability Pc of interception was given by Pc"3/2
(rp/rb)2 where rp and rb are the spherical radii of the
particle and bubble, respectively. The original equa-
tions have been approximately corrected to allow for
gravitational effects in terms of a parameter
K"2 (
pr2pU)/(9f rb) where U is the bubble rise velo-
city in water of viscosity f and
p is the particle
density. Pc indicates some interesting dependencies on
K, e.g. (i) there is a Kc which is approximately 1,
below which no collision will occur; the smallest size
galena particle that can collide with a 1.5-mm bubble
is 30
m; (ii) from the deRnition of K, when it is
'Kc, an increase in bubble size will increase the
collision efRciency and; (iii) very Rne particles
follow the streamlines exactly and will only collide if
the streamline brings them within 1 particle radius of
the bubble. Maximum velocities of 0.9, 1.5, 2.2 and
2.7 mm bubbles have been observed to be respective-
ly 25.0, 36.5, 3.47 and 32.0 cm s\1. However caution
has to be used using predictions based on K as
anomalies have been observed, and more accurate
equations are now available. Surface-active chemical
frothers, e.g. Dowfroth 250 and methyl isobutyl
ketone (MIBK), are normally added at concentrations
of the order of 30 ppm, these are to stabilize the Rnal
froth. The effect of the presence of these frothers
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
is to reduce the bubble rise velocity by about 50%.
The effect of the surfactants is to adsorb on the
bubble surface, increasing the surface viscosity of the
water near the bubble, and the bubble rigidity.
Method 7: The Recovery of a Speci\c
Particle Size
The effect of particle size is dependent on inter-
actions with various factors. The ultimate objective is
to remove particles of size dlib which is the size at
which maximum liberation occurs. Even if the dry
separation gives material with a perfect size distribu-
tion as the mill feed, there will be a distribution of
sizes in the product from the mills, which is the feed
to the Sotation circuit. If this particle size distribution
is f(dp), the fractional mass of the size dp, as character-
ized by their valuable content, then the Sotation cir-
cuit should aim to maximize the recovery of particles
of this size and sizes close to it. Clearly f(dlib) should
have a maximum and a small standard deviation.
Ideally this could take the form of a delta function
giving f(dlib)"(dlib)"1.0. The following step is to
ensure what spread in the size distribution in the
removal of these valuable particles occurs in the Sota-
tion circuit. This will depend on the speciRcity of the
interception/attachment efRciency achievable for
these particles, with an f(d) size distribution, in a So-
tation circuit.
The calculation for a speciRc circuit will depend on
the performance of its cells. Consider the speciRcity of
removal of particles in a single mechanical cell. In
addition to the interception/attachment in the turbu-
lent zone which has a spread of efRciences, there
is also attachment of particles in the quiescent (pulp)
zone which may have a different distribution.
There is also a restriction on the new upSow rate
through the pulp; if it is too high, non-attached par-
ticles may be entrained, and if too low the
bubble}particle aggregates may settle. The design
problem for the cells is Rrstly to design an
impeller which will produce to the optimum size
range of valuable particles, while pumping water at
a sufRcient rate into the pulp for the necessary
upSow in the quiescent zone, which as an optimum
bubble size distribution, to maximize the recovery of
f(dlib).
The bubble}particle contact in mechanical cells is
sensitive to variations in the size of both, and the cell
has to be designed for optimal removal of particle
sizes with a high degree of liberation. It is clear that
the requirement of the maximum removal of valuable
material is not easily met. The mill product will have
a distribution of particle sizes, each size with a vary-
ing valuable content. The Sotation circuit separation
4611
should also separate the most valuable particles of
a size close to dlib. The design challenge is to design
milling and Sotation circuits capable of producing
and removing a speciRc size. It is also the responsibil-
ity of the operators to ensure that the system is oper-
ated continuously at its optimum level. These are
realistic objectives.
The separations achievable in columns is poten-
tially far more selective than that of the mechanical
cells but are restricted by a narrow operating range.
The most effective particle size in a column is of
the order of 75
m with bubble sizes varying from
about 0.8 to 2.5 mm. This size range may not be
consistent with the peak liberation size. The height of
the columns is determined by the need for bubbles to
accelerate from the bottom to their terminal velocity,
where interception is a maximum. At these conditions
the loaded bubbles will entrain into the froth signiR-
cant amounts of solution containing gangue. The
most effective feature of the columns is the re-
moval of these waste solids by adding wash water to
the top of the deep froth column. The froth at its top
surface will overSow with a maximum grade. In in-
dustrial practice, columns are used to upgrade (clean)
the froth concentrate from the primary cells
(roughers).
Method 8: Pulp Microprocesses
Attachment following the interception of a particle by
a bubble depends on the magnitude of the surface
forces between them. The characterization of these
forces to generate selective attachment is possibly the
key factor in the separation.
The particles are always dispersed in water and the
nature of the wetting of their surfaces determines the
effectiveness of air bubble attachment. For hydro-
phobic surfaces, the water Rlm is weakly bound and
would fail easily after impact, thus causing attach-
ment. The selectivity of the separation can be en-
hanced by the adsorption of a surface-active agent,
a conditioner. These have a polar end which attaches
to the solid surface, and a non-polar end which sticks
into the water making the surface hydrophobic. In
a bubble}particle interception, the particle has a time
of contact, the sliding time after the initial intercep-
tion during which the particle and the bubble will be
separated by a thin water Rlm. The sliding time will
depend on the initial displacement of the particle
from the line of centres of the rising bubble and the
settling particle. The rate of thinning will depend on
the kinetic energy dissipation after impact and
London}van der Waals dispersive forces, electrostatic
interactions and capillary forces following distortion
of the bubble surface during impact. If the Rlm thins
4612 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
to a critical thickness, it will fail, which results in
a successful attachment. Both the thinning process
and the critical thickness depend on the interaction
energy at the particle}water interface.
The requirement of hydrophobicity as a basis for
attachment is justiRed by the following theoretical
treatment. The dispersion energy (hx)"!(1/
12)A/h2x. The attraction stress between the two sur-
faces is then ((hx)/hx)"A/(6h3x) dynes cm\2. A is
the Hamaker constant which is of the order of 10\12
ergs. For a condensed system to describe the interac-
tion stress, A is replaced by A132 which is a linear
combination of two surface interactions only.
A12 represent the interfacial energy between a particle
1 and a bubble 2 separated by a vacuum, A13 and
A23 are the Hamaker constants representing the par-
ticle}water and bubble}water interaction energies re-
spectively. A33 is the interaction energy between
water molecules. The linear combination
A132"A12!(A13#A23!A33). If A13 and A23 repre-
senting the particle}water and the bubble}water in-
teractions are low, then A132 will be greater than A12.
The enhanced attraction between particle and air
bubble in the presence of water represents hydropho-
bic bonding.
Experimental characterization of hydrophobicity
can be done using the contact angle. If a liquid droplet
is placed on a solid surface at the three-phase point of
contact, it will form a deRnite angle  between the
liquid and solid surfaces, which is called the contact
angle. If G0 is the change in free energy of the
three-phase contact, following a small change AS in
the area of contact, then G0"AS(SL!0SV)#
LV cos(!) at equilibrium; this leads to the
Young equation, SL!SV#LV cos "0. Since
0SV is the energy of contact of a solid with a saturated
vapour of partial pressure p0 there is also present on
the surface a Rlm of condensed vapour with its own
surface energy 0. Then the total surface energy of the
solid surface is S"0SV#0. Substituting in the
Young equation for SL gives LV cos "S!SL!0
in the Dupre equation, the work of adhesion
wSLV"LV(1#cos )#0. This relates an increase in
 to a reduction in adhesion energy or alternatively an
increase in the surface hydrophobicity.
Electrical Effects
The charge on a solid surface can vary from mineral
to mineral and forms a basis for selective separation.
The charge is strongly bound close to the surface in
the Stern layer while further from the surface the
layer is diffuse. If the particle moves in an ap-
plied electrical Reld, a lower potential will be ob-
served which is that at the border of the Stern and the
diffuse layers. This is an electrodynamic effect
called the potential. The potential is a strong
function of pH; the point where the potential is zero
is called the PZC (point of zero charge). For goethite
(FeOOH) the PZC is 6.7 and at acid pH the zeta
potential is positive. Using RSO3 which has a negative
polar group, the recovery of goethite is 100% below
pH 4.5 and 0% above pH 6.7, while if dodecylam-
monium chloride which has a positive polar group is
used, the recovery of goethite in acid solution is 0%
while above pH 9.5 the recovery is 100%.
The attachment of bubbles to charged electrical
surfaces is underresearched. The bubbles are stabil-
ized by surfactants (frothers) whose polar end may be
anionic, cationic or non-ionic. The hydrophobic end
of their molecule will be inside the bubble while the
polar end is in the water. After impact, the previous
treatment suggests that if the electrical forces between
the bubble and the particle surface are attractive then
A132 will increase and conversely, if the two surfaces
repel each other, A132 will decrease. This is mere
speculation as experimental conRrmation is held up
by the extremely small frother concentrations on the
bubble surface, and the difRculty in the deter-
mination of the bubbles charge.
Method 9: Froth Microprocesses
As the particle}bubble aggregates rise to the top of
the pulp they entrain with them in their boundary
layers a small but signiRcant amount of the pulp
suspension which contains both unattached valuable
and waste particles. The bubbles loaded with valu-
able solids pass through the upper level of the pulp
and form a froth. As they pass into the froth, they
entrain with them pulp water. In the froth the void
volume of the pulp suspension is reduced from that in
the pulp, owing to closer packing of the loaded bub-
bles. This forms the interface between the pulp and
the froth which is readily observable (Figure 1). The
sharp reduction of water in the froth at the point is
a preliminary drainage effect. Bubbles will rise
through the froth until they either burst at the top
surface or, as unbroken bubbles loaded with valuable
particles, overSow the concentrate weir. This is the
Rnal product of a single cell; the unSoated waste
product will settle to the bottom of the quiescent
region and leave the cell in the tailings stream.
The performance of the froth depends on the stabil-
ity of the bubble passing through it. This in turn
depends on the amount of frother added. At high
levels of frother, the individual bubbles will remain as
small spheres and will leave over the concentrate weir
essentially unchanged from their condition at the
bottom of the froth. Their water content will there-
fore be unchanged and there will be no upgrading of
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION
Figure 1 Pulp}froth interface.
the solid product. They will not burst at the top
surface of the froth and will Sow easily over the weir.
With a lower level of frother addition, the spherical
bubbles will deform to non-coalescing dodecahedra.
The Sat faces of the bubbles will be separated by
lamellae with a size in microns (Figure 2). Water
squeezed down the lamellae, Sows into plateau bor-
ders which are of millimetre size; the three plateau
borders at their ends join to form nodes down a net-
work of which the entrained froth water drains back
to the pulp. It has been reported that for two-phase
foams, that the volumetric water content of the foam
will fall from 0.26, at 5 mm from the pulp}foam
4613
interface to 0.016, at 12 cm from the bottom of the
foam. After that the value of 0.016 will remain con-
stant through the foam. For coalescing foams the
bubbles moving towards the weir will continue to
grow, giving a lower volumetric water contents of
(0.01. The effect of coalescence in a three-
phase froth, will therefore be to increase the Grade,
G, but with a decrease in fractional recovery, R,
because of bubble breakage at the top surface of the
froth.
Finally it may be observed that cells may poten-
tially be controlled automatically with online image
processing cameras.
4614 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN FLOTATION

Figure 2 Control of flotation cells by image processing the upper surface of their froths. (A) The bubble structure in inverted light and
the same structure with bubble boundaries sharply demarcated. (B) and (C) A comparison between the image-processed bubble
structures, in inverted and reflected light, respectively. (D) The bubble structure in reflected light with image-processed boundaries
superimposed.

See also: I/Flotation. II/Flotation: Bubble-Particle Ad-
herence: Synergistic effect of Reagents; Bubble-Particle
Capture; Column Cells; Cyclones for Oil/Water Separ-
ations; Dissolved Air; Foam Fractionation; Froth Pro-
cesses and the Design of Column Flotation Cells;
Historical Development; Hydrophobic Surface State Flota-
tion; Intensive Cells: Design; Oil and Water Separ-
ation; Reagent Adsorption on Phosphates. III/De-inking
of Waste Paper: Flotation.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY
ESSENTIAL GUIDES TO METHOD DEVELOPMENT
IN GAS CHROMATOGRAPHY
C. F. Poole, Wayne State University,
Detroit, MI, USA
Copyright ^ 2000 Academic Press
Introduction
Separations are possible in gas chromatography if
the solutes differ in their vapour pressure and/or
intensity of solute}stationary}phase interactions. As
a minimum requirement the sample, or some conve-
nient derivative of it, must be thermally stable at the
temperature required for vaporization. The funda-
mental limit for sample suitability is established by
the thermal stability of the sample and system suit-
ability by the thermal stability of column materials. In
contemporary practice an upper temperature limit of
about 4253C and a sample molecular weight less than
1250 is indicated with only minor exceptions. A large
number of general and selective derivatizing reagents
are available for sample modiRcation to enhance
compound thermal stability, improve sample separ-
ation properties, and to provide compound-selective
detection.
Column Types
Wall-coated open tubular columns (WCOT col-
umns), or simply capillary columns, and classical
packed columns dominate the practice of gas}liquid
chromatography. Porous layer open tubular columns
(PLOT columns) and classical packed columns dom-
inate the practice of gas}solid chromatography. Clas-
sical packed columns are usually 0.5}3 m long with
an internal diameter greater than 2 mm packed with
adsorbent or liquid-coated support particles of
100}250
m diameter. WCOT columns are typically
up to 100 m long with internal diameters of capillary
dimensions coated with a thin, and usually immobi-
lized, Rlm of stationary phase leaving an open interior
passageway down the centre of the column. PLOT
columns are identical to WCOT columns with the
liquid phase replaced by a layer of Rne adsorbent
particles. WCOT and PLOT columns are the Rrst
choice for analytical separations because of their su-
perior peak capacity and greater chemical inertness.
Packed columns offer a lower cost choice for
some applications, are easier to use, are relatively
tolerant of thermally unstable and involatile sample
components and are better suited to isolating prep-
4615
arative-scale quantities of materials. Only a limited
number of poly(siloxane) and poly(ethylene glycol)
stationary phases have been successfully immobilized
in WCOT columns compared to the larger number
and variety of stationary phases available for use in
packed columns.
Column Properties
The high permeability of WCOT columns allows long
columns to be used to provide very high total plate
numbers, as indicated in Table 1. Narrow-bore and
thin-Rlm columns are intrinsically the most efR-
cient and are selected for fast chromatography. Since
resolution increases only as the square root of the
plate number, and also the column length, large
values for the plate number are required for difR-
cult separations. Such large numbers are avail-
able in gas chromatography, albeit at the expense
of separation time, allowing separations to be
achieved with only minimal differences in selec-
tivity. In contrast to other chromatographic meth-
ods, separations in gas chromatography are often
achieved through kinetic optimization, allowing
many separations to be obtained on a limited number
of stationary phases. A favourable feature of kinetic
optimization is that the outcome is readily predictable
from simple arithmetic calculations once some in-
formation of peak order has been established in a trial
separation.
At a given temperature the partition coefRcient
is constant and the observed retention factor will
depend on the phase ratio. The phase ratio is given by
the column volume accessible to the mobile phase
divided by the volume of stationary phase. Columns
with a large phase ratio provide small retention fac-
tors for volatile compounds and require inconven-
iently large plate numbers to provide adequate resolu-
tion. Columns with a low phase ratio, that is, thick
Rlm columns, have a lower intrinsic efRciency
than thin Rlm columns, but provide better resolution
of volatile compounds, because they provide more
favourable retention factors. They also allow separ-
ations of volatile compounds at a higher and more
convenient temperature range than is possible with
thin Rlm columns. For volatile compounds this often
means at temperatures above room temperature as
opposed to cryogenic temperatures. For high boiling
compounds, columns with a low phase ratio are not
4616 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY

Table 1 Characteristic properties of some representative columns

Column type Length (m) Internal diameter Film thick- Phase ratio Column plate Plates per metre
(mm) ness (
m) number

Classical packed 2 2.16 10%(w/w) 12 3 640 1 820

2 2.16 5%(w/w) 26 4 000 2 000
WCOT 30 0.10 0.10 249 480 000 16 000
30 0.10 0.25 99 368 550 12 285
25 0.25 0.25 249 160 000 6 400
50 0.25 0.25 249 320 000 6 400
100 0.25 0.25 249 640 000 6 400
30 0.32 0.32 249 150 000 5 000
30 0.32 0.50 159 131 330 4 380
30 0.32 1.00 79 102 080 3 400
100 0.32 1.00 79 304 200 3 400
30 0.32 5.00 15 68 970 2 300
10 0.53 1.00 132 23 500 2 340
30 0.53 1.00 132 70 420 2 340
10 0.53 5.00 26 14 700 1 470
30 0.53 5.00 26 43 940 1 470
50 0.53 5.00 26 73 200 1 470

useful because they lead to long separation times.
Increasing the phase ratio by reducing the Rlm thick-
ness lowers the retention factors to a value within the
optimum range so that there is little deterioration in
resolution and faster separations are obtained.
Packed columns have low phase ratios compared to
most WCOT columns. For compounds of moderate
and low volatility, separation times on packed col-
umns are generally longer. Since several combina-
tions of Rlm thickness and column radius can be used
to generate the same phase ratio, there are other
factors that need to be considered for selecting these
variables for a particular separation.
Mobile-Phase Selection
Nearly all separations are achieved with hydrogen,
helium or nitrogen as the carrier gas. At temperatures
and pressures typical of normal operation in gas
chromatography these gases behave almost ideally,
providing a transport mechanism for the sample
without inSuencing selectivity. The exception is
gas}solid chromatography where the carrier gas par-
ticipates in the retention process through competition
with the sample for stationary-phase adsorption sites.
Differences between hydrogen, helium and nitrogen
are not usually large but absolute retention and reten-
tion order can change as a function of the carrier gas
type and average carrier gas pressure. Heavier carrier
gases, such as carbon dioxide, are more effective at
inSuencing retention in gas}solid chromatography
than the common carrier gases.
Although the choice of carrier gas does not signiR-
cantly inSuence selectivity in gas}liquid chromato-
graphy, it can still inSuence resolution through its
effect on efRciency. This results from differences
in gas diffusivity. The separation time is also
affected because the optimum carrier gas velo-
city decreases with solute diffusion rates. In pres-
sure-limiting conditions, gas viscosity differences
are important as well. Nitrogen provides lower plate
heights but at a lower optimum velocity
(Figure 1), leading to long separation times. Close to
the optimum plate height region, the ascending por-
tions of the curves are shallower for hydrogen and
helium. Thus, for separations at mobile-phase vel-
ocities higher than the optimum velocity, hydrogen
and, to a lesser extent, helium provide faster separ-
ations than nitrogen with little loss in efRciency.
For thick-Rlm columns ('0.5
m), diffusion in
the stationary phase is a signiRcant factor in zone
broadening and the relative contribution of the car-
rier gas to separation performance and time are not as
great. Thick-Rlm columns should be operated close to
the optimum velocity, with the choice of carrier gas
being less signiRcant. Nitrogen is often the preferred
carrier gas for these columns. For packed columns
nitrogen provides (slightly) higher efRciency at
low temperatures and Sow rates, while hydrogen is
superior at higher temperatures and at above opti-
mum velocities. Hydrogen is preferred in pressure-
limited conditions because of its lower viscosity.
A considerable difference in the relative cost of
helium in the USA and Europe has resulted in dif-
ferent preferences on the two continents. For WCOT
columns, helium is widely used in the USA for safety
rather than theoretical considerations, while hydro-
gen is commonly used in Europe.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY 4617

Figure 1 Influence of the choice of carrier gas on the efficiency of (A) a thin-film and (B) a thick-film WCOT column (k is the retention
factor).

Stationary-Phase Selection 3. liquid organic salts

4. poly(siloxane) liquid phases
Given the nonsolvating properties of the mobile
phase in gas chromatography, selectivity optimiza-
Representative examples and their useful temperature
tion is a matter of stationary-phase selection. Over
operating range are summarized in Tables 2 and 3.
the years thousands of substances have been evalu-
Most can be considered useful for general applica-
ated as stationary phases and most abandoned in
tions, except for the perSuorocarbon liquid phases
favour of a smaller number of liquids and adsorbents
that are used for the speciation of perSuorocarbon
with favourable temperature operating ranges, kin-
compounds or the separation of reactive compounds
etic properties and possibility of immobilization if
(metal Suorides, interhalogen compounds and non-
used in WCOT columns.
metal halides) that tend to destroy conventional
Packed-column liquid phases can be roughly cat-
phases. The family of poly(siloxanes) provides the
egorized into four groups:
widest range of favourable stationary-phase proper-
1. hydrocarbon and perSuorocarbon liquid phases ties and variation in selectivity. They are the most
2. ether and ester liquid phases widely used stationary phases for packed-column

Table 2 Characteristic properties of some liquid phases used in packed-column gas chromatography

Name Structure Temperature range (3C)

Miminum Maximum

Hexadecane C16H34 (20 50

Squalane 2,6,10,15,19,23-hexamethyltetracosane (20 120
Apolane-87 (C18H37)2CH(CH2)4C(C2H5)2(CH2)4CH(C18H37)2 30 280
Fomblin YR }[OCF(CF3)CF2)n(OCF2)m]} 30 (255
PPE-5 C6H5O(C6H5O)3C6H5 20 200
Dioctyl phthalate C6H4(COOC8H17)2 (20 160
EGS HO(CH2)2[OOCCH2CH2COO(CH2)2]nOH 100 210
DEGS HO(CH2)2O(CH2)2[OOCCH2CH2COO(CH2)2O(CH2)2]nOH 20 200
Carbowax 20M HO(CH2CH2O)nCH2CH2OH 60 225
FFAP 50 250
1,2,3-Tris(2-cyanoethoxy)propane (CH2OCH2CH2CN)3 20 170
Tetrabutylammonium perfluorooctanesulfonate (20 220
Tetrabutylammonium 4-toluenesulfonate 55 200
Tetrabutylammonium tetrafluoroborate 162 290
Ethylammonium 4-toluenesulfonate 121 220
Tetrabutylphosphonium chloride 83 230

PPE-5, Poly(phenyl ether); EGS, poly(ethylene glycol succinate); DEGS, poly(diethylene glycol succinate); Carbowax 20M,
poly(ethylene glycol); FFAP, Carbowax 20M treated with 2-nitroterephthalic acid.
4618 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY

Table 3 Characteristic properties of some poly(siloxane) liquid phases used for packed-column gas chromatography

Name Structure Temperature operating range (3C)

Minimum Maximum

OV-1 Dimethylsiloxane (gum, molecular weight'106) 100 350

OV-101 Dimethylsiloxane (oil, molecular weight 3;104) (20 350
OV-7 Phenylmethyldimethylsiloxane 80% methyl and 20% phenyl (20 350
OV-17 Phenylmethylsiloxane 50% methyl and 50% phenyl (20 350
OV-25 Phenylmethyldiphenylsiloxane 25% methyl and 75% phenyl (20 300
OV-210 Trifluoropropylmethylsiloxane 50% methyl and 50% 3,3,3-trifluoropropyl (20 275
OV-225 Cyanopropylmethylphenylmethylsiloxane 50% methyl, 25% (20 250
phenyl and 25% 3-cyanopropyl
Silar 7CP Cyanopropylphenylsiloxane 75% 3-cyanopropyl and 25% phenyl 50 250
OV-275 Di(cyanoalkyl)siloxane 70% 3 cyanopropyl and 30% 2-cyanoethyl 250
Silar 10CP Di(3-Cyanopropyl)siloxane 50 250

gas}liquid chromatography and, because of their ease
of immobilization, are the dominant stationary
phases used to prepare WCOT columns (Table 4).
With todays technology the only other family of
stationary phases that can be immobilized for WCOT
columns are the poly(ethylene glycols).
The selectivity of the stationary phases is of more
interest than their chemical structure for method de-
velopment. Liquid stationary phases have been classi-
Red based on their solvent strength (polarity) and
selectivity. ClassiRcation based on the idea of polarity
has had to be abandoned because of the lack of
a working deRnition. Selectivity is deRned as the rela-
tive capacity of a stationary phase for speciRc inter-
molecular interactions, such as dispersion, induction,
orientation and complexation (including hydrogen
bond formation). Unlike solvent strength (polarity) it
should be feasible to devise experimental scales of
stationary-phase selectivity. Modern approaches to
stationary-phase classiRcation by selectivity are based
on the cavity model of solvation. This model assumes
that the transfer of a solute from the gas phase to
solution in the stationary phase involves three steps.
Initially a cavity is formed in the stationary phase of
the same size as the solute. The solute is then transfer-
red to the cavity with reorganization of solvent
molecules around the cavity and the set-up of
solute}solvent interactions. Retention in gas}liquid
chromatography, therefore, will depend on the cohe-
sive energy of the stationary phase, represented by the

Table 4 Rough guide to the temperature operating range for bonded poly(siloxane) liquid phases in open tubular columns

Type Temperature range (3C) High temperature

version
Minimum Maximum

Dimethylsiloxane !60 325 420

Dimethyldiphenylsiloxane (5% diphenyl) !60 325 420
Dimethyldiphenylsiloxane (35% diphenyl) 40 300 340
Dimethyldiphenylsiloxane (50% diphenyl) 40 325 390
Methylphenylsiloxane 0 280
Dimethyldiphenylsiloxane (65% diphenyl) 50 260 370
3,3,3-Trifluoropropylmethylsiloxane (50% trifluoropropyl) 45 240 300
3-Cyanopropylphenyldimethylsiloxane (6% cyanopropylphenyl and 20 280
84% dimethyl)
3-Cyanopropylphenyldimethylsiloxane (25% cyanopropylphenyl and 40 240
75% dimethyl)
3-Cyanopropylphenyldimethylsiloxane (50% cyanopropylphenyl and 40 230
50% dimethyl)
3-Cyanopropyl-silphenylene co-polymer (equivalent to 70% dicyanopropyl) 290
Poly(ethylene glycol) 20 250 280
FFAP 40 250

FFAP, Poly(ethylene glycol) treated with 2-nitroterephthalic acid.

 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY
free energy required for cavity formation, the forma-
tion of additional dispersion interactions of a
solute}solvent type, and on selective solute}solvent
polar interactions dependent on the complementary
character of the polar properties of the solute and
stationary phase. Quantitatively, these interactions
are described by the solvation parameter model set
out below in the form suitable for stationary-phase
classiRcation:
log k"c#rR2#sH
2 #a
2 #b
2 #l log L
H H 16
[1]
where k is the retention factor. The remainder of the
equations is made up of product terms called system
constants (r, s, a, b, l) and solute descriptors (R2, H,


H2 ,
2 , log L ). Each product term represents
H 16
a contribution from a deRned intermolecular interac-
tion to the solute property. The solute descriptors are
free energy-related solute properties known for about
4000 compounds with others available by estimation
or from experiment. They are not of immediate inter-
est to us here except to note that once the system
constants are established, the retention property of
any solute with known or easily estimated solute
descriptors can be estimated for that system by simple
arithmetic calculation using the model described. The
system constants (also called phase constants in gas
chromatography) contain the information of the sta-
tionary-phase properties and provide an unambigu-
ous means of classiRcation. The r phase constant
refers to the ability of the stationary phase to interact
with solute n} or }electron pairs. The s phase con-
stant to the ability of the stationary phase to take part
in dipole-type interactions. The a phase constant is
a measure of stationary-phase hydrogen bond basic-
ity and the b phase constant a measure of stationary-
phase hydrogen bond acidity. The l phase constant
describes (in part) the contribution of cavity forma-
tion and dispersion interactions to retention and,
more speciRcally, indicates the ability of the station-
ary phase to separate members of a homologous
series. The phase constants for any stationary phase
can be determined through the method of multiple
linear regression analysis by measurement of a reten-
tion property for a series of varied solutes with
known solute descriptors.
The stationary phase constants for a number of
common liquid phases at a reference temperature of
1213C are summarized in Table 5. The system con-
stants in Table 5 are only loosely scaled to each other
so that changes in magnitude in any column can be
read directly, but changes in magnitude along rows
must be interpreted more cautiously. Most stationary
4619
phases possess some capacity for lone-pair electron
interactions (r constant), but selectivity for this inter-
action is all but nonexistent among common station-
ary phases. Fluorine-containing stationary phases
have negative values of the r constant representing the
tighter binding of electron pairs in Suorocarbon com-
pared to hydrocarbon groups. Lone pair electron in-
teractions do not usually make a signiRcant contribu-
tion to retention in gas}liquid chromatography and
are not considered a primary means of selectivity
optimization. The most striking feature of Table 5 is
the paucity of stationary phases with signiRcant hy-
drogen bond acidity (b constant). In the case of
EGAD, DEGS and TCEP, the small b phase constants
indicated in Table 5 are probably a product of impu-
rities and thermal modiRcation of the stationary
phases during use rather than a fundamental property
of the stationary phases themselves.
A few novel stationary phases with strong hydro-
gen bond acid properties have been synthesized re-
cently, but none of these are commercially available.
Stationary-phase hydrogen bond acid interactions,
therefore, do not contribute signiRcantly to method
development strategies for the commonly used sta-
tionary phases. The only practical exception seems
to be the poly(triSuoropropylmethylsiloxane)
WCOT column stationary-phase DB-210, which
exhibits some weak hydrogen bond acidity, presum-
ably acquired through the immobilization process
that is absent from the structurally similar packed-
column stationary phase QF-1. That leaves the most
important stationary-phase properties for selectivity
optimization as their cohesive energy and capacity for
dipole-type and hydrogen bond base interactions.
Cluster analysis provides a visual picture of the
difference in selectivity for different station-
ary phases and a classiRcation of their properties into
groups of similar selectivity (Figure 2). The stationary
phases most similar to each other are next to each
other and are connected. Connections at the extreme
left-hand side of the dendrogram occur for phases
with similar properties and those towards the right-
hand side with greater degrees of difference. Sta-
tionary phases with no paired descendents are singu-
lar phases with properties that cannot be duplicated
by the other phases. The stationary phases are classi-
Red into Rve groups with three phases behaving inde-
pendently. Group 1 contains squalane, Apolane-87,
OV-3. OV-7, SE-30 and OV-105. These are phases of
low cohesive energy with minimal capacity for polar
interactions. The second group of stationary phases
contains OV-22, OV-25, OV-11, OV-17, PPE-5 and
DDP. These phases have low cohesive energy and are
weakly dipolar and hydrogen bond basic. QF-1 is
loosely connected to this group but is signiRcantly
4620 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY

Table 5 System constants derived from the solvation parameter model for common stationary phases at 1213C

Stationary phase System constant

r s a b l

Hydrocarbon phases
Squalane 0.13 0.01 0 0 0.58
Apolane-87 0.17 0 0 0 0.56

Ether and ester phases

Poly(phenyl ether) 5 rings (PPE-5) 0.23 0.83 0.34 0 0.53
Carbowax 20M (CW20M) 0.32 1.26 1.88 0 0.45
Poly(ethylene glycol) Ucon 50 HB 660 0.37 0.63 1.28 0 0.50
Nitroterephthalic acid modified poly(ethylene glycol) (DB-FFAP) 0.21 1.42 2.08 0 0.43
1,2,3-Tris(2-cyanoethoxypropane) (TCEP) 0.12 2.09 2.10 0.26 0.37
Didecylphthalate (DDP) 0 0.75 0.77 0 0.56
Poly(ethylene glycol adipate) (EGAD) 0.13 1.39 1.82 0.21 0.43
Poly(diethylene glycol succinate) (DEGS) 0.23 1.57 2.11 0.17 0.41

Liquid organic salts

Tetrabutylammonium 4-toluenesulfonate (QBApTS) 0.16 1.58 3.30 0 0.46
Tetrabutylammonium tris(hydroxymethyl)methyl-amino-2-hydroxy-1-propanesulfonate 0.27 1.96 3.06 0 0.32
(QBATAPSO)
Tetrabutylammonium 4-morpholinepropanesulfonate (QBAMPS) 0 1.75 3.54 0 0.55
Tetrabutylammonium methanesulfonate (QBAMES) 0.33 1.45 3.76 0 0.44

Poly(siloxane) phases
Poly(dimethylsiloxane) (SE-30) 0.02 0.19 0.13 0 0.50
Poly(dimethyldiphenylsiloxane) (DB-5) (5 mol% diphenylsiloxane) 0 0.28 0.19 0 0.51
Poly(dimethylmethylphenylsiloxane) (OV-3) (10 mol% phenyl) 0.03 0.33 0.15 0 0.50
Poly(dimethylmethylphenylsiloxane) (OV-7) 0.06 0.43 0.17 0 0.51
Poly(dimethylmethylphenylsiloxane) (OV-11) (35 mol% phenyl) 0.10 0.54 0.17 0 0.52
Poly(methylphenylsiloxane) (OV-17) 0.07 0.65 0.26 0 0.52
Poly(dimethyldiphenylsiloxane) (HP-50) (50 mol% diphenylsiloxane) 0.16 0.62 0.28 0 0.47
Poly(methylphenyldiphenylsiloxane) (OV-22) (65 mol% phenyl) 0.20 0.66 0.19 0 0.48
Poly(methylphenyldiphenylsiloxane) (OV-25) 0.28 0.64 0.18 0 0.47
Poly(cyanopropylmethyldimethylsiloxane) (10 mol% cyanopropylmethylsiloxane) 0 0.36 0.41 0 0.50
(OV-105)
Poly(cyanopropylmethylphenylmethylsiloxane) (OV-225) 0 1.23 1.07 0 0.47
Poly(cyanopropylphenyldimethylsiloxane) (50 mol% cyanopropylphenylsiloxane) 0 1.21 1.18 0 0.44
(DB-225)
Poly(dicyanoalkylsiloxane) (OV-275) 0.21 2.08 1.99 0 0.29
Poly(trifluoropropylmethylsiloxane) (QF-1) !0.45 1.16 0.19 0 0.42
Poly(trifluoropropylmethylsiloxane) (DB-210) !0.27 1.15 0 0.19 0.43
Poly(dimethylsiloxane)-poly(ethylene glycol) copolymer (OV-330) 0.10 1.06 1.42 0 0.48

more dipolar and has a more signiRcant and opposite
capacity for lone pair electron interactions. The third
group contains OV-330 and OV-225 with UH50B
loosely connected to this group. Compared to the
second group these stationary phases are more dipo-
lar and hydrogen bond basic and slightly more cohe-
sive. They represent an increase in the intensity of the
same range of interactions as the group 2 stationary
phases. The fourth group contains the liquid organic
salts with QBATAPSO distinguished within this
group by its greater cohesive energy. Phases in this
group are dipolar and the strongest hydrogen bond
bases. The Rfth group of solvents is divided into two
subgroups. TCEP and OV-275 are strongly dipolar,
hydrogen bond basic and have high cohesive energy.
EGAD, CW20M and DEGS have a similar range of
polar interactions, but not quite as intense, and have
a lower cohesive energy. For selectivity optimization
in packed-column gas chromatography, a single
phase is initially selected from each group. Sub-
sequently, for Rne-tuning additional phases are se-
lected from within the group, identiRed as possessing
the desired separation properties.
For historical reasons stationary phases are classi-
Red at a common reference temperature of about
1203C. The capacity of a stationary phase for speciRc
intermolecular interactions determined at one tem-
perature can be misleading for selectivity optimiza-
tion at other distant temperatures. The broad outlines
in Table 5 and Figure 2 remain true but changes in
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY
Figure 2 Nearest neighbour complete link cluster dendrogram for some common stationary phases. The abbreviations for the
stationary phases are identified in Tables 2, 3 and 5.
rank order due to cross-overs occur at different
temperatures. Also, selectivity differences be-
tween individual stationary phases are enhanced at
low temperatures with phases becoming more alike at
higher temperatures. Information on the contribution
of polar interactions to retention at high temperatures
is scarce. These contributions could be small and
stationary-phase selectivity differences rather limited
at high temperatures.
Gas}solid chromatography is used for a narrower
range of separations than gas}liquid chromatogra-
phy. Because of higher retention, typical applications
are the separation of Rxed gases, volatile hydrocar-
bons, halocarbons, organic solvents and sulfur gases.
The presence of immobilized active centres enhances
the separation of isomers and isotopes. These separ-
ations are often difRcult or impossible with liquid
phases. A rough guide to the selection of sorbents for
particular applications is given in Table 6. PLOT
columns provide higher efRciency, faster separ-
ations and faster column regeneration compared to
packed columns. Surface coating with inorganic salts
and small amounts of liquid phase are used to extend
the molecular weight separation range with inorganic
oxide and carbon sorbents and to optimize selectivity.
4621
PLOT columns generally require greater care in their
use than WCOT columns. Other features include
lower efRciency than WCOT columns, limited
sample capacity and high activity.
General Elution Problem
In gas chromatography there is an approximate ex-
ponential relationship between retention time and
solute boiling point at a constant (isothermal) column
temperature. Consequently, it is impossible to estab-
lish a suitable compromise temperature for the separ-
ation of mixtures with a boiling point range exceed-
ing about 1003C. This is generically referred to as the
general elution problem and is characterized by long
separation times, poor separations of early eluting
peaks and poor detectability of late eluting peaks due
to zone broadening. The general solution to this prob-
lem is the use of programmed temperature and Sow
separation modes. Neither constant nor programmed
modes are superior to each other. They are com-
plementary, with the properties of the sample decid-
ing which approach is adopted.
Temperature programming is the most popular pro-
grammed separation mode in gas chromatography.
4622 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY

Table 6 General applications of PLOT columns in gas chormatography

Stationary phase Maximum operating Typical applications

temperature (3C)

Alumina oxide 200 Alkanes, alkenes, alkynes and aromatic hydrocarbons from C1 to C10. C1 and C2
halocarbons
Silica gel 250 Hydrocarbons (C1 to C4), inorganic gases, volatile ethers, esters and ketones
Carbon 350 Inorganic gases and hydrocarbons (C1 to C5)
Carbosieves 150 C1 to C6 compounds
Molecular sieves 350 Hydrogen, oxygen, nitrogen, methane and noble gases
(5A and 13X) Hydrocarbons (C1 to C3) on 5A with higher alkanes on 13X (up to C12) but not isomer
separations
Porous polymers
Q 310 Hydrocarbons (C1 to C14), halocarbons (C1 and C2), volatile oxygenated solvents
S 250 (C1 to C6), thiols, amines, nitro compounds, nitriles, water and inorganic gases
U 190

Q, Poly(divinylbenzene-styrene); S, poly(divinylbenzene-vinylpyridine); U, poly(divinylbenzene-ethylene glycol dimethacrylate).

Stationary phases of high thermal stability allow wide
temperature ranges to be used and temperature is
easily adjusted and controlled. Temperature pro-
gramme techniques are the most useful approach for
scouting the properties of an unknown sample and
are compatible with the large volume injection modes
employing low temperature solute refocusing used in
trace analysis. Flow programming is easily achieved
with instruments Rtted with electronic pressure con-
trol but is limited by the narrow pressure range which
is usually available. It can be used to separate ther-
mally labile compounds at a lower temperature than
required for temperature-programmed separations.
On the other hand, Sow programming results in a loss
of efRciency for late eluting peaks and presents
difRculties in calibrating Sow-sensitive detectors.
A temperature programme consists of a series of
changes in the oven temperature and includes isother-
mal and controlled temperature rise segments. In es-
sence, most programmes are simple, consisting of an
initial isothermal period, a linear temperature rise
segment, Rnal isothermal period at the temperature
reached at the end of the rise segment, and a cool-
down period to return the oven to the starting tem-
perature. The initial and Rnal isothermal periods are
optional, the temperature rise segment can be selected
over a wide range (0.1 to c. 1003C min\1), nonlinear
changes in temperature may be used (extremely rare)
and for complex mixtures, several linear programmes
may be used in sequence to optimize the separation.
The initial oven temperature is selected with due
consideration to the resolution of the earliest eluting
peaks in the chromatogram. If the temperature
chosen is too high, the resolution of the initial peaks
may be inadequate and, if it is too low, resolution
may be acceptable but the separation time will be
extended needlessly. The Rnal temperature should be
selected so that the termination of the temperature
rise segment and elution of the last sample compon-
ent coincide unless the last few eluting peaks are
particularly difRcult to separate and require an
isothermal period. Peaks eluting after completion of
the temperature rise segment will be wider than those
eluted during the programme. The selection of the
heating rate represents a compromise between the
necessity of maintaining a minimum acceptable res-
olution for the sample and the desire to reduce the
separation time. This is governed largely by the com-
plexity of the sample and its boiling point range. For
samples containing components of different po-
larity temperature-induced changes in selectivity
make the prediction of the resolution of closely
spaced peaks a problem. Certain generalities can be
made however. For the most difRcult separations
a slow heating rate will usually provide the optimum
resolution. The separation time of weakly retained
solutes is more readily adjusted by changing the Sow
rate of the carrier gas than the heating rate. For
strongly retained solutes increasing the heating rate
causes a proportional decrease in the separation time
at a constant carrier gas Sow rate. The retention time
of well-retained solutes are less affected by
changing Sow rates in temperature-programmed gas
chromatography.
The lack of an exact mathematical model to de-
scribe temperature-programmed separations makes
computer simulation for their rapid optimization dif-
Rcult. Simplex optimization of experimental variables
and a model based on the linear elution strength
approximation have been used with some success.
The linear elution strength approach has the ad-
vantage that it only requires experimental data
from two temperature-programmed separations of a
sample using different programme rates. A series
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY 4623

of empirical equations are then employed to predict
optimum separation conditions using relative resolu-
tion maps.
Generic Method Development
Method development commences with a deRnition of
the problem and a review of available resources.
Some pointers are given in Table 7. The separation of
enantiomers requires special stationary phases and
some separations of isomers use liquid crystal station-
ary phases that are not common laboratory items.
Fast separations require special equipment and truly
fast separations are only possible for simple mixtures.
The separation of complex mixtures can be speed-
optimized but not necessarily performed quickly on
the same timescale used for simple mixtures. Prepara-
tive separations are usually performed with packed
columns unless only a small amount of material
is to be isolated. Most analytical separations are per-
formed using WCOT and PLOT columns, except for
reasons noted earlier. A general guide to the selection
of column types and dimensions is given in Table 8.
For a sample of unknown composition a generic
method for sample evaluation is provided in Table 9.
These instruction sets will work for most simple mix-
tures and provide a starting point for difRcult
samples.
Injection and Detection
Considerations
The choice of sample inlet depends on the injection
volume, concentration of analytes, thermal stability
of the analytes, concentration of involatile matrix
components, volatility range of the analytes, relative
volatility difference between the analytes and

Table 7 Defining the problem and utilizing available resources to formulate a solution

Problem definition
How many detectable compounds are present in the sample?
This is the only way to know that a separation is complete. It indicates the complexity of the problem, if only because, statistically, as the
number of components requiring separation increases, so does the difficulty of achieving the separation. Fast separations are not easy
for complex mixtures
Are all components equally relevant?
The only separation required is that of the compounds of interest from each other and all other compounds in the sample. The latter
compounds can be considered as matrix components and need not be individually separated. This reduces the difficulty of providing
a separation fit for the defined purpose
Are standards available for the compounds of interest?
This enables peaks to be tracked through initial trial separations and links difficult-to-separate compound pairs to their structure so that
informed changes to the separation system can be made. It is required for calibration if quantification is needed
What is the concentration range of relevant compounds?
Trace components may initially be missed because of inadequate dynamic range if only the major components are considered.
Particular injection techniques and selective detectors may be required to detect some compounds at anticipated concentrations
Is identi\cation of unknowns required?
If unknown compounds are to be identified, retention information alone will be inadequate in most cases. Coupling to mass or infrared
spectroscopic detectors is usually required to achieve the desired level of confidence in the result

Resources
Literature describing similar separations
Substantial databases of the chromatographic literature in an electronic searchable form are available. Column manufacturers
catalogues contain information on the separation of common sample types and certified columns may be available for mixtures subject
to routine analysis. Official methods controlled by regulatory agencies usually specify appropriate columns for the separation
Past experience with similar samples
Life is a learning experience and there is no substitute for a good memory. What worked in a previous case for a similar sample is
probably a good starting point for the new sample. Colleagues may have an informed opinion based on a different lifetime experience
Availability of certain columns and equipment
Resources are restricted to available columns and equipment and an evaluation of whether they will provide the information required is
performed. Additional resources may have to be purchased or informed decisions made of the suitability of substituting available for
desired resources
Compound information from handbooks
Structures, molecular weight, boiling point (or vapour pressure), solubility in common solvents is useful information that can be found in
handbooks for many compounds which are identical or similar to the compounds of interest. This is useful for stationary-phase
selection, derivatization strategies and detector selection
4624 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY

Table 8 Guide to the selection of WCOT columns

Column internal diameter

E Use 0.25 mm i.d. columns for normal split and splitless injection unless sample overloading is a problem
E Use 0.32 mm i.d. columns for splitless and on-column injection, especially when injecting large sample amounts
E Use 0.53 mm i.d. columns as a replacement for packed columns, for the separation of samples containing (30 components, or
samples with components spanning a wide concentration range
E Use 0.18 mm i.d. (or less) columns when the maximum efficiency is required and for high speed separations (modifications to
standard instruments may be needed)

Film thickness
E Standard film thicknesses are used for most applications (0.25
m for 0.25 and 0.32 i.d. columns)
E Use thin film columns (0.1}0.25
m) for solutes of low volatility (e.g. waxes, triglycerides, steroids, etc.)
E Use thick film columns (1}5
m) for volatile solutes (e.g. solvents, gas-purgeable compounds)
E Choose columns with a similar phase ratio to obtain similar retention (larger phase ratios reduce retention)

Stationary phase
E If the sample composition is unknown, begin with a nonpolar stationary phase such as a poly(dimethylsiloxane) or poly(dimethyl-
diphenylsiloxane) with a low mol fraction of diphenylsiloxane groups that separate mainly by differences in volatility
E To improve selectivity, choose a stationary phase whose polarity best matches that of the solutes (similar dipolarity or complement-
ary hydrogen bond interactions). See Figure 2 for the systematic identification of suitable stationary phases
E Consider using a PLOT column for the separation of light hydrocarbons and gases (other applications are indicated in Table 6)

Set-up conditions
Internal diameter (mm) 0.18 0.25 0.32 0.53
Flow rate (mL min\1)
Hydrogen, u"40 cm s\1 0.6 1.4 2.4 5.2
Helium, u"20 cm s\1 0.3 0.7 1.2 2.6
Sample capacity (
g) (0.05 0.05}0.1 0.4}0.5 1.2
Separation number 40 30 25 15
Separation efficiency (n m\1) 5300 3300 2700 1600

solvent, and the required accuracy and precision.
Split injection is commonly used for evaluating separ-
ation conditions, even if a different injection tech-
nique is used for routine applications. Split injection
involves ofSine vaporization and mixing of the
sample vapours with the gas phase, a portion of
which is the carrier gas Sow for the column and is
responsible for transporting a fraction of the sample
into the column. Sample bands are narrow, preserv-
ing the resolving power of the column. Split injection
can handle samples containing involatile matrix com-
ponents and is the preferred method for injecting
gases and volatile samples such as solvents. Ac-
curacy and precision are often poor compared to

Table 9 Generic exploratory conditions for the separation of a sample of unknown composition

Stationary phase Nonpolar poly(dimethylsiloxane) of poly(dimethyldiphenylsiloxane) wiht 5 mol% diphenylsiloxane groups

Column Length 10}30 m
Internal diameter 0.25 or 0.32 mm
Film thickness 0.25
m (1.0
m for volatile compounds)
Flow rate uopt
Temperature Programme from 50 to 3003C at 203C min\1 (or to temperature limit for phase)
Note the elution temperature (TE) and range of TE values
TE range (253C isothermal analysis
TE range '253C programmed analysis
Isothermal Optimize range of retention factors (k). Topt found from plot of log k versus 1/T
Programmed From original programme:
1 Select T initial (TE!20 for first component)
2 Select T final (TE#20 for last component)
3 Programme rate selected based on complexity. Simple mixture 103C min\1 or higher and complex mixture
1}23C min\1
Injector/detector Initially high (c. 3503C). Reset, based on findings in trial chromatograms
Temperature c. 253C higher than the final column temperature
Injector Split with a split ratio 1 : 50 to 1 : 100
Detector Universal (flame ionization detector)
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN GAS CHROMATOGRAPHY 4625

Table 10 Characteristic properties of common detectors

Detector Minimum detectable amount Linear response range Selectivity

Thermal conductivity 3;10\9 g mL\1 104

Flame ionization 10\12 g s\1 106
Thermionic ionization 10\13 g s\1(N) 104 4;104 gC/gN
10\14 g s\1(P) 7;104 gC/gP
0.5 gN/gP
Photoionization 10\12 g mL\1 107
Helium ionization 4;10\14 g s\1 104
Electron capture 10\13 g mL\1 104
Flame photometric 10\11 g s\1 (S) Nonlinear 103}106 gC/gS
10\12 g s\1 (P) 105 5;105 gC/gP
Sulfur chemiluminescence 4;10\13 g s\1 (S) 103}104 106}107 gC/gS
Microwave plasma 1}75;10\13 g s\1 104 Large
Electrolytic conductivity 10\12 g s\1 (N) 103}105 104}109 gC/g(N, Cl or S)
10\13 g s\1 (Cl)
5;10\13 g s\1 (S)

other injection techniques and sample information is
not preserved for mixtures of a wide volatility range.
Splitless injection allows larger sample volumes to be
injected for trace analysis but requires an effec-
tive refocusing mechanism using cold trapping or
solvent effects. Optimization of injection condi-
tions is relatively complicated and time-consuming
but accuracy and precision are good for favourable
cases. On-column injection is the most accurate and
precise injection technique but is limited to small
sample volumes and requires relatively clean extracts.
Programmed temperature vaporization injection is
able to emulate all of the above injection methods as
well as allowing large volume injections in the sol-
vent-venting mode. Injection techniques using Sash
vaporization place the greatest thermal stress on the
sample and are unsuitable for labile compounds. All
methods can be automated, resulting in improved
accuracy and precision compared to manual injection
techniques.
Gas chromatography is blessed by a number of
reliable and near universal and selective detectors
(Table 10). Interfacing of gas chromatography to
spectroscopic detectors for structural elucidation as
well as quantiRcation is straightforward and reduced
to routine practice. For general applications the Same
ionization detector is difRcult to eclipse. It is
sensitive, rugged, has a wide linear range, and it has
a near universal response to carbon-containing com-
pounds. It has a poor response to the noble gases and
certain simple organic compounds containing a single
carbon atom bonded to nitrogen, oxygen or sulfur.
Thermal conductivity or helium ionization detection
can be used for these compounds. A wide range of
element-selective detectors and structure-selective de-
tectors has been developed for particular applications
demanding matrix discrimination, low sample detec-
tability, or for portable instruments. There are few
situations encountered in gas chromatography where
the identiRcation of a suitable detector is the limit to
progress.
Further Reading
Abraham MH, Poole CF and Poole SK (1999) ClassiRcation
of stationary phases and other materials by gas
chromatography. Journal of Chromatography A 842:
79}114.
Bautz DE, Dolan JW and Snyder LR (1991) Computer
simulation as an aid in method development for gas
chromatography. I. The accurate prediction of separ-
ation as a function of experimental conditions. Journal
of Chromatography 541: 1}20.
Jayatilaka A and Poole CF (1993) Computer-assisted op-
timization of the gas chromatographic separation of
equine estrogens. Journal of Chromatography 617:
19}27.
Jennings W, Mittlefehldt E and Stremple P (1997) Analyti-
cal Gas Chromatography. San Diego: Academic Press.
Ji Z, Majors RE and Guthrie EJ (1999) Porous layer open-
tubular capillary columns: preparations, applications
and future directions. Journal of Chromatography
A 842: 115}142.
Poole CF and Poole SK (1991) Chromatography Today.
Amsterdam: Elsevier.
Rotzsche H (1991) Stationary Phases in Gas Chromatogra-
phy. Amsterdam: Elsevier.
Villalobos R and Annino R (1991) Computer-aided design
and optimization of an on-line gas chromatographic
procedure for the analysis of purgeable compounds in
waste water. Journal of High Resolution Chromatogra-
phy 14: 681}685.
4626 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY
ESSENTIAL GUIDES TO METHOD DEVELOPMENT
IN LIQUID CHROMATOGRAPHY
J. W. Dolan and L. R. Snyder, LC Resources Inc.,
CA, USA
Copyright ^ 2000 Academic Press
Introduction: Steps in Method
Development
Development of a method for a high performance
liquid chromatography (HPLC) separation can be
a major undertaking. Before the separation can be
made, the sample must be in a suitable form to inject,
and pretreatment steps are often required to remove
major interferences or materials that might shorten
the column life. After conditions for adequate separ-
ation are determined, some level of method validation
is usually performed. Sample pretreatment and
method validation are beyond the scope of the pres-
ent discussion, which concentrates on achieving sep-
aration. This article describes only the major steps
that are required for most samples. For additional
information, the reader is urged to consult the refer-
ence by Snyder et al. (see Further Reading) which
covers HPLC method development in detail. Addi-
tional method development information can be found
in the other monographs listed.
General Approach
There are different approaches to HPLC method
development, but we will follow the steps outlined
in Table 1 and discussed below. For most samples,
this approach provides the highest probability of
success with the minimum investment in time and
effort.
The Rrst step in HPLC method development is to
choose a chromatographic mode or method type. The
most common modes are reversed-phase, normal-
phase, ion exchange and size exclusion. User surveys
over the last 10 years consistently show that most
separations are performed using reversed-phase col-
Table 1 General approach to HPLC method development
Select HPLC method
Obtain minimal separation
Check for and correct peak shape and width problems
Fine-tune primary variable
Change additional variables
Adjust column conditions
umns. For the present discussion, reversed-phase sep-
aration is assumed. The following section gives a brief
description of the use of alternative HPLC modes for
special samples.
Once a mode is selected, the next step is to Rnd
conditions that will provide a separation of most
of the sample components. When this has been
achieved, it is then possible to estimate the effort
that will be required to obtain an adequate separation
of all components. This Rrst step can be accomplished
using either gradient or isocratic elution. We favour
an initial gradient run, because all peaks are likely to
elute in a deRned time with reasonable separation of
both early and late peaks. Usually several isocratic
runs are required to achieve a similar result, and often
no isocratic conditions will provide an acceptable
separation. From the initial gradient run it is possible
to estimate whether isocratic elution is possible. If
this is the case, it is also possible to estimate condi-
tions that give reasonable separation of most sample
components.
As soon as this minimal separation is obtained, the
chromatogram should be examined for problems re-
lated to peak shape. Most obvious are peak tailing
problems. Although perfectly symmetrical peaks are
preferred, many separations (usually for samples that
contain basic compounds) will have one or more
peaks that exhibit tailing. Most workers will accept
peaks with asymmetry factors, As42.0 (United
States Pharmacopeia (USP) tailing factor, Tf41.7).
More severe tailing suggests the presence of un-
wanted sample interactions with the stationary phase.
The most common Rxes for tailing bands, in order of
decreasing usefulness, are:
1. the use of columns designed for the separation of
basic samples (based on very pure, type B silica);
2. adjustment of pH;
3. addition of triethylamine as a tailing suppressor;
4. use of ion pairing;
5. switching to a nonsilica (e.g. polymeric) column.
Symmetrical peaks that are too broad can also
signal poor chromatographic behaviour; e.g. when
column plate numbers, N, for the sample are (60%
of the column manufacturers test report. Broad
peaks can result from the use of too strong a sample
solvent, injection volumes that are too large, column
overload or column problems. Usually it is advisable
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY
Rrst to repeat the separation on a new column, to be
sure that the problem is caused by a bad column.
Reducing the injection volume to (25
L, keeping
the injected mass (10
g, matching the injection
solvent with the mobile phase and increasing the
column temperature are some possible approaches to
sharpening broad peaks.
Once acceptable peak shape is obtained, the next
step is to Rne-tune the primary variable: the percent-
age of organic solvent in the mobile phase, %B, for
isocratic separations, or gradient time, tG, for gradi-
ent elution. In general, weaker (lower %B) isocratic
mobile phases of shallower (larger tG) gradients will
increase resolution at the expense of longer run times
and broader peaks (with lower detection sensitivity).
The best separation depends on the relative import-
ance of peak resolution, run time and detection sensi-
tivity, and will usually correspond to an intermediate
value of %B or tG.
An example of the effect of isocratic solvent
strength (%B) on retention and selectivity is seen in
Figure 1 for the simulated separations of eight aro-
matic compounds. It is seen that retention and band-
width increase inversely with %B. In general, Rs also
increases, but not for every peak pair } only seven out
of the eight peaks are visible at 70% and 50%B. Note
the relative forward movement of benzene from
70%B, where it co-elutes with 2-nitrotoluene to
50%B, where it co-elutes with 2,6-dinitrotoluene; at
intermediate solvent strengths it is resolved from
neighbouring peaks.
Figure 2 shows the effect of gradient time (tG)
on retention and selectivity for simulated separations
of a proprietary mixture of 11 herbicides. Retention
and bandwidth increase with increasing tG. The over-
all resolution increases with longer gradients, but
note that peak 7, which elutes after peak 6 in the
20 min gradient, moves ahead of peak 7 with longer
gradient times.
When satisfactory separation cannot be obtained
by adjustment of the primary variable (%B or tG), the
usual problem is one of overlapping bands or selectiv-
ity. In the latter case, other conditions (mobile phase,
column packing, temperature) can be varied. For
example, we recommend starting with acetonitrile as
the B solvent. Changes in selectivity are often ob-
served if methanol or tetrahydrofuran is used instead
of acetonitrile. Other variables worth examining are
column temperature, pH (for ionic samples), use of
ion-pairing reagents (ionic samples) and different
types of stationary phase (e.g. change from C18 to
a cyano or phenyl phase).
The Rnal step in method development is to adjust
so-called column conditions: Sow rate, column di-
mensions and/or packing particle size. Typically,
4627
sample resolution increases only slowly with decrease
in Sow rate or increase in column length, while run
time increases much faster. If resolution is greater
than required, this means that an increase in Sow rate
and/or decrease in column length can be used for
a signiRcant decrease in run time with acceptable loss
in resolution. Smaller particle columns are typically
used in shorter lengths; these small particle columns
can provide shorter run times without loss in resolu-
tion or increase in column pressure. The column
pressure drop (or system pressure) increases with
higher Sow rates, longer columns and smaller par-
ticles. Since it is desirable to maintain a system pres-
sure (200 atm, this places a further constraint on
the latter column conditions.
The simulated chromatograms of Figure 3 show
the effect of changes in column conditions on the
aromatic sample of Figure 1. The lower run is the
same as the middle run of Figure 1, using a 250 mm,
5
m particle column with a Sow rate of 2 mL min\,
generating Rs"2.0 in 11 min with 100 bar back
pressure. By changing to a 150 mm, 3.5
m column
at the same Sow rate, the run time is reduced to
6 min. For many applications, the narrower peaks
(and thus lower detection limits) and shorter run time
will be worth the minimal loss in resolution and
increase in pressure (Rs"1.85, 120 bar back pres-
sure). If lower resolution is acceptable, a shorter col-
umn (75 mm, 3.5
m) at a higher Sow rate
(4 mL min\1) will reduce the run time to (2 min, as
shown in Figure 3C (Rs"1.15, 120 bar back pres-
sure).
Choice of HPLC Mode
Reversed-phase HPLC will prove adequate for most
samples. Sample types requiring other chromato-
graphic methods are summarized in Table 2. For
samples that fall in one of these categories, consult
the Further Reading section for detailed instructions.
Choice of Starting Conditions
A recommended set of starting conditions is sum-
marized in Table 3. A C8 or C18 column is chosen,
with no particular preference for either phase. The
150;4.6 mm column size packed with 5
m particles
is capable of achieving most separations; with Sow
rates of 1}2 mL min\1, run times are usually
(15 min. One of the newer type B (low metal)
silicas is strongly recommended for optimum peak
shape and better column-to-column reproducibility.
Acetonitrile}water is recommended as mobile
phase, because of its lower viscosity (and lower pres-
sure drop), as well as its ability to be used with low
4628 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY

Figure 1 Simulated chromatograms for isocratic separations of aromatic compounds on a C18 column using water}acetonitrile
mobile phases. (A) 70%; (B) 65%; (C) 60%; (D) 55%; (E) 50% acetonitrile. Samples: nitrobenzene, 2,6-dinitrotoluene, benzene,
2-nitrotoluene, 4-nitrotoluene, 3-nitrotoluene, 2-nitro-1,3-xylene, and 4-nitro-1,2-xylene (in retention order).

wavelength UV detection (5190 nm; required for
assay of some samples). If ionizable compounds are
present in the sample, a buffer should be used.
Phosphate at pH 2.5 is recommended for the initial
separation, but note its reduced solubility for '80%
acetonitrile}buffer. When a volatile buffer is needed
for liquid chromatography}mass spectrometry (LC-
MS) applications, 0.1% triSuoroacetic acid (pH 1.9),
formic acid or ammonium acetate can be used.
The column should be thermostatted to maintain
constant temperature and retention times; 5}153C
above room temperature is recommended for the in-
itial separation. Temperature can be further adjusted
to change selectivity if necessary. A sufRcient
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY 4629

weight of sample must be injected to obtain adequate

detection sensitivity, but weights '10
g should be
avoided initially. Similarly, sample volume should be
(50
L to avoid excess band broadening.

Control of the Separation:

Selection of Conditions
The selection of conditions for an HPLC method is
expedited by a systematic approach. Because the goal
of most separation development is to establish resolu-
tion for some or all peaks in a chromatographic run,
we will use the fundamental resolution equation
(eqn [1]) as a guideline:

Figure 2 Simulated chromatograms for gradient separations of Rs" (N0.5)(
!1)(k/(1#k)) [1]

a proprietary herbicide mixture with a buffer}acetonitrile mobile 
phase. 5}80% acetonitrile in (A) 20; (B) 30; (C) 40 min. i ii iii

Figure 3 Simulated chromatograms for the sample shown in Figure 1 (60% acetonitrile) when column conditions are varied. (A)
250 mm, 5
m particle column at a flow rate of 2 mL min1; (B) 150 mm, 3.5
m at 2 mL min1; (C) 75 mm, 3.5
m at 4 mL min1 (expanded
scale in upper right).
4630 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY

Table 2 Preferred HPLC methods and columns for different samples

Sample characteristics Preferred HPLC method/column

High molecular weight Special columns usually required

Size exclusion and ion exchange HPLC often preferred
Optical isomers (enantiomers) present Special chiral columns required
Other isomers (stereo-, position, etc.) Normal-phase often best, especially with unmodified silica
Mixtures of inorganic salts Ion chromatography
Carbohydrates Amino-bonded phase columns with reversed-phase conditions; ion exchange resins
Biological samples Special conditions often required for life science samples; may not require different approach
Hydrocarbon mixtures Normal-phase with unmodified silica

where Rs is the resolution, N is the column plate graphic performance, separations in which
number,
is the separation factor (selectivity), and 0.5(k(20, or better 1(k(10, are preferred. If
k is the retention factor. The inSuence of each of these k is too small (low retention), resolution is often poor
variables on the separation is discussed below. because peaks tend to bunch at t0, while interferences
from unretained sample components can also be
Control of Retention a problem. When k is too large, run times are excess-
ive, and detection sensitivity suffers because of
Term iii of eqn [1] varies with solvent strength. For
wide peaks. Because of the major effect of sol-
reversed-phase separations, increased %B increases
vent strength on separation, the selection of an ac-
solvent strength, reduces sample retention (values of
ceptable value of %B should be the Rrst priority. As
k), and reduces the size of term iii (and the value of Rs).
will be seen in the following section, separation selec-
The retention factor, k, is calculated using eqn [2]:
tivity may also be affected by %B. Examples of the
effect of %B on the separation were discussed earlier
k"(tR!t0)/t0 [2] in conjunction with the chromatograms of Figure 1.
For isocratic method development, the rule of three
where tR is the retention time and t0 is the column can be used as a guideline to adjust retention by
dead time. In general, as k increases, resolution and varying %B. The rule of three states that retention (or
run time increase while bandwidth increases and peak k) changes about threefold for a 10% change in
height (sensitivity) decreases. For the best chromato- mobile-phase %B. Thus, a change from 50% meth-
anol to 60% methanol will reduce retention times by
Table 3 Experimental conditions for initial isocratic HPLC about three times. Similarly, a 20%B change will
separation cause a 3;3 or about 10-fold change in retention.
A convenient way to select a value of %B for isocratic
Separation variable Preferred initial choice separations is to start at 90% or 100% B and reduce
%B in 10% steps until retention is in a reasonable
Column
Dimensions (length, i.d.) 150;4.6 mm range, then carry out Rnal small adjustments in this
Particle size 5
m variable.
Stationary phase C8 or C18
Control of Selectivity
Mobile phase
Solvents A/B Water}acetonitrile The selectivity term, ii, of eqn [1] is based on the
%B Variable separation factor
, deRned in eqn [3]:
Buffer 25 mmol L\1 phosphate, pH 2.5 or
0.1% trifluoroacetic acid
Additivesa (e.g. ion pair As necessary
"k2/k1 [3]
reagents, amines)
Flow rate 1}2 mL min\1
where k1 and k2 are values of k for the Rrst and second
Temperature 403C peaks of interest, respectively. Because
is related to
k, changing k by varying %B often results in changes
Sample size in selectivity as well. Because the optimization of
Volumeb (50
L
term iii of eqn [1] depends on the adjustment of %B,
Massb (100
g
corresponding changes in selectivity (term ii) are con-
a
Mainly affecting separation of ionized compounds. veniently made at the same time (by small further
b
Assumes 150;4.6 mm reversed-phase column. adjustments in %B). Note that acceptable values of
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY 4631

term iii can often be achieved by any value of %B most proRtable use of term i (by change in column
within a 5}10% range. The ease of changing %B conditions).
is a further reason for using this variable to vary
Gradient Elution
selectivity.
Although selectivity is inSuenced by %B, changes Most workers prefer isocratic methods for routine
in other conditions can have a much larger effect use. If an isocratic separation is not feasible because
on values of
. Changes in the organic solvent type, of too broad a sample retention range (0.5(k(20
pH, or use of additives are usually the next choice, not possible for all peaks), gradient elution is re-
once mobile-phase %B is optimized in terms of both quired. Even where a Rnal isocratic method is poss-
k and
. For simplicity, we recommend changes in ible, it is still advantageous to begin the method
organic solvent Rrst for neutral compounds. After development process with a gradient run. Thus,
starting with acetonitrile, change next to methanol, a single gradient run can be carried out which will
while reserving tetrahydrofuran as a last choice for provide an attractive value of term iii for every
solvent type. Changes in temperature can provide sample peak, thus avoiding problems in isocratic elu-
further changes in selectivity, especially for acid and tion that are caused by values of %B that are too
base samples. If the sample contains acids and/or large or too small (poor resolution, long run times,
bases, changes in mobile-phase pH can be the most wide peaks and poor detection sensitivity). With iso-
powerful means to control selectivity. If none of these cratic separation, several runs (and several hours)
approaches is successful, mobile-phase additives, may be required to Rnd conditions for which
such as ion-pairing reagents may be helpful, or a dif- 0.5(k(20 for the sample. In contrast, equivalent
ferent kind of column (C18, cyano, phenyl) can be conditions for gradient elution can be determined in
tried. advance and the Rrst run can generate a reasonable
separation. Furthermore, by using method develop-
Control of Column Ef\ciency
ment software with gradient input runs, it is easy to
We have noted that sample resolution is not much convert a gradient method to an isocratic one and to
increased by changes in column conditions (except evaluate the trade-offs between an isocratic and
for a large increase in run time). For the same reason, gradient Rnal method.
the column plate number N and term i of eqn [1] Typical starting conditions for gradient elution are
usually cannot provide a large increase in resolution, given in Table 4. In general, longer gradient times
once terms ii and iii have been optimized. This means (smaller %B min\1 changes) will give the same re-
that it is important to select initial conditions which sults (increased resolution and run time, broader and
provide a value of N that will be sufRcient for
most separations. For most samples, we recommend
one of the newer columns based on a low metal Table 4 Recommended gradient elution starting conditions
silica, often termed type B or base-deactivated silica. Separation variable Preferred initial choice
These columns reduce unwanted chemical interac-
tions that cause peak tailing and column-to-column Column
variations. Columns packed with spherical 5, 3.5 or Dimensions (length, i.d.) 150;4.6 mm
3
m particles are preferable. We favour either Particle size 5
m
Stationary phase C8 or C18
(a) 150;4.6 mm, 5
m or (b) 75;4.6 mm, 3.5
m
columns as a starting point. Column (a) is the Rrst Mobile phase
choice of many users because it is robust and has Solvents A/B Water}acetonitrile
sufRciently low back pressure to allow operation %B 5}100% B in 20 mina
at 2 mL min\1, resulting in short run times. Smaller Buffer 25 mmol L\1 phosphate, pH 2.5 or
0.1% trifluoroacetic acid
particles give a better compromise of plate num- Additivesb (e.g. ion pair As necessary
ber versus run time (for the same column pressure reagents, amines)
drop), but are more prone to problems such as Flow rate 1}2 mL min\1
blockage by particulates in the sample or mobile
phase. Temperature 403C
If the adjustment of conditions for optimization of Sample size
terms ii (
) and iii (k) in eqn [1] has been successful, Volumec (50
L
Massb (100
g
not infrequently sample resolution will be greater
than required. In this case, run time can often be a
With phosphate buffers and acetonitrile, use 5}80% B in 15 min.
substantially reduced by increasing Sow rate while b
Mainly affecting separation of ionized compounds.
decreasing column length. The latter represents the c
Assumes 150;4.6 mm reversed-phase column.
4632 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY
shorter peaks) as using a weaker mobile phase (lower
%B) in isocratic separations. The earlier discussion of
the chromatograms of Figure 2 illustrated the ef-
fect of gradient time on the separation. For more
detailed instructions in the use of gradient elution,
consult the Further Reading section.
Method Development Strategies
The goal for method development is to obtain a ro-
bust method with acceptable resolution and run time.
The strategy to reach this goal has two components.
First, selectivity variables should be chosen for testing
in an order that is most likely to give a successful
separation in the minimum amount of development
time. Second, for robustness in routine use and
method transfer, the separation conditions should be
as simple as possible and avoid potentially unstable
parameters.
Most neutral or ionic samples can be separated
successfully with reversed-phase columns and simple
binary mobile phases of water or buffer in com-
bination with one of the three primary organic sol-
vents } acetonitrile, methanol or tetrahydrofuran.
A choice of conditions should be made by investigat-
ing the parameter most likely to succeed, then moving
to the next most likely, and so forth. If a satisfactory
separation cannot be obtained by single-parameter
optimization, the use of (simultaneous) two-para-
meter optimization can be explored.
Choice of Selectivity Variables
Selectivity variables should be examined one at a time
in a systematic manner. Starting conditions generally
will correspond to those listed in Table 3 or 4. Sys-
tematic method development can be approached by
proceeding in order through the variables listed in
Table 5, as described below.
Optimize %B First adjust the mobile-phase %B for
reasonable k values and Rne-tune for selectivity, as
described previously. Acetonitrile is generally the Rrst
choice for organic solvent, but methanol is an accept-
Table 5 Choice of selectivity variables
Optimize %B using acetonitrile or methanol
Solvent type (acetonitrile versus methanol)
Temperature
pH (ionics)
Column type (C8"C18'CN'amide'phenyl)
Additives
Tetrahydrofuran
Separation mode
Figure 4 Solvent nomograms for reversed-phase separations.
Convert percentage acetonitrile (ACN) to percentage methanol
(MeOH) or percentage tetrahydrofuran (THF) by moving vertically
between scales.
able alternative. Use water as the A solvent for neu-
trals or a low pH buffer for ionics.
Solvent type If a separation cannot be obtained with
acetonitrile, change to methanol and repeat the op-
timization experiments. A mobile phase that gives
approximately the same retention time can be se-
lected with the help of the nomogram shown in
Figure 4. For example, 60% acetonitrile}water is
roughly equivalent to 70% methanol}water. If nei-
ther solvent provides satisfactory separations, some
workers will try tetrahydrofuran at this point, but we
recommend delaying the use of this solvent until later.
Tetrahydrofuran readily forms peroxides and is less
desirable for other reasons.
Figure 5 illustrates both the use of the nomogram
of Figure 4 and the selectivity that may be obtained
with different organic solvents using simulated
chromatograms of a 10-component steroid mix-
ture on a C18 column. Figure 5A shows the best separ-
ation with acetonitrile (50%); the resolution of
the critical pairs is shown. Figure 4 indicates that
50% acetonitrile is equivalent to about 60% meth-
anol, and the methanol separation is shown in
Figure 5B. The separation in methanol is signiRcantly
improved, but still unsatisfactory, so tetrahydrofuran
is tried next. Figure 5C shows the 40% tetra-
hydrofuran separation, with baseline resolution
of all peaks. Note that the nomogram is not perfect
} the indicated tetrahydrofuran concentration re-
sulted in a longer run time than for acetonitrile or
methanol.
Temperature Column temperature should be con-
trolled so that retention times do not drift. In general,
a 1}2% change in retention will be observed for a 13C
change in temperature, but many workers do not
appreciate that selectivity also often changes with
temperature. A second experiment run at 20}303C
higher temperature will indicate if improved selectiv-
ity can be obtained by varying column temperature.
While changes in
as a result of a change in temper-
ature are smaller than for other changes in condi-
tions, this disadvantage is offset by the greater
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY 4633

Figure 5 Comparison of selectivity changes with different organic solvents using an 11-component steroid sample and a C18 column
(simulated chromatograms); resolution of critical peak pair(s) shown as call-outs. (A) 50% acetonitrile; (B) 60% methanol; (C) 40%
tetrahydrofuran.

convenience of a change in temperature, without any
offsetting disadvantages.
Mobile-phase pH If the sample contains ionizable
compounds (acids or bases), mobile-phase pH rep-
resents a powerful variable for changing selectiv-
ity. Thus, values of k for ionized species will generally
be much smaller than for the nonionized compound.
As a result, both absolute and relative retention for
acids or bases can change dramatically with small
changes in pH, when the pKa value of the compound
is within 1}1.5 units of the mobile-phase pH. Be
sure to use buffers within their effective buf-
fering range ($1 unit from the buffer pKa).
When optimizing pH, changes in steps no more
than about 0.5 pH units are recommended. Note that
silica-based columns are generally limited for use at
2(pH(8.
On the other hand, the choice of a pH(3 is
advantageous for several reasons: Rrst, the pKa values
of both acids and bases will differ from the
mobile phase pH by '2 units, so that sample reten-
tion will not vary with small changes in pH; i.e., the
method will be more robust. Second, bases are usu-
ally best separated at low pH, because undesirable
interactions between sample molecules and the col-
umn packing (i.e. silanols) are suppressed, thereby
minimizing peak tailing and maximizing column
plate numbers. However, the choice of a pH(3
means that very little change in selectivity can be
expected as a result of intentional changes in pH (e.g.
for 2(pH(3).
Column type The initial separation will usually be
done on a C8 or C18 column. For changes in selectiv-
ity, it is seldom fruitful to change the bonded-phase
chain length (e.g. C8 to C18 or C4). Similarly, although
changes in selectivity may be observed with the same
phase obtained from different manufacturers,
the magnitude of such changes is generally small.
Rather, if the column is to be changed in order to
change selectivity, it is recommended to change
4634 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY
to a stationary phase with signiRcantly different
chemistry.
After a C8 or C18 column has been tried, a cyano
(CN) phase is usually the next choice. Because cyano
columns are more polar, similar retention requires the
use of 10}20% less organic solvent; e.g. similar reten-
tion might be obtained with 35%B on a cyano col-
umn as with 50%B on a C8 column.
A phenyl column is usually the next choice, if
a cyano phase does not work. However, recently
developed columns with an amide or carbamate func-
tion (e.g. Symmetry Shield, Zorbax Bonus RP or
Discovery Amide) have proven to have unique selec-
tivity that is also worth exploring when examining
column-type effects. Additional information re-
garding column selection and chemistry can be found
in the Further Reading section.
Additives Mobile-phase additives (in addition to
buffers) can be used to enhance selectivity with
some sample types. For example, ion pairing may be
used to advantage when the sample contains both
acidic and basic components. While large changes in
selectivity are possible by varying the concentration
of an ion pair reagent, ion pairing often results in long
equilibration times when changing the mobile phase,
as well as other problems. Optimization of additives
is beyond the scope of the present discussion.
Tetrahydrofuran Tetrahdyrofuran as the B solvent
often gives signiRcant selectivity changes when com-
pared to acetonitrile or methanol. Problems related to
slow equilibration, equipment memory effects,
excessive UV background at low wavelengths, insta-
bility and unpleasant odour make most workers delay
the use of tetrahydrofuran until it is unavoidable. In
spite of these potential problems, tetrahydrofuran
does have unique selectivity characteristics and will
often provide separations when acetonitrile or meth-
anol have failed. If tetrahydrofuran is to be used, use
Figure 4 to select starting mobile-phase conditions
based on previous experiments with acetonitrile or
methanol.
Separation mode When efforts at obtaining a
successful reversed-phase separation prove un-
successful, one should consider other separation
modes. Several other separation modes are shown in
Table 2.
Single-Variable Optimization
Traditionally, a single variable is optimized at a time
during HPLC method development. In most cases,
one can proceed through the list of variables in
Table 5 in order, stopping when an adequate separ-
ation is achieved. A convenient procedure is to
optimize the Rrst variable, then hold that condition
constant while changing the next parameter. This
sequential optimization of parameters is a straightfor-
ward approach. Once all the desired variables have
been optimized, it is a good idea to make small and
reasonable changes around the Rnal conditions to
check robustness. For example, change $3}5%B,
$0.5 pH units, $53C, and so forth to make sure
the separation is not adversely affected by such
changes.
Multi-Variable Optimization
An alternate approach to single-variable optimization
is to change two or more parameters at once. Two
different approaches are suggested: the method
development triangle and the simultaneous optimiza-
tion of solvent strength (or gradient time) with a sec-
ond variable. In both of these cases, one should
choose variables that change selectivity in differ-
ent ways, ideally orthogonal to each other in terms of
their selectivity effects.
Method development triangle The method develop-
ment triangle shown in Figure 6 is a widely used
approach for selecting the optimum organic solvent
or mixture of organic solvents. This is a logical next
step when acetonitrile, methanol and tetrahydrofuran
have been optimized individually, and the least re-
solved peak pairs are different for at least two
solvents. Each corner of Figure 6 represents a binary-
solvent mobile phase with a %B value (for each of
these three B solvents) that gives acceptable isocratic
Figure 6 Solvent selectivity optimization for reversed-phase
HPLC. Experiments 1, 2 and 3 are binary (water and one organic)
mobile phases; 4, 5 and 6 are 1 : 1 ternary (water and two
organics) blends of corner compositions, and 3; 7 is 1 : 1 : 1 quat-
ernary (water plus three organics) blend of corner compositions.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN LIQUID CHROMATOGRAPHY
separation; e.g. so as to give k+10 for the last peak
in each separation. These mobile phases (1, 2 and 3 in
Figure 6) then are blended 1 : 1 or 1 : 1 : 1 for the
remaining experiments. Once all seven experiments
are run, the chromatograms can be spread out in the
same grid pattern and examined for changes in selec-
tivity between conditions. Further adjustments in
solvent blends may beneRcial. For simplicity and
robustness, mobile phases with fewer solvents are
preferred (binary'ternary'quaternary). As with
other optimization strategies, the Rnal conditions
should be varied in a systematic manner to determine
the robustness of the chosen mobile phase.
Two-variable optimization The method develop-
ment triangle approach (shown in Figure 6) is one
example of the simultaneous variation of two vari-
ables. Other two-variable optimization procedures
are now possible with the recent availability of
appropriate computer simulation software. This new
software (DryLab version 3.0, LC Resources) facili-
tates the simultaneous optimization of either isocratic
%B or gradient time and any second variable (e.g.
temperature, pH, additive concentration). The com-
bination of two variables having different selec-
tivity actions can help identify separation conditions
that are unlikely to be found using more traditional
approaches. With the use of optimization software,
four to six input runs allow the user to model the
separation under any combination of the two
variables.
An example of the results of a two-variable optim-
ization is shown in Figure 7 using a 150 mm C18 col-
Figure 7 Simulated chromatograms for separation of 10 ben-
zoic acids and anilines using a 150 mm, 5
m particle C18 column
with acetonitrile}buffer mobile phases. (A) Optimum separation at
pH 3.0 (15% acetonitrile); (B) optimized %B-pH conditions (pH
3.4, 25% acetonitrile) using computer-assisted method develop-
ment (DryLab).
4635
Table 6 DryLab software optimization modes
Parameter Input
experiments
Isocratic %B 2 or 3
Gradient time 2
Normal phase 3
pH 3
Ternary solvent 3
Ionic strength 3
Additive concentration 3
Temperature 2
Gradient time versus temperature 4
Isocratic %B or gradient time versus any variable 4}9
umn with buffered acetonitrile to separate a 10-
component mixture of benzoic acids and anilines.
The best single-variable separation (at an arbitrary
starting pH of 3.0) is shown in Figure 7A with
Rs'2, but the %B must be held within $1% and
the pH within $0.05 units to maintain Rs'1.5,
so the method is not robust. Using six experimental
runs and computer optimization, the lower
chromatogram (Figure 7B) was obtained, offer-
ing Rs'2 for $5%B and $0.1 pH units in half the
run time.
Computer-Assisted Method Development
Many of the above changes in separation as a func-
tion of conditions can be described in theoretical or
empirical equations. The fundamental relationships
deRned in eqn [1] form the basis of algorithms used to
predict resolution. For example, term i of eqn [1] can
be calculated from Rrst principles, term ii is deRned in
eqn [3], and log(k) is linearly related to %B (term iii).
This means that two experiments differing only
in %B can be used to predict resolution at any other
%B. Similarly, basic theory can relate isocratic and
gradient separations in terms of the same retention
relationships. A computer program (software) can
therefore be used to predict separation as a function
of isocratic %B, gradient conditions, and/or column
conditions, using two gradient runs to calibrate the
sample and initial conditions. The example of Fig-
ure 7 required six runs (3 pH values at each of two
gradient times) for optimization of isocratic %B and
pH. Some of the other variables available for optim-
ization with one of these programs (DryLab soft-
ware) are shown in Table 6 along with the number of
input experiments required. The use of optimization
software is strongly recommended in order to reduce
method development time, achieve more robust
separations, and gain a better understanding of the
separation.
4636 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION

Table 7 Common HPLC separation problems

Observation Problem source Solution

Poor peak shape Wrong silica type Use type B silica

Blocked frit or column void Replace frit, backflush column
Silanol interactions Use amine additives, change pH,
use end-capped stationary phase
Excessive peak width Bad column Replace column
Column overload Reduce injection volume or mass
High molecular weight Normal
Unresolved peaks Improve separation
Inadequate retention Mobile phase too strong Use lower %B
Column too weak Switch to C18
Samples ionized Change pH
Samples too polar Change to normal phase
Gradient starting too strong Start at lower %B
Excessive retention Mobile phase too weak Use higher %B
Column too retentive Switch to C8, C4 or CN
Samples too hydrophobic Change to normal phase
Gradient stops too soon Stop at higher %B
Excessive retention range Acids and bases or bases and neutrals in sample Use ion pairing
Too broad of polarity for isocratic method Use gradient elution
Inadequate resolution Retention too short Increase k
Poor selectivity Change

Plate number too low Use longer column or smaller
particle size

Troubleshooting Common Problems HorvaH th Cs (ed.) (1980}86) High Performance Liquid

Table 7 highlights some of the commonest causes of
chromatographic problems likely to be encountered
in the HPLC method development.
Further Reading
Dolan JW and Snyder LR (1989) Troubleshooting HPLC
Systems. Clifton, NJ: Humana Press.
Chromatography. Advances and Perspectives, vols 1}4.
New York: Academic Press.
Neue UD (1997) HPLC Columns. Theory, Technology and
Practice. New York: Wiley-VCH.
Snyder LR, Kirkland JJ and Glajch JL (1997) Practical
HPLC Method Development, 2nd edn. New York:
Wiley-Interscience.

ESSENTIAL GUIDES TO METHOD

DEVELOPMENT IN SOLID-PHASE
EXTRACTION
M. J. M. Wells, Tennessee Technological University,
Cookeville, TN, USA
Copyright ^ 2000 Academic Press
Solid-phase extraction (SPE) is a sample preparation
technique combining nonlinear modes of chromato-
graphy for the separation, puriRcation, concen-
tration and/or solvent exchange of analytes of
interest. SPE is the removal of chemical constituents
from a Sowing liquid sample via retention on a solid
sorbent, and the subsequent recovery of selected
constituents by elution from the sorbent. SPE was
developed as an heterogeneous (two-phase) alterna-
tive to homogeneous (one-phase) liquid}liquid
extraction (LLE) for the isolation of solutes from
solution.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION 4637

Background
The modern era of SPE began in October 1977 when
prepackaged, disposable cartridges/columns contain-
ing bonded silica sorbents were introduced by Waters
Associates. This technique was featured on the cover
of Laboratory Equipment in May 1978 and the Rrst
peer-reviewed method was published in the Journal
of Chromatography that same year. The term solid-
phase extraction wasnt actually popularized until the
early 1980s.
Unlike the meagre resources available to early SPE
researchers, there are currently thousands of publica-
tions for analytes of pharmaceutical and environ-
mental interest that may be consulted for examples of
developed SPE methods. Many manufacturers pub-
lish bibliographies of methods developed using their
products that are available in print or via the Internet.
However, it is sometimes difRcult to recognize the
process used to arrive at published protocol, and, it is
still common that an SPE method for the solute}
matrix combination required has not been previously
developed. Even if an appropriate method exists, it is
advisable to understand thoroughly the principles of
SPE method development in order to evaluate proper-
ly published methods. Figure 1 Solid-phase extraction consists of four basic steps:
Elsewhere in this volume, there are articles (A) conditioning, (B) retention, (C) rinsing and (D) elution. (A)
Conditioning the sorbent prior to sample application ensures re-
dealing with speciRc SPE topics (Table 1) that should producible retention of the compound of interest (the isolate). (B)
be consulted for detailed descriptions. This contribu- Squares, adsorbed isolate; circles, undesired matrix constituents;
tion addresses method development issues in SPE. triangles, other undesired matrix components. (C) Triangles, rinse
the columns to remove undesired matrix components. (D) Circles,
undesired components remain; squares, purified and concen-
Principles trated isolate is ready for analysis. (Reproduced with permission
SPE method development requires exploitation of from http://www.varianinc.com/spp/shared/4step.jpg at http://
www.varianinc.com/spp/solphase.html Copyright 1999 Varian, Inc.)
analyte properties, selection of the appropriate

Table 1 Articles on solid-phase extraction appearing in the

sorbent and recognition of limitations imposed on the
Encyclopedia of Separation Science analysis by the sample matrix. The distribution of the
analyte between the sample matrix and the sorbent in
Article SPE is determined by physical and chemical proper-
Extraction ties of the analyte and the sorbent. The sample matrix
Solid-phase extraction (SPE) can be manipulated to inSuence the distribution. In
Automation of SPE SPE method development, identiRcation of the char-
Bioanalytical applications of SPE acteristic properties of the analytes of interest is a
(excluding drugs of abuse)
Classical SPE
necessary Rrst step before the sorbent can be selected
Covalent SPE using immobilized boronic acids or the sample matrix can be modiRed to effect
Disk approach to SPE the recovery.
Drugs of abuse SPE consists of a basic four-step approach
Herbicides (Figure 1):
Insecticides
Medical applications (treatment of blood): SPE
Mixed-mode SPE 1. Sorbent preparation or pre-wash: stationary phase
Molecular imprints for SPE conditioning;
Polycyclic aromatic hydrocarbons 2. Retention: analyte adsorption;
Phenols 3. Sorbent post-wash: removing undesirable con-
Restricted-access media (SPE)
Sorbent selection for SPE
taminants;
4. Elution: analyte desorption.
4638 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION
The four-step process can be as simple as this
or may become more involved as one or more of
these stages includes additional phases, such as
selective adsorption or selective desorption. SPE
method development can be tedious because the
retention and recovery processes are interdepen-
dent. Retention and elution are confounded dur-
ing method development because the overall analyte
recovery is dependent upon both the retention
efRciency and the elution efRciency.
the analyte is only recovered in part by an SPE tech-
nique, it will be initially unclear whether the problem
lies with the retention process or with the elution
process.
Because of this quandary, SPE method devel-
opment can be approached in two ways. An iter-
ative approach to protocol development emulating
the classical analytical approach to change one
variable at a time can be used. Retention para-
meters are held constant at selected values while
optimizing the elution process (Table 2). Once
elution is optimized, then the most favourable
conditions for retention are determined. The pro-
cedure is repeated until the desired results are ob-
tained.
Alternatively, a factorial experimental design ap-
proach to determine extraction efRciency is an
efRcient method development technique. Para-
meters important to SPE, such as sample pH, elution
solvent strength, ionic strength of the sample, addi-
tion of organic modiRer to the sample, elution by
gravity or vacuum, sorbent retentivity, selection of
sorbent mass, sample volume, elution volume and
sample concentration, representing effects on
both retention and elution, may be selected as
factors that inSuence analyte recovery. The factors
are usually tested at two or three levels. As a screen-
ing procedure, the factorial design can quickly
pinpoint signiRcant effects important to SPE.
The objective of the factorial design approach is to
obtain as much information as possible from few
analyses.
Table 2 Factors affecting SPE retention and elution
Retention
Analyte character
Sorbent type
Matrix additives
Sample volume
Sorbent mass
Elution
Eluting solvent identity
Eluting solvent volume
Elution rate
Table 3 SPE sorbent}analyte interaction mechanisms
Primary mechanism Sorbents
Van der Waals Octadecyl, octyl, ethyl, phenyl,
cyclohexyl, styrene-divinylbenzene
Polar}dipole/dipole Cyano, silica, alumina, Florisil
Hydrogen-bonding Amino, diol
Electrostatic Cation exchange, anion exchange
If
Fundamentals
Stationary-phase Conditioning
Each different SPE sorbent requires conditioning
or pretreatment in order to activate or prepare the
sorbent to retain the analyte. Conditioning of hydro-
phobic stationary phases requires a two-step process
of treatment with organic solvent followed by an
aqueous wash. The conditioning solvent is prepared
to mimic the chemistry of the sample matrix, that is,
matching the pH and/or the ionic strength of the
matrix. The volume of each conditioning solvent
passed through the sorbent is usually about Rve times
the dead volume of the sorbent. In addition to acti-
vating the sorbent, the conditioning solvent(s) also
removes undesirable contamination potentially re-
maining in the sorbent during manufacture.
Retention
Originally, hydrophobic bonded-silica sorbents were
the Rrst materials introduced speciRcally for SPE,
but currently, a suite of sorbents in varying
formats are available with nonpolar, intermediate
nonpolar/polar, polar, strong and weak anion and
cation exchange, and steric exclusion (restricted ac-
cess) properties. SPE sorbents are designed to retain
analytes by a primary mechanism (Table 3) but often
exhibit a secondary mechanism as well. For example,
bonded-phase ion exchange sorbents primarily ex-
hibit anionic or cationic exchange mechanisms but
the analyte also experiences nonpolar interactions
with the bonded ligand. Also, nonpolar bonded sil-
icas exhibit a secondary polar interaction due to the
silica backbone and unreacted surface silanol groups.
Knowledge of the dual retention mechanisms encoun-
tered in SPE can work to the analysts advantage.
Mixed-mode sorbents (different ligands on the
same sorbent) capitalize on dual retention mecha-
nisms by design.
Understanding the mechanism(s) of retention and
the selection of an appropriate sorbent depends on
a thorough understanding of the character of the
analyte. The solute properties of principal importance
to retention by SPE are hydrophobicity, polarity and
ionogenicity.
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION
Unionized chemicals Many organic compounds are
not ionized in water. An unionized organic com-
pound is formed entirely of covalent bonds. SPE
sorbent selection and recovery of unionized com-
pounds is known generally to depend on the hydro-
phobicity of the analyte. The most widely measured
parameter used to represent solute hydrophobicity
is the octanol/water partition coefRcient (Kow),
which is the ratio of the analyte concentration in
octanol (o) relative to its concentration in an aqueous
(w) phase. The logarithm of Kow, also referred to as
log P, ranges from !3 to 7 for organic chemicals.
Analytes with log Kow values less than 1 are hy-
drophilic; analytes with log Kow values greater than
4 are highly hydrophobic; analytes with log Kows be-
tween 1 and 4 are intermediate in hydrophobicity/
hydrophilicity.
Aliphatic hydrocarbons and mono- and polycyclic
aromatic hydrocarbons (PAHs) are nonpolar com-
pounds that tend to increase in hydrophobicity
(log Kow) with increasing molecular weight; that is,
they tend to be more distributed in the organic,
octanol phase, rather than in the aqueous phase.
As oxygen- and/or nitrogen-containing functional
groups are added to these compounds, they may (but
not always) become more polar relative to the parent
compound and would distribute more equally be-
tween the octanol and water phases, or even prefer to
exist in the aqueous phase, thereby decreasing the
value of log Kow.
Sorbents for extraction of unionized compounds
fall into two general categories: nonpolar extraction
sorbents and polar extraction sorbents, depending on
the polarity of the functional groups present. Com-
mercially available reversed-phase bonded-silica
sorbents for SPE are produced in ranging polarities.
The identity of the hydrocarbon covalently bonded to
the silica gel backbone may be varied. Common non-
polar ligands bonded on the silica gel surface include
aliphatic hydrocarbons of one, two, eight, or 18
methylene, cyclohexyl or phenyl groups. The greater
the number of methylene groups in the aliphatic
chain, the greater the hydrophobicity of the sorbent
generated, i.e. C18'C8'C2. For nonionized
compounds, the hydrophobicity of the analyte and
the hydrophobicity of the sorbent selected for SPE are
inversely related; that is, less hydrophobic sorbents
are used for highly hydrophobic analytes; conversely,
the most hydrophobic sorbents should be used for
more polar analytes.
Bonded-phase silica sorbents are known to exhibit
mixed-mode retention mechanisms due to silano-
philic (silanol) sites that remain on the sorbent after
the initial hydrocarbon is bonded to the surface. The
presence of silanol groups reduces the hydrophobic
4639
character of the surface. Silanol groups on the surface
of bonded silicas interact with electron-rich hydroxyl,
carbonyl, nitrile and nitro functional groups in
analytes. Some of the silanol groups on the surface
may be masked by a subsequent reaction with a short
chain hydrocarbon in a manufacturing process
termed end-capping. Consequently, reversed-phase
sorbents that are end-capped are more hydrophobic
in character than those that are not end-capped.
When a less retentive, less hydrophobic sorbent is
desired, a nonend-capped product should be tested.
Other polar sorbents are produced by adding oxygen-
or nitrogen-containing functional groups such as
cyano, hydroxyl or amino to the hydrocarbon bonded
phase, or functionalized polymeric phases are
produced to enhance polarity. UnmodiRed silica,
alumina and Florisil sorbents are polar extraction
sorbents.
Matrix additives inSuence retention in SPE. For
many nonionized chemical compounds, increasing
the ionic strength of the sample matrix by the addi-
tion of sodium chloride decreases analyte solubility in
the sample matrix and increases adsorption on to
the nonpolar sorbent via a salting-out effect.
Increased salt content in the sample matrix may also
produce silanol masking. Silanol groups remaining on
the surface can also be deactivated through ion pair-
ing by the addition of masking reagents such as
tetrabutylammonium hydrogen sulfate to the sample
matrix.
Adding water-miscible organic solvents such as
methanol or acetonitrile to the sample matrix reduces
the surface tension of the sample matrix, thereby
decreasing the retentiveness of highly hydrophobic
compounds. The addition of salt to increase the ionic
strength of the sample matrix and the concomitant
addition of methanol to decrease surface tension can
be useful in developing methods for samples having
multiple compounds that vary widely in their hydro-
phobic character.
Ionized chemicals Organic chemical compounds
that undergo ionization are comprised of deriva-
tives of acidic organic alcohols, carboxylic acids
or basic amine functional groups. Substances that
ionize dissociate into their conjugate acid or base
in aqueous solution. Ionizable compounds, includ-
ing organic acids such as acetic acid or benzoic
acid, and organic bases such as ethylamine or aniline,
are weak electrolytes and are incompletely disso-
ciated in water. The acid dissociation constant,
Ka, expresses the ratio of ionized to unionized
analyte; therefore, the greater the value of Ka (or the
logarithm of Ka, the pKa) the more ionized material
is present. Just as in LLE, the dependence of SPE
4640 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION
recovery on sample pH is a function of the pKa of the
analyte.
The relative per cent of unionized analyte to that in
the ionized form exhibits a pH-dependent dissocia-
tion that can be exploited by SPE methodology. The
relative concentrations of dissociated and non-
dissociated forms of ionizable analytes in aqueous
solution are equal when the solution pH is equal to
the pKa. Therefore, in aqueous solution an organic
acid is 99% unionized when the pH of the sample is
two log units below the pKa; and it is 99% ionized
when the pH of the sample is two log units above the
pKa. Two log units is a good rule of thumb for
considering retention of ionizable compounds. How-
ever, if the sorbent (whether nonpolar, polar or ion
exchange) exhibits primary or secondary, nonpolar
van der Waals-type interactions, the overall hydro-
phobicity and size of the ionized form of the analyte
can also have an effect upon recovery and inSu-
ence the two log units generalization.
In SPE method development for ionogenic com-
pounds, the decision to be made is whether to retain
the analyte as the dissociated or the undissociated
form. If a compound is ionizable, the extraction may
be performed using ion exchange of the dissociated
form. Alternatively, if the analyte can be converted to
an undissociated form by ion suppression or ion pair-
ing, then SPE can be conducted on nonpolar sorbents,
as described in the earlier section on unionized
chemicals.
Organic acids lose protons in aqueous solution,
ionizing to form anions, and are retained on cation
exchange sorbents. Organic bases gain protons in
aqueous solution, ionizing to form cations, and are
retained on anion exchange sorbents. Strong and
weak ion exchange sorbents are available for SPE. Ion
exchange sorbents developed for SPE retain analytes
not only by ionic (electrostatic) attraction, but also
through secondary van der Waals (nonpolar) interac-
tions between the analyte and the atoms comprising
the bridge that links the charged functional group to
the silica gel backbone.
Ion suppression refers to adjusting the pH of the
sample matrix to inSuence the chemistry of the
analyte of interest. If the ionizability of the analyte is
suppressed by controlling the pH of the sample
matrix relative to the pKa of the analyte, then the
analyte can be retained in the unionized form and
nonpolar or polar extraction sorbents are used in-
stead of ion exchange sorbents.
Ion pairing involves using a reagent added to the
mobile phase to accomplish two objectives: to neu-
tralize the analyte charge by combining with an
oppositely charged counterionic solute; and to use
a hydrophobic, bulky group on the counterion to
form an ion pair that is hydrophobic enough to be
retained on nonpolar extraction sorbents. The ion-
pairing strategy applies to ionized organics as well as
metal cations.
Ionizable and nonionizable compounds may co-
exist in the same sample to be analysed. Selective
adsorption of either type of chemical in the presence
of the other can be accomplished by controlling the
sorbent through a mixed-mode approach or by
chromatographic mode sequencing (the use of dif-
fering SPE sorbents in tandem), such as using both ion
exchange and nonpolar mechanisms to extract the
analytes. The use of selective adsorption can be ap-
plied to compounds differing in hydrophobicity,
charge and structure.
In tandem mode, ionizable and nonionizable com-
pounds may be fractionated by adsorbing nonioniz-
able analytes on nonpolar extraction sorbents and
ionized analytes on ion exchange sorbents. Once re-
tained on different sorbents in tandem, they can
be physically separated and eluted individually, there-
by separating the compounds. Such fractionation of-
ten improves the ease of subsequent chromatographic
analyses.
Selective adsorption of nonionized compounds in
the presence of ionogenic compounds can also be
accomplished with pH control of the sample matrix.
By selecting a sample pH at which ionogenic com-
pounds exist in the ionized form, it may be possible
selectively to retain either the ionized or the
nonionized components depending on the type of
sorbent selected.
Values for hydrophobic parameters (log P) and
acid dissociation constants (pKa) can often be ob-
tained from the analytes manufacturer. Alterna-
tively, they can be measured in the laboratory or
predicted by numerical estimation methods.
Sample volume and sorbent mass SPE retention is
dependent on the relationship between sorbent mass
and sample size. The strength of the interaction
(whether nonpolar/polar or ion exchange) between
the analyte and the sorbent, as inSuenced by the
sample matrix solvent strength, determines the
amount of the sample that may be passed through
the sorbent before analyte breakthrough occurs. As
the strength of the interaction increases and as the
sorbent amount increases, the breakthrough volume
increases. Breakthrough can be controlled and
monitored by attaching a second check cartridge in
tandem with the primary extraction cartridge, and
eluting them separately.
To establish the dependence of retention on sample
loading volume, variable volume samples (each of
which comprises a constant molar amount loaded)
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION
are passed through a constant sorbent mass and per
cent recovery is plotted as a function of sample load-
ing volume. Repeating this procedure for differ-
ent sorbent masses will establish the dependence of
retention on sorbent mass. Factorial design experi-
ments can also be used to screen for the sorbent mass
required relative to sample size. Optimizing the
amount of sorbent necessary for the analysis will
control analytical costs.
Adsorbed Contaminant Removal
During the retention process, undesirable con-
taminants in the sample matrix may become asso-
ciated with the sorbent, or may remain behind in
the interstitial spaces between sorbent particles.
When the post-wash solvent is identical to the
conditioning solvent and to the sample matrix,
adsorbed contaminants are not likely to be removed.
However, matrix components remaining in the
interstitial spaces between sorbent particles will
be Sushed from the sorbent. The volume of the post-
wash solvent should be at least equal to or preferably
twice the void volume of the sorbent to ensure that
the pore space is entirely replaced with the desired
solvent.
If a blank, i.e. uncontaminated, sample matrix is
available, it can be used to screen for potential
column post-wash solvents. To remove undesirable
contaminants in the sample matrix that became
associated with the sorbent during the retention
process, solvents of greater eluotropic or eluting sol-
vent strength than the conditioning solvent must
be used. Even a small amount of the elution solvent
can be a post-wash solvent if the effects on the
retained analytes of interest are monitored.
Elution
Elution solvent strength and volume The ability of
a solvent to overcome the interaction between the
analyte and a chromatographic sorbent, thereby
causing elution to occur, is known as the solvents
eluotropic strength. Charts of eluotropic series
can be consulted to determine relative solvent
strength, which is roughly equivalent to polarity. The
eluotropic strength of elution solvents on a non-
polar adsorbent (e.g. reversed-phase) increases in re-
verse order to that measured on polar sorbents such
as silica or alumina. On reversed-phase sorbents,
the eluting power increases as the solvent polarity
decreases.
Many different solvents are used for elution of
the analytes from sorbents in SPE. Elution by acetic
acid, methanol, acetonitrile, acetone, ethyl acetate,
diethyl ether, methyl-tert-butyl ether, methylene
4641
chloride, benzene and hexane, and aqueous buf-
fers containing appropriate counterions have been
reported. Miscible solvent mixtures produce elution
solvents of intermediate eluotropic strength.
After candidate elution solvents are selected, an
elution solvent screen is conducted. Elution solvents
tested at a constant volume are compared for poten-
tial to elute the analytes of interest. When selecting
a desorption solvent, the effect it will have upon
contaminants adsorbed from the sample matrix must
be considered. A control sample matrix should also
be screened if possible. The solvent demonstrating the
most desirable results in the elution screen is further
examined for the variation of recovery as a function
of the volume of eluting solvent. Generally, the elu-
tion solvent selected is the one for which the smallest
volume produces acceptable recovery. However, elu-
tion with a larger volume of lower eluting strength
solvent can have the advantage of leaving strongly
retained contaminants on the sorbent as the analyte
of interest is desorbed. Solvent selection must also be
compatible with the analytical instrumentation used
for Rnal analysis.
The elution solvent screen may reveal that selective
desorption is possible for an analysis. Selective
desorption uses differences in the eluotropic
strength of the elution solvents to produce serial de-
sorption. Class separation or distinct fractionation of
analytes may be possible if the chemical properties of
the analytes differ enough that they respond dif-
ferently to weak and strong elution solvents.
Elution rate The rate of elution can affect the
SPE recovery of solutes from the sorbent. Particularly
for highly hydrophobic compounds, there can be
slow mass transfer from the stationary phase into the
mobile phase. The problem can be overcome by re-
ducing the Sow rate during elution, even to the point
of allowing elution to occur by gravity if necessary.
Sample concentration independence Finally, any
protocol developed must be independent of sample
concentration in the range of samples to be analysed.
Care must be taken not to exceed the maximum
loadability of the sorbent, but that is generally not
a problem since SPE is primarily used for trace
enrichment.
Applications
Early in SPE history, a series of simple yet ingenious
experiments were developed by Bidlingmeyer and
Warren that illustrate the principles of SPE. This
author has used these experiments as practical demon-
strations to introduce SPE method development to
4642 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION

Table 4 SPE isolation of food colours in grape drink

A B C D E F

Sorbent C18 C18 C18 Silica Silica Silica

Mechanism Van der Waals Van der Waals Van der Waals Polar}dipole/ Polar}dipole/ Polar}dipole/
isocratic selective ion pairing/ dipole isocratic dipole ion dipole
separation desorption silanol masking separation suppression selective
isocratic desorption
separation
Pre-wash 1) IPA (70%) 1) IPA (70%) 1) IPA (70%) Water Distilled white Water
2) Water 2) Water 2) Cetylpyridinium vinegar
chloride
Retention Drink mixa Drink mix Drink mix Drink mix Drink mix Drink mix

Post-wash Water Water Cetylpyridinium Water Distilled white Water

chloride vinegar
Elution IPA (18%) 1) IPA (5%) 1) IPA (18%) IPA (18%) IPA (16% in 1) Water
2) IPA (25%) 2) IPA (70%) vinegar) 2) IPA (15%)
3) IPA (70%)

IPA, isopropyl alcohol.

a
Mixture of FD&C Blue 1 and FD&C Red 40.

new users from elementary school students to adult
analysts. These experiments are outlined here
(Table 4) as practical applications of the foregoing
discussion. They are a useful method development
learning aid for novice users because:
1. they demonstrate the four-step process of SPE;
2. the dyes concentrated and separated in these
experiments can be observed by the naked
eye, clearly revealing extraction and recovery
processes;
3. the stepwise recovery of dyes from the sorbent
demonstrates the ability of SPE selectively to frac-
tionate samples.
The experiments, A}F in Table 4, demonstrate the
use of SPE to isolate food colours using inexpensive
reagents. The analytes in these experiments are the
dyes FD&C Blue 1 and FD&C Red 40 prepared in an
aqueous mixture by dissolving grape-Savoured drink
mix (Kool-Aid), in distilled water. Other reagents
necessary for these experiments include vinegar, rub-
bing alcohol and mouthwash. The original source
(Bidlingmeyer and Warren) should be consulted for
details.
The types of sorbents used for the experiments are
a hydrophobic, reversed-phase sorbent, C18 (experi-
ments A}C), and a polar sorbent, silica (experiments
D}F). The same two analyte dyes are retained in each
experiment, albeit via different mechanisms de-
pending on the sorbent: by van der Waals forces on
the C18 sorbent and polar dipole}dipole interactions
with the silica sorbent.
On the reversed-phase sorbent (C18), conditioning
involves exposure of the sorbent to a water-miscible
organic solvent followed by an aqueous wash (experi-
ments A}C). The preparation of the sorbent surface
to accept the analyte must be done in this order.
Measuring exact amounts of solvents is not necessary
during the pre-wash step. The sorbent is not allowed
to dry between column preparation steps and the
sample loading step. If it does dry out during column
preparation, before the sample has begun to be
loaded, the process should begin again. The silica
sorbent is activated by a single aqueous pre-wash of
water (experiments D and F) or distilled white
vinegar (experiment E).
As the sample is loaded on to the sorbent, the
organic dyes extracted are observed to become
concentrated at the leading edge of the sorbent. After
the sample is loaded, drying of the column is not
crucial, and in fact it is useful in some analyses to dry
the sorbent with vacuum before eluting the analytes.
FD&C Blue 1 (Brilliant Blue FCF, CASC3844-45-
9) and FD&C Red 40 (Allura Red AC, CASC 25956-
17-6) are large dye molecules (formula weights of
approximately 800 and 500, respectively) that have
negatively charged sulfonate groups in aqueous solu-
tion. Although ionogenic, the molecular size and
hydrophobicity of the dyes permit retention on the
C18 sorbent even without pH control (experiments
A and B).
Two of the experiments demonstrate the effect
of matrix additives on retention (experiments C and
E). In experiment E, distilled white vinegar is used for
ion suppression by reducing the pH and its addition
results in reversal of the elution order of the dyes. In
experiment C, an ion-pairing reagent, cetylpyridin-
ium chloride (Cepacol mouthwash), is added as
 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SOLID-PHASE EXTRACTION
a bulky, hydrophobic, positively charged counterion
to pair with the negatively charged sulfonate groups
of the dyes. In this case, cetylpyridinium chloride also
behaves as a silanol masking agent.
Post-washes Sush any remaining sample matrix
from the interstitial pores of the sorbent, and in each
experiment the post-wash solvent is the same as the
last pre-wash solvent. Additives in the drink mix that
are not retained (sugars, acids) will elute during
sample loading and during the post-wash.
Elution is accomplished by varying the concentra-
tion of isopropyl alcohol (commercially available
rubbing alcohol is approximately 70% isopropyl al-
cohol in water). The isopropyl alcohol is mixed with
either water (experiments A, B, C, D or F) or vinegar
(experiment E). Sample desorption volumes are mea-
sured for quantitative purposes. The volume re-
covered is always less than the volume added. The
desorption can be done in one isocratic separation
process (experiments A, D and E). Performed in
stages, experiments B, C and F are examples of selec-
tive desorption. SPE columns generate around
20}200 theoretical plates and this is sufRcient in
many cases to produce a fractionation of components
and chemical classes.
Future Developments
The history of SPE has already been marked by
continuous advances in sorbents and the formats in
which SPE sorbents are utilized. So, its a fairly safe
prediction that in the future there will be continued
development of new SPE sorbents and modes of deliv-
ery. Along that trend, recent developments of molecu-
larly imprinted sorbents for SPE and the 96-well plate
format for SPE are gaining attention and are expected
to be actively developed in the near future. Molecu-
larly imprinted sorbents for SPE are polymeric phases
formed with the analyte of interest as a print molecu-
le. The template thus formed exhibits selectivity for
the imprinted molecule. The 96-well microassay plate
collection format is designed to utilize robotics for
simultaneous extraction of 96 samples.
In addition to continued development of specialty-
phase sorbents and formats, advances in performing
more sophisticated chemistry associated with extrac-
tion are predicted. Applications of analyte derivatiz-
ation in conjunction with SPE are already reported,
with the chemical reaction occurring at the sorbent
surface. Methods apply either to solid-supported re-
actants or solid-supported analytes. Selective adsorp-
tion schemes that utilize tandem chromatographic
mode sequencing approaches will be used to solve
complex multiclass/multiresidue extractions. More
intricate mixed-mode adsorption mechanisms that
4643
mimic analyte}receptor lock-and-key approaches to
extraction are expected.
See also: I/Extraction. II/Extraction: Analytical Extrac-
tions; Solid-Phase Extraction; Solid-Phase Microextrac-
tion. III/Airborne Samples: Solid-Phase Extraction.
Bioanalytical Applications: Solid-Phase Extraction.
Drugs of Abuse: Solid-Phase Extraction. Environ-
mental Applications: Solid-Phase Microextraction.
Herbicides: Solid-Phase Extraction. Immobilised
Boronic Acids: Extraction. Immunoaffinity Extraction.
Insecticides: Solid-Phase Extraction. Molecular Im-
prints for Solid-Phase Extraction. Solid-Phase Extrac-
tion with Cartridges. Solid-Phase Extraction with
Discs. Solid-Phase Matrix Dispersion: Extraction.
Solid-Phase Microextraction: Biomedical Applications;
Environmental Applications; Food Technology Applica-
tions; Overview. Sorbent Selection for Solid-Phase
Extraction. Appendix 2: Essential Guides to Method
Development in Extraction.
Further Reading
Bidlingmeyer BA and Warren FV (1984) An inexpensive
experiment for the introduction of high performance
liquid chromatography. Journal of Chemistry Education
61: 716.
Hansch C and Leo A (1995) Exploring QSAR: [1]. Funda-
mentals and Applications in Chemistry and Biology.
Washington, DC: American Chemical Society.
Hansch C, Leo A and Hoekman DH (1995) Exploring
QSAR: [2] Hydrophobic, Electronic, and Steric Con-
stants. Washington, DC: American Chemical Society.
Lyman WJ, Reehl WF and Rosenblatt DH (1990) Hand-
book of Chemical Property Estimation Methods: Envir-
onmental Behavior of Organic Compounds. Washing-
ton, DC: American Chemical Society.
Nakamura M, Nakamura M and Yamada S (1996) Condi-
tions for solid-phase extraction of agricultural chemicals
in waters by using n-octanol}water partition coefR-
cients. Analyst 121: 469
Simpson NJK (ed.) (2000) Solid-Phase Extraction: Prin-
ciples, Strategies, and Applications. New York: Marcel
Dekker.
Simpson NKJ and Van Horne KC (1993) Handbook of
Sorbent Extraction Technology, 2nd edn. Palo Alto, CA:
Varian Associates.
Subden RE, Brown RG and Noble AC (1978) Determina-
tion of histamines in wines and musts by reversed-phase
high-performance liquid chromatography. Journal of
Chromatography 166: 310.
Thurman EM and Mills MS (1998) Solid-Phase Extraction:
Principles and Practice. New York: John Wiley.
Zief M, Crane LJ and Horvath J (1982) Preparation of
steroid samples by solid-phase extraction. American
Laboratory 14: 120.
Zief M, Crane LJ and Horvath J (1982) Preparation of
steroid samples by solid-phase extraction. International
Laboratory 12: 102.
4644 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY

ESSENTIAL GUIDES TO METHOD

DEVELOPMENT IN SUPERCRITICAL
FLUID CHROMATOGRAPHY
P. Schoenmakers, Shell Research and Technology
Centre (SRTCA), Amsterdam, The Netherlands and
University of Amsterdam, Amsterdam, The Netherlands
Copyright ^ 2000 Academic Press
Introduction
Supercritical-Suid chromatography (SFC) is deRned
as a mode of chromatography in which both the
temperature and the pressure in the column exceed
the critical values of the mobile phase. This deRnition
is exact, but rather arbitrary, as there is no phase
transition between gases (or liquids) and supercritical
Suids. Technically, a gas chromatograph operated
above 2.24 atm with He as the carrier gas, is an SFC
instrument according to this deRnition. We normally
speak of gas chromatography when retention is large-
ly controlled by the oven temperature (and largely
determined by analyte volatility). We speak of SFC
when retention is largely controlled by the mobile-
phase density (and largely determined by analyte in-
teraction with the mobile phase). Supercritical-Suid
chromatography (another name for it is dense-gas
chromatography) was Rrst developed in the 1960s
by Klesper in Aachen, shortly followed by Sie and
Rijnders in Amsterdam. The technique subsided into
oblivion during the rapid advent of modern high
pressure liquid chromatography (HPLC) in the
1970s. It experienced a second youth in the 1980s.
During this period, some researchers optimistically
claimed that SFC combined the advantages of gas
chromatography (GC) and HPLC. Although state-
ments of this kind still appear in the literature today,
the chromatographic community as a whole has come
to accept that SFC holds a position somewhere
in between, rather than above GC and LC. SFC
offers an } occasionally favourable } compro-
mise between the two mainstream chromatographic
techniques (see Table 1).
Why opt for SFC?
Although this article deals speciRcally with SFC, we
are treating it as a niche technique. In real life, GC
and HPLC are more commonly available. When GC
can readily be used, SFC offers few advantages
other than a lower operating temperature. When

Table 1 General considerations when considering SFC as a possible chromatographic separation methodH

Parameter GC SFC LC

(Most) suitable application
range
Operating temperature
Suitable columns
Suitable detectors
Gases and volatile materials
All but the most polar analytes
High (related to analyte boiling
point)
Packed columns (10}50
m
particle diameters)
Open-tubular columns
(100}500
m internal diameter)
Vacuum detectors (MS)HH
Gas-phase detectors (FID,
NPD, ECD, etc.)R
Low to marginally volatile
materials
Low to moderately polar
analytes
Low to moderate
Packed columns (3}10
m
particle diameters)
Open-tubular columns
(10}50
m internal diameter)
Vacuum detectors (MS)HHH
Gas-phase detectors (FID,
NPD, ECD, etc.)R
Liquid-phase detectors (UV,
fluorescence)
Low-volatile and non-volatile
materials
All polarities (non-polar to ionic)
Low
Packed columns (1}5
m
particle diameters)HH
Open-tubular columns (1}5
m
internal diameter)
Vacuum detectors (MS)HHH
Liquid-phase detectors (UV,
fluorescence, refractive index,
etc.)

HThe most suitable technique is given in italics.

HHMonolithic columns are an emerging alternative to packed columns.
HHHMS"mass spectrometry.
RFID"flame-ionization detector; NPD"nitrogen}phosphorus or thermionic detector; ECD"electron-capture detector.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY 4645

Table 2 Possible mobile phases for SFC and their compatibility with different detection principles

Mobile phase Polarity Tc (3C) pc (atm) Detection compatibility

FID UV MS IR

Carbon dioxideH Low 31.05 72.9 ## ## # $

with modifier:
Methanol High 239.4 79.9 ! ## # !
Formic acid High # # ! !
Nitrous oxide Low 36.4 71.5 $ # # $
Sulfur hexafluoride Low 45.5 37.1 $HH # $ $
n-Butane Low 152.0 37.5 ! ## $ !
Xenon Very low 16.6 57.6 # ## # ##
AmmoniaHHH High 132.4 111.3 # # # !
WaterHHH Very high 374.1 217.6 # # # !

HMost suitable mobile phase for most applications.

HHFeasible, but highly corrosive.
HHHHighly corrosive and hardly feasible.

HPLC may readily be used, SFC } when applicable
} may offer shorter analysis times and a greater
choice of detectors. HPLC can be applied to a much
greater variety of samples and analytes than SFC.
Table 1 lists some general considerations for con-
sidering or discarding SFC as a possible (analytical)
separation technique. The most important reasons for
selecting SFC are described below in more detail.
Universal detection When using carbon dioxide
(CO2) as the mobile phase, SFC allows the use of
Same-based detectors (see Table 2). The Same-ioniz-
ation detector can be applied almost universally. Even
more importantly, it shows an approximately equal
response within a class of analytes. As a result, refer-
ence standards within each class (rather than for each
individual compound) sufRce for calibrating
a quantitative method.
Because universal detectors are available in GC,
but not in LC, there are potentially, two directions in
which relevant SFC}FID methods can be developed:
E analysis of non-volatile materials, that cannot be
analysed by GC (including the high-temperature
version, HT-GC); and
E achieving separations with a (type of) selectivity
that cannot be achieved in GC.
Applications in the former direction are quite rare.
Some thermally labile components, such as explosives
and peroxides, have been analysed by SFC. However,
due to the highly inert nature of GC mobile phases
(e.g. helium), the increased inertness of modern GC
columns, and the increased Sexibility of injection
systems (e.g. cold on-column injection), such com-
ponents can often be analysed with good integrity
by GC.
Some components that are not sufRciently vol-
atile for analysis by HT-GC may be amenable to
analysis by SFC. However, in the authors experience
this is limited to highly apolar materials, such as
saturated hydrocarbons. For moderately polar
analytes, such as aromatic hydrocarbons, HT-GC ap-
pears to allow materials with higher boiling points
(lower volatility) to be eluted in comparison with
SFC.
The most successful SFC}FID methods follow the
second approach, using a unique kind of selectivity.
In GC, retention is determined by two factors, viz. the
pure-analyte vapour pressure and the interactions of
the analyte with the stationary phase. In SFC the
effect of the vapour pressure can be minimized
by working with high-density mobile phases, while
the interaction with the stationary phase can be maxi-
mized by using stationary phases with large, active
surface areas. This allows a so-called group-type
selectivity to be achieved, in which the sample is
separated (or classiRed) into a limited number of
distinct groups (or classes) of analytes. Within a class,
the size (and thus volatility) of the analyte molecules
varies, but the chemical structure (functional groups)
remains similar. Examples of such methods include
the following.
E Separation of complex hydrocarbon mixtures, for
example the separation of middle-distillate fuels
(diesel or kerosene) into saturates, mono-aromatics
and di-aromatics; the determination of the total
amount of oleRns in gasoline-type fuels. Both these
examples concern highly successful applications
of SFC. A group-type separation method for
middle distillates is standardized as ASTM D-
5186. An ASTM standard method for oleRns in
4646 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY
gasoline by SFC is expected to be approved by June
2000.
E Separation of (low-molecular-mass) polymers into
fractions representing different end-groups.
In both cases, we try to achieve retention that is
affected by the chemical structure of the molecules
(the functionality) but irrespective of their size
(molecular mass). This type of chromatography is
} somewhat confusingly } referred to as critical
chromatography, or as supercritical-Suid chro-
matography at the critical conditions.
DifVcult separations SFC possesses some fa-
vourable fundamental characteristics, especially in
comparison with liquid chromatography. The mo-
lecular diffusion is about an order of magnitude
greater than in liquids (but three orders worse than in
gases) and the viscosity is about a factor hundred
lower than that of a typical liquid. Thus, SFC has
advantages in terms of mass transfer and column
pressure drop. This may result in higher efRcien-
cies per unit length of column, while longer columns
may sometimes be used. Therefore, SFC may be at-
tractive for some difRcult separations.
SFC has proved a rather attractive alternative to
normal-phase LC for the separation of stereoisomers.
Like in normal phase LC, CO2-based SFC features
a polar stationary phase and a non-polar mobile
phase. Organic modiRers may be added to modify the
mobile-phase polarity and detergent-like molecules
have been added to help create adequate selectivities.
The advantages of SFC in this context are sum-
marized in Table 3.
SFC is seen to score well in all categories, except its
Sexibility in dealing with a variety of samples. Re-
verse-phase LC (RPLC) also scores well in the table.
SFC appears to be more attractive as an alternative to
normal-phase LC (NPLC) than to RPLC. The latter
technique is compatible with almost all sample sol-
Table 3 General advantages of (packed-column) SFC in comparison with reversed-phase (RPLC) and normal-phase (NPLC) liquid
chromatographyH
Property Related to:
Efficiency per unit time Mass transfer (Dm, )
Maximum efficiency Eluent viscosity ()
Sample capacity Surface homogeneity
Eluent strength
Equilibration time Surface activity
Mass transfer
Flexibility (range of samples) Mobile phase
Surface activity
HMost important effects are in italics.
HHVery high plate numbers have been reached in SFC but operating conditions close to the critical point must be avoided.
vents, ranging from water to quite non-polar organic
solvents, such as tetrahydrofuran. NPLC on unmodi-
Red silica surfaces can be used in combination with
solvents of low-to-medium polarity. When using
polar-bonded phases, again a great variety of sample
solvents may be introduced on the column. In both
cases (RPLC and NPLC), strongly acidic and strongly
basic samples cause problems. SFC is typically
restricted to solvents and analytes of low to moderate
polarity, especially in case FID detection is to be
used.
Preparative separations Carbon dioxide is an out-
standing solvent for preparative chromatography. It
is available in high purities at a relatively low cost and
it can easily be removed from the efSuent by
evaporation. In fact, the latter characteristic implies
that it is somewhat more difRcult to collect frac-
tions than is the case in preparative LC.
The main disadvantage of CO2 for preparative
chromatography is its low polarity, which seriously
limits its applicability as a chromatographic eluent.
Packed columns, with large surface areas and thus
high sample capacities, are desirable for preparative
separations. Without organic modiRers, only compo-
nents of little or no polarity can be eluted from such
columns using CO2. When substantial amounts of
modiRers need to be used, the advantages of using
CO2 diminish.
Hyphenated systems
SFC}MS Although it would seem that the use of
CO2 is also advantageous when coupling a dense-
phase chromatograph to a mass spectrometer (MS),
successful SFC}MS systems have hardly materialized.
In what are now the most common LC}MS interfaces
(electrospray, ESI; and atmospheric-pressure chem-
ical ionization, APCI), a high mobile-phase polarity
is preferable. Only a small niche remains where
RPLC SFC NPLC
Second First Last
Second FirstHH Second
First Second Last
First Second Last
First Last Second
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY
SFC}MS may compete with LC}MS, i.e. components
of low volatility and low polarity. This implies that
there is little incentive for the further development of
SFC}MS.
SFC}NMR CO2 is a perfect eluent when 1H-NMR
is to be coupled with a chromatographic separation
device. LC}NMR has received a good deal of atten-
tion in recent years and some workers have extended
this work to include SFC}NMR. The main instru-
mental difference is that a high-pressure Sow-cell
(or probe) is required. However, the inherent sensi-
tivity of NMR is so low that fractionation followed
by ofSine spectroscopy is usually the preferred
approach.
SFE}SFC Very elegant hyphenated systems may
arise from a combination of two separation tech-
niques that both involve supercritical Suids. Such
systems include SFC}SFC and SFE}SFC. The latter
approach, where the extraction serves as an online
sample-preparation technique, has been especially in-
vestigated by several groups. Unfortunately, the high
expectations surrounding SFE around 1990 have not
quite materialized. Current interest in SFE}SFC has
waned.
Types of SFC Columns
There are traditionally two approaches to SFC. One
involves packed columns, the other open-tubular
(capillary) columns. This situation is not differ-
ent from that experienced in GC and LC. In the
former technique, open columns are strongly prefer-
red. In the latter, open columns are ideal in theory,
but virtually impossible to use in practice. The opti-
mum internal diameter of open columns used in
chromatography is essentially determined by the dif-
fusion coefRcients of the analytes in the mobile
phase. As a rule of thumb, the required analysis time
is given by
Nreqhd 2
tR" (1#k)
Dm
where tR is the required analysis time for a separation
with Nreq theoretical plates and a solute retention
factor of k, h is the reduced plate height, d the column
diameter,  the reduced (average) velocity and Dm
is the diffusion coefRcient of the analyte(s) in
the mobile phase. Both Nreq and k are essentially
determined by the retention of the analytes. The
greater the selectivity (differences in retention),
the lower the required number of plates. Neither
4647
Nreq nor k are affected by the dimensions (length
and diameter) of the column. The reduced plate
height (h) and the reduced (average) linear velocity ()
have typical values for packed and open-tubular col-
umns. Typically, for packed columns h"3 and
"10, so that h/"0.3. For open-tubular columns
h"4.5 and "45 are good values, so that
h/"0.1. All things being equal, open-tubular col-
umns are expected to be about three times faster than
packed columns.
Dm is the parameter that suggests SFC may allow
faster separations than HPLC. However, the dif-
fusion coefRcient must be balanced against the
characteristic dimension (d) of the column. For
packed columns, d is the particle diameter (dp), while
for capillary columns it is the internal diameter of the
column (dc). It follows from the equation that similar
performance (in terms of analysis times) can be
expected in different forms of open-tubular
chromatography when the ratio d 2c/Dm is kept con-
stant. With Dm,gas+1000;Dm,SF+104;Dm,liquid, we
Rnd for the optimum diameters of open tubular col-
umns dc,GC+30;dc,SFC+100;dc,LC. Since GC col-
umns have internal diameters between 100 and
500
m, we anticipate optimal internal diameters for
SFC columns to be of the order of 10
m and for LC
columns to be 1}5
m. Because very many practical
problems are associated with the use of such extreme-
ly small columns, open-tubular SFC (OT-SFC) has
typically been performed with somewhat larger col-
umns (50 or 100
m). However, this has led to a
modest efRciency and speed.
Table 4 provides a summary of some of the advant-
ages and disadvantages of using packed and capillary
columns in GC and LC, with a more extensive sum-
mary of the characteristics of packed and open-tubu-
lar SFC. In SFC, open-tubular columns with optimal
diameters are difRcult to use. As a result, packed-
column SFC is the more robust and more practical
technique.
Most Important Parameters
The parameters that are most important in the devel-
opment of SFC methods are as follows.
Mobile-Phase Density
The outstanding parameter in SFC is the mobile-
phase density. This factor plays a role similar to the
temperature in GC and the solvent strength in LC.
Density gradients (typically increasing density lin-
early with time) in SFC are the common equivalent of
temperature gradients in GC and mobile-phase com-
position gradients in LC. When the density increases,
4648 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY
Table 4 Advantages (O) and disadvantages (P) of packed and open-tubular (capillary) columns in GC, SFC and LC
GC
Packed columns O Fast analysis
O Large sample capacity
O Broad dynamic range
O Reliable and robust
O Allows preparative separations
0 Perceived to be old-fashioned
P Low permeability
P Limited maximum efficiency
Open-tubular O High permeability
columns
O High maximum efficiency
O Inert surface for polar analytes
0 Many different injectors
P Limited sample capacity
P Limited dynamic range
P Sensitive to
(large volumes of) solvents
P Sensitive to (liquid) water
HIn practice, the efficiencies obtained in open-tubular SFC are well below the theoretically expected values.
interactions between the mobile phase and the
analytes increase. The analytes are better dissolved in
the mobile phase. Retention typically decreases expo-
nentially with increasing density.
The column inlet and outlet pressures are signiR-
cant parameters, but their effect on retention is
indirect, as the pressure affects the density. In
this context, the pressure closest to the critical value is
most important. In SFC this is the column-outlet
pressure.
SFC LC
O Compatible with back-pressure O Compatible with many
regulators detectors
O Broad range of optimum k values O Fast analysis
O Programming often not needed O Large sample capacity
O Fast analysis O Broad dynamic range
O Large sample capacity O Reliable and robust
O Broad dynamic range O Allows preparative separations
O Reliable and robust 0 Columns still not perfect
O Easy online solvent mixing P Very small particles required
O Routine loop injections P Low permeability
O Allows preparative separations P Limited maximum efficiency
0 Columns optimized for LC P High pressure drop
P Low permeability
P Limited maximum efficiency
P High pressure drop
P Active stationary-phase surface
P Modifiers often required
P FID often not possible
O Inert surface for polar analytes P High permeability
O Modifiers not often needed O High maximum efficiency
O FID can usually be used 0 Extremely low flow rates
O High permeability P Very few detection options
O Low column pressure drop P Extremely small diameters
O High theoretical efficiencyH required
0 Very low mobile-phase flow P Extremely small sample
rates volumes
P Very small diameters required P High detection limits
P Sub-optimal (too large) columns P Extremely small dynamic
commonly used range
P Little tolerance for extra-column P No tolerance for extra-column
dispersion dispersion
P Very small sample volumes
P Proper injections are difficult (very
small volumes and time splitting)
P Rather high detection limits
P Very small dynamic range
P Hard to combine with MS
P Narrow range of optimum k values
P Sensitive to (large volumes of
solvents
P Programming usually required
P Requires fixed restrictor (no
adequate flow control)
Pressure, temperature and density are connected
through an equation of state. Different equations
can be used that provide good estimates for the den-
sity of pure supercritical Suids. However, when
mixed eluents are used (e.g. CO2 with a modiRer such
as methanol), no reliable equation is available that
provides the density as a function of pressure and
temperature. Nevertheless, when the latter two para-
meters and the composition of the eluent are estab-
lished, the density is in principle deRned.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY 4649

Temperature Rers give rise to substantially different selectiv-

The second most important parameter is the temper-
ature. Like the pressure, the temperature has a signiR-
cant effect through its effect on the density.
However, at constant density, an increased temper-
ature may lead to a lower retention, especially for
relatively volatile analytes. Apart from the temper-
ature of the column oven, the temperature of the
injector or injection valve can also be quite signiR-
cant. In order to test the feasibility of eluting certain
analytes by SFC, it is worthwhile to perform some
experiments at an increased injector temperature.
Stationary Phase
The stationary phase. The column used plays a major
role in SFC, especially when using non-polar mobile
phases, such as carbon dioxide. The stationary phase
has a large effect on the retention and an often
prevailing effect on the selectivity. In addition,
the stationary phase has a very large effect on
peak shape and peak width (efRciency). Again,
this effect is strongest when using non-polar elu-
ents (CO2).
Studying the effects of different stationary
phases can be (very) expensive and time consuming.
There are often practical limitations and when the
option to use modiRers is available, this may be at-
tempted Rrst, provided UV detection is adequate.
ities.
Method Development
The Sow chart for developing an SFC method shown
as Figure 1 follows logically from the discussion on
different types of columns and the overview of
main parameters given in previous sections above.
Carbon dioxide will almost always be the eluent of
choice. This is assumed to be the case in Figures 1 and
2. Instrument availability is the obvious Rrst consid-
eration. It greatly affects all the other decisions
taken. Selecting packed columns is very attractive (see
Table 4), but this requires a compatible instrument,
with a pumping system capable of delivering substan-
tial Sow rates. It is quite possible to use microbore
(1-mm i.d.) and packed-capillary (40.5 mm i.d.) in
SFC, but some of the advantages of packed-column
SFC are then lost. Most importantly, miniaturized
columns do not allow the use of controllable back-
pressure regulators. In this case, there is no adequate
Sow control, a problem that is especially serious
when the mobile-phase density (in practice the pres-
sure and/or the temperature) is programmed during
the run.
When a novel sample is being subjected to SFC,
the recommended strategy is to rapidly establish

Mobile-Phase Composition
The mobile-phase composition may have dramatic
effects on retention, selectivity, efRciency and
peak shape in SFC. However, adding modiRers has
some signiRcant disadvantages, especially with regard
to detector compatibility. Therefore, changing the
mobile-phase composition is not the Rrst option in
developing an SFC method. There are two modiRers
that allow FID detection to be used, i.e. water and
formic acid. Both have been investigated, but neither
has found many applications in practice. The ef-
fect of a modiRer tends to be greatest at low concen-
trations. In this range, the modiRer mainly acts by
competing with the analytes for strong adsorption
sites on the stationary-phase surface. A small amount
of modiRer (often well below 1%) may lead to
a much reduced analysis time and a much increased
column efRciency. The use of modiRers often
leads to much sharper and much more symmetrical
peaks. At higher concentrations, modiRers may still
affect retention and selectivity, through increas-
ing the polarity and the density of the mobile phase.
However, these effects are much smaller and the
variations become more gradual. At these high con-
centrations it is more likely that different modi- Figure 1 Flow chart for the development of an SFC method.
4650 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY
whether the analytes can be eluted. Because retention
decreases with increasing density, high densities must
be tried Rrst. Once sharp peaks have been obtained
for the analytes it will be easy to increase the reten-
tion by lowering the density. In open-tubular SFC
a mobile-phase density gradient with a high Rnal
density will typically be used. In packed-column SFC,
where retention times are typically of the order of
minutes, constant elution conditions (isobaric, iso-
thermal, isochoric and isocratic) will be preferred for
initial scanning experiments.
Despite the high praise for FID, it is extremely
valuable to have an informative detector available at
this stage. A UV detector, especially a multichannel
diode-array (DAD) instrument, will provide much
on-line information on the progress of the method
development. It is very much easier to know the
whereabouts of different analyte peaks in the
chromatogram if DAD and FID data are obtained
simultaneously. In open-tubular SFC, a DAD cannot
be used and SFC}MS is the obvious choice. However,
SFC}MS is not an easily accessible practical tool.
If the analytes cannot be eluted at the highest pos-
sible (or highest practical) CO2 density, it may yet be
worthwhile to attempt elution at elevated temper-
atures. Increasing the temperature may lead to lower
densities, but this may be compensated by an
increased analyte volatility, especially for analytes
with a signiRcant vapour pressure. In addition,
adsorption effects (including analyte-stationary-
phase interactions) may be reduced. The temperature
of the injector plays a different role. Sometimes
it has proven beneRcial to inject at temperatures
well above the column temperature. In many cases,
the sample (or sample solvent) is a limiting factor.
When loop injection is used, the temperature must
usually be kept well below the boiling point of the
solvent.
If the analytes are not eluted as sharp, symmetrical
peaks at high densities, nor at increased temperatures,
then the use of modiRers may be attempted if this is
an option. Using pre-mixed CO2-based mobile phases
is not attractive for reasons of accuracy and reproduc-
ibility as the composition in the cylinder will vary
with time. Also, the Sexibility regarding the possible
concentrations is very limited. However, this is often
the only choice in miniaturized (open-tubular) sys-
tems. In some cases, mixtures have been prepared
inside the pump head of a syringe pump, which is
preferred in terms of accuracy and Sexibility. Many
packed-column SFC systems allow more convenient
online mixing, which makes it relatively easy to inves-
tigate the possible advantages of using modiRers. Un-
less experiments are performed with water or formic
acid as a modiRer (neither being very practical), the
use of FID is not feasible at this point. It may be useful
to attempt a few different modiRers. However,
the chances of obtaining good peaks become very
small once the addition of 10% methanol has proven
inadequate for the purpose.
The scanning experiments suggested so far can
typically be performed within one or two days. This
is what is referred to when it is claimed that method
development in SFC can be very rapid. If at this
stage the results are unsatisfactory, an important
decision needs to be made. It is quite possible that
better results will be obtained on a different
column. However, when it is decided to investigate
the use of different columns the amount of work
needed will be multiplied. If there are still good
reasons to opt for SFC, then it is most realistic
to identify the column with the most inert surface
(for example, a column packed with polysiloxane-
coated particles) and repeat the sequence outlined
above. If this attempt is not successful, then an alter-
native separation technique must be seriously con-
sidered.
Method Improvement and
Troubleshooting
If at any stage during the method development rap-
idly eluting, sharp peaks have been obtained for the
analytes, then the separation can be optimized. The
actions that may be taken are summarized in
Figure 2.
From the initial results it may be concluded
whether the retention should be increased. The ap-
propriate action depends on the stage at which suc-
cess was obtained. If high-density CO2 at a low tem-
perature proved successful, then reducing the density
will sufRce. When elevated temperatures were
Figure 2 Flow chart for the optimization of an SFC method. Rs
denotes resolution (i.e. ratio of distance between two peaks and
their average base width).
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN SUPERCRITICAL FLUID CHROMATOGRAPHY 4651

used, the temperature may be lowered and/or the When this is the case, proper functioning of the equip-
pressure may be decreased to achieve optimal elution ment should Rrst be veriRed. The mobile-phase
conditions. If a modiRer was used then the concentra- density (pressure and temperature), Sow rate and
tion of modiRer may be lowered or the mobile-phase composition may all be veriRed. If variation in either
density may be decreased. of these parameters is excluded, then a change in
When moving the retention of the analytes into the the stationary-phase surface is a probable diagnosis.
optimum range, it will become apparent whether or The column may be simply replaced at this stage,
not programmed elution is needed. In packed-column but a few other options are open. These are listed
SFC this will only be the case if the last analyte has below.
a high retention factor (say k'15) when the Rrst
E It is possible that the column is contaminated with
eluting analyte has k"1. In marginal cases, it may be
very heavy (high molecular weight) or very polar
possible to use a different (less selective) station-
material from the sample or the solvent, that can-
ary phase to avoid programmed analysis. In open-
not be eluted under SFC conditions. In this case it
tubular SFC, programmed analysis is often needed
may be possible to wash the column with a liquid
just to deal with the excess of solvent introduced with
solvent such as 2-propanol, to recondition it in the
the analytes. Resolution in programmed analysis can
SFC instrument (without the FID connected), and
typically be increased by lowering the eluent strength
to use it again for the application.
(in SFC typically the density) at the start of the pro-
E It is possible that the column is irreversibly altered
gram and by lowering the slope (increasing the dura-
by the presence of sample or solvent components.
tion of the gradient segment of the program). In either
Water on a silica column is the most obvious
case, this leads to a longer analysis time. In open-
example. Water may be removed by drying a col-
tubular SFC (or when using a Rxed restrictor in
umn overnight at a high temperature (e.g. 2503C)
packed-column SFC) the Sow rate may be relatively
under a small Sow of an inert gas (N2, H2 or He).
high, especially in later parts of the program. In this
A GC oven is very useful for this purpose.
case lowering the Sow rate (by preparing a smaller
E In case non-programmed conditions are used, it
restrictor) may be more rewarding, either by itself, or
may be advantageous to program the column to
in combination with lowering the gradient slope.
different conditions at the end of each analy-
If a separation under non-programmed conditions
sis, each series of samples, or each working day to
leads to abundant resolution between the relevant
avoid column contamination.
analytes, it may be possible to decrease the retention
E Some columns may change gradually in a truly
time. In order of decreasing rewards, this may be
irreversible manner. The use of amino-derivatized
achieved by decreasing the column length, increasing
columns is not recommended in combination with
the Sow rate, or increasing the eluent strength (in-
CO2, due to the anticipated formation of carba-
creasing the density or the modiRer concentration). In
mates. If such a column is to be used, a gradual
the more important case in which the achieved resolu-
change of the stationary phase may necessitate
tion is inadequate, the pressure and/or temperature
gradual adaptation of the mobile-phase density or
may be altered, but this often affects retention
composition to maintain adequate resolution. Less
much more than selectivity (i.e. the retention factors
dramatic changes of the surface may occur with
of the various analytes tend to be affected in the
different stationary phases (e.g. a gradual loss
same way). If modiRers are being used, different
of some chemically bonded ligands from the sur-
modiRers may lead to different selectivities.
face) and these may also be counteracted by small
However, the most likely road to success is to attempt
changes in the conditions, rather than by fre-
different stationary phases at this stage.
quently replacing the column.
When we considered the use of different sta-
tionary phases at the end of the method-development
stage, this was thought not to be very promising. See also: II/Chromatography: Supercritical Fluid:
However, in the present situation, at the method- Historical Development; Instrumentation; Large-Scale
optimization stage, it has already been demonstrated Supercritical Fluid Chromatography; Theory of Supercriti-
that SFC is a feasible technique for eluting the ana- cal Fluid Chromatography.
lytes, but not yet for separating them. Trying dif-
ferent stationary phases with greatly different
selectivities may be a rewarding option at this stage. Further Reading
The actions outlined here may also be relevant Anton K and Berger C (eds) (1998) Supercritical-Fluid
when the separation deteriorates at some stage during Chromatography in Packed Columns: Techniques and
the development or application of an SFC method. Applications. New York: Marcel Dekker.
4652 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Berger TA (1995) Packed-Column SFC, pp. 102}136.
London: Royal Society of Chemistry.
Berger TA (1997) Separation of polar solutes by packed
column supercritical-Suid chromatography. Journal of
Chromatography A 785: 3}33.
Jinno K (1992) Hyphenated Techniques in Supercritical-
Fluid Chromatography and Extraction. Amsterdam:
Elsevier.
Markides KE, Lee ML and Later DW (1989) Capillary
supercritical-Suid chromatography: practical aspects.
In: Yang FJ (ed.) Microbore Column Chromatography:
A UniTed Approach to Chromatography, pp. 239}266.
New York: Marcel Dekker.
Mulcahey LJ, Rankin CL and McNally MP (1994) Envir-
onmental applications of supercritical-Suid chromato-
graphy. Advances in Chromatography 34: 251}308.
Petersson P and Markides KE (1994) Chiral separations
performed by supercritical-Suid chromatography.
Journal of Chromatography A 666: 381}394.
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN THIN-LAYER (PLANAR)
CHROMATOGRAPHY
S. Nyiredy, Research Institute for Medicinal Plants,
Budakala& sz, Hungary
Copyright ^ 2000 Academic Press
Introduction
One of the most critical steps of qualitative and
quantitative planar (thin-layer) chromatographic
(TLC) analysis is development of a method result-
ing in sufRcient separation. The main steps of
method development are summarized in Figure 1.
The Rrst stage is selection of the stationary phase, the
vapour phase, and suitable solvents. This stage is the
sine qua non of method development, and the selec-
tion of these can occasionally immediately result in
a suitable separation. For most real separation prob-
lems the second stage, optimization of the mobile
phase is also necessary. The third part of method
development is selection of the Rnal conditions, for
example the mode of development, transfer of the
mobile phase to an appropriate forced-Sow method,
and last but not least, the selection of suitable oper-
ating parameters. This paper gives essential guides to
method development in planar chromatography and
draws attention to the most important considerations.
Schoenmakers PJ (1988) Supercritical-Suid chromatogra-
phy: open columns vs. packed columns. In: Smith
RM (ed.) Supercritical-Fluid Chromatography, pp.
102}136. London: Royal Society of Chemistry.
Schoenmakers PJ and Uunk LGM (1989) Mobile and sta-
tionary phases for supercritical-Suid chromatography.
Advances in Chromatography 30: 1}80.
Smith RM (ed.) (1988) Supercritical-Uuid Chromatogra-
phy. London: Royal Society of Chemistry.
Smith RM and Hawthorne SB (eds) (1997) Supercritical
Fluids in Chromatography and Extraction. Oxford:
Elsevier.
White CM (ed.) (1988) Modern Supercritical-Fluid
Chromatography. Heidelberg: HuK thig.
Wilson ID and Davis RJ (1993) Supercritical-Suid
chromatography and extraction of pharmaceuticals. In:
Dean J (ed.) Application of Supercritical Fluids in Indus-
trial Analysis, pp. 74}103. Glasgow: Blackie.
Stationary Phase Selection
TLC separations can be performed on modiRed,
unmodiRed, and impregnated stationary phases, be-
cause of differences between the chemical proper-
ties of the sorbent material and those of compounds
present in the sample to be separated. Different
types of chromatographic process (normal-phase,
reversed-phase, partition, and ion exchange
chromatography) can be distinguished on the basis
of the types of interactions involved. Although more
than 90% of TLC separations are performed on sil-
ica, chemically bonded phases have recently become
increasingly popular for solving special separation
problems.
In normal-phase chromatography the hydroxyl
groups on the surface of the silica are the polar, active
centres which result in the interactions leading to
the retention of the compounds to be separated.
These interactions are mainly hydrogen-bonding and
induced dipole}dipole interactions. The stationary
phase can generally be characterized in terms of its
speciRc surface area, speciRc pore volume, and mean
pore diameter.
UnmodiRed stationary phases include silicas,
aluminas, kieselguhr, silicates, controlled-porosity
glass, cellulose, starch, gypsum, polyamides, and
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY 4653

chitin. For TLC separations silica is manufactured

by spontaneous polymerization and dehydration of
aqueous silicic acid, which is prepared by adding acid
to a solution of sodium silicate. The product of this
process is an amorphous, porous solid, the speciRc
surface area of which can vary over a wide range (200
to more than 100 m2 g\, as can the average pore
diameter (10}1500 A> ).
ModiRed silicas can be nonpolar or polar adsor-
bents. The former include silicas bearing alkane or
alkene chains or phenyl groups, whereas the polar
modiRed silicas contain cyano, diol, amino, or thiol
groups or substance-speciRc complexing ligands. The
structures of some chemically modiRed silicas are
shown in Figure 2.
Almost all the stationary phases used in normal-
and reversed-phase column liquid chromatography
are also available for TLC. The dimensions of com-
mercially available analytical thin-layer plates are
10;10, 10;20 or 20;20 cm; the layer thickness is
20 or 25
m. It is generally accepted that better
resolution is obtained on thinner layers (10
m), de-
pending on the mode of detection. The silica mater-
ials commonly used for precoated plates have an
average particle size of ca. 11
m, the size range is
from 3 to 18
m; for analytical layers prepared in the
users laboratory the average particle size is 15
m
and the range of particle sizes is much greater. The
average particle size of precoated high-performance
TLC (HPTLC) plates is now 5}6
m and the range of
Figure 1 Schematic diagram of method development.
particle sizes is very small.

Figure 2 The structures of some commercially available surface-modified silicas.

4654 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Figure 3 Flow chart illustrating a systematic approach for the selection of the appropriate separation technique and stationary phase.
Precoated analytical layers with a preadsorbent
zone are also commercially available for linear devel-
opment. This zone serves to hold the sample until
development begins. Compounds soluble in the sol-
vent system pass through the preadsorbent zone and
are concentrated in a narrow band on entering the
chromatographic layer; this improves resolution.
Figure 3 gives a decision Sow chart for the systematic
selection of the appropriate separation technique and
stationary phase.
Vapour Phase Selection
In planar chromatography the separation process oc-
curs in a three-phase system of stationary, mobile,
and vapour phases, all of which interact both with
each other and with the operating conditions.
Selection of chamber type and vapour space is a vari-
able offered only by planar chromatography as the
third dimension of the chromatographic parameters.
The role of the vapour phase in TLC is well known,
although little attention is given to this in practice.
In planar chromatography two basic types of
chromatographic chamber must be distinguished.
In the common normal (N) chamber the distance
between the layer and the wall of the chromato-
graphic tank is more than 3 mm. If this distance is
smaller, the chamber is said to have the S conRgura-
tion. Both types of chamber can be used for un-
saturated or saturated systems. As a rule of thumb, if
the sample contains fewer than seven compounds to
be quantitatively determined, saturated N chambers
must be selected for method development. If the
sample contains more than seven substances, or the
separation is very difRcult, S chambers must be
selected which enable transfer of the optimized
mobile phase by forced-Sow.
Often the separation problem cannot be solved by
use of conventional TLC with solvent migration by
capillary action, because of the relatively modest sep-
arating power of the method. In such circumstances
use of one of the different forced-Sow techniques
is necessary; this must be considered during selection
of the vapour phase. The chambers used for forced-
Sow planar separations can be also assigned to the
above two categories. The chambers used for over-
pressured layer chromatography (OPLC) are un-
saturated S chambers, theoretically and practically
devoid of any vapour space. This must be considered
in the selection of appropriate solvents and during the
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY 4655

Figure 4 The saturation grade of different forced-flow methods, in comparison with the N and S chambers.

optimization of the solvent system. In rotation planar
chromatography (RPC) the size, and thus the extent
of saturation, of the vapour phase can be varied. In
RPC the micro and ultramicro chambers belong to
the S-chamber type. Because in microchamber RPC
the plate rotates with the small chromatographic
chamber, and the distance between the layer and the
lid of the chamber is less than 2 mm, the vapour space
is rapidly saturated. In ultramicrochamber RPC the
lid of the chamber is placed directly on the plate and
so in practice there is no vapour space, as in OPLC.
When a mobile phase is transferred from a chro-
matographic tank separation to a forced-Sow tech-
nique, the vapour phase can be characterized on the
basis of the saturation grade (SG). The SG value of
a given chromatographic chamber can be calculated
by dividing the sum of the hRF values of the three
furthest-migrating substances by the sum of the hRF
values of all the components, subtracting the result
from 1, and multiplying the answer by 100. The
saturation grade can be used as a measure of the
reproducibility of separations with given stationary
and mobile phases and at different temperatures
and humidity; this enables transfer of the mobile
phase to other vapour-phase conditions. Figure 4
shows the saturation grade of the different
chromatographic chambers. The lines indicate sug-
gested mobile phase transfer possibilities; the dotted
line indicates other mobile phases which might be
used, but with less probability of success.

Figure 5 Flow chart illustrating a systematic approach for the selection of the appropriate chromatographic chamber and vapour
phase.
4656 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY

Among the forced-Sow methods the highest separ-
ating power is obtained with OPLC, because of the
optimum mobile phase velocity on the HPTLC plate
and the greater separation distance. If, therefore, the
quality of the Rnal separation is likely to be deter-
mined by the separation distance, OPLC and, for the
preassay, the unsaturated S chamber must be selected.
If RPC equipment is available for improving the ef-
Rciency of the Rnal separation, the choice of
chromatographic tank for the preassay depends on
the types of compound to be separated. If the acidic
or basic character of the vapour phase is important
for the separation, a saturated S-chamber (micro-
chamber) should be used; if this is not available,
a saturated N chamber is the right selection for the
TLC pre-assay. If the mobile phase is to be transferred
to a U-RPC separation, an unsaturated S chamber
(ultramicro chamber) must be chosen. These consid-
erations are summarized in Figure 5.
Selection of Suitable Solvents
The modern strategy of solvent selection is based
on the solvent classiRcation by Snyder, who classiRed
more than 80 solvents into eight groups for normal-
phase chromatography according to their properties
as proton acceptors (xa) and proton donors (xd), and
their dipole}dipole interactions (xn).
For, selection of suitable solvents, preliminary experi-
ments are performed on silica TLC plates with the nine
solvents indicated by stars in Table 1, which lists the
solvents commonly used in planar chromatography.

Table 1 Solvent classification based on solvent strength and selectivity values

Group Solvent Solvent strength (Si ) Xe Xd Xe

Sv"
Xd

} n-Hexane 0 } } 0.10H
I n-Butyl ether 2.1 0.44 0.18 2.44
Diisopropyl ether 2.4 0.48 0.14 3.43
Methyl-t-butyl ether 2.7 0.49 0.14 3.50
Diethyl ether* 2.8 0.53 0.13 4.08
II i-Pentanol 3.7 0.56 0.19 2.95
n-Butanol 3.9 0.56 0.19 2.95
i-Propanol 3.9 0.55 0.19 2.89
n-Propanol 4.0 0.54 0.19 2.84
Ethanol* 4.3 0.52 0.19 2.74
Methanol 5.1 0.48 0.22 2.18
III Tetrahydrofuran* 4.0 0.38 0.20 1.90
Pyridine 5.3 0.41 0.22 1.86
Methoxyethanol 5.5 0.38 0.24 1.58
Methylformamide 6.0 0.41 0.23 1.78
Dimethylformamide 6.4 0.39 0.21 1.86
Dimethylsulfoxide 7.2 0.39 0.23 1.70
IV Acetic acid* 6.0 0.39 0.31 1.26
Formamide 9.6 0.36 0.23 1.57
V Dichloromethane* 3.1 0.29 0.18 1.61
1,1-Dichloroethane 3.5 0.30 0.21 1.43
Benzyl alcohol 5.7 0.40 0.30 1.33
VI Ethyl acetate* 4.4 0.34 0.23 1.48
Methyl ethyl ketone 4.7 0.35 0.22 1.59
Dioxane 4.8 0.36 0.24 1.50
Acetone 5.1 0.35 0.23 1.52
Acetonitrile 5.8 0.31 0.27 1.15
VII Toluene* 2.4 0.25 0.28 0.89
Benzene 2.7 0.23 0.32 0.72
Nitrobenzene 4.4 0.26 0.30 0.87
Nitromethane 6.0 0.28 0.31 0.90
VIII Chloroform* 4.1 0.25 0.41 0.61
Dodecafluoroheptanol 8.8 0.33 0.40 0.83
Water 10.2 0.37 0.37 1.00

*Approximate value.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
After these initial TLC experiments with the neat
solvents, the solvent strength (Si) must either be re-
duced or increased so that the substance zones are
distributed between hRF 20 and 80. The two theoret-
ical situations are depicted in Figure 6 (A and P in
Figure 6). If the compounds to be separated migrate
in the upper third of the plate (A-a in Figure 6) the
solvent strength must be reduced by dilution with
hexane. If the neat solvents do not cause migration of
Figure 6 Strategy for the selection of a suitable TLC solvent.
4657
the substances, the solvent strength must be increased
(P-a in Figure 6) by the addition of water. In both
circumstances the solvent strength should be varied
so that better distribution of the substance zones is
obtained. Consequently, the structures and properties
of the compounds to be separated do not have to be
known. Their classiRcation as apolar (A) or polar (P)
compounds can be made in accordance with their
behaviour in these TLC experiments.
4658 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
If solvents result in good separation, their homo-
logues or other solvents of the same group can also
be tested, as indicated by A-b and P-b in Figure 6.
After these experiments the solvents giving the best
separations are chosen for further optimization of
the separation of apolar compounds. For optimiza-
tion of the mobile phase for separation of polar
compounds, suitable solvents are again selected; the
solvent mixture should contain one solvent in which
the compounds do not migrate; this is necessary for
the transfer of the mobile phase to certain forced-
Sow techniques. In certain circumstances a suitable
separation can be achieved with this solvent-selection
strategy. The individual steps of this method of
solvent selection are depicted in a Sow chart in
Figure 7.
Thus the structures and properties of the com-
pounds to be separated do not have to be known for
these experiments. After these experiments, the sol-
vents giving adequate separations are chosen for
optimization of the mobile phase.
Figure 7 Flow chart illustrating a systematic approach for the selection of suitable solvents.
Mobile Phase Optimization
Mobile phase optimization is based both on modi-
Rcation of published data, on experience with the
analytes, and on intuition. As sample composition
becomes more complex, however, systematic solvent
optimization becomes increasingly important. For sys-
tematic mobile phase optimization four methods are
generally used in planar chromatography:
} window diagram
} sequential simplex method
} Geisss structural approach
} the PRISMA model.
Because only the PRISMA model is currently suit-
able for both manual and automatic mobile phase
optimization, this method is summarized below.
After the selection of suitable solvents the construc-
tion of the actual PRISMA model is begun. In gen-
eral between two and Rve solvents might be selected
for the construction of the model; modiRers might
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Figure 8 The PRISMA system for the systematic optimization of a planar chromatographic method.
also be added. The actual PRISMA model is a three
dimensional geometrical design which correlates the
solvent strength with the selectivity value of the mo-
bile phase. The tripartite model (see the central part
of Figure 8) consists of an irregular top part (light
grey), a regular middle part (white) and the lower
part (dark grey) symbolizing the modiRer(s). When
working with silica as the stationary phase, the upper
frustum is generally used for the optimization of
mobile phases, with or without modiRer, for the
separation of polar compounds. The regular centre
portion of the prism is used for the optimization of
mobile phases, with or without modiRer, for the sep-
aration of apolar compounds. The construction of the
model, the role of solvent strength, and the character-
ization of the selectivity points (PS) are described
extensively in the literature.
The selectivity points on the vertical planes of the
regular part of the prism can be obtained by diluting
the solvent mixtures with a solvent-strength regula-
tor. Solvent-strength (ST) values decrease from top to
bottom; at the base of the prism ST is zero. If sections
are taken across the regular prism parallel to the base,
triangles of different ST levels are obtained. Ob-
viously, the solvent strength is identical at all points
on one of these triangles, and all points on a vertical
straight line correspond to the same selectivity point.
For normal-phase chromatography hexane (Si"0)
is the regulator. If reversed-phase plates must be used
4659
for the separation, the regular part of the model is
used for the separation, irrespective of the polarity of
the compounds to be separated. In these circumstan-
ces water, rather than hexane, must be used as the
solvent-strength regulator.
The solvent-strength values of the modiRer(s) are
treated by the PRISMA model as additive terms. For
the sake of simplicity, the solvent-strength values of
the modiRers are neglected, because they are usually
present at low, constant concentrations (generally
between 0.1 and 3%, e.g. acids, ion pairs).
Manual Optimization Procedure
The four basic selectivity points within the regular
part of the prism (PS"333, 811, 181, 118) for four
solvent mixtures and the three basic selectivity points
on the side of the prism (PS"550, 75}25, 25}75) for
three solvent mixtures are emphasized in Figure 9.
The black points symbolize mixtures of one solvent
and the solvent strength regulator (binary systems);
the dark grey points symbolize mixtures of two sol-
vents and the regulator (ternary systems); and the
three-digit numbers symbolize mixtures of three sol-
vents and the regulator (quaternary systems).
If three solvents were selected for the separation of
apolar compounds, optimization is performed within
the regular part of the model with the help of the four
basic selectivity points. The steps for optimizing the
solvent combination for apolar compounds are
4660 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Figure 9 Favoured selectivity points for mobile phase optimization.
depicted in a Sow chart in Figure 10. If two solvents
were selected, the optimization is performed along
the side of the prism. In both circumstances the sol-
vent strength is adjusted and then different selec-
tivity points are tested. If three to Rve solvents
are selected as best, the number of solvents is reduced
on the basis of criteria such as the number of com-
pounds separated and the
RF values obtained. If the
solvent combinations tested with this strategy do not
result in a sufRcient separation, or at least the
beginnings of a separation, of important pairs of sub-
stances, other solvents must be selected and the process
must be repeated, as indicated in the Sow chart.
For the separation of apolar compounds the op-
timization is generally a rapid process because a few
experiments are sufRcient to evaluate the optimum
mobile-phase composition.
For polar compounds, the optimization is always
started on the top irregular triangle of the model,
either within the triangle, when three solvents were
selected, or along one side, when two solvents were
selected. Water is usually used as a modiRer to in-
crease solvent strength and reduce tailing; if water
is used, several selectivity points cannot be tested
because of immiscibility problems (especially near
PS"811).
Changing the selectivity points on the top triangle
also changes the solvent strength; thus a small change
in the selectivity point might result in a large dif-
ference in resolution, especially when the solvent
strength of the selected solvents differs substan-
tially. The subsequent procedure is similar to that
for the apolar compounds, but the solvent strength
must be adjusted after a suitable selectivity is found.
The Sow chart for the optimization of the solvent
combination for polar compounds is shown in
Figure 11.
In contrast to the separation of apolar compounds,
optimization is a longer process for polar substances
because of the simultaneous change in solvent
strength and selectivity. When water, in particular, is
one of the solvents selected for the construction of the
triangle, a small change in selectivity results in ex-
treme changes in resolution. More chromatographic
experience is, therefore, necessary if the separation
problem is to be solved rapidly.
Manual optimization of the mobile phase must
be performed until at least the beginnings of a separ-
ation of the compounds is obtained. This can usually
be achieved with the Rrst PRISMA combination,
assuming the individual solvents were selected cor-
rectly.
Automatic Optimization Procedure
The basis of automatic mobile-phase optimization,
the correlation between mobile-phase composition
and resolution for saturated TLC systems, can be
described by mathematical functions. The correlation
between hRF values and the selectivity points at a
constant solvent strength level can be expressed by
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY 4661

Figure 10 Flow chart illustrating a systematic approach for the optimization of the mobile phase for the separation of nonplanar
compounds.

the function: Because the vertical correlation can be linearized,

hRF"a(PS)2#b(PS)#c
For quaternary solvent systems, the correlation be-
tween hRF values and solvent strength at a constant
selectivity point can be expressed by the function:
ln hRF"d(ST)#e
measurements on three solvent-strength levels are
needed to calculate the hRF values for all selectivity
points in the spatial design. These correlations are
also relevant when modiRers are used in constant
amounts, for different classes of substance. From
these correlations of hRF values with the selectivity of
the mobile phase, the chromatographic behaviour of
4662 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY

Figure 11 Flow chart illustrating a systematic approach for the optimization of the mobile phase for the separation of polar
compounds.

substances to be separated can be predicted for all
selectivity values in saturated chromatographic
chambers.
The separation quality of predicted chromato-
grams can be assessed by use of a chromatographic
response function (CRF). The optimum composition
can be found by a simple mathematical proced-
ure which maximizes the CRF by monitoring its
dependence upon mobile-phase composition. Twelve
measurements are necessary to discover a local opti-
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Table 2 Required measurements for automatic mobile phase
optimization to achieve the global optimum
Solvent strength Selectivity points
S T1 811 631 118 343 136 181
S T2 811 433 118 316 361 181
S T3 811 613 118 334 163 181
mum, and Rfteen for the global optimum. To increase
the accuracy, six measurements at three different
solvent strength levels (18 experiments) are necessary,
as is seen in Table 2.
Selection of the Mode of Development
Planar chromatography differs from all other chro-
matographic methods in that it enables selection of
the optimum mode of development; the linear mode
of development is used most frequently. Because as-
cending development has no theoretical advantage
over horizontal development, the latter, being more
adaptable, has become increasingly common in re-
cent years.
The advantage of circular development, where the
solvent system migrates radially from the centre of
the plate to the periphery, is well known for the
separation of compounds in the lower RF range.
Working with the same mobile phase, the resolution
is about 4}5 times higher in circular than in linear
development mode, as is seen in Figure 12. This state-
ment is only valid if the samples are spotted exactly at
the centre (x"0 cm). If the distance between the
sample and the mobile phase inlet is, e.g. 2 cm, there
is no signiRcant difference in the lower RF range
between circular and linear development (see
Figure 12). Development can, however, be started
at a point displaced from the centre if a Rlter-paper
ring is used to achieve higher mobile-phase velocity.
Under these conditions many samples can be applied
and the advantages of circular development can be
exploited.
Figure 12 Effect on the
R F value of the distance between
mobile phase inlet and sample.
4663
In anticircular development the mobile phase is
applied to the layer as a circle and Sows towards
the centre. Because the solvent Sow velocity decreases
with the square of the distance, but the area wetted
also decreases with the square of the distance
travelled, the speed of mobile-phase migration is
practically constant. Although anticircular develop-
ment is rarely used, it is an accepted approach in
TLC if resolution must be increased in the higher
RF range.
The multiple development (MD) techniques, UMD
(unidimensional MD) and IMD (incremental MD)
can also be used to increase separating power in the
lower RF range. UMD is the repeated development of
the plate over the same development distance with
mobile phase of the same composition; between de-
velopment steps the mobile phase is removed from
the layer by careful drying and the dried plate is
returned to the development chamber for development
under the same chromatographic conditions as pre-
viously. IMD is an alternative version of this technique
in which successive chromatographic developments
are performed over increasing development distances
with mobile phase of the same composition. In the
IMD Rrst development distance is the shortest and
subsequent development steps are over longer distances;
the development distance usually increases by equal
increments. The last migration distance, the longest,
corresponds to the useful development length of
the plate (but can depend on the mobile phase
employed).
The advantages of the different modes of de-
velopment can be summarized as follows:
} circular development increases resolution in the
lower RF range
} anticircular development increases resolution in
the higher RF range
} UMD is most effective at improving separ-
ation in the lower RF range
} IMD improves zone-centre separation.
A comparison of these modes of development is pre-
sented in Figure 13.
Mobile Phase Transfer
There are two reasons for transferring the optimized
TLC mobile phase. The Rrst is that the separation is
not sufRciently good and better resolution might
be achieved by use of forced-Sow methods. The opti-
mized TLC mobile phase is, therefore, transferred
without alteration to the U-RPC or OPLC technique.
When the latter is used, a prerun must be performed.
For separation of nonpolar compounds the prerun
can be performed with hexane; for separation of polar
4664 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
Figure 13 Effect of linear, circular, UMD, and IMD development modes on RF values in the lower R F range.
substances the prerun can be performed with any
component of the mobile phase in which the compo-
nents do not migrate. The selection of this solvent
Figure 14 Possibilities of transferring the optimized TLC mo-
bile phase to the different forced-flow planar chromatographic
methods, and to preparative column liquid chromatographic
techniques.
might be considered during optimization of the mo-
bile phase. Highly effective separation can be
achieved by use of HPTLC plates and forced-Sow
techniques.
The second reason for transferring an optimized
TLC mobile phase is when scaling up to the various
preparative chromatographic systems. As a result of
the characterization of the different saturation
grade of chromatographic chambers (see Figure 4),
excellent mobile phase transfer between analytical
and preparative planar chromatographic methods
and analytical HPLC can be achieved. The transfer
can be performed on the basis of the chromatographic
conditions used. Dry-Rlled preparative columns (for
Sash, low-pressure liquid, and medium-pressure
liquid chromatography) can be equilibrated with the
solvent used for the prerun in analytical OPLC,
whereas if the column is Rlled by the slurry
technique, the slurry must be prepared from the
same solvent as was used for the OPLC prerun. In
both of these, air bubbles can be eliminated by pas-
sage of an appropriate amount of the solvent used
for the prerun; preparative separation can then be
started with the optimized unsaturated TLC mobile
phase.
The possibilities of mobile-phase transfer between the
different solid}liquid chromatographic methods
are comprehensively summarized in Figure 14, which
demonstrates the possibilities of direct transfer. Dif-
ferent lines show those applicable to the different
methods; dotted lines and thin lines are indicative
of ofSine and online methods, respectively,
whereas thick lines indicate the possibility of transfer
of the optimized mobile phase without change be-
tween different solid}liquid planar and column
chromatographic techniques, both ofSine and on-
line.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY 4665

Selection of Other Operating
Parameters
In conventional TLC the solvent velocity cannot, in
principle, be inSuenced by the chromatographer. The
enhanced efRciency of forced-Sow techniques,
compared with TLC driven by capillary action only,
results from the constant linear mobile phase velocity.
Forced-Sow techniques guarantee optimum H/u
values. In OPLC the upper limit of velocity depends
on the applied external pressure and on the viscosity.
In RPC, the greater the speed of rotation, the faster

Figure 15 Flow chart illustrating a systematic approach for the selection of the mode of development, development distance, and
forced-flow technique.
4666 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN THIN-LAYER (PLANAR) CHROMATOGRAPHY
the migration of the mobile phase. The local mobile-
phase velocity can be inSuenced by the mode of
development selected.
In TLC separation efRciency improves with the
square root of the separation distance. The optimum,
however, depends on the quality of the plate (average
particle size and size distribution of the stationary
phase), the vapour space, the mode of development,
and the properties of the compounds to be separated.
The Rrst of these cannot be inSuenced by the user of
precoated plates. The maximum length of commer-
cially available precoated plates is 20 cm. Thus, the
maximum separation distance in linear development
is 18 cm. The efRciency and rapidity of planar chro-
matography can be increased by the use of a novel
category of multilayer OPLC, long-distance OPLC,
by use of which the separation efRciency is in-
creased signiRcantly. In this technique the end of
the Rrst plate has a slit-like perforation through
which the mobile phase is transferred to a second
layer. Clearly, on this basis, a very long separation
distance can be achieved by combining one plate with
another.
Sample application is one of the most important
stages of successful planar chromatography. The
amount of applied sample depends on the determina-
tion method. Generally,
g and ng quantities of sam-
ple can be determined, but even less than 100 pmol
substance per chromatogram zone has been reported.
During method development, the separation dis-
tance always depends on the mode of development
and the forced-Sow technique used, and on the devel-
opment distance; this is summarized in Figure 15 in
the form of a Sow-chart.
In normal circumstances alteration of temperature
is not an effective means of modifying selectivity
and maximizing resolution. If two compounds are
unresolved at a given temperature, they normally
remain unseparated at other temperatures, irrespect-
ive of whether N- or S-chambers are used. It can
generally be stated that in saturated chromatographic
chambers, which are most commonly used, the tem-
perature does not have a great inSuence on separ-
ations. A change of $53C results in a change in hRF
of less than 3. Nevertheless, in the interest of reproduc-
ibility in duplicate separations it is important to note
the working temperature. Remarkably, temperature
is now being found to play an important role in the
selectivity and efRciency of OPLC separations.
Strategy of Method Development
The PRISMA optimization system is a strategy for
method development in liquid chromatography. Fig-
ure 8, which shows the PRISMA system for planar
chromatography, consists of three parts. The Rrst part is
the selection of the basic parameters; stationary and
vapour phases and suitable solvents, the last according
to the Snyder classiRcation. The second part is the
optimization of the mobile phase, using the PRISMA
optimization model. The third part is the selection of the
Rnal parameters; the mode of development, transfer of
the mobile phase to the appropriate forced-Sow method
and, last but not least, the selection of suitable operating
conditions. The PRISMA system enables the combi-
nation of the appropriate mode of development with
the appropriate forced-Sow technique by the use of
a mobile phase of optimized composition; this of-
fers special possibilities for solving difRcult separ-
ation problems. This system provides guidelines for
method development in planar chromatography.
See also: II/Chromatography: Thin-Layer (Planar)
Chromatography: Historical Development; Instrumenta-
tion; Layers; Modes of Development; Conventional;
Modes of Development; Forced Flow Overpressured
Layer and Centrifugal Chromatography.
Further Reading
Geiss F (1987) Fundamentals of Thin Layer Chromatogra-
phy (Planar Chromatography). Heidelberg: HuK thig.
Nyiredy Sz (1992) Planar chromatography. In: Heftmann
E (ed.) Chromatography, 5th edition, pp. A109}150,
Amsterdam: Elsevier.
Nyiredy Sz (1997) Solvent classiRcation for liquid chromato-
graphy. In: Kaiser O, Kaiser RE, Gunz H and GuK nter
W (eds) Chromatography, pp. 231}239. DuK sseldorf :
InCom Sonderband.
Nyiredy Sz, Botz L, Sticher O (1989) ROTACHROM.
A new instrument for rotation planar chromatography
(RPC). Journal of Planar Chromatography 2: 53}61.
Nyiredy Sz, Dallenbach-Toelke K and Sticher O (1988) The
PRISMA optimization system in planar chromatogra-
phy. Journal Planar Chromatography 1: 336}342.
Nyiredy Sz, FateH r Zs, Botz L and Sticher O (1992) The role
of chamber saturation in the optimization and transfer
of the mobile phase. Journal Planar Chromatography 5:
308}315.
Schoenmakers PJ (1986) Optimization of Chromato-
graphic Selectivity. Amsterdam: Elsevier.
Sherma J and Fried B (eds) (1995) Handbook of Thin-Layer
Chromatography. New York: Dekker.
Szepesi G and Nyiredy Sz (1995) Pharmaceuticals and
drugs. In: Fried B and Sherma J (eds) Handbook of Thin
Layer Chromatography, pp. 819}876. Marcel Dekker:
New York.
TyihaH k E and Mincsovics E (1988) Forced-Sow planar
liquid chromatographic techniques. Journal of Planar
Chromatography 1: 6}19.
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS
ESSENTIAL GUIDES TO METHOD
DEVELOPMENT IN TWO-DIMENSIONAL
ELECTROPHORESIS
M. J. Dunn, Imperial College School of Medicine,
Harefield Hospital, Middlesex, UK
Copyright ^ 2000 Academic Press
Introduction
Most one-dimensional (1-D) methods of polyacryl-
amide gel electrophoresis are limited to the resolution
of 100 or so protein zones. These techniques are
therefore not suitable for the analysis of complex
mixtures containing several thousands of proteins,
such as total protein homogenates of whole cells and
tissue. In addition, they are only able to separate
proteins on the basis of a single physico-chemical
property. For example, the observation of a particu-
lar zone following SDS-PAGE does not imply protein
heterogeneity, but simply indicates that any proteins
present in that zone have nearly identical size proper-
ties, while their charge (and other) properties could
be very different.
The best approach to this problem is to combine
two different 1-D methods into a 2-D procedure.
Ideally, the methods used for each dimension should
be selected by their ability to separate proteins ac-
cording to different properties in each dimen-
sion. Thus, if each method when used alone is able to
resolve 100 protein zones, it would be expected that
up to 10 000 proteins might be resolved when these
methods are used orthogonally. This level of resolu-
tion has rarely been achieved in practice, but never-
theless 2-D has become the method of choice for the
analysis of patterns of protein expression in whole
cells, tissues and organisms; the area now known as
proteomics.
History of 2-D
The Rrst protein separation by 2-D is attributed to
Smithies and Poulik who in 1956 described a combina-
tion of paper and starch gel electrophoresis for the
separation of serum proteins. Since that time, sub-
sequent advances in electrophoresis, such as the use of
polyacrylamide gels, discontinuous buffer sys-
tems, gradient gels, SDS-PAGE, and isoelectric focus-
ing (IEF) have all resulted in the development of
improved methods of 2-D. These developments cul-
4667
minated in the 1970s with publications from several
independent groups describing a combination of a Rrst-
dimension separation by IEF under denaturing condi-
tions with a second dimension separation by SDS-
PAGE. This coupling of IEF with SDS-PAGE resulted in
a method of 2-D which separates proteins according to
two independent parameters, charge and size.
The OFarrell Method of 2-D
The method described by OFarrell in 1975 has for-
med the basis of almost all subsequent developments
in 2-D, and several thousand papers have been pub-
lished using this technique in the 25 years following
its publication. This method was optimized for the
separation of the proteins of Escherichia coli (E. coli)
and used a combination of IEF in cylindrical gels (cast
in glass capillary tubes) containing 8 M urea and 2%
w/v of the non-ionic detergent, Nonidet P-40 (NP-
40), with the SDS-PAGE system of Laemmli. This
method was able to resolve around 500 proteins from
E. coli. It has subsequently been applied to a wide
variety of samples.
Limitations of the OFarrell Method
The main problem with the 2-D method of OFarrell
is associated with the synthetic carrier ampholytes
(SCA) which are used to generate the pH gradient in
the IEF dimension. SCA are produced by a complex
synthetic process which is difRcult to control re-
producibly. This results in considerable batch-to-
batch variability and limits the reproducibility and
consistency of 2-D separations. Perhaps more impor-
tantly SCA are relatively small molecules, which are
not Rxed within the IEF gel. As a consequence, the
electroendosmotic Sow of water that occurs during
IEF results in migration of the SCA molecules to-
wards the cathode.
This process, known as cathodic drift, results in
pH gradient instability and is exacerbated using tube
gels due to the negatively charged groups present on
the walls of the glass capillaries. In practice, pH
gradients using the OFarrell method of 2-D rarely
extend far beyond pH 7, with the resultant loss of the
basic proteins. This problem was recognized by
OFarrell, who developed an alternative procedure,
4668 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS

known as non-equilibrium pH gradient electrophor- 2-D Using IPG IEF

esis (NEPHGE), for the 2-D separation of basic pro-
teins. In this method, separation occurs on the basis IPG IEF gels are prepared using Immobilines (Amer-
of protein mobility in the presence of a rapidly form- sham Pharmacia Biotech), a series of eight acrylamide
ing pH gradient, but reproducibility is extremely dif- derivatives with the structure CH2"CH}
Rcult to control. Fortunately, this problem was solved CO}NH}R, where R contains either a carboxyl or
with the development of immobilized pH gradient tertiary amino group. These form a series of buf-
(IPG) IEF. fers with different pK values distributed through-

Figure 1 Schematic diagram of the procedure of 2-D using IPG IEF. (A) Assembly of the polymerization cassette for the preparation
of IPG and SDS gels cast on plastic backings, (B) casting of IPG and gradient SDS gels, (C) cutting of washed and dried IPG gels into
individual IPG strips, (D) rehydration of IPG strips, (E) IEF in individual IPG strips, (F) equilibration of IPG strips prior to SDS-PAGE, (G)
transfer of IPG strip onto surface of laboratory-made horizontal SDS gel along cathodic wick, (H) transfer of IPG strip onto surface of
commercial horizontal SDS gel along cathodic buffer strip, (I) loading of IPG strip onto the surface of a vertical SDS gel. (Courtesy of
A. GoK rg, Technical University, Munich, Germany).
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS 4669

out the pH range 3 to 10. The appropriate IPG
reagents, calculated according to published recipes,
are added to the mixture used for gel polymerization.
Thus, during polymerization, the buffering
groups which will form the pH gradient are covalent-
ly attached via vinyl bonds to the polyacrylamide
backbone. IPG generated in this way are, therefore,
immune to the effects of electroendosmosis, so
that they provide the opportunity to carry out IEF
separations which are extremely stable, allowing the
true equilibrium state to be attained.
Initial attempts to implement the IPG technology to
2-D separations encountered several problems. Fortu-
nately, largely due to the work of GoK rg and her
colleagues, these problems have been solved and IPG
IEF has become the method of choice for the Rrst
dimension separation of 2-D. The method is shown
schematically in Figure 1. BrieSy, IPG slab gels of the
desired pH range are cast (Figure 1(B)) according to
the extensive library of published recipes. After poly-
merization, the gels are washed, dried and stored at
!203C. The required number of gel strips (3}5 mm
wide) for 2-D are cut off of the slab using a paper
cutter (Figure 1(C)). Alternatively, a range of ready-
made strips is available commercially from Amer-
sham Pharmacia Biotech. IPG strips of any desired
length can be used, but is should be remembered that,
in general, the larger the separation area of a 2-D gel,
the more proteins can be resolved. Strips of 18 cm are
usually employed for high-resolution separations,
while shorter strips (7 or 11 cm) are used for rapid
screening applications.
A choice of a linear pH gradient from 3.5 to 10 is
often useful for the initial analysis of a new type of
sample. However, for many samples this can result in
loss of resolution in the region of pH 4 to 7, in which
the pI values of many proteins occur. This problem
can be overcome to some extent with the use of
a non-linear pH 3.5}pH 10 IPG IEF gel, in which the
pH 4}7 region contains a much Satter gradient than
in the pH 7}10.5 region. This allows good separation
in the pH 4}7 region while still resolving the majority
of the more basic species (Figure 2). However, use of
a pH 4}7 IPG IEF gel will result in even better protein
separation (Figure 3). Commercial IPG strips are
available for these pH ranges (Amersham Pharmacia
Biotech). Laboratory-made IPG strips with either
very narrow pH gradients (spanning 1 pH or less) can
be useful for separating components with very similar
pI values, while very basic pH gradients can be used
to advantage for certain types of sample, such as
ribosomal proteins and nuclear proteins.

Figure 2 A 2-D separation of 100
g heart proteins using a nonlinear pH 3.5 to 10 IPG IEF gel in the first dimension. The protein
pattern was visualized by silver staining. The scale at the top indicates the nonlinear pH gradient obtained using an IPG 3-10 NL strip for
the first dimension IEF separation. The scale at the left indicates the size separation in the range 15 to 150 kDa using a 15%
SDS-PAGE gel in the second dimension.
4670 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS

Figure 3 A 2-D separation of 100
g heart proteins using a linear pH 4 to 7 IPG IEF gel in the first dimension. The protein pattern was
visualized by silver staining. The scale at the top indicates the linear pH gradient obtained using an IPG 4-7 strip for the first dimension
IEF separation. The scale at the left indicates the size separation in the range 10 to 150 kDa using a 10% SDS-PAGE gel in the second
dimension.

For use in 2-D, the strips are rehydrated in a re-
swelling cassette (Figure 1(D)) in a solution containing
8 M urea, 0.5% non-ionic (e.g. NP-40, Triton X-100)
or zwitterionic (CHAPS) detergent (3-[(cholamido-
propyl)dimethylammonio]-1-propanesulfonate),
15 mM DTT and 0.2% synthetic carrier ampholyte
(SCA) of the appropriate pH range. The strips are
then placed directly on the surface of the cooling plate
of a horizontal Sat-bed electrophoresis apparatus
(Figure 1(E)). A convenient alternative is to use the
special strip tray available from Pharmacia (Fig-
ure 1(E)). This tray is Rtted with a corrugated plastic
plate which contains grooves allowing easy alignment
of the IPG strips. In addition, the tray is Rtted with
bars carrying the electrodes and a bar Rtted with
sample cups allowing the application of samples at
any desired point on the gel surface. This tray is Rlled
with silicone oil which protects the gel from the
effects of the atmosphere during IEF. Horizontal
streaking can often be observed at the basic end of
2-D protein proRles, particularly when IPG 6}10 is
used for the Rrst dimension. This problem can be
resolved by applying an extra electrode strip soaked
in 15 mM DTT on the surface of the IPG strip along-
side the cathodic electrode strip. This has the advant-
age that the DTT within the gel, which migrates
towards the anode during IEF, is replenished by the
DTT released from the strip at the anode. An alterna-
tive approach is to use the non-charged reducing
agent, tributyl phosphine (TBP), which does not mi-
grate during IEF and has been found to greatly im-
prove protein solubility during IEF.
Sample Preparation
There is no universal method of sample preparation
for 2-D due to the diverse nature of samples which
can be analysed. Whatever method is used, it is essen-
tial to minimize protein modiRcations which can re-
sult in artefactual spots on 2-D protein patterns. In
particular, samples containing urea should not be
heated as this will lead to charge heterogeneity as
a result of protein carbamylation by isocyanate ions
formed from the decomposition of urea. Proteases
present within samples can also readily result in arte-
factual spots, so that samples should be subjected to
minimal handling and kept cold at all times. Protease
inhibitors can also be added.
Liquid samples containing a relatively high protein
concentration (e.g. serum, plasma) require little or no
APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS
pre-treatment prior to 2-D. However, less concen-
trated solutions (e.g. urine, cerebrospinal Suid (CSF),
amniotic Suid) often require concentration by
methods such as lyophilization, or precipitation with
trichloroacacetic acid (TCA) or acetone. Solid tissue
samples must usually be disrupted in the presence of
solubilization solution. For small samples this is read-
ily achieved by crushing the sample in liquid nitrogen
using a pestle and mortar, while larger tissue samples
must be homogenized using a suitable device. Cell
suspensions can be readily harvested by centrifu-
gation, while cells adherent to a substrate, such as
a tissue culture Sask or dish, should be collected by
scraping (the use of proteases should be avoided to
prevent possible sample degradation). Alternatively,
the cells can be detached by lysis directly in a small
volume of sample solubilization solution.
Sample Solubilization
The most popular method for protein solubilization
for 2-D is that originally described by OFarrell, using
a mixture of 9.5 M urea, 4% w/v NP-40, 1% w/v
DTT and 2% w/v SCA. While this method works
well for the majority of samples, it is not universally
applicable, with membrane proteins representing
a particular challenge. The zwitterionic detergent,
CHAPS has been found to be effective for the
solubilization of membrane proteins, particularly
when used at a concentration of 4% w/v in combina-
tion with a mixture of 2 M thiourea and 8 M urea.
Linear sulfobetaine detergents, such as SB 3}10 or
3}12, are also effective solubilizing agents, but
these are not compatible with high concentrations of
urea. This can be overcome by using these reagents at
2% w/v in combination with 5 M urea, 2 M thiourea
and 2% CHAPS.
The presence of nucleic acids can be problematic
during IEF. This is due to an increase in the viscosity
of the sample and in some cases formation of com-
plexes with the sample proteins, leading to artefactual
migration and streaking. If problems of this type are
suspected, it is best to degrade the nucleic acid by the
addition of a suitable pure (i.e. protease free) en-
donuclease to the sample solubilization solution.
Sample Reduction
Protein disulRde bonds are normally reduced with
free thiol-containing reagents such as DTT or
-mer-
captoethanol. However, reagents such as DTT are
charged so that they migrate out of the gel during IEF,
leading to reoxidation of the sample proteins which
can result in loss of sample solubility. It has recently
been reported that replacing the thiol-containing re-
4671
ducing agents with a non-charged reducing agent
such as tributyl phosphine (TBP) can greatly increase
protein solubility during the IEF dimension and result
in increased transfer to the second dimension gel.
Sample Application and
Running Conditions
Samples are usually applied into silicone rubber
frames or special sample cups (Figure 1(E)) placed
either at the anodic or cathodic end of the IPG strips,
the optimum position being determined empirically
for each type of sample. The initial voltage should be
limited to 150 V for 30 min to allow maximal sample
entry and then progressively increased until 3500 V
is attained. The time required for the run depends on
several factors, including the type of sample, the
amount of protein applied, the length of the IPG
strips, and the pH gradient. The IEF run should be
performed at 203C, as at lower temperatures there is
a risk of urea crystallization and higher temperatures
have been found to result in alterations in the relative
positions of some proteins on the Rnal 2-D patterns.
Some typical running conditions are given in Table 1.
This method of sample application can result in
protein precipitation and the effect is more pro-
nounced when high protein loadings (1 mg or more)
are used. The problem can be overcome by reswelling
the IPG strips directly in the solution containing the
protein sample to be analysed. Very high protein
loads ('10 mg) have been successfully separated us-
ing this method, but there can be a selective loss of
high molecular weight, very basic and membrane
proteins. Recently a new integrated instrument,
named the IPGPhor (Amersham Pharmacia Biotech),
has been developed to simplify the IPG IEF dimension
2-D. This instrument features a strip holder that pro-
vides for the rehydration of individual IPG strips with
or without sample, optional separate sample loading,
and subsequent IEF, all without handling the strip
after it is placed in the ceramic strip holder. The
Table 1 Suggested running conditions for 18 cm IPG strips for
the first, IEF dimension of 2-D. The strips should be run at
0.05 mA per strip (2 mA maximum total), 0.5 W maximum, 203C
Voltage (maximum) IPG strip (pH range) Time
150 All 30 min
300 All 60 min
1500 All 60 min
3500 4}7 42 000 Vh
4}8 35 000 Vh
4}9 30 000 Vh
6}10 35 000 Vh
3}10.5 25 000 Vh
4672 APPENDIX 2 / ESSENTIAL GUIDES TO METHOD DEVELOPMENT IN TWO-DIMENSIONAL ELECTROPHORESIS
instrument can accommodate up to 12 individual
strip holders and incorporates Peltier solid-state cool-
ing and a programmable 8000 V, 1.5 mA power
supply.
Equilibration Between Dimensions
After the IPG IEF dimension, strips can be used im-
mediately for the second dimension. Alternatively,
strips can be stored between two sheets of plastic Rlm
at !803C for periods of several months. Prior to the
second-dimension separation, it is essential that the
IEF gels are equilibrated to allow the separated pro-
teins to interact fully with sodium dodecyl sulfate
(SDS) so that they will migrate properly during SDS-
PAGE (Figure 1(F)). The recommended protocol is to
incubate the IPG IEF gel strips for 15 min in 50 mM
Tris buffer, pH 8.8 containing 2% w/v SDS, 1%
w/v DTT, 6 M urea and 30% w/v glycerol. The urea
and glycerol are used to reduce electroendosmotic
effects which otherwise result in reduced protein
transfer from the Rrst to the second dimension. This is
followed by a further 15 min equilibration in the
same solution containing 5% w/v iodoacetamide in
place of DTT. The latter step is used to alkylate any
free DTT, as otherwise this migrates through the
second-dimension SDS-PAGE gel, resulting in an ar-
tefact known as point-streaking which can be ob-
served after silver staining. An alternative procedure,
allowing equilibration to be achieved in a single step,
is to replace the DTT in the equilibration buffer
with 5 mM TBP, which is uncharged and so does not
migrate during SDS-PAGE.
The Second Dimension
After equilibration, the Rrst dimension IEF gels are
applied directly to the surface of the second-dimen-
sion SDS-PAGE gels. The SDS-PAGE gels can be of
any appropriate single or gradient polyacrylamide,
and can be used either in a vertical (Figure 1(I)) or
horizontal format (Figure 1(G), 1(H)). The use of
vertical formats enables multiple gels to be run simul-
taneously, which improves reproducibility, while the
use of horizontal, 0.5 mm thin SDS gels cast on plas-
tic supports improves the ease of handling the gels
and gives rapid separations.
Resolution of 2-D
The resolving capacity of 2-D gels is usually con-
sidered to be proportional to the total gel area avail-
able for the separation. Using 18 cm long IPG IEF gels
in combination with 20 cm long second-dimension
SDS-PAGE gels, around 2000 proteins can be readily
resolved. Only a few hundred proteins can be separ-
ated using mini-gel formats, but these are much
quicker to run and can be useful for rapid screening
purposes. For maximum resolution of very complex
mixtures, very large format gels ('30 cm in each
dimension) can be used. These are reported to be able
to separate as many as 5000 to 10 000 proteins from
whole cell lysates, but this is achieved at the expense
of the ease of gel handling and processing.
Reproducibility of 2-D
Until recently reproducibility was a major problem
limiting the more widespread application of 2-D.
Using the tube gel technique of OFarrell, it was often
difRcult to obtain reproducible separations of
a particular type of sample even within a single labor-
atory, while comparison of 2-D separation patterns
generated in different laboratories was often
considered to be impossible. The use of dedicated
equipment for 2-D, such as the ISO-DALT (Amer-
sham Pharmacia Biotech) and the Investigator (ESA
Inc) systems, helps in this regard as it allows the
simultaneous electrophoresis of large numbers
(between 5 and 20) of 2-D gels under reproducibly
controlled conditions. More importantly, inter-
laboratory studies of various types of sample (heart,
barley, yeast) have unequivocally demonstrated
that 2-D using IPG IEF results in 2-D protein separ-
ations with very high spatial and quantitative repro-
ducibility.
Proteomics
2-D separation has now matured into a technique
which is capable of separating reproducibly thou-
sands of proteins present in samples such as cells,
tissues and even whole organisms. Recent develop-
ments in methods for the microchemical characteriza-
tion of proteins, particularly techniques for the analy-
sis of proteins and peptides by mass spectrometry,
now make it possible to identify and characterize
proteins spots directly from 2-D gels. This has made
2-D separation an ideal tool to use in studies designed
to determine the nature and function of the large
number of structural genes being identiRed in various
genome initiatives. This area has become known as
proteomics and is the subject of a separate article.
Further Reading
Blomberg A, Blomberg L, Norbeck J et al. (1995) Inter-
laboratory reproducibility of yeast protein patterns ana-
lyzed by immobilized pH gradient two-dimensional gel
electrophoresis. Electrophoresis 16: 1935}1945.
 APPENDIX 3 / ABBREVIATIONS 4673

Corbett JM, Dunn MJ, Posch A and GoK rg A (1994)
Positional r!eproducibility of protein spots in two-di-
mensional polyacrylamide gel electrophoresis using im-
mobilised pH gradient isoelectric focusing in the Rrst
dimension: An interlaboratory study. Electrophoresis
15: 1205}1211.
Dunn MJ (1987) Two-dimensional polyacrylamide gel elec-
trophoresis. In: Chrambach A, Dunn MJ and Radola BJ
(eds) Advances in Electrophoresis, Vol. 1, pp. 1}109.
Weinheim: VCH.
Dunn MJ (1993) Gel Electrophoresis: Proteins. Oxford:
BIOS ScientiRc.
GoK rg A, Postel W and GuK nther S (1988) The current state of
two-dimensional electrophoresis with immobilized pH
gradients. Electrophoresis 9: 531}546.
GoK rg A, Boguth G, Obermaier C, Posch A and Weiss
W (1995) Two-dimensional polyacrylamide gel elec-
trophoresis with immobilized pH gradients in the Rrst
dimension (IPG-Dalt): The state of the art and the con-
troversy of vertical versus horizontal systems. Elec-
trophoresis 16: 1079}1086.
GoK rg A, Obermaier C, Boguth G et al. (1997) Very alkaline
immobilized pH gradients for two-dimensional elec-
trophoresis of ribosomal and nuclear proteins. Elec-
trophoresis 18: 328}337.
Herbert BR, Sanchez JC and Bini L (1997) Two-dimen-
sional electrophoresis: The state of the art and future
directions. In: Wilkins MR, Williams KL, Appel RD
and Hochstrasser DF (eds) Proteome Research: New
Frontiers in Functional Genomics, pp. 13}33. Berlin:
Springer.
Humphery-Smith I, Cordwell SJ and Blackstock WP (1997)
Proteome research: Complementarity and limitations
with respect to the RNA and DNA worlds. Electrophor-
esis 18: 1217}1242.
Klose J and Kobalz U (1995) Two-dimensional electrophor-
esis of proteins: An updated protocol and implications
for a functional analysis of the genome. Electrophoresis
16: 1034}1059.
OFarrell PH (1975) High resolution two-dimensional elec-
trophoresis of proteins. Journal of Biological Chemistry
250: 4007}4021.
Pennington SR, Wilkins MR, Hochstrasser DF and Dunn
MJ. Proteome analysis: From protein characterization
to biological function. Trends in Cell Biology 7:
168}173.
Rabilloud T, Adessi C, Giraudel A and Lunnardi J (1997)
Improvement of the solubilization of proteins in two-
dimensional electrophoresis with immobilized pH gradi-
ents. Electrophoresis 18: 307}316.
Wilkins MR, Williams KL, Appel RD and Hochstrasser DF
(eds) (1997) Proteome Research: New Frontiers in
Functional Genomics. Berlin: Springer.

3. ABBREVIATIONS

2,3,4,6-TeCP 2,3,4,6-tetrachlorophenol
2,3-DMP 2,3-dimethylpentane
2,4,5-T 2,4,5-trichlorophenoxyacetic acid
2,4,5-TP 2-(2,4,5-trichlorophenoxy)propionic acid
2,4,6-TCP 2,4,6-trichlorophenol
2,4-D 2,4-dichlorophenoxyacetic acid
2,4-DCP 2,4-dichlorophenol
2,4-DMP 2,4-dimethylphenol
2-CP 2-chlorophenol
2-DNP 2-dinitrophenol
2-M-4,6-DNP 2-methyl-4,6-dinitrophenol
2-NP 2-nitrophenol
4-NP 4-nitrophenol
AA amino acid
AAA amino acid compositional analysis
AAS atomic absorption spectrometry
ACN acetonitrile
ADAM 9-anthryldiazomethane
ADC analog-to-digital converter
AD-CSP amylose tris-(3,5-dimethylphenylcarbamate) CSP
AE alcohol ethoxylates
AEDA aroma extract dilution analysis
AFM atomic force microscopy
AGP 1-acid glycoprotein
4674 APPENDIX 3 / ABBREVIATIONS

AMD automated multiple development

AMW acidic mine waters
ANN artiRcial neural network
APCI atmospheric pressure chemical ionization
APE alkylphenol ethoxylates
API atmospheric pressure ionization
AQC 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate
ASE accelerated solvent extraction
ASE Archimedean screw effect
ASP amnesic poisoning
ASPEC automated SPE clean-up
ASTM American Society for Testing and Materials
b.p. boiling point
BE benzoylecgonine
BF batch factor
BF Best Foods
BGE background electrolyte
BOD biological oxygen demand
BPR back-pressure regulator
BSA bovine serum albumin
BSTFA N,O-bis-(trimethylsilyl)triSuoroacetamide
BTEX benzene, toluene, ethylbenzene, xylene
BTX benzene, toluene, xylene
C(GC)2 comprehensive gas chromatography
c.m.c. critical micellar concentration
CB Contaminants Branch
CBH I cellobiohydrolase I
CCC countercurrent chromatography
CCDs charged coupled devices
CCP chiral coated phase
CD cyclodextrin
CDCCC centrifugal droplet countercurrent chromatography
CDR chiral derivatization reagent
CE capillary electrophoresis
CEC capillary electrochromatography
CF-FAB continuous Sow } fast atom bombardment
CI chemical ionization
CID collision-induced dissociation
CLA conjugated linoleic acid
CLD chemiluminescence detector
CMA chiral mobile-phase additive
CMP chiral mobile phase
CMR continuous membrane reactor
CoMFA comparative molecular Reld analysis
CPC coil planet centrifuges
CPF co-current permeate Sow
CPT cone penetrometer
cSFC supercritical Suid chromatography with a capillary column
CSP chiral stationary phase
CTA cellulose triacetate
CTAB cetyltrimethylammonium bromide
CV coefRcient of variation
CXC cation exchange chromatography
CZE capillary zone electrophoresis
DA domoic acid
 APPENDIX 3 / ABBREVIATIONS 4675

DABITC dimethylaminoazobenzene isothiocyanate

DABS dimethylaminonaphthalene-5-sulfonyl
DAG diacetonegulonic acid
DANS 1-N, N-dimethylaminonaphthalene-5-sulfonyl
DB-5 5% poly(diphenyldimethylsiloxane)
DBT dibenzothiophene(s)
DBV divinylbenzene
DCCC droplet countercurrent chromatography
DCTFA 1,2-dichlorotetraSuoroacetone
DEHPA di(2-ethylhexyl)phosphoric acid
DETA diethylenetriamine
DIA deisopropylatrazine
DIC diisopropylcarbodiimide
DIGE difference gel electrophoresis
DMAPA dimethylaminopropylamine
DMCS dimethylchlorosilane
DME
,-dicarboxylic acid methyl esters
DMOX 4,4-dimethyloxazoline
DNP dinitrophenyl
DNPH 2,4-dinitrophenylhydrazine
DNPU 3,5-dinitrophenyl urethane
DP degree of polymerization
DRI differential refractive index
DRIFT diffuse reSectance Fourier transform infrared
DSC N,N-disuccinimidylcarbonate
DSP diarrhoeic poisoning
DTX dinophysistoxins
ECD electron-capture detector
ECF ethyl chloroformate
ECL equivalent chain length
ECN equivalent carbon number
ED electrodialysis
ED extractive distillation
EDMA ethylene glycol dimethacrylate
EDTA ethylenediaminetetraacetic acid
ee enantiomer excess
EG ethylene glycol
EGA ethylene glycol adipate
EHPA mono-2-ethylhexyl ester
EI electron ionization
ELCD electrolytic conductivity detector
ELSD evaporative light-scattering detector
EOF electroosmotic Sow
EPA Environmental Protection Agency
ESCA electron spectroscopy for chemical analysis
ESI electrospray ionization
EtG ethyl glucoronide
FAB fast atom bombardment
FAEE fatty acid ethyl ester
FAME fatty acid methyl ester
FDA Food and Drug Administration
FDNB 1-Suoro-2,4-dinitrobenzene
FFA free fatty acid
FFPPC forced Sow PPC
FID Same ionization detector
4676 APPENDIX 3 / ABBREVIATIONS

FMOC Suoronylmethyl chloroformate

FPD Same photometric detector
FTD Same thermionic detector
FT-IR Fourier transform infrared spectrometry
FT-Raman Fourier transform Raman
GC gas chromatography
GC-FTIR gas chromatography } Fourier transform infrared spectrometry
GC-IRMS gas chromatography } isotope ratio mass spectrometry
GC-MS gas chromatography } mass spectrometry
GC-MS/MS GC with coupled or tandem MS
GFAA graphite furnace atomic absorption
GFC gel Rltration chromatography
GLP Good Laboratory Practice
GPC gel permeation chromatography
HAS human serum albumin
HDC hydrodynamic chromatography
HDEHP di-(2-ethylhexyl)orthophosphoric acid
HETEs hydroxyeicosatetraenes
HETP height equivalent to one theoretical plate
HFB heptaSuorobutyryl
HI hydrophobic interaction
HIC hydrophilic interaction chromatography
HIC-CXC hydrophilic interaction } cation exchange chromatography
HMDS hexamethyldisilazine
HOM humic organic matter
HOMO highest occupied molecular orbital
HPA heteropolyacid
HPALC high performance afRnity liquid chromatography
HPIC high performance ion chromatography
HPLC high performance liquid chromatography
HPLC-CSP HPLC-chiral stationary phases
HPTLC high performance thin-layer chromatography
HRGC high resolution gas chromatography
HS humic substances
HSCCC high speed countercurrent chromatography
HSES hydrostatic equilibrium system
HS-GC headspace-gas chromatography
HSSI N-hydroxysulfosuccinimide
HS-SPME headspace-solid-phase microextraction
HTGC high temperature gas chromatography
HVS high volume sampling
i.d. internal diameter
IAM immobilized artiRcial membrane
IBCF isobutyl chloroformate
IBOC N-isobutyloxycarbonyl
ICP-AES inductively coupled plasma-atomic emission spectroscopy
ICP-MS inductively coupled plasma-mass spectrometry
ICR ion cyclotron resonance
IE ion exchange
IEC ion exchange chromatography
IEF isoelectric focusing
IIR ion interaction reagent
IP-TLC ion pair-thin-layer chromatography
IR infrared
IS internal standard
 APPENDIX 3 / ABBREVIATIONS 4677

ITD ion trap detector/detection

IXISS ion exchange isothermal supersaturation
JT Joule-Thompson
L/B length-to-breadth
LAS linear alkylbenzenesulfonate
LC liquid chromatography
LDH lactic acid dehydrogenase
LD-OPLC long-distance overpressured-layer chromatography
LEC liquid exclusion chromatography
LFER linear free-energy relationship
LLE liquid-liquid extraction
LOD limit of detection
LOX liquid oxygen
LSC liquid-solid chromatography
LSER linear solvation energy relationship
LSIMS liquid secondary ion mass spectrometry
LTB4 leukotriene B4
LUMO lowest unoccupied molecular orbital
LVS low volume sampling
MA macrocyclic antibiotic
MAC multistage air compressor
MAGIC monodisperse aerosol generator interface for chromatography
MALDI matrix-assisted laser desorption ionization
MALDI-MS matrix-assisted laser desorption ionization-mass spectrometer
MASE microwave-assisted solvent extraction
MBTH 3-methyl-2-benzothiazolinone hydrazone-HCl
MCF methyl chloroformate
MCTA microcrystalline cellulose triacetate
MDGC multidimensional gas chromatography
MDMA 3,4-methylenedioxymethamphetamine
ME methyl ester
MEKC micellar electrokinetic chromatography
MF microRltration
MIBK methylisobutylketone
MID multiple ion detection
MIPs molecular imprinted polymers
MLL mean list length
ML-OPLC multi-layer overpressured-layer chromatography
MP mobile phase
MPA mobile phase additive
MPA 3-mercaptopropionic acid
MPLC medium pressure liquid chromatography
MS mass spectrometry
MSD mass selective detector/detection
MS-MS tandem mass spectrometry
MSPD matrix solid phase dispersion
MSTFA N-methyl-N-trimethylsilyltriSuoroacetamide
MTBSTFA N-t-butyldimethylsilyl-N-methyltriSuoroacetamide
MWD molecular weight distribution
MWPC multiwire proportional counters
NAC N-acetyl-L-cysteine
NBP 4-(p-nitrobenzyl)pyridine
NCA National Council on Alcoholism
NCI-GC-MS negative ion chemical ionization-GC-MS
nd not determined
4678 APPENDIX 3 / ABBREVIATIONS

NF nanoRltration
NFM N-formylmorpholine
NHYD ninhydrin
NIR near-infrared spectroscopy
NMP N-methylpyrrolidone
NMR nuclear magnetic resonance
NN neural network
NP normal phase
NPD nitrogen-phosporus detector
NQS 1,2-naphthoquinone-4-sulfonate
NRTL nonrandom two liquids
NSAID nonsteroidal anti-inSammatory drugs
o.d. outer diameter
OA okadaic acid
ODS octadecylsilica
ODS-1 commercial octadecylsilica phase
OGCHI ovoglycoprotein from chicken egg whites
OMCHI chicken ovomucoid
OMCTS octamethylcyclotetrasiloxane
OMTKY turkey ovomucoid
OPA/MCE o-phthalaldehyde/-mercaptoethanol
O-PFBO O-pentaSuorobenzyloxime
OPLC overpressured-layer liquid chromatography
OPTLC overpressured TLC
OV-225 commercial phase
oxo-ETEs oxo-eicosatetraene
P&T purge-and-trap
PA photoacoustic
PA polyacrylate
PAD pulsed amphometric detector/detection
PAH polyaromatic hydrocarbons
PAR 4-(2-pyridylazo)resorcinol
PAS photoacoustic spectroscopy
PB particle beam
PCA principal component analysis
PCB polychlorinated biphenyl
PCDD polychlorinated dibenzo-p-dioxin
PCDF polychlorinated dibenzofuran
PCP pentachlorophenol
PDB Pee Dee Belimnite
PDCA pyridine-2,6-dicarboxylate
PDMS poly(dimethysiloxane)
PED pulsed electrochemical detection
PEEK polyetheretherketone
PEG poly(ethylene glycol)
PEI poly(ethyleneimine)
PEO poly(ethylene oxide)
PETRA pentaerythritol triacrylate
PFB pentaSuorobenzyl
PFBBr pentaSuorobenzyl bromide
PFBOA pentaSuorobenzyloxime
PFP pentaSuoropropionyl
PFPH pentaSuorophenylhydrazine
PGC porous graphitic carbon
PGM platinum group metals
 APPENDIX 3 / ABBREVIATIONS 4679

PHDC packed column hydrodynamic chromatography

PLOT porous-layer open-tubular
PLS partial least squares
PMMA poly(methyl methacrylate)
PNBX potassium n-butylxanthate
PNP purine nucleoside phosphorylase
PPC preparative planar thin-layer chromatography
PPO 2,5-diphenyloxazole
PS poly(styrene)
PSD particle size distribution
PS-DVB poly(styrene-divinylbenzene)
PSFC packed-column supercritical Suid chromatography
PSP paralytic poisoning
PTFE poly(tetraSuoroethylene)
PTH phenylthiohydantoin
PV pervaporation
QSAR quantitative structure-activity relationship
QSERR quantitative structure enantioselective retention relationship
QSRR quantitative structure-retention relationship
R/D reSux-to-overhead ratio
RI refractive index
RIA radioimmunoassay
RLCCC rotation locular countercurrent chromatography
RMM relative molar mass
RO reverse osmosis
RP reversed-phase
RPC reversed-phase chromatography
RS Raman scattering
RSD relative standard deviation
S/F solvent-to-feed ratio
S/N signal-to-noise ratio
SAC strong acid cation
SARA saturates, aromatics, resins and asphaltenes
SB/CD short bed/continuous development
SBC strong base anion
SBF separation by Sow
SCA synthetic carrier ampholyte
SCOT support-coated open-tubular
SD standard deviation
SDE simultaneous distillation-extraction
SDGlu N-dodecanoyl-L-glutamate
SDS sodium dodecyl sulfate
SDS-PAGE sodium dodecyl sulfate-polyacrylamide gel electrophoresis
SDVal sodium N-dodecanoyl-L-valinate
SDVB styrene-divinylbenzene
SEC size exclusion chromatography
SERRS surface-enhanced resonance Raman scattering
SERS surface-enhanced Raman scattering
SF solvent front
SF6 sulfur hexaSuoride
SFC supercritical Suid chromatography
SFE supercritical Suid extraction
SIM selected ion monitoring
SIM single ion monitoring
SMB simulated moving bed
4680 APPENDIX 3 / ABBREVIATIONS

SN separation number
SP stationary phase
SPE solid-phase extraction
SPM simultaneous pyrolysis/methylation
SPME solid-phase microextraction
SRB sulfate-reducing bacteria
SS supersaturated solution
STC sodium taurocholate
STDC sodium taurodeoxycholate
STX saxitoxin
TA time-to-amplitude
TAA tetraalkylammonium
TAB N-triSuoroacetyl-n-butyl ester
TBDMS t-butyldimethylsilyl
TCD thermal conductivity detector
TDS total dissolved solids
TEA thermal energy analyser
TEAA triethylammonium acetate
TEAP triethylammonium phosphate
TEG triethylene glycol
TEPA tetraethylenepentamine
TFA triSuoroacetyl/triSuoroacetic acid
TGA thermogravimetric analyser
THBC 1,2,3,4-tetrahydro--carbolines
THCA 11-nor-delta9-tetrahydrocannabinol-9-carboxylic acid
THF tetrahydrofuran
TIQ 1,2,3,4-tetrahydroisoquinoline
TLC thin-layer chromatography
TLC-MS thin-layer chromatography-mass spectrometry
TLRC thin-layer radiochromatography
TLV threshold limit value
TMAH tetramethylammonium hydroxide
TMCS trimethylchlorosilane
TMPA trimethylphenylammonium hydroxide
TMS trimethylsilyl
TOC total organic carbon
TOEDA tetraoctylethylenediamine
TRIM trimethylolpropane trimethacrylate
TSI thermospray ionization
TTF tetrathiafulvalene
UF ultraRltration
UHP ultrahigh purity
UNIQUAC universal quasichemical
UrdPase uridine phosphorylase
UTP uniform transmembrane pressure
UV/Vis ultraviolet-visible
VC total column capacity
VFA volatile fatty acids
VLDL very low-density lipoproteins
VLE vapour-liquid equilibrium
VOCs volatile organic chemicals
VPO vapour pressure osmometry
VR retention volume of the solute
VSF retention volume of solvent front
WAC weak acid cation
 APPENDIX 4 / ANALYTICAL CHIRAL SEPARATION METHODS 4681

WBA weak base anion

WCOT wall-coated open-tubular
WPC whey protein concentrates
XE-60 commercial phase
XRF X-ray Suorescence
ZDDP zinc dialkyldithiophosphate
-CD -cyclodextrin

4. ANALYTICAL CHIRAL SEPARATION

METHODS (IUPAC RECOMMENDATIONS 1997)
Prepared for publication by
V. A. Davankov, Russian Academy of Sciences, Moscow, Russia
^ 1997 IUPAC

Abstract
In recent years there has been considerable interest in the synthesis and separation of enantiomers of organic
compounds especially because of their importance in the biochemistry and pharmaceutical industry. Fre-
quently the methods used for the separations, for monitoring the progress of an asymmetric synthesis or
optical purity of the products are chromatographic with either liquids, gases, or supercritical Suids as the
mobile phase. More recently capillary electrophoresis has been added as an analytical chiral separation
method.
These applications have led to a number of terms and expressions in addition to those commonly used or
recently recommended for the chemistry and physical properties of chiral compounds. This Nomenclature
provides the descriptions and deRnitions for additional terms particularly related to analytical separation
methods, and to the formation and enantiomeric purity of chiral products

Introduction
Enantiomers are two chemically identical molecular species which differ from each other as non-
superposable mirror images. The most simple and vivid model for enantiomeric structures is the two hands,
left and right. Enantiomers, in addition to diastereomers and cis-trans-isomers, are thus a special case of
stereoisomers.
The chirality (handedness) of enantiomeric molecules is caused by the presence of one or more chirality
elements (chirality axis, chirality plane, or chirality centre, e.g. asymmetric carbon atom) in their structure.
The chirality sense and optical activity of the enantiomers are determined by their absolute conRguration, i.e.
the spatial arrangement of the atoms in the molecule. In contrast to their conformation, the conRguration of
enantiomers cannot be changed without a change in the connectivity of constituent atoms. Designation of the
conRguration of enantiomers should be made in accordance with the Cahn-Ingold-Prelog R, S-system. The
Delta-Lambda designations for enantiomers of octahedral complexes and the D,L Fischer-Rosanoff
designations for amino acids and sugars are also in use.
Conventional chemical synthesis, in contrast to asymmetric synthesis, deals mostly with the transformations
of achiral compounds. If these reactions result in the formation of a chirality element in the molecule, the
reaction product appears to be an equivalent mixture of a pair of enantiomers, a racemate, which is optically
inactive. Racemates are also formed through racemisation of chiral compounds. Racemates crystallize in the
form of a racemic compound or, less frequently, as a conglomerate.
Separation of the enantiomers comprising the racemate, i.e. the resolution of the racemate, is a common
problem in stereochemical research as well as in the preparation of biologically active compounds, in
particular, drugs. The problem is that in contrast to distereomers and all the other types of isomeric species,
4682 APPENDIX 4 / ANALYTICAL CHIRAL SEPARATION METHODS

enantiomers, in an achiral environment, display identical physical and chemical properties. (Energetic
inequivalence of enantiomeric species, which can arise from the violation of parity by the weak interactions
[1], is negligibly small- of the order of 10\14 J mol\1).
One approach to separate enantiomers, sometimes referred to as indirect enantiomeric resolution, involves
the coupling of the enantiomers with an auxiliary chiral reagent to convert them into diastereomers. The
diastereomers can then by separated by any achiral separation technique.
Nowadays, direct separation methods are commonly used in which the enantiomers are placed in a chiral
environment. As a matter of principle, only chiral selectors or chiral irradiation (e.g. a polarized light beam
which consists of two chiral circular-polarised components) can distinguish between two enantiomers. Chiral
selectors can be an appropriate chiral molecule or a chiral surface (e.g. a chiral seed crystal). Due to the
enantioselectivity (a special case of stereoselectivity) of the interaction with the two enantiomers, the chiral
selector either transforms the enantiomers at a different rate into new chemical entities (kinetic enan-
tioselectivity) or forms labile molecular adducts of differing stability with the enantiomers (thermodyn-
amic enantioselectivity). Enzymic selective transformation of L-enantiomers of racemic D,L-amino acids is
a typical example of a kinetically enantioselective process (kinetic resolution). Enantioselective (chiral)
chromatography does not modify the enantiomeric species to be separated and thus represents an example of
a thermodynamically enantioselective process.
Direct enantiomeric resolutions are only feasible in chromatographic systems which contain an appropriate
chiral selector. The latter can be incorporated into the stationary phase (chiral stationary phase) or be
permanently bonded to or coated onto the surface of the column packing material (chiral bonded and chiral
coated stationary phases). In all these cases it is appropriate to refer to the chromatographic column as an
enantioselective (chiral) column. Enantioselective chromatography can also be performed on achiral
chromatographic columns using the required chiral selector as a chiral mobile phase or a chiral mobile phase
additive. Combinations of several chiral selectors in the mobile phase [2] as well as mobile and stationary
phases [3] are also feasible.
In the case of chiral stationary phases, the enantiomer that forms the more stable association with the chiral
selector will be the more strongly retained species of the racemate. The enantioselectivity of the chiral
chromatographic system is then expressed as the ratio of the retention factors of the two enantiomers. This
ratio may approach the value of the thermodynamic enantioselectivity of the association of the chiral selector
with the enantiomers. This situation occurs when the association with the chiral selector governs the retention
of the enantiomers in the chromatographic system and other, non-selective types of solute-sorbent interactions
are negligible. On other hand, a chiral mobile phase reduces the retention of the solute enantiomer which
forms a stronger association with the chiral selector. Here again, the limit for the enantioselectivity of the
chiral chromatographic system is set by the enantioselectivity of the selector-solute association (in the mobile
phase). However, in the majority of chiral mobile phase systems, the chiral selector as well as its associates
with the solute enantiomers are distributed between the mobile and stationary phases. The effective
enantioselectivity of the chromatographic system will therefore be proportional to the ratio of the enan-
tioselectivities of the association processes in the stationary and mobile phases [4].
Interaction of the chiral selector of the system with the enantiomers of the solute results in the formation of
two labile diastereomers. These differ in their thermodynamic stability, provided that at least three active
points of the selector participate in the interaction with corresponding sites of the solute molecule. This
three-point interaction rule is generally valid for enantioselective chromatography, with the extension to the
rule, stating that one of the required interactions may be mediated by the adsorption of the two components of
the interacting pair onto the sorbent surface [5].
Because of the multiplicity and complexity of the interactions of the enantiomers to be separated with the
chiral selector, sorbent surface and other components of the chromatographic system, the total enantioselec-
tivity can depend strongly on the composition, pH and temperature of the mobile phase. Therefore, in papers
on enantioselective chromatography, it is important to deRne these parameters.
Enantioselective chromatography and capillary electrophoresis are extensively employed in the analysis of
the enantiomeric composition (enantiomeric excess, optical purity) of chiral compounds. Liquid and super-
critical Suid chromatography are also used for the isolation of chiral compounds from racemic mixtures on
a preparative scale.
Enantioselective separations have been realised in all possible separation techniques, including gas chromato-
graphy, column liquid chromatography, thin-layer chromatography, supercritical Suid chromatography,
as well as electromigration methods, countercurrent liquid chromatography and liquid-liquid extractions.
 APPENDIX 4 / ANALYTICAL CHIRAL SEPARATION METHODS 4683

Numerous review papers and special monographs [6}15] describe the technical details as well as the
achievements and potential of these important modern separation techniques.
In the following glossary of deRnitions and terms related to the chromatographic and capillary-elec-
trophoretic separation of chiral compounds some terms (those marked with asterisks) were deRned in the Basic
Terminology of Stereochemistry, recently published by the IUPAC Joint Working Party on Stereochemical
Terminology [16]. Some of these deRnitions contain further cross references which are to be found in the
original paper.

Terms and De\nitions

General Terms Related to Chirality

HChirality The geometric property of a rigid object (or spatial arrangement of points or atoms) of being non-
superposable on its mirror image; such an object has no symmetry elements of the second kind (a mirror plane,
"S1, a centre of inversion, i"S2, a rotation-reSection axis, S2n). If the object is superposable on its mirror
image the object is described as being achiral. See also handedness.

Diastereoisomers (Diastereomers) see diatereoisomerism

Diastereoisomerism Stereoisomerism other than enantiomerism and cis-trans isomerism. Diastereoisomers
(or diastereomers) are stereoisomers not related as mirror images. Diastereoisomers are characterised by
differences in physical properties, and by differences in chemical behaviour towards achiral as well as
chiral reagents.

HEnantiomer One of a pair of molecular entities which are mirror images of each other and non-superpos-
able. See also enantiomorph.

HStereoisomers Isomers that possess identical constitution but which differ in the arrangement of their
atoms in space. See enantiomer, diastereomer, cis-trans-isomers.

Terms Related to the Separated Process

Chiral additive The chiral selector which has been added as a component of a mobile phase or elec-
trophoretic medium.

Chiral mobile phase A mobile phase containing a chiral selector.

Chiral selector The chiral component of the separation system capable of interacting enantioselectively with
the enantiomers to be separated.

Chiral stationary phase A stationary phase which incorporates a chiral selector. In not a constituent of the
stationary phase as a whole, the chiral selector can be chemically bonded to (chiral bonded stationary phase)
or immobilized onto the surface of a solid support or column wall (chiral coated stationary phase), or simply
dissolved in the liquid stationary phase.

Enantioselective chromatography (electrophoresis) The separation of enantiomeric species due to the enan-
tioselectivity of their interaction with the chiral selector(s) of a chromatographic (electrophoretic) system. Also
called Chiral chromatography (electrophoresis).

Enantioselective column A chromatographic column containing a chiral stationary phase. Also called
a chiral column.

Enantioselectivity (in chiral separations) The preferential interaction with the chiral selector of one enan-
tiomer over the other.
4684 APPENDIX 4 / ANALYTICAL CHIRAL SEPARATION METHODS

Enantioselectivity of a chromatographic (electrophoretic) system The ratio of the retention factors of two
solute enantiomers in a chiral chromatographic (electrophoretic) system.
Terms Related to the Chiral Purity of the Sample
HDiastereoisomer excess/Diastereoisomeric excess This is deRned by analogy with enantiomer excess, as
D1!D2 [and the percent diastereoisomer excess as 100(D1!D2)], where the mole fractions of the two
diastereoisomers in a mixture or the fractional yields of two diastereoisomers formed in a reaction are D1 and
D2 (D1#D2"1). The term is not applicable if more than two diastereoisomers are present. Frequently this
term is abbreviated to d.e. See stereoselectivity; diastereoisomerism.

HEnantiomer excess/Enantiomeric excess For a mixture of (#) and (!) enantiomers, with composition
given as the mole or weight fractions F(#) and F( ) (where F(#)#F( )"1) the enantiomeric excess is deRned
\ \
as F(#)!F( ) (and the percent enantiomer excess by 100F(#)!F( )). Frequently this term is abbreviated as
\ \
e.e. See optical purity.

HEnantiomeric purity see Enantiomer excess

HOptical purity The ratio of the observed optical rotation of a sample consisting of a mixture of enantiomers
to the optical rotation of one pure enantiomer. See enantiomeric excess.

References
1. S. F. Mason and G. E. Tranter, The electroweak origin of bimolecular handedness, Proc. R. Soc. London, A 397,
45}65 (1985).
2. D. Sybilska, A. Bielejewska, R. Nowakowski, K. Duszczyk and J. Jurczak, Improved chiral recognition of some
compounds via the simultaneous use of beta-cyclodextrin and its permethylated derivative in a reversed-phase
high-performance liquid chromatographic system, J. Chromatogr., 625, 349}352 (1992).
3. K. J. Duff, H. L. Gray, R. J. Gray and C. C. Bahler, Chiral stationary phases in concert with homologous chiral
mobile phase additives: Push/pull model, Chirality, 5, 201}206 (1993).
4. V. A. Davankov, A. A. Kurganov and T. M. Ponomareva, Enantioselectivity of complex formation in ligand-exchange
chromatographic systems with chiral stationary and/or chiral mobile phases, J. Chromatogr., 452, 309}316 (1988).
5. V. A. Davankov, V. R. Meyer and M. Rais, A vivid model illustrating chiral recognition induced by achiral structures,
Chirality, 2, 208}210 (1990).
6. A. M. Krstulovic, Editor, Chiral Separations by HPLC, Applications to Pharmaceutical Compounds, Ellis Horwood,
1989, 548 pp.
7. V. A. Davankov, A. A. Kurganov and A.S. Bochkov, Resolution of racemates by high-performance liquid chromatog-
raphy, Adv. Chromatogr., 22, 71}116 (1983).
8. P. Schreier, A. Bernreuther and M. Huffer, Analysis of Chiral Organic Molecules, Walter de Gruyter & Co., 1995,
331 pp.
9. D. W. Armstrong and S. M. Han, Enantiomeric Separations in Chromatography, CRC Critical Reviews in Analytical
Chemistry, 19, 175}224 (1988).
10. W. H. Pirkle and T. C. Pochapsky, Consideration of chiral recognition relevant to the liquid chromatographic
separation of enantiomers, Chem. Rev., 89, 347}362 (1989).
11. Chiral Separations by Liquid Chromatography (ACS Symposium Series, No. 471), ed. by S. Ahuja, American
Chemical Society, Washington, DC, 1991, 239 pp.
12. W. A. Koenig, Gas Chromatographic Enantiomer Separation with ModiTed Cyclodextrins, HuK thig, Heidelberg,
1992, 168 pp.
13. A Practical Approach to Chiral Separations by Liquid Chromatography, ed. by G. Subramanian, VCH, Weinheim
(FRG), 1994.
14. S. Allenmark, Chromatographic Enantioseparation, Ellis Horwood, New York, 2nd ed./1991.
15. E. Francotte, Contribution of preparative chromatographic resolution to the investigation of chiral phenomena,
J. Chromatogr. A, 666, 565}601 (1994).
16. G. P. Moss, Basic Terminology of Stereochemistry (IUPAC Recommendations 1996), Pure Appl. Chem., 68,
2193}2222 (1996).
APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology 4685

5. BIOLOGICAL BUFFERS

Table 1 This table of frequently used buffers gives the pKa value at 253C and the useful pH range of each buffer. The buffers are
listed in order of increasing pH

Acronym Name Mol. wt. pKa Useful

pH range

MES 2-(N-Morpholino)ethanesulphonic acid 195.2 6.1 5.5}6.7

BIS TRIS Bis(2-hydroxyethyl)iminotris(hydroxymethyl)methane 209.2 6.5 5.8}7.2
ADA N-(2-Acetamido)-2-iminodiacetic acid 190.2 6.6 6.0}7.2
ACES 2-[(2-Amino-2-oxoethyl)amino]ethanesulphonic acid 182.2 6.8 6.1}7.5
PIPES Piperazine-N,N -bis(2-ethanesulphonic acid) 302.4 6.8 6.1}7.5
MOPSO 3-(N-Morpholino)-2-hydroxypropanesulphonic acid 225.3 6.9 6.2}7.6
BIS TRIS PROPANE 1,3-Bis[tris(hydroxymethyl)methylamino]propane 282.3 6.8a 6.3}9.5
BES N,N-Bis(2-hydroxyethyl)-2-aminoethanesulphonic acid 213.2 7.1 6.4}7.8
MOPS 3-(N-Morpholino)propanesulphonic acid 209.3 7.2 6.5}7.9
HEPES N-(2-Hydroxyethyl)piperazine-N -(2-ethanesulphonic acid) 238.3 7.5 6.8}8.2
TES N-Tris(hydroxymethyl)methyl-2-aminoethanesulphonic acid 229.2 7.5 6.8}8.2
DIPSO 3-[N,N-Bis(2-hydroxyethyl)amino]-2-hydroxypropanesulphonic acid 243.3 7.6 7.0}8.2
TAPSO 3-[N-Tris(hydroxymethyl)methylamino)-2-hydroxypropanesulphonic acid 259.3 7.6 7.0}8.2
TRIZMA Tris(hydroxymethyl)aminomethane
HEPPSO N-(2-hydroxyethyl)piperazine-N -(2-hydroxypropanesulphonic acid) 121.1 8.1 7.0}9.1
POPSO Piperazine-N,N -bis(2-hydroxypropanesulphonic acid) 268.3 7.8 7.1}8.5
EPPS N-(2-Hydroxyethyl)piperazine-N -(3-propanesulphonic acid) 362.4 7.8 7.2}8.5
TEA Triethanolamine 252.3 8.0 7.3}8.7
TRICINE N-Tris(hydroxymethyl)methylglycine 149.2 7.8 7.3}8.3
BICINE N,N-Bis(2-hydroxyethyl)glycine 179.2 8.1 7.4}8.8
TAPS N-Tris(hydroxymethyl)methyl-3-aminopropanesulphonic acid 163.2 8.3 7.6}9.0
AMPSO 3-[(1,1-Dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulphonicacid 243.3 8.4 7.7}9.1
227.3 9.0 8.3}9.7
CHES 2-(N-Cyclohexylamino)ethanesulphonic acid 207.3 9.3 8.6}10.0
CAPSO 3-(Cyclohexylamino)-2-hydroxy-1-propanesulphonic acid 237.3 9.6 8.9}10.3
AMP 2-Amino-2-methyl-1-propanol 89.1 9.7 9.0}10.5
CAPS 3-(Cyclohexylamino)-1-propanesulphonic acid 221.3 10.4 9.7}11.1

a
pKa"9.0 for the second dissociation stage.
The table is reprinted with permission of Sigma Chemical Company, St. Louis, Mo.

6A. CLASSIFICATION AND CHARACTERIZATION

OF STATIONARY PHASES FOR
LIQUID CHROMATOGRAPHY
(IUPAC RECOMMENDATIONS 1997)

Descriptive Terminology
Prepared for publication by
R. M. Smith, Loughborough University, Loughborough, Leicestershire, UK
A. Marton, University of VeszpreH m, VeszpreH m, Hungary
^ 1997 IUPAC
APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology 4685

5. BIOLOGICAL BUFFERS

Table 1 This table of frequently used buffers gives the pKa value at 253C and the useful pH range of each buffer. The buffers are
listed in order of increasing pH

Acronym Name Mol. wt. pKa Useful

pH range

MES 2-(N-Morpholino)ethanesulphonic acid 195.2 6.1 5.5}6.7

BIS TRIS Bis(2-hydroxyethyl)iminotris(hydroxymethyl)methane 209.2 6.5 5.8}7.2
ADA N-(2-Acetamido)-2-iminodiacetic acid 190.2 6.6 6.0}7.2
ACES 2-[(2-Amino-2-oxoethyl)amino]ethanesulphonic acid 182.2 6.8 6.1}7.5
PIPES Piperazine-N,N -bis(2-ethanesulphonic acid) 302.4 6.8 6.1}7.5
MOPSO 3-(N-Morpholino)-2-hydroxypropanesulphonic acid 225.3 6.9 6.2}7.6
BIS TRIS PROPANE 1,3-Bis[tris(hydroxymethyl)methylamino]propane 282.3 6.8a 6.3}9.5
BES N,N-Bis(2-hydroxyethyl)-2-aminoethanesulphonic acid 213.2 7.1 6.4}7.8
MOPS 3-(N-Morpholino)propanesulphonic acid 209.3 7.2 6.5}7.9
HEPES N-(2-Hydroxyethyl)piperazine-N -(2-ethanesulphonic acid) 238.3 7.5 6.8}8.2
TES N-Tris(hydroxymethyl)methyl-2-aminoethanesulphonic acid 229.2 7.5 6.8}8.2
DIPSO 3-[N,N-Bis(2-hydroxyethyl)amino]-2-hydroxypropanesulphonic acid 243.3 7.6 7.0}8.2
TAPSO 3-[N-Tris(hydroxymethyl)methylamino)-2-hydroxypropanesulphonic acid 259.3 7.6 7.0}8.2
TRIZMA Tris(hydroxymethyl)aminomethane
HEPPSO N-(2-hydroxyethyl)piperazine-N -(2-hydroxypropanesulphonic acid) 121.1 8.1 7.0}9.1
POPSO Piperazine-N,N -bis(2-hydroxypropanesulphonic acid) 268.3 7.8 7.1}8.5
EPPS N-(2-Hydroxyethyl)piperazine-N -(3-propanesulphonic acid) 362.4 7.8 7.2}8.5
TEA Triethanolamine 252.3 8.0 7.3}8.7
TRICINE N-Tris(hydroxymethyl)methylglycine 149.2 7.8 7.3}8.3
BICINE N,N-Bis(2-hydroxyethyl)glycine 179.2 8.1 7.4}8.8
TAPS N-Tris(hydroxymethyl)methyl-3-aminopropanesulphonic acid 163.2 8.3 7.6}9.0
AMPSO 3-[(1,1-Dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulphonicacid 243.3 8.4 7.7}9.1
227.3 9.0 8.3}9.7
CHES 2-(N-Cyclohexylamino)ethanesulphonic acid 207.3 9.3 8.6}10.0
CAPSO 3-(Cyclohexylamino)-2-hydroxy-1-propanesulphonic acid 237.3 9.6 8.9}10.3
AMP 2-Amino-2-methyl-1-propanol 89.1 9.7 9.0}10.5
CAPS 3-(Cyclohexylamino)-1-propanesulphonic acid 221.3 10.4 9.7}11.1

a
pKa"9.0 for the second dissociation stage.
The table is reprinted with permission of Sigma Chemical Company, St. Louis, Mo.

6A. CLASSIFICATION AND CHARACTERIZATION

OF STATIONARY PHASES FOR
LIQUID CHROMATOGRAPHY
(IUPAC RECOMMENDATIONS 1997)

Descriptive Terminology
Prepared for publication by
R. M. Smith, Loughborough University, Loughborough, Leicestershire, UK
A. Marton, University of VeszpreH m, VeszpreH m, Hungary
^ 1997 IUPAC
4686 APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology

Abstract
A wide range of stationary phases and column packing materials have been developed over the years for liquid
chromatography and these need to be described accurately and unambiguously. The present paper, which is
the Rrst of a series planned for this area, recommends terms for the description of the stationary phase
materials and their properties and expands the list of terms given in Nomenclature of Chromatography [PAC,
1993, 65, 819}872.]. It concentrates on the chemical properties and chromatographic role of the materials.
Many of the terms to describe their physical properties as particles have been discussed in a recent paper on the
characterization of porous solids [PAC, 1994, 66, 1739}1758].

Index
Introduction
General descriptive terms for the stationary phase
Terms for the nature of the stationary phase material
Modes of application of stationary phase materials
Physical properties of the stationary phase materials
References
Appendix. Terms from the Nomenclature for Chromatography [1] which have been redeRned

Introduction
One of the major problems throughout analytical liquid (and supercritical Suid) chromatography has been in
reproducibly transferring methods between columns and systems. One of the main factors is that there
are large differences in the chemical and physical properties of stationary phase materials, even between
those which are nominally the same, such as octadecylsilyl (ODS)-bonded silicas. In addition, the methods
used to describe the physical and chemical properties of these stationary phase materials are not standardized
and even the terms used have not been agreed. In the laboratory, the behaviour of the stationary phases
is also dependent on the nature of the interaction of speciRc mobile phases and analytes with the stationary
phase.
The present discussions will concentrate on chromatographic applications of stationary phases. The
standardization of analytical stationary phase materials falls into two areas.

a) Terms needed to describe the physical and chemical properties of the stationary phase materials.
b) Methods and tests needed to describe the operational properties of these materials. This is an important
area but one where research is still active and as yet no consensus of approaches and techniques has been
reached. This topic still requires much experimental work and was considered inappropriate for the
Commission to consider at this time. It is also the subject of work by ASTM committees and others who are
in a better position to carry out experimental work and to organise comparative studies.

The present paper considers the Rrst of these areas and recommends a number of terms for the description of
stationary phase materials and their chemical and physical properties. Some descriptive terms for the
stationary phase have already been deRned in the Nomenclature of Chromatography (NC) [1] and in the
recently published Nomenclature for Analytical Chiral Separation methods (CS) [2]. In addition many of the
Recommendations for the Characterization of Porous Solids published by the Physical Chemistry Division [3]
are relevant to the physical description of stationary phase materials.
Amended and expanded deRnitions are now recommended for a number of the terms. The original versions
are included in the Appendix.

General Descriptive Terms for the Stationary Phase

A number of these terms are generally applicable throughout chromatography and were deRned previously in
the Nomenclature for Chromatography (NC) [1]. However, in a number of cases, an inappropriate capitaliza-
tion was used in NC, particularly of the second word in terms and these have been corrected in the deRnitions
reproduced here.
APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology 4687

Stationary Phase [Replaces NC 1.1.05]

One of the two phases forming a chromatographic system. It is the part of a chromatographic system
responsible for the retention of the analytes, which are being carried through the system by the mobile phase. It
may be a solid, a gel or a liquid. If a liquid, it may be distributed on a solid support. This solid support may or
may not contribute to the separation process. The liquid may also be chemically bonded to the solid (bonded
phase) or immobilized onto it (immobilized phase).
The expression chromatographic bed or sorbent may be used as a general term to denote any of the
different forms in which the stationary phase is used.
Note: Particularly in gas chromatography where the stationary phase is most often a liquid, the term liquid
phase is used for it as compared to the gas phase, i.e. the mobile phase. However, particularly in the early
development of liquid chromatography, the term liquid phase had also been used to characterize the mobile
phase as compared to the solid phase i.e. the stationary phase. Due to this ambiguity the use of the term
liquid phase is discouraged. If the physical state of the stationary phase is to be expressed, the use of the
adjective forms, such as liquid stationary phase and solid stationary phase, bonded stationary phase or
immobilized stationary phase, are recommended.

Packing Material, Stationary Phase Material [Replaces NC 3.1.07]

The packing is the active solid, stationary phase plus solid support or swollen gel which is contained in the
chromatographic column. In liquid chromatography the usage of the terms packing material and stationary
phase material are often synonymous. The term packing material is preferred as a general term for a loose,
usually particulate, material intended for chromatographic use before it is packed into the column. Once it is
packed and in contact with the mobile phase, it becomes the stationary phase as one of the two chromato-
graphic phases. The stationary phase usually consists of a speciRc stationary phase material, which has been
packed into a column. Both are typically given the same description.

Solid Support (NC 3.1.03)

A solid that holds the stationary phase but, ideally, does not contribute to the separation process.

Continuous Bed Packing

A column packing, which is a single entity, rather than being composed of individual particles.

Carbon Loading (of the Packing Material)

Mass fraction of the packing material which is carbon. Usually taken as a guide to the extent of alkyl
substitution on the surface. Usually reported as percentage carbon determined using elemental analysis.

Terms for the Nature of the Stationary Phase Material

Immobilized Stationary Phase (Material) [Replaces, NC 1.1.05.2]
A stationary phase which has been immobilized on the support particles, or on the inner wall of the column
tubing, e.g. by either a physical attraction (coated stationary phase), by chemical bonding (bonded stationary
phase), or by in situ polymerisation (cross-linked stationary phase) after coating.

Coated stationary phase (material) A material in which a stationary phase is immobilized by a physical
attraction to the surface of the solid support.

Filled stationary phase (material) An immobilized stationary phase (material) in which a liquid Rlls the pores
of the solid phase.

Bonded stationary phase (material) [Replaces NC 1.1.05.1] A stationary phase which is covalently bonded
to solid support particles or to the inside wall of the column tubing. Sometimes referred to as a bonded phase
(material). The bonded stationary phase (material) may be monomeric, polymeric or polymer-grafted phase
(material) and the stationary phase (material) can also receive additional treatment to give a capped
(end-capped) stationary phase (material).
4688 APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology

Bonded phase See Bonded stationary phase [Replaces NC 1.1.05.1]

Monomeric-bonded stationary phase (material) Bonded stationary phase (material) prepared using a re-
agent, usually monofunctional, which reacts with single sites on the surface of the solid support.

Polymeric-bonded stationary phase (material) Bonded stationary phase (material) prepared using a poly-
functional reagent which can react both with the surface of the solid support and/or with additional reagent
molecules.

Polymer-grafted stationary phase (material) Bonded stationary phase (material) in which a pre-formed
polymer has been bound to the surface by a chemical bond.

Capped stationary phase (material) (also known as end-capped stationary phase (material)) Bonded station-
ary phase (material) which has been treated with a second (usually less bulky) reagent, which is intended to
react with remaining functional (e.g. silanol) groups which have not been substituted by the original reagent
because of steric hindrance.

Alkyl-bonded stationary phase (material) Bonded stationary phase (material) in which the group bound to
the surface contains an alkyl chain (usually between C1 and C18).

Phenyl-bonded stationary phase (material) Bonded stationary phase (material) in which the group bound to
the surface contains a phenyl group.

Cyano-bonded stationary phase (material) Bonded stationary phase in which the group bound to the surface
contains a cyanoalkyl}(}[CH2]n}CN) group.

Diol-bonded stationary phase (material) Bonded stationary phase in which the group bound to the surface
contains a vicinal dihydroxyalkyl (}[CH2]n}CHOH}CH2OH) group.

Amino-bonded stationary phase (material) Bonded stationary phase in which the group bound to the surface
contains an aminoalkyl- (usually a }[CH2]n}NH2) group.

Internal surface reversed-phase (ISRP) materials) Bonded stationary phase in which the external surface of
the solid support carries different bonded groups from the internal pores (usually an external hydrophilic
layer with a more hydrophobic internal layer). Examples include restricted-access stationary phase (material)
in which polar macromolecules are excluded from the internal pores.

Cross-linked Stationary Phase (Material) A stationary phase (material) in which the liquid phase coating on
a solid support has been polymerized or cross-linked after coating to make it insoluble in the mobile phase.

Polymeric Stationary Phase (Material)

Stationary phase (material) based on particles of a cross-linked organic polymeric material. Typical materials
are polystyrene divinylbenzene copolymers (PS-DVB) and modiRed PS-DVB materials.

Liquid-coated Stationary Phase (Material)

A material in which a liquid stationary phase is coated on the surface of the solid support.

Modes of Application of Stationary Phase Materials

Stationary phases are often deRned in terms of the mode of chromatography being employed in the separation.

Size Exclusion Chromatographic Phases

These phases are described in Compendium of macromelecular nomenclature [4] (term 3.4.6). or in the
Nomenclature for Chromatography section 6 [1].
APPENDIX 6A / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Descriptive Terminology 4689

Ion-exchange Stationary Phases

The principle terms have already been deRned in NC (section 5) and are included here for comparison.

Cation exchanger (NC 5.302) Ion-exchanger with cations as counter-ions. The term cation-exchange resin
may be used in the case of solid organic polymers.

Anion exchanger (NC 5.3.03) Ion-exchanger with anions as counter-ions. The term anion-exchange resin
may be used in the case of solid organic polymers.
Chiral Stationary Phase (CS 2.4 [2])
A stationary phase which incorporates a chiral selector. If not constituent of the stationary phase as a whole,
the chiral selector can be chemically bonded to (chiral bonded stationary phase) or immobilized onto the
surface of a solid support or column wall (chiral coated stationary phase), or simply dissolved in the liquid
stationary phase.
Af\nity Stationary Phase (Material)
Bonded stationary phase (material) containing attached (adsorbed or covalently bonded) ligand molecules
with a speciRc biological interaction for a particular molecule or small group of related molecules.
Perfusion Stationary Phase (Material)
Stationary phase in which the mobile phase primarily travels through the pores of the stationary phase.

Physical Properties of the Stationary Phase Material

Recommendations for the characterization of porous solids have recently been published by the Commission
on Colloid and Surface Chemistry [3] and many of these are relevant to the characterization of stationary
phase materials. In particular the conclusions presented in that paper should be noted, especially, that in many
cases absolute values of the parameters such as pore diameter and surface area cannot be obtained. The
measured value frequently depends on the method of measurement (and this should always be stated) and the
selection of a method of characterization starts from the intended use of the material. These comments would
also apply to the determination of particle diameter. In addition, the calculation methods for average particle
diameters, such as number average or weight average, must be reported.
Previously De\ned General Terms
Particle diameter (dp) (NC 3.1.08) The average diameter of the solid particles.

Pore radius (rp) (NC 3.1.09) The average radius of the pores within the solid particles.

References
1. L. S. Ettre, Nomenclature for chromatography (IUPAC Recommendations 1993), Pure Appl. Chem., 1993, 65,
819}872.
2. V. A. Davankov, Analytical Chiral Separation Methods. (IUPAC Recommendations 1997), Pure Appl. Chem., 1997,
69, 1469}1474.
3. J. Rouquerol, D. Avnir, C. W. Fairbridge, D. H. Everett, J. H. Haynes, N. Pernicone, J. D. F. Ramsay, K. S. W. Sing and
K. K. Unger, Recommendations for the Characterization of Porous Solids, Pure Appl. Chem., 1994, 66, 1739}1758.
4. W. V. Metanomski, Compendium of Macromelecular Nomenclature, Blackwell, Oxford, 1991.

Appendix 1.
Terms from the Nomenclature for Chromatography [1] which have been RedeRned.
1.1.05. Stationary phase
One of the two phases forming a chromatographic system. It may be a solid, a gel or a liquid. If a liquid it may
be distributed on a solid. This solid may or may not contribute to the separation process. The liquid may also
be chemically bonded to the solid (bonded phase) or immobilized onto it (immobilized phase).
4690 APPENDIX 6B / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Ion Exchange

The expression chromatographic bed or sorbent may be used as a general term to denote any of the
different forms in which the stationary phase is used.
Note: Particularly in gas chromatography where the stationary phase is most often a liquid, the term liquid
phase is used for it as compared to the gas phase, i.e., the mobile phase. However, particularly in the early
development of liquid chromatography, the term liquid phase had also been used to characterize the mobile
phase as compared to the solid phase, i.e. the stationary phase. Due to this ambiguity the use of the term
liquid phase is discouraged. If the physical state of the stationary phase is to be expressed the use of the
adjective forms, such as liquid stationary phase and solid stationary phase, bonded phase or immobilized
phase, are recommended.

1.1.05.1. Bonded phase

A stationary phase which is covalently bonded to the support particles or to the inside wall of the column
tubing.

1.1.05.2. Immobilized phase

A stationary phase which immobilized on the support particles, or on the inner wall of the column tubing, e.g.
by in situ polymerization (cross-linking) after coating.

3.1.07. Packing
The active solid, stationary phase plus solid support or swollen gel contained in a tube.

6B. Characterization of Ion Exchange Chromatographic

Stationary Phases

Prepared for publication by

A. Marton, University of VeszpreH m, VeszpreH m, Hungary
^ 1997 IUPAC

Abstract
In order to characterize ion exchange chromatographic stationary phases the thermodynamic exchange
constant and the free energy interaction parameters are recommended. These parameters are calculated from
the experimentally available corrected selectivity coefRcient vs. exchanger phase composition functions.
The equations used for the calculations have been obtained by introducing the Friedman equation (developed
for the calculation of the excess free energy change) into the thermodynamic derivation. The suggested
parameters also make possible the estimation of the value of the selectivity coefRcient at an arbitrary
exchanger phase composition. The characteristic parameters of the ion exchange resins and the equations in
a directly suitable form for the estimation of the selectivity coefRcient are calculated and presented for
several systems.

Introduction
Parameters for the physical and chemical characterization of chromatographic stationary phases including ion
exchangers have already been deRned [1]. The purpose of this paper is to introduce sensitive numerical
parameters for the comparison of operation ion exchange chromatographic stationary phases based on their
selectivity coefRcient exhibited in a particular ion exchange equilibria. For two competing counter ions
(e.g. A# and Bz#) the problem arises not only because various selectivity coefRcients may be assigned to
the various commercially available products, but also because the exact value of the selectivity coefRcient
 APPENDIX 6B / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Ion Exchange 4691

may vary considerably with the degree of conversion of the exchanger phase as the ion exchange reaction (1)
proceeds.
z ) RA#Bz# RzB#z ) A# (1)

Here R represents the (usually monovalent) functional group covalently bond to a solid phase. The so-called
corrected selectivity coefRcient (K
) is an experimentally available parameter deRned for the above
equilibria as:
xN B ) azA
K
" z (2)
xN A ) aB

Throughout this article, symbols with overbars refer to the resin phase where the standard and reference
states of RA and RzB are taken to the respective mono-ionic forms of the exchanger in equilibrium with water.
Symbols without a bar refer to the solution phase where the Henryan standard and reference states are
accepted in accordance with conventional practice [2]. xN denotes mole or, if z'1, equivalent fraction (in
general xN i"zi ) mN i/ zi ) mN i and the summation is carried out over all counterion molalities mN i), a and the
parameter aN (see below) are the activities in the solution and resin phases respectively. For the ultimate
characterization and comparison of the selectivity of the exchange reactions the more exactly deRned
thermodynamic exchange constant, KT is recommended [2,3]:

aN B ) azA
KT" z (3)
aN A ) aB

An extensive compilation of the KT data for the various ion exchange equilibria has been made in an earlier
report of the IUPAC [4]. Since the ion exchange equilibrium constant is directly related to the distribution
coefRcient of the ion studied its knowledge in the calculation of the retention volume, or generally in the
design of ion exchange separations, is indispensable. Considering however, that in the majority of analytical
ion exchange separations practically either one or the other end to the mole fraction scale is utilized (xN A+1 or
xN B+1), the thermodynamic constant could be quite far from the actual (operational) value of the selectivity
coefRcient. The suggested characterization method is meant to provide a solution for these seemingly
conSicting aspects.
Using the concentrated electrolyte solution model of the ion exchange resins, equations were derived for the
composition dependence of the selectivity coefRcient (see equations on pages 104 and 105 of reference
[5]). From the experimentally available functions ln K
vs. xN B the derived relationships make possible both the
calculation of the thermodynamic exchange constant and the so-called free energy interaction parameters
which, in turn, could be used to calculate the selectivity coefRcient at any value of xN B. It was proved that
the free energy interaction parameters are related to the selectivity controlling properties of the ion exchanger
phase, such as the crosslinking of the polymer matrix, the type of the functional group and the size of the
exchanging counter ions [5]. The purpose of the suggested method is to characterize the ion exchangers with
these parameters in connection with their actual ion exchange reaction. It may also be considered as an
operational characterization which uses both the thermodynamic constant and the above mentioned free
energy interaction parameters to estimate the value of the corrected selectivity coefRcient at an arbitrary
exchanger phase composition.

Theoretical Background
When reaction (1) takes place a mixture of the concentrated electrolytes is always formed. The composi-
tion of this mixture (xN B) varies as the resin is converted from A# and Bz# form. According to H.L. Friedman
[6] the excess free energy change (GE) accompanying the formation of a two component electrolyte solution
mixture at constant ionic strength I (containing a common cation or anion) can be approximated by the
equation:
GE"R ) T ) I2 ) xA ) xB ) [g0#g1 ) (xA!xB)] (4)

Here xA and xB are the mole fractions of the components (e.g. RA and RzB) and g0 and g1 are the so called
free energy interaction parameters independent of the composition. These terms have been introduced by
4692 APPENDIX 6B / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Ion Exchange

Table 1 Summary of the equations used for the calculations

lnK
"a0#a1 ) xN B#a2 ) xN B2 (5)

a2(0 a2'0

a1#a2 a1#a2
gN 0" (6) gN 0" (9)
2)z 2)z

!a2 a2
gN 1" (7) gN 1" (10)
6)z 6)z

lnK T"a0#z ) gN 0#z ) gN 1 (8) lnK T"a0#z ) gN 0!z ) gN 1 (11)

Friedman to account for the strength of pair and triplet interactions respectively in a concentrated two
component electrolyte solution mixture. The existence of a similar mixture of concentrated electrolyte
solutions is supposed to be present in the exchanger phase too. It has been pointed out [5] that by introducing
eqn. (4) into the thermodynamic derivation the composition dependence of the selectivity coefRcient can
be expressed conveniently by these free energy interaction parameters. Equations suggested for the calculation
of the characteristic ion exchange parameters (gN 0, gN 1 and ln KT) were taken from reference [5] and are
summarized for our purpose in Table 1.
Although a direct, a priori, calculation of the free energy interaction parameters for the concentrated
electrolyte solution of the exchanger phase is still not feasible a detailed analysis of several literature data
proved that their value is dramatically inSuenced by the crosslinking of the inert (polymer) matrix, by the type
and density of the active group of the resin and by the type of the counter ion [5]. Consequently, these
parameters by themselves are characteristic for the ion exchange chromatographic stationary phase i.e.
a difference in their values for the two compared stationary phases indicate differences in relevant
structural parameters governing ion exchange selectivity.

Source of Experimental Data and Examples for the Suggested Ion Exchange
Stationary Phase Characterization
An important criteria for the application of the equations shown in Table 1 is that the exchange reaction should be
completely reversible, where the exchange capacity is freely accessible to the competing counter ions. The exchange
equilibrium should be studied at a constant temperature and ionic strength in the full range of mole fraction
scale and the corrected selectivity coefRcient, K
deRned by eqn. (2) should be calculated at each exchanger
phase composition. As an application of the above equations, we can consider the following experimental data
obtained by Bonner [7] for the Na#/H# exchange reaction on a strongly acidic Dowex 50;8 resin:
xN Na: 0.12 0.22 0.32 0.42 0.58 0.70 0.75 0.86
ln K
: 0.470 0.438 0.438 0.439 0.451 0.405 0.343 0.270
When eqn. (5) of Table 1 is Rtted to the above lnK
vs. xN Na data pairs then the following coefRcients are
obtained: a0"0.400, a1"0.398, a2"!0.624 (the curve Rtting program used for the calculation is given in
reference [8]). Since a2(0 eqns. (6, 7 and 8) of Table 1 can be used for the calculation of the characteristic
parameters of the studied exchange equilibria. The obtained values are: gN 0"!0.113, gN 1"0.104,
ln KT"0.39.
The function describing the dependence of ln K
on the exchanger phase composition can therefore be given
by the so-called selectivity polynomial:

ln K
"0.400#0.398 ) xN Na!0.624 ) xN 2Na (12)

If in the actual exchange process the estimated value of the stationary phase loading is e.g. 0.1 (or at the
other extreme end of the mole fraction scale is e.g. 0.9) then the calculated ln K
value is 0.433 (or 0.253)
 APPENDIX 6B / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Ion Exchange 4693

which are certainly more realistic values for the design of an ion exchange separation process than the value of
ln KT (0.39).
If the experimentally obtained (ln K
vs. xN B) function is concave i.e. a2'0 then eqns. (9, 10 and 11) of
Table 1 should be used for the calculation of the above parameters. It may also happen that the experimental
data Rts well with a straight line. In this case the above equations are also valid but, of course, now a2"0. The
choice between the linear or the quadratic Rtting procedures can be made by the comparison of the goodness
of Rt parameters. It is, in fact, automatically calculated by the referred curve Rtting program and the
improvement in the goodness of Rt can be seen immediately when the degree of the polynomial is changed (e.g.
from one to two).
In order to illustrate the wide scope of applicability of the suggested characterization method Tables 2 and 3
show the calculated values of the above discussed equilibrium parameters for a set of systems. The equilibria

Table 2 Free energy interaction parameters (gN 0 and gN 1) and the selectivity polynomial for some ion exchnage equilibria

B# DVB% gN 0 gN 1 ln K T ln K
Ref.
# #
RH#B  RB#H
Li# 4 0.020 0.061 0.263 0.304!0.327 xN B#0.368xN B2 9
8 !0.030 0.101 0.222 0.353!0.669 xN B#0.608xN B2 9
16 !0.242 0.142 0.353 0.737!1.338 xN B#0.854xN B2 9
Na# 4 0.01 0.088 0.139 0.041#0.548 xN B!0.528xN B2 9
8 !0.113 0.104 0.390 0.400#0.398 xN B!0.624xN B2 9
16 !0.423 0.215 0.445 0.653#0.444 xN B!1.290xN B2 9
K# 4 !0.612 0.102 0.474 0.534#0.289 xN B!0.613xN B2 9
8 !0.314 0.144 0.729 0.899#0.235 xN B!0.863xN B2 9
16 !1.039 0.106 1.019 1.952!1.439 xN B!0.640xN B2 9
Rb# 4 0.354 0.073 0.494 0.775!0.270 xN B!0.439xN B2 10
8 !0.607 0.035 0.856 1.458!1.423 xN B#0.209xN B2 10
16 !1.196 0.241 1.059 2.014!0.946 xN B!1.446xN B2 10
Cs# 4 !0.398 0.083 0.587 0.893!0.282 xN B!0.496xN B2 10
8 !0.891 0.006 0.832 1.717!1.746 xN B!0.035xN B2 10
16 !1.487 0.093 1.082 2.476!2.417 xN B!0.557xN B2 10
NH#
4 4 !0.156 * 0.303 0.460!0.313 xN B 9
8 !0.333 * 0.576 0.909!0.667 xN B 9
16 !0.647 * !0.763 1.411!1.295 xN B 9
N(Me)# 4 7 !0.649 * 0.081 0.730!1.298 xN B 11
N(Et)#
4 7 !4.245 * !0.489 3.756!8.491 xN B 11
#
N(Pr)4 7 !5.565 * !0.870 4.695!11.130 xN B 11
N(Bu)#4 7 !9.371 * !0.785 8.586!18.742 xN B 11
RNa#B\  RB#Na#
#
Cs 8 !0.094 0.003 0.813 0.445!0.173 xN B!0.016xN B2 12
Ag# 8 !0.184 0.033 1.308 0.720!0.173 xN B!0.196xN B2 12
Tl# 8 !0.154 0.045 1.660 0.810!0.081 xN B!0.389xN B2 12
RCl#B\  RB#Cl\
Br\ 2 !0.048 * 0.896 0.945!0.097 xN B 13
4 !0.085 * 1.028 1.114!0.171 xN B 13
8 !0.123 * 1.118 1.311!0.245 xN B 13
10 !0.214 * 1.411 1.625!0.427 xN B 13
I\ 2 !0.087 * 1.219 1.327!0.278 xN B#0.213xN B2 14
4 !0.121 * 1.487 1.567!0.004 xN B!0.246xN B2 14
8 !0.197 * 2.297 2.549!0.724 xN B#0.329xN B2 14
10 !0.287 * 2.946 3.143!0.029 xN B!0.546xN B2 14
NO\
3 2 !0.084 0 0.643 0.727!0.168 xN B 14
4 !0.061 0 0.841 0.903!0.123 xN B 14
8 !0.149 0 1.171 1.321!0.299 xN B 14
10 !0.169 0 1.456 1.653!0.393 xN B 14
ClO\4 8 0.698 0.409 3.924 5.033!3.855 xN B#2.458xN B2 15
ClO\3 8 0.195 0 0.891 0.694#0.393 xN B 15
BrO\3 8 !0.077 0 0.354 0.431#0.154 xN B 15
IO\
3 8 !0.093 0.054 !1.272 !1.233#0.136 xN B!0.323xN B2 15
HCO\ 3 8 !0.073 0.100 !1.010 !0.840!0.751 xN B#0.605xN B2 15
OH\ 8 !0.315 0.118 !2.371 !2.174#0.077 xN B!0.708xN B2 15
SCN\ 8 !0.481 0.142 3.29 3.91!1.816 xN B#0.854xN B2 15
4694 APPENDIX 6B / CHARACTERIZATION OF STATIONARY PHASES FOR LC / Ion Exchange

Table 3 Free energy interaction parameters (gN 0 and gN 1) and the selectivity polynomial for some anion exchange equilibria

B2\ F.G. gN 0 gN 1 ln K T ln K
Ref.

2RCl#B2\  R2B#2Cl\
Ox2\ TMA# 0.187 1.012 !2.290 !0.644!11.4 xN B#12.15xN B2 16
Ma2\ 0.460 0.565 !3.048 !2.838!4.942 xN B#6.782xN B2 16
Su2\ 0.288 0.017 !3.760 !4.303#0.945 xN B#0.207xN B2 16
Gl2\ 0.07 0.44 !4.050 5.07#5.00 xN B!5.28xN B2 16
Ad2\ !0.562 0.956 !4.975 !5.763#9.233 xN B!11.47xN B2 16
Pi2\ !0.922 0.102 !5.384 !5.58#8.590 xN B!12.28xN B2 16
Ox2\ !2.09 1.52 !7.10 !5.96#9.90 xN B!18.29xN B2 16
Ma2\ !0.845 0.81 !5.46 5.39#6.30 xN B!9.68xN B2 16
Su2\ #0.097 0.17 !4.05 !4.59#2.44 xN B!2.05xN B2 16
Gl2\ TEA# #0.387 0.32 !3.29 !3.43!2.29 xN B#3.89xN B2 16
Ad2\ #0.006 0.07 !4.25 !4.42!0.95 xN B!0.926xN B2 16
Pi2\ !0.515 0.438 !4.96 !4.81#3.20 xN B!5.26xN B2 16

B2\:Ox2\"Oxalic, Ma2"Malonic, Su2"Succinic, Gl2\"Glutaric, Ad2\"Adipic, Pi2\"Heptanedioic (Pimelic) acid anion F.G.:
functionalities of the Amberlite resin, tetramethyl and tetraethylammonium groups (TMA#, TEA#).

selected, mostly from the classics of the ion exchange literature are meant to represent both inorganic and
organic cation and anion exchange reactions, where the crosslinking of the polymer matrix, the size of the
active group and the size of the counter ion varies considerably.

Conclusion
The nonideal behaviour of the exchanger phase is recognized to be highly characteristic for the ion exchanger
as a chromatographic stationary phase. As a quantitative measure of this nonideality the free energy
interaction parameters are calculated from the data of equilibrium measurements. These data are then applied
to construct the so-called selectivity polynomial which, in turn, can be used to estimate the selectivity
coefRcient at any required composition of the exchanger phase. Beyond the highly speciRc, numerically
sensitive feature of these parameters their recommendation for the characterization of ion exchange chromato-
graphic stationary phases is further supported by their connection with thermodynamic equilibrium constant
of the exchange reaction.

References
1. R. M. Smith, A. Marton, ClassiRcation and characterization of stationary phases for liquid chromatography Part I.
Descriptive Terminology, Pure and Appl. Chem. (under publication).
2. F. Helfferich, Ion Exchange, McGraw Hill, New York, 1962, p.95.
3. H. M. N. H. Irving, Recommendations on Ion Exchange Nomenclature, Pure and Appl. Chem., 29, 619}623 (1972).
4. Y. Marcus, D. H. Howery, Ion Exchange Equilibrium Constants, IUPAC Commission V/6 Equilibrium Data, (1972).
5. A. Marton, J. InczeH dy, Application of the concentrated electrolyte solution model in the evaluation of ion exchange
equilibria Reaction Polymers, 7, 101}109 (1988).
6. H. L. Friedman, Ionic Solution Theory, Interscience, New York, 1962, p.225.
7. O. D. Bonner, A selectivity scale for some monovalent cations on Dowex 50, J. Phys. Chem., 58, 318}320 (1954).
8. J. D. Lee, T. D. Lee, Statistics and Computer Methods in BASIC, Van Nostrand Reinhold, Wokingham, 1982, p.121.
9. O. D. Bonner, A selectivity scale for some monovalent cations on Dowex 50, J. Phys. Chem., 58, 318}320 (1954).
10. O. D. Bonner, Ion exchange equilibria involving rubidium, cezium and thallous ions, J. Phys. Chem., 59, 719}721
(1955).
11. J. R. Millar, D. G. Smith, W. E. Marr, T. R. E. Kressman, Solvent modiRed polymer networks. Part III, J. Chem. Soc.,
2740}2746 (1964).
12. A. JaH sz, T. Lengyel, Investigation of binary ion exchange equilibria with radioisotopes (in Hungarian), Magy. Ke& m.
Foly., 67, 351}369 (1961).
13. B. Soldano, D. Chesnut, Osmotic approach to ion exchange equilibrium, J. Am. Chem. Soc., 77, 1334}1339 (1955).
14. H. P. Gregor, G. J. Belle, R. A. Marcus, Studies on ion exchange resins XIII, J. Am. Chem. Soc., 77, 2713}2719 (1955).
15. A. Marton, Relation of the free energy interaction parameters to some structural properties of ion exchange resins,
Talanta, 41, 1127}1132 (1994).
16. S. Subramonian, D. Clifford, Monovalent/divalent selectivity and the charge separation concept, Reactive
Polymers, 9, 195}209 (1988).
 APPENDIX 7 / CONVERSION OF UNITS 4695

7. CONVERSION OF UNITS
The table below gives conversion factors from a variety of units to the corresponding SI unit. For each
physical quantity the name is given, followed by the recommended symbol(s). Then the SI unit is given,
followed by the esu, emu, Gaussian unit (Gau), atomic unit (au), and other units in common use, with their
conversion factors to SI. The constant
which occurs in some of the electromagnetic conversion factors is the
(exact) pure number 2.997 924 58 ;1010"c0/(cms\1).
The inclusion of non-SI units in this table should not be taken to imply that their use is to be encouraged.
With some exceptions, SI units are always to be preferred to non-SI units. However, since many of the units
below are to be found in the scientiRc literature, it is convenient to tabulate their relation to the SI.
For convenience units in the esu and Gaussian systems are quoted in terms of the four dimensions length,
mass, time, and electric charge, by including the franklin (Fr) as an abbreviation for the electrostatic unit of
charge and 40 as a constant with dimensions (charge)2/(energy;length). This gives each physical quantity
the same dimensions in all systems, so that all conversion factors are pure numbers. The factors 440 and
the Fr may be eliminated by writing Fr"esu of charge"erg1/2cm1/2"cm3/2g1/2s\1, 40"(ir) 0 "1
Fr2 erg\1 cm\1"1, to recover esu expressions in terms of three base units. The symbol Fr should be regarded
as a compact representation of (esu of charge).
Conversion factors are either given exactly (when the"sign is used), or they are given to the approximation
that the corresponding physical constants are known (when the +sign is used). In the latter case the
uncertainty is always less than $5 in the last digit quoted.

Name Symbol Relation to SI

Length, l
metre (SI unit) m
centimetre (cgs unit) cm "10\2 m
bohr (au) a0, b "40
2/mee2+5.291 77;10\11 m
a ngstroK m A> "10\10 m
micron  "m"10\6 m
x unit X +1.002;10\13 m
fermi f, fm "fm"10\15 m
inch in "2.54;10\2 m
foot ft "12 in"0.3048 m
yard yd "3 ft"0.9144 m
mile mi "1760 yd"1609.344 m
nautical mile "1852 m

Area, A
square metre (SI unit) m2
barn b "10\28 m2
acre +4046.856 m2
are a "100 m2
hectare ha "104 m2

Volume,V
cubic metre (SI unit) m3
litre l, L "dm3"10\3 m3
lambda  "
l"10\6 dm3
barrel (US) +158.987 dm3
gallon (US) gal (US) "3.785 41 dm3
gallon (UK) gal (UK) "4.546 09 dm3
4696 APPENDIX 7 / CONVERSION OF UNITS

Name Symbol Relation to SI

Mass, m
kilogram (SI unit) kg
gram (cgs unit) g "10\3 kg
electron mass (au) me +9.109 39;10\31 kg
uniRed atomic mass unit, dalton u, Da "ma(12C)/12+1.660 540;10\27 kg
tonne t "Mg"103 kg
pound (avoirdupois) lb "0.453 592 37 kg
ounce (avoirdupois) oz +28.3495 g
ounce (troy) oz (troy) +31.1035 g
agrain gr "64.798 91 mg

Time, t
second (SI, cgs unit) s
au of time
/Eh +2.41888;10\17 s
minute min "60 s
hour h "3600 s
daya d "86 400 s
yearb a +31 556 952 s
svedberg Sv "10\13 s

Acceleration, a
SI unit m s\2 "9.806 65 m s\2
standard accleration of free fall gn
gal, galileo Gal "10\2 m s\2

Force, F
newton (SI unit)c N "kg m s\2
dyne (cgs unit) dyn "g cm s\2"10\5 N
au of force Eh/a0 +8.238 73;10\8 N
kilogram-force kgf "9.806 65 N

Energy, U
joule (SI unit) J "kg m2 s\2
erg (cgs unit) erg "g cm2 s\2"10\7 J
hartree (au) Eh "
2/mea20+4.359 75;10\18 J
rydberg Ry "Eh/2+2.179 87;10\18 J
electronvolt eV "e;V+1.602 18;10\19 J
calorie, thermochemical calth "4.184 J
calorie, international calIT "4.1868 J
153C calorie cal15 +4.1855 J
litre atmosphere 1 atm "101.325 J
British thermal unit Btu "1055.06 J

Pressure, p
pascal (SI unit) Pa "N m\2"kg m\1 s\2
atmosphere atm "101 325 Pa
bar bar "105 Pa
torr Torr "(101 325/760) Pa+133.322 Pa
millimetre of mercury (conventional) mmHg "13.5951;980.665;10\2Pa+133.322 Pa
pounds per square inch psi +6.894 757; 103 Pa
 APPENDIX 7 / CONVERSION OF UNITS 4697

Name Symbol Relation to SI

Power, P
watt (SI unit) W "kg m2 s\3
horse power hp "745.7 W

Action, L, J (angular momentum)

SI unit Js "kg m2 s\1
cgs unit erg s "10\7J s
au of action
"h/2+1.054;10\34 J s

Dynamic viscosity, 
SI unit Pa s "kg m\1 s\1
poise P "10\1 Pa s
centipoise cP "mPa s

Kinematic viscosicty, 
SI unit m2 s\1 "10\4m2 s\1
stokes St

Thermodynamic temperature, T
kelvin (SI unit) K
degree Rankined 3R "(5/9) K

Entropy, S
Heat capacity, C
SI unit J K\1
clausius Cl "calth/K"4.184 J K\1

Molar entropy, Sm
Molar heat capacity, Cm
SI unit J K\1 mol\1
entropy unit e.u. "calth K\1mol\1"4.184 J K\1mol\1

Molar volume, Vm
SI unit m3 mol\1
amagat5 amagat "Vm of real gas at 1 atm and 273.15 K
+22.4;10\3 m3 mol\1

Amount density, 1/Vm

SI unit mol m\3
amagate amagat "1/Vm of a real gas at 1 atm and 273.15 K
+44.6 mol m\3

Plane angle, 
radian (SI unit) rad
degree 3 "rad;2/360+(1/57.295 78) rad
minute "degree/60
second
"degree/3600
grade grad "rad;2/400+(1/63.661 98) rad
4698 APPENDIX 7 / CONVERSION OF UNITS

Name Symbol Relation to SI

Radioactivity, A
becquerel (SI unit) Bq "s\1
curie Ci "3.7;1010 Bq

Absorbed dose of radiationf

gray (SI unit) Gy "J kg\1
rad rad "0.01 Gy

Dose equivalent
sievert (SI unit) Sv "J kg\1
rem rem +0.01 Sv

Electric current, I
ampere (SI unit) A
esu, Gau (10/)A +3.335 64;10\10 A
biot (emu) Bi "10 A
au eEh/
+6.623 62;10\3 A

Electric charge, Q
coulomb (SI unit)
franklin (esu, Gau)
emu (abcoulomb)
proton charge (au)
C "A s
Fr "(10/)C+3.335 64;10\10 C
"10 C
e +1.602 18;10\19 C+4.803 21;10\10 Fr

Charge density, 
SI unit C m\3
esu, Gau Fr cm\3 "107 \1C m\3+3.33564;10\4 C m\3
au ea\
o
3
+1.081 20;10\12 C m\3

Electric potential, V,

volt (SI unit) V "JC\1"J A\1 s\1
esu, Gau erg Fr\1 "Fr cm\1/40"299.792 458 V
cm\1g e cm\1/40 +1.439 97;10\7 V
au e/40a0 "Eh/e+27.2114 V
mean international volt "1.000 34 V
US international volt "1.000 330 V

Electric resistance, R
ohm (SI unit)  "V A\1"m2 kg s\3 A\2
mean international ohm "1.1000 49 
US international ohm "1.000 495 

Electric Teld, E
SI unit V m\1 "J C\1 m\1
esu, Gau Fr cm\2/40 "2.997 924 58;104 V m\1
cm\2g e cm\2/40 +1.439 97;10\5 V m\1
au e/40a20 "5.142 21;1011 V m\1

Electric Teld gradient, E , q

?@ ?@
SI unit V m\2 "J C\1 m\2
esu, Gau Fr cm\3 /40 "2.997 924 58;106 V m\2
 APPENDIX 7 / CONVERSION OF UNITS 4699

Name Symbol Relation to SI

cm\3g e cm\3/40 +1.439 97;10\3 V m\2

au e/40a30 +9.717 36;1021 V m\2

Electric dipole moment, p, 

SI unit Cm
esu, Gau Fr cm +3.335 64;10\12 C m
debye D "10\18 Fr cm+3.335 64;10\30 C m
cm dipole lengthg e cm +1.602 18;10\21 C m
au ea0 +8.478 36;10\30 C m

Electric quadrupole moment,

Q ,  , eQ
?@ ?@
SI unit C m2
esu, Gau Fr cm2 +3.335 64;10\14 C m\2
cm2, e cm2 +1.602 18;10\23 C m2
quadrupole areag
au ea20 +4.486 55;10\40 C m2

Polarizability, 
SI unit J\1 C2 m2 "F m2
esu, Gau, cm3 40 cm3 +1.112 65;10\16 J\1 C2 m2
polarizability volumeg
A> 3g 40 A> 3 +1.112 65;10\40 J\1 C2 m2
au 40a30 +1.648 78;10\41 J\1 C2 m2

Electric displacement, D
(Volume) polarization, P
SI unit C m\2
esu, Gau Fr cm\2 "(105/)C m\2+3.33564;10\6C m\2
(But note: the use of the esu or Gaussian unit for electric displacment usually implies that the irrational
displacement is being quoted, D(ir)"4D.)

Magnetic Uux density, B

(magnetic Teld)
tesla (SI unit) T "J A\1 m\2"V s m\2"Wb m\2
gauss (emu, Gau) G "10\4 T
au
/ea20 +2.350 52;105 T

Magnetic Uux,

weber (SI unit) Wb "J A\1"V s
maxwell (emu, Gau) Mx "G cm\2"10\8 Wb

Magnetic Teld, H
(Volume) magnetization, M
SI unit A m\1 "C s\1 m\1
oersted (emu, Gau) Oe "103 A m\1
(But note: in practice the oersted, Oe, is only used as a unit for H(ir)"4H; thus when H(ir)"1 Oe,
H"(103/4) A m\1.)
4700 APPENDIX 7 / CONVERSION OF UNITS

Name Symbol Relation to SI

Magnetic dipole moment, m, 

SI unit A m2 "J T\1
emu, Gau erg G\1 "10 A cm2"10\3 J T\1
Bohr magnetonh B "e
/2me+9.274 02;10\24 J T\1
au e
/me "2B+1.854 80;10\23 J T\1
nuclear magneton N "(me/mp)B+5.050 79;10\27 J T\1
Magnetizability,
SI unit J T\2 "C2 m2 kg\1
au e2a20/me +7.891 04;10\29 J T\2
Magnetic susceptibility, , 
SI unit 1
emu, Gau 1
(But note: in practice susceptibilities quoted in the context of emu or Gaussian units are always values for
(ir)"/4; thus when (ir)"10\6, "4;106)
Molar magnetic susceptibility, m
SI unit m3 mol\1
emu, Gau cm3 mol\1 "10\6 cm3 mol\1
(But note: in practice the units cm3 mol\1 usually imply that the irrational molar susceptibility is being
quoted, (ir)
m "m/4; for example if m "!15;10\ cm mol\ , which is often written as
(ir) 6 3 1

!15 cgs ppm, then m"!1.88;10\ m mol\ .10 3 1

a
Note that the day is not exactly deRned in terms of the second since so-called leap-seconds are added or subtracted from
the day semiannually in order to keep the annual average occurrence of midnight at 24:00 on the clock.
b
The year is not commensurable with the date and not a constant. Prior to 1967, when the atomic standard was introduced,
the tropical year 1900 served as the basis for the deRnition of the second. For the epoch 1900.0. it amounted to
365.242 198 79 d+31 556 925.975 s and it decreases by 0.530 seconds per century. The calendar years are exactly
deRned in terms of the day:
Julian year"365.25 d
Gregorian year"365.2425 d.
The deRnition in the table corresponds to the Gregorian year. This is an average based on a year of length 365 days, with
leap years of 366 days; leap years are taken either when the year is divisible by 4 but is not divisible by 100, or when the
year is divisible by 400.
c
1 N is approximately the force exerted by the earth upon an apple.
d
T/3R"(9/5)T/K. Also, Celsius temperature  is related to thermodynamic temperature T by the equation:

/3C"T/K!273.15

Similarly Fahreheit temperature F is related to Celsius temperature  by the equation:

F/3F"(9/5)(/3C)#32

e
The name amagat is unfortunately used as a unit for both molar volume and amount density. Its value is slightly
different for different gases, reSecting the deviation from ideal behaviour for the gas being considered.
f
The unit roK ntgen, employed to express exposure to X or  radiations, is equal to: R"2.58;10\4 C kg\1.
g
The units in quotation marks for electric potential through polarizability may be found in the literature, although they are
strictly incorrect; they should be replaced in each case by the units given in the symbol column. Thus, for example, when
a quadrupole moment is quoted in cm2, the correct unit is e cm2; and when a polarizability is quoted in A> 3, the correct
unit is 40 A> 3.
h
The Bohr magneton B is sometimes denoted BM (or B.M.), but this is not recommended.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd edn.
Oxford: Blackwell ScientiRc Publications.)
 APPENDIX 8 / DEFINITIONS AND SYMBOLS FOR UNITS 4701

8. DEFINTIONS AND SYMBOLS FOR UNITS

The International System of Units (SI)

The International System of units (SI) was adopted by the 11th General Conference on Weights and Measures
(CGPM) in 1960. It is a coherent system of units built from seven SI base units, one for each of the seven
dimensionally independent base quantities: they are the metre, kilogram, second, ampere, kelvin, mole, and
candela, for the dimensions length, mass, time, electric current, thermodynamic temperature, amount of
substance, and luminous intensity, respectively. The SI derived units are expressed as products of powers of the
base units, analogous to the corresponding relations between physical quantities but with numerical factors
equal to unity.
In the International System there is only one SI unit for each physical quantity. This is either the appropriate
SI base unit itself or the appropriate SI derived unit. However, any of the approved decimal preRxes, called SI
preTxes, may be used to construct decimal multiples or submultiples of SI units.
It is recommended that only SI units be used in science and technology (with SI preRxes where appropriate).
Where there are special reasons for making an exception to this rule, it is recommended always to deRne the
units used in terms of SI units.

De\nitions of the SI Base Units

Metre: The metre is the length of path travelled by light in vaccum during a time interval of 1/299 792 458 of
a second (17th CGPM, 1983).

Kilouram: The kilogram is the unit of mass; it is equal to the mass of the international prototype of the
kilogram (3rd CGPM, 1901).

Second: The second is the duration of 9 192 631 770 periods of the radiation corresponding to the transition
between the two hyperRne levels of the ground state of the caesium-133 atom (13th CGPM, 1967).

Ampere: The ampere is that constant current which, if maintained in two straight parallel conductors of
inRnite length, of negligible ciruclar cross-section, and placed 1 metre apart in vacuum, would produce
between these conductors a force equal to 2;10\7 newton per metre of length (9th CGPM, 1948).

Kel*in: The kelvin, unit of thermodynamic temperature, is the fraction 1/273.16 of the thermodynamic
temperature of the triple point of water (13th CGPM, 1967).

Mole: The mole is the amount of substance of a system which contains as many elementary entities as there are
atoms in 0.012 kilogram of carbon-12. When the mole is used, the elementary entities must be speciRed and
may be atoms, molecules, ions, electrons, other particles, or speciRed groups of such particles (14th CGPM,
1971).

Examples of the use of the mole

1 mol of H2 contains about 6.022;1023 H2 molecules, or 12.044;1023 H atoms

1 mol of HgCl has a mass of 236.04 g
1 mol of Hg2Cl2 has a mass of 472.08 g
1 mol of Hg2#
2 has a mass of 401.18 g and a charge of 192.97 kC
1 mol of Fe0.91S has a mass of 82.88 g
1 mol of e\ has a mass of 548.60
m and a charge of !96.49 kC
1 mol of photons whose frequency is 5;1014 Hz has energy of about 199.5 kJ

Candela: The candela is the luminous intensity, in a given direction, of a source that emits monochromatic
radiation of frequency 540;1012 hertz and that has a radiant intensity in that direction of (1/683) watt per
steradian (16th CGPM, 1979).
4702 APPENDIX 8 / DEFINITIONS AND SYMBOLS FOR UNITS

Names and Symbols for the SI Base Units

The symbols listed here are internationally agreed and should not be changed in other languages or scripts.
Physical quantity Name of SI unit Symbol for SI unit
Length metre m
Mass kilogram kg
Time second s
Electric current ampere A
Thermodynamic temperature kelvin K
Amount of substance mole mol
Luminous intensity candela cd
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd
edn. Oxford: Blackwell ScientiRc Publications.)

SI Derived Units with Special Names and Symbols

Physical quantity Name of Symbol for Expression in terms of

SI unit SI unit SI base units
Frequencya hertz Hz s\1
Force newton N m kg s\2
Pressure, stress pascal Pa N m\2 "m\1 kg s\2
Energy, work, heat joule J Nm "m2 kg s\2
Power, radiant Sux watt W J s\1 "m2 kg s\3
Electric charge coulomb C As
Electric potential, electromotive force volt V J C\1 "m2 kg s\3A\1
Electric resistance ohm
V A\1 "m2 kg s\3A\2
Electric conductance siemens S
\1 "m\2 kg\1 s3 A2
Electric capacitance farad F C V\1 "m\2 kg\1 s4 A2
Magnetic Sux density tesla T V s m\2 "kg s\2 A\1
Magnetic Sux weber Wb Vs "m2 kg s\2 A\1
Inductance henry H V A\1 s "m2 kg s\2 A\2
Celsius temperatureb degree Celsius 3C K
Luminous Sux lumen lm cd sr
Illuminance lux lx cd sr m\2
Activitityc becquerel Bq s\1
(radioactive)
Absorbed dosec gray Gy J kg\1 "m2 s\2
(of radiation)
Dose equivalentc sievert Sv J kg\1 "m2 s\2
(dose equivalent index)
Plane angled radian rad 1 "m m\1
Solid angled steradian sr 1 "m2 m\2
a
For radial (angular) frequency and for angular velocity the unit rad s\1, or simply s\1, should be used, and this
may not be simpliRed to Hz. The unit Hz should be used only for frequency in the sense of cycles per second.
b
The Celsius temperature  is deRned by the equation /3C"T/K!273.15.
The SI unit of Celsius temperature is the degree Celsius, 3C, which is euqal to the kelvin, K. 3C should be treated
as a single symbol,with no space between the 3 sign and the letter C. (The symbol 3K, and the symbol 3, should no
longer be used).
c
The units becquerel, gray and sievert are admitted for reasons of safeguarding human health.
d
The units radian and steradian are described as SI supplementary units. However, in chemistry, as well as in
physics, they are usually treated as dimensionless derived units, and this was recognized by CIPM in 1980. Since
they are then of dimension 1, this leaves open the possiblity of including them or omitting them in expressions of
SI derived units. In practice this means that rad and sr may be used when appropriate and may be omitted if
clarity is not lost thereby.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd
edn. Oxford: Blackwell ScientiRc Publications.)
 APPENDIX 8 / DEFINITIONS AND SYMBOLS FOR UNITS 4703

SI Derived Units for Other Quantities

This table gives examples of other SI derived units; the list is merely illustrative.

Physical quantity Expression in terms of SI base units

Area m2
volume m3
Speed, velocity m s\1
Angular velocity s\1, rad s\1
Acceleration m s\2
Moment of force Nm "m2 kg s\2
Wavenumber m\1
Density, mass density kg m\3
SpeciRc volume m3 kg\1
Amount concentraiona mol m\3
Molar volume m3 mol\1
Heat capacity, entropy J K\1 "m2 kg s\2 K\1
Molar heat capacity, molar entropy J K\1 mol\1 "m2 kg s\2 K\1 mol\1
SpeciRc heat capacity, speciRc entropy J K\1 kg\1 "m2s\2 K\1
Molar energy J mol\1 "m2 kg s\2 mol\1
SpeciRc energy J Kg\1 "m2s\2
Energy density J m\3 "m\1 kg s\2
Surface tension N m\1"J m\2 "kg s\2
Heat Sux density, irradiance W m\2 "kg s\3
Thermal conductivity W m\1 K\1 "m kg s\3 K\1
Kinematic viscosity, diffusion coefRcient m2 s\1
Dynamic viscosity N s m\2"Pa s "m\1 kg s\1
Electric charge density C m\3 "m\3 s A
Electric current density A m\2
Conductivity S m\1 "m\3 kg\1 s3 A2
Molar conductivity S m2 mol\1 "kg\1 mol\1s3 A2
Permittivity F m\1 "m\3 kg\1 s\4 A2
Permeability H m\1 "m kg s\2 A\2
Electric Reld strength V m\1 "m kg s\3 A\1
Magnetic Reld strength A m\1
Luminance cd m\2
Exposure (X and  rays) C kg\1 "kg\1 s A
Absorbed dose rate Gy s\1 "m2 s\3
a
The words amount concentration are an abbreviation for amount-of-substance concentration. When there
is not likely to be any ambiguity this quantity may be called simply concentration.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

SI Pre\xes
To signify decimal multiples and submultiples of SI units the following preRxes may be used.

Submultiple PreRx Symbol Multiple PreRx Symbol

10\1 deci d 10 deca da

10\2 centi c 102 hecto h
10\3 milli m 103 kilo k
10\6 micro
106 mega M
10\9 nano n 109 giga G
10\12 pico p 1012 tera T
4704 APPENDIX 8 / DEFINITIONS AND SYMBOLS FOR UNITS

Submultiple PreRx Symbol Multiple PreRx Symbol

10\15 femto f 1015 peta P

10\18 atto a 1018 exa E
10\21 zepto z 1021 zetta Z
10\24 yocto y 1024 yotta Y

(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

PreRx symbols should be printed in roman (upright) type with no space between the preRx and the unit
symbol.
Example kilometre, km
When a preRx is used with a unit symbol, the combination is taken as a new symbol that can be raised to any
power without the use of parentheses.
Examples 1 cm3"(0.01 m)3"10\6 m3
1
s\1"(10\6 s)\1"10\6 s\1
1 V/cm"100 V/m
1 mmol/dm3"1 mol m\3
A preRx should never be used on its own, and preRxes are not to be combined into comnpound preRxes.
Example pm, not

m
The names and symbols of decimal multiples and submultiples of the SI base unit of mass, the kg, which
already contains a preRx, are constructed by adding the appropriate preRx to the word gram and symbol g.
Examples mg, not
kg; Mg, not kkg
The SI preRxes are not to be used with 3C.
ISO has recommended standard representations of the preRx symbols for use with limited character sets.

Units in Use Together with the SI

These units are not part of the SI, but it is recognized that they will continue to be used in appropriate contexts.
SI preRxes may be attached to some of these units, such as millilitre, ml; millibar, mbar; megaelectronvolt,
MeV; kilotonne, kt. A more extensive list of non-SI units, with conversion factors to the corresponding SI
units, is given in the appendix, Conversion of Units.

Physical quantity Name of unit Symbol for unit Value in SI units

Time minute min 60 s

Time hour h 3600 s
Time day d 86 400 s
Plane angle degree 3 (/180) rad
Plane angle minute  (/10 800) rad
Plane angle second  (/648 000) rad
Length a ngstroK ma A> 10\10 m
Area barn b 10\28 m2
Volume litre l, L dm3"10\3 m3
Mass tonne t Mg"103 kg
Pressure bara bar 105Pa"105 N m\2
Energy electronvoltb eV("e;V) +1.60218;10\19 J
Mass uniRed atomic mass unitb,c u("ma(12C)/12 +1.66054;10\27 kg
a
The a ngstroK m and the bar are approved by CIPM for temporary use with SI units, until CIPM makes
a further recommendation. However, they should not be introduced where they are not used at present.
 APPENDIX 8 / DEFINITIONS AND SYMBOLS FOR UNITS 4705

b
The values of these units in terms of the corresponding SI units are not exact, since they depend on the values
of the physical constants e (for the electronvolt) and NA (for the uniRed atomic mass unit), which are
determined by experiment. See appendix, Fundamental Constants.
c
The uniRed atomic mass unit is also sometimes called the dalton, with symbol Da, although the name and
symbol have not been approved by CGPM.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

Atomic Units
For the purposes of quantum mechanical calculations of electronic wavefuntions, it is convenient to regard
certain fundamental constants (and combinations of such constants) as though they were units. They are
customarily called atomic units (abbreviated: au), and they may be regarded as forming a coherent system of
units for the calculation of electronic properties in theoretical chemistry, although there is no authority from
CGPM for treating them as units. The Rrst Rve atomic units in the table below have special names and
symnbols. Only four of these are independent; all others may be derived by multiplication and devision in the
usual way, and the table includes a number of examples.
The relation of atomic units to the corresponding SI units involves the values of the fundamental physical
constants, and is therefore not exact. The numerical values in the table are based on the estimates of the
appendix, Fundamental Constants. The numerical results of calculations in theoretical chemistry are fre-
quently quoted in atomic units, or as numerical values in the form (physical quantity)/(atomic unit), so that the
reader may make the conversion using the current best estimates of the physical constants.

Physical quantity Name of unit Symbol for unit Value of unit in SI

mass electron rest mass me 9.109 3897 (54);10\31 kg

charge elementary charge e 1.602 177 33 (49);10\19 C
action Planck constant/2a
1.054 572 66 (63);10\34 J s
length bohra a0 5.291 772 49 (24);10\11 m
energy hartreea Eh 4.359 7482 (26);10\18 J
time
/Eh 2.418 884 3341 (29);10\17 s
velocityb a0Eh/
2.187 691 42 (10);106 m s\1
force Eh/a0 8.238 7295 (25);10\8 N
momentum, linear
/a0 1.992 8534 (12);10\24 N s
electric current eEh/
6.623 6211 (20);10\3 A
electric Reld Eh/ea0 5.142 2082 (15);1011 V m\1
electric dipole moment ea0 8.478 3579 (26);10\30 C m
magnetic Sux density
/ea20 2.350 518 08 (71);105 T
magnetic dipole momentc e
/me 1.854 803 08 (62);10\23 J T\1
a

"h/2; a0"40
2/mee2; Eh"
2/m3a20.
b
The numerical value of the speed of light, when expressed in atomic units, is equal to the reciprocal of the Rne
structure constant ; c/(au of velocity)"c
/a0Eh"\1+137.035 9895 (61).
c
The atomic unit of magnetic dipole moment is twice the Bohr magneton, B.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

Dimensionless Quantities
Values of dimensionless physical quanitites, more properly called quantities of dimension one, are often
expressed in terms of mathematically exactly deRned values denoted by special symbols or abbreviations, such
as %( per cent) and ppm (part per million). These symbols are then treated as units, and are used as such in
calculations.

Fractions (Relative Values, Yields, Ef\ciencies)

Fractions such as relative uncertainty, mole fraction x (also called amount fraction, or number fraction), mass frac-
tion w, and volume fraction
, are sometimes expressed in terms of the sumbols summarized in the table below.
4706 APPENDIX 9 / FUNDAMENTAL PHYSICAL CONSTANTS

Name Symbol Value Examples

percent % 10\2 The isotopic abundance of carbon-13 expressed as

a mole fraction is x"1.1%
part per million ppm 10\6 The relative uncertainty in the Planck constant
h("6.626 0755(40);10\34 J s) is 0.60 ppm
The mass fration of impurities in a sample of
copper was found to be less than 3 ppm,
w(3ppm

(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

These multiples of the unit one are not part of the SI and ISO recommends that these symbols should never
be used. They are also frequently used as units of concentration without a clear indication of the type of
fraction implied (e.g. mole fraction, mass fraction or volume fraction). To avoid ambiguity they should only be
used in a context where the meaning of the quantity is carefully deRned. Even then, the use of an appropriate SI
unit ratio may be preferred.

Deprecated Usage
Adding extra labels to ppm and similar symbols, such as ppmv (meaning ppm by volume) should be avoided.
Qualifying labels may be added to symbols for physical quantities, but never to units.
The symbols % and ppm should not be used in combination with other units. In table headings and in
labelling the axes of graphs the use of % and ppm in the denominator is to be avoided. Although one would
write x(13C)"1.1%, the notation 100 x is to be preferred to x/% in tables and graphs.
The further symbols listed in the table below are also to be found in the literature, but their use is to be
deprecated. Note that the names and symbols for 10\9 and 10\12 in this table are based on the American
system of names. In other parts of the world a billion sometimes stands for 1012 and a trillion for 1018. Note
also that the symbol ppt is sometimes used for part per thousand, and sometimes for part per trillion.
To avoid ambiguity the symbols ppb, ppt and pphm should not be used.

Name Symbol Value Examples

part per hundred pph 10\2 (Exactly equivalent to percent, %)


part per thousand ppt 10\3 Atmospheric carbon dioxide is depleted in
permillea  10\3 carbon-13 mass fraction by 7 (or 7 ppt)
relative to ocean water
part per hundred million pphm 10\8 The mass fraction of impurity in the metal
was less than 5 pphm
part per billion ppb 10\9 The air quality standard for ozone is a vol-
ume fraction of
"120 ppb
part per trillion ppt 10\12 The natural background volume fraction of
NO in air was found to be
"140 ppt
part per quadrillion ppq 10\15
a
The permille is also spelled per mille, per mill, permil or pro mille.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

9. FUNDAMENTAL PHYSICAL CONSTANTS

The following values were recommended by the CODATA Task Group on Fundamental Constants in 1986.
For each constant the standard deviation uncertainty in the least signiRcant digits is given in parentheses.
4706 APPENDIX 9 / FUNDAMENTAL PHYSICAL CONSTANTS

Name Symbol Value Examples

percent % 10\2 The isotopic abundance of carbon-13 expressed as

a mole fraction is x"1.1%
part per million ppm 10\6 The relative uncertainty in the Planck constant
h("6.626 0755(40);10\34 J s) is 0.60 ppm
The mass fration of impurities in a sample of
copper was found to be less than 3 ppm,
w(3ppm

(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

These multiples of the unit one are not part of the SI and ISO recommends that these symbols should never
be used. They are also frequently used as units of concentration without a clear indication of the type of
fraction implied (e.g. mole fraction, mass fraction or volume fraction). To avoid ambiguity they should only be
used in a context where the meaning of the quantity is carefully deRned. Even then, the use of an appropriate SI
unit ratio may be preferred.

Deprecated Usage
Adding extra labels to ppm and similar symbols, such as ppmv (meaning ppm by volume) should be avoided.
Qualifying labels may be added to symbols for physical quantities, but never to units.
The symbols % and ppm should not be used in combination with other units. In table headings and in
labelling the axes of graphs the use of % and ppm in the denominator is to be avoided. Although one would
write x(13C)"1.1%, the notation 100 x is to be preferred to x/% in tables and graphs.
The further symbols listed in the table below are also to be found in the literature, but their use is to be
deprecated. Note that the names and symbols for 10\9 and 10\12 in this table are based on the American
system of names. In other parts of the world a billion sometimes stands for 1012 and a trillion for 1018. Note
also that the symbol ppt is sometimes used for part per thousand, and sometimes for part per trillion.
To avoid ambiguity the symbols ppb, ppt and pphm should not be used.

Name Symbol Value Examples

part per hundred pph 10\2 (Exactly equivalent to percent, %)


part per thousand ppt 10\3 Atmospheric carbon dioxide is depleted in
permillea  10\3 carbon-13 mass fraction by 7 (or 7 ppt)
relative to ocean water
part per hundred million pphm 10\8 The mass fraction of impurity in the metal
was less than 5 pphm
part per billion ppb 10\9 The air quality standard for ozone is a vol-
ume fraction of
"120 ppb
part per trillion ppt 10\12 The natural background volume fraction of
NO in air was found to be
"140 ppt
part per quadrillion ppq 10\15
a
The permille is also spelled per mille, per mill, permil or pro mille.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

9. FUNDAMENTAL PHYSICAL CONSTANTS

The following values were recommended by the CODATA Task Group on Fundamental Constants in 1986.
For each constant the standard deviation uncertainty in the least signiRcant digits is given in parentheses.
 APPENDIX 9 / FUNDAMENTAL PHYSICAL CONSTANTS 4707

Quantity Symbol Value

Permeability of vacuuma
0 4
;10\7 H m\1 (deRned)

Speed of light in vacuum c0 299 792 458 m s\1 (deRned)
Permitttivity of vacuuma 0"1/
0c20 8.854 187 8162;10\12 F m\1
Plank constant h 6.626 075 5 (40);10\34 J s

"h/2
1.054 572 66 (63);10\34 J s
Elementary charge e 1.602 177 33 (49);10\19 C
Electron rest mass me 9.109 389 7 (54);10\31 kg
Proton rest mass mp 1.672 623 1 (10);10\27 kg
Neutron rest mass mn 1.674 928 6 (10);10\27 kg
Atomic mass constant, (uniRed atomic mass unit) mu"1 u 1.660 540 2 (10);10\27 kg
Avogadro constant L, NA 6.022 136 7 (36);1023 mol\1
Boltzmann constant K 1.380 658 (12);10\23 J K\1
Faraday constant F 9.648 530 9 (29);104 C mol\1
Gas constant R 8.314 510 (70) J K\1 mol\1
Zero of the Celsius scale 273.15 K (deRned)
Molar volume, ideal gas, p"1 bar, "03C 22.711 08 (19) l mol\1
Standard atmosphere atm 101 325 Pa (deRned)
Fine structure constant "
0e2c0/2h 7.297 353 08 (33);10\3
\1 137.035 989 5 (61)
Bohr radius a0"4
0
2/mee2 5.291 772 49 (24);10\11 m
Hartree energy Eh"
2/mea20 4.359 748 2 (26);10\18 J
Rydberg constant R "Eh/2hc0 1.097 373 153 4 (13);107 m\1

Bohr magneton
B"e
/2me 9.2740154 (31);10\24 J T\1
Electron magnetic moment
e 9.284 770 1 (31);10\24 J T\1
LandeH g-factor for free electron ge"2
e/
B 2.002 319 304 386 (20)
Nuclear magnetcon
N"(me/mp)
B 5.050 786 6 (17);10\27 J T\1
Protron magnetic moment
p 1.410 607 61 (47);10\26 J T\1
Proton magnetogyric ratio p 2.675 221 28 (81);108 s\1 T\1
Magnetic moment of protons in H2O,
p
p/
B 1.520 993 129 (17);10\3
Proton resonance frequency per Reld in H2O p/2
42.576 375 (13) MHz T\1
Stefan-Boltzmann constant "2
5k4/15h3c20 5.670 51 (19);10\8 W m\2 K\4
First radiation constant c1"2
hc2o 3.741 7749 (22);10\16 W m2
Second radiation constant c2"hco/k 1.438 769 (12);10\2 m K
Gravitational constant G 6.672 59
(85);10\11 m3 kg\1s\2
Standard acceleration of free fall gn 9.806 65 m s\2 (deRned)
a
H m\1"N A\2"N s2 C\2; F m\1"C2 J\1 m\1; 0 may be calculated exactly from the deRned values of

0 an c0.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)

Values of Common Mathematical Constants

Mathematical constant Symbol Value

Ratio of circumference to diameter of a circle
3.141 592 653 59

Base of natural logarithms e 2.718 281 828 46
Natural logarithm of 10 ln 10 2.302 585 092 99

(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry,
2nd edn. Oxford: Blackwell ScientiRc Publications.)
4708 APPENDIX 10 / IMPORTANT PEAKS IN THE MASS SPECTRA OF COMMON SOLVENTS

10. IMPORTANT PEAKS IN THE MASS SPECTRA

OF COMMON SOLVENTS
The following table gives the most important peaks that appear in the mass spectra of the most common
solvents which might occur as an impurity in organic samples. The solvents are classiRed in ascending order of
thier M# peaks. The highest intensity peaks are indicated with (100%).1}3

Important peaks in the mass spectra of common solvents

Solvents Formula M# Important peaks (m/z)

Water H2O 18 (100%) 17

Methanol CH3OH 32 31 (100%), 29, 15
Acetonitrile CH3CN 41 (100%) 40, 39, 38, 28, 15
Ethanol CH3CH2OH 46 45, 31 (100%), 27, 15
Dimethyl ether CH3OCH3 46 (100%) 45, 29, 15
Acetone CH3COCH3 58 43 (100%), 42, 39, 27, 15
Acetic acid CH3CO2H 60 45, 43, 18, 15
Ethylene glycol HOCH2CH2OH 62 43, 33, 31 (100%), 29, 18, 15
Furan C4H4O 68 (100%) 42, 39, 38, 37, 29, 18
Tetrahydrofuran C4H8O 72 71,43, 42 (100%), 41, 40, 39, 27,
18, 15
n-Pentane C5H12 72 57, 43 (100%), 42, 41, 39, 29, 28,
27, 15
Dimethyformamide (DMF) HCON(CH3)2 73 (100%) 58, 44, 42, 30, 29, 28, 18, 15
Diethyl ether (C2H5)2O 74 59, 45, 41, 31 (100%), 29, 27, 15
Methyl acetate CH3CO2CH3 74 59, 43 (100%), 42, 32, 29, 28, 15
Carbon disulphide CS2 76 (100%) 64, 44, 38, 32
Benzene C6H6 78 (100%) 77, 52, 51, 50, 39, 28
Pyridine C5H5N 79 (100%) 80, 78, 53, 52, 51, 50, 39, 26
Dichloromethane CH2Cl2 84 86, 51, 49 (100%), 48, 47, 35, 28
Cyclohexane C6H12 84 69, 56, 55, 43, 42, 41, 39, 27
n-Hexane C6H14 86 85, 71, 69, 57 (100%), 43, 42, 41,
39, 29, 28, 27
p-Dioxane C4H8O2 88 (100%) 87, 58, 57, 45, 43, 31, 30, 29, 28
Tetramethylsilane (TMS) (CH3)4Si 88 74, 73, 55, 45, 43, 29
1,2-Dimethoxyethane (CH3OCH2)2 90 60, 58, 45 (100%), 31, 29
Toluene C6H5CH3 92 91 (100%), 65, 51, 39, 28
Chloroform CHCl3 118 120, 83, 81, (100%), 47, 35, 28
Chloroform-d1 CDCl3 119 121, 84, 82 (100%), 48, 47, 35, 28
Carbon tetracholoride CCl4 152 121, 119, 117 (100%), 84, 82, 58.5,
(not seen) 47, 35, 28
Tetrachloroethene CCl2"CCl2 164 168, 166 (100%), 165, 164, 131,
(not seen) 128, 129, 95, 94, 82, 69, 59, 47,
31, 24

Reprinted from T.J. Bruno and P.D.N. Svoronos, CRC Handbook of Basic Tables for Chemical Analysis, CRC Press, Boca
Raton, FL, 1989, p. 357.
 APPENDIX 11 / NOMENCLATURE AND TERMINOLOGY FOR ANALYTICAL PYROLYSIS 4709

References
1. Clerce, J. T., Pretsch, E., and Seibl, J., Studies in Analytical Chemistry, Vol. I. Structural Analysis of Organic
Compounds by Combined Application of Spectroscopic Methods, Elsevier, Amsterdam, 1981.
2. McLafferty, F. W., Interpretation of Mass Spectra, Universtiy Science Books, Mill Valley, CA, 1980.
3. Pasto, D. J. and Johnson, C. R., Organic Structure Determination, Prentice-Hall, Englewood Ciffs, NJ, 1969.

11. NOMENCLATURE AND TERMINOLOGY

FOR ANALYTICAL PYROLYSIS
(IUPAC RECOMMENDATIONS 1993)
Prepared for publication by
P. C. Uden, University of Massachusetts, Amherst, MA, USA
^ 1993 IUPAC

Abstract
This paper deRnes terms and deRnitions used in analytical methods of pyrolysis and includes expressions for
coupled systems and for the description of the temperature proRles and the products that are obtained.

Introduction
Thermal degradation under controlled conditions is often used an part of an analytical procedure, either to
render a sample into a suitable form for subsequent analysis by gas chromatography, mass spectrometry or
infrared spectroscopy or by direct monitoring as an analytical technique in its own right. A range of terms and
expression have been used in the Reld and this nomenclature brings these together in a systematic manner and
assigns each a speciRc meaning.

Analytical Pyrolysis
Analytical Pyrolysis
The characterization, in an inert atmosphere, of a material or a chemical process by a chemical degradation
reaction(s) induced by thermal energy.

Catalytic Pyrolysis
A pyrolysis that is inSuenced by the addition of a catalyst.

Char
A solid carbonaceous pyrolysis residue.

Coil Pyrolyser
A pyrolyser in which the sample (sometimes located in a tubular vessel) is placed in a metal coil that is heated
to cause pyrolysis.

Continuous Mode (Furnace) Pyrolyser

A pyrolser in which the sample is introduced into a furnace preheated to the Rnal temperature.
4710 APPENDIX 11 / NOMENCLATURE AND TERMINOLOGY FOR ANALYTICAL PYROLYSIS

Curie-Point Pyrolyser
A pyrolyser in which a ferromagnetic sample carrier is inductively heated to its Curie point.

Filament (Ribbon) Pyrolyser

A pyrolyser in which the sample is placed on a metal Rlament (ribbon) that is resistively heated to cause
pyrolysis.

Final Pyrolysis Temperature (T(f,Py) )

The Rnal (steady state) temperature which is attained by a pyrolyser. (The terms equilibrium temperature and
pyrolysis temperature may be used when referring to an isothermal pyrolysis; they are not recommended for
use with a non-isothermal pyrolysis.)

Flash Pyrolysis
A pyrolysis that is carried out with a fast rate of temperature increase, of the order of 10 000 K/s.

Fractionated Pyrolysis
A pyrolysis in which the same sample is pyrolysed at different temperatures for different times in
order to study special fractions of the sample.

In-Source Pyrolysis
A pyrolysis in which the reactor is located within the ion source of a mass spectrometer.

IR-Pyrogram
Chromatogram of a pyrolysate detected by infrared spectrometry.

Isothermal Pyrolysis
A pyrolysis during which the temperature is essentially constant.

Maximum Pyrolysis Temperature (T(max,Py) )

The highest temperature in a temperature/time proRle.

MS-Pyrogram
Chromatogram of a pyrolysate detected by mass spectrometry.

Off-Line Pyrolysis
A pyrolysis in which the products are trapped before analysis.

Oxidative Pyrolysis
A pyrolysis that occurs in the presence of an oxidative atmosphere.

Pressure Monitored Pyrolysis

A pyrolysis technique in which the pressure of the volatile pyrolysates is recorded as the sample is heated.

Pulse Mode Pyrolyser

A pyrolyser in which the sample is introduced into a cold furnace which is then heated rapidly.

Pyrogram
A chromatogram of a pyrolysate.
 APPENDIX 11 / NOMENCLATURE AND TERMINOLOGY FOR ANALYTICAL PYROLYSIS 4711

Pyrolysate (Pyrolyzate)
The products of pyrolysis.

Pyrolyser (Pyrolyzer)
A device for performing pyrolysis.

Pyrolysis
A chemical degradation reaction that is caused by thermal energy. The term pyrolysis generally refers to an
inert environment.)

Pyrolysis-Gas Chromatography (Py-GC)

A pyrolysis technique in which the volatile pyrolysates are directly conducted into a gas chromatograph for
separation and detection.

Pyrolysis-Gas Chromatograph-Mass Spectrometry (Py-GC-MS)

A pyrolysis technique in which the volatile pyrolysates are separated and analysed by on-line gas chromatogra-
phy-mass spectrometry.

Pyrolysis-Gas Chromatography-Infrared Spectroscopy (Py-GC-IR)

A pyrolysis technique in which the volatile pyrolysates are separated and analysed by on-line gas chromatogra-
phy-infrared spectroscopy.

Pyrolysis-Infrared Spectroscopy (Py-IR)

A pyrolysis technique in which the pyrolysates are detected and analysed by on-line infrared spectroscopy.

Pyrolysis-Infrared Spectrum
Infrared spectrum obtained from pyrolysis-infrared spectroscopy.

Pyrolysis-Mass Spectrometry (Py-MS)

A pyrolysis technique in which the volatile pyrolysates are detected and analysed by on-line mass spectro-
metry.

Pyrolysis-Mass Spectrum
Mass spectrum obtained from pyrolysis-mass spectrometry.

Pyrolysis Reactor
That portion of the pyrolyser in which the pyrolysis takes place.

Pyrolysis Residue
That portion of the pyrolysate that does not leave the reactor.

Pyrolysis Thermogram
The result of a temperature programmed pyrolysis in which the detector signals, e.g. total ion current or single
ions, total absorbance or a GC-detector, are plotted against time or temperature.

Reductive Pyrolysis
A pyrolysis which occurs in the presence of a reducing atmosphere.

Sequential Pyrolysis
A pyrolysis in which the same initial sample is repetitively pyrolysed under indentical conditions (Rnal
pyrolysis temperature, temperature rise time and total heating time).
4712 APPENDIX 12A / NOMENCLATURE / Chromatography

Stepwise Pyrolysis
A pyrolysis in which the sample temperature is raised stepwise. The pyrolysis products are recorded between
each step.

Tar
A liquid pyrolysis residue.

Temperature-Programmed Pyrolysis
A pyrolysis during which the sample is heated at a controlled rate within a temperature range in which
pyrolysis occurs.

Temperature Rise Time (TRT)

The time required for a pyrolyser temperature to be increased from its initial to its Rnal temperature.

Temperature Time Pro\le (TTP)

A graphical representation of temperature versus time for a particular pyrolysis experiment or pyrolyser.

Total heating time (THT)

The time between the onset and conclusion of the sample heating in a pyrolysis experiment.

Volatile Pyrolyzate
That portion of the pyrolystate which has adequate vapour pressure to reach the detector.

List of Symbols
T(f,Py) Final pyrolysis temperature
T(max,Py) Maximum pyrolysis temperature

Index of Acronyms
Py-GC Pyrolysis-gas chromatography
Py-GC-IR Pyrolysis-gas chromatography-infrared spectroscopy
Py-GC-MS Pyrolysis-gas chromatography-mass spectrometry
Py-IR Pyrolysis-infrared spectroscopy
Py-MS Pyrolysis-mass spectrometry
THT Total heating time
TRT Temperature rise time
TTP Temperature time proRle

12A. NOMENCLATURE

Chromatography
(IUPAC Recommendations 1993)

Prepared for publication by

L. S. Ettre, Yale University, New Haven, CT, USA
^ 1993 IUPAC
4712 APPENDIX 12A / NOMENCLATURE / Chromatography

Stepwise Pyrolysis
A pyrolysis in which the sample temperature is raised stepwise. The pyrolysis products are recorded between
each step.

Tar
A liquid pyrolysis residue.

Temperature-Programmed Pyrolysis
A pyrolysis during which the sample is heated at a controlled rate within a temperature range in which
pyrolysis occurs.

Temperature Rise Time (TRT)

The time required for a pyrolyser temperature to be increased from its initial to its Rnal temperature.

Temperature Time Pro\le (TTP)

A graphical representation of temperature versus time for a particular pyrolysis experiment or pyrolyser.

Total heating time (THT)

The time between the onset and conclusion of the sample heating in a pyrolysis experiment.

Volatile Pyrolyzate
That portion of the pyrolystate which has adequate vapour pressure to reach the detector.

List of Symbols
T(f,Py) Final pyrolysis temperature
T(max,Py) Maximum pyrolysis temperature

Index of Acronyms
Py-GC Pyrolysis-gas chromatography
Py-GC-IR Pyrolysis-gas chromatography-infrared spectroscopy
Py-GC-MS Pyrolysis-gas chromatography-mass spectrometry
Py-IR Pyrolysis-infrared spectroscopy
Py-MS Pyrolysis-mass spectrometry
THT Total heating time
TRT Temperature rise time
TTP Temperature time proRle

12A. NOMENCLATURE

Chromatography
(IUPAC Recommendations 1993)

Prepared for publication by

L. S. Ettre, Yale University, New Haven, CT, USA
^ 1993 IUPAC
 APPENDIX 12A / NOMENCLATURE / Chromatography 4713

Abstract
This report presents deRnitions of terms and symbols used in all chromatographic separations. The reports
covers gas, liquid, size-exclusion, ion-exchange and supercritical-Suid chromatography and both column and
planar modes of separation. DeRnitions are included for the description of the separation process, the
chromatographic system and equipment and the properties of detectors.

Introduction
The Commission on Analytical Nomenclature of IUPAC has been active for a long time in establishing
nomenclatures for chromatography. After proposing suitable nomenclatures for gas chromatography [1}2]
and ion exchange [3}4] the Commission developed a uniRed nomenclature for chromatography [5}6]. Parallel
to these activities other standardization bodies and scientists have also dealt with nomenclatures on gas
chromatography [7}15], supercritical-Suid chromatography [16], liquid chromatography [17}20], exclusion
chromatography [21}23] and planar chromatography [24].
The original activities of the IUPAC Commission on Analytical Nomenclature aimed to create a uniRed
nomenclature applicable to all forms of chromatography, took place over 20 years ago. Since that time
chromatographic techniques have advanced signiRcantly. Based on these developments it was decided to
prepare a new, up-to-date universal chromatography nomenclature, which also considers the recommenda-
tions incorporated in the various other nomenclatures developed since the original work of IUPAC.
The present nomenclature was prepared by Dr. L. S. Ettre originally for the Commission on Analytical
Nomenclature. Following the reorganization of the Commissions of the Analytical Division at the General
Assembly in Lund in 1989, this project became the responsibility of the Commission on Chromatography and
Other Analytical Separations (LLTC). The Nomenclature considers all the previous nomenclatures referenced
above as well as the four publications dealing with these nomenclatures [25}27].
The present nomenclature deals with all chromatographic terms and deRnitions used in the major chromato-
graphic techniques such as gas, liquid and supercritical-Suid chromatography, column and planar chromato-
graphy, partition, adsorption, ion-exchange and exclusion chromatography. However, it does not include terms
related to the results calculated from chromatography data such as e.g. the various molecular weight terms
computed from the primary data obtained by exclusion chromatography. Also it does not deal with detailed
information related to detection and detectors or the relationships between chemical structure and chromato-
graphic retention.

General Rules
In developing the uniRed nomenclature the rules and recommendations set up by IUPACs Division of Physical
Chemistry [28] were followed. According to these, the following symbols should be used for major physical
and physico-chemical quantities and units:

area . . . . . . . . . . . .............. A
density . . . . . . . . . . ..............

diameter . . . . . . . . . .............. d
diffusion coefRcient . .............. D
equilibrium constant .............. K
mass (weight) . . . . . .............. W
pressure . . . . . . . . . .............. p or P
radius . . . . . . . . . . .............. r
rate constant . . . . . . .............. k
temperature (kelvin) .............. T
time . . . . . . . . . . . . .............. t
velocity . . . . . . . . . .............. u
viscosity . . . . . . . . . .............. 
volume . . . . . . . . . . .............. V

The only deviation from the rules set by the Division of Physical Chemistry of IUPAC is the use of L (instead
of l) for length. The reason for this is the easy interchangeability in a printed, and particularly typed, text of the
4714 APPENDIX 12A / NOMENCLATURE / Chromatography

letter l with the numeral one. Additional basic symbols accepted were F for the volumetric Sow rates and
w for the peak widths. Also, differentiation has been made between p (for pressures) and P (for relative
pressure).
In addition to these basic rules the following additional rules are followed in the present proposal:

(a) Except for a few superscripts further differentiation is always made by using subscripts and never
composite symbols.
(b) Superscripts are used for various retention times and volumes and to speciRcally indicate data obtained
in programmed-temperature conditions.
(c) Subscripts referring to the physical conditions or the phase are capitalized, e.g. M and S for the mobile
and stationary phases respectively, or, in gas chromatography, G for the gas and L for the liquid phase.
Thus, e.g. the diffusion coefRcient in the mobile phase is DM and not Dm.
(d) In addition to those mentioned above, a few capitalized subscripts are used such as R for retention (as in
tR and VR), N for net (as in tN and VN) and F in RF, the retardation factor used in planar chromatogra-
phy.
(e) Compound subscripts are avoided. If a given compound is indicated and there is already a subscript, and
if the compound is characterized by more than a simple number or letter, then the new subscript should
be in parentheses. Thus, while it is tRi, it should be tR(st) or tR(z#1).
(f) In addition to reference to the outlet of a column, subscript o is also used in a number of terms to
describe some fundamental values. Similarly subscript i has various meanings, depending on the term in
which it is used.
(g) Physical parts of the system are generally characterized by lower-case subscripts such as, c for column,
p for particles or pores, and f for Rlm.

Three tables follow the nomenclature, listing alphabetically the terms, symbols and acronyms included in
the text.

Contents

1. General Terminology
1.1. Basic DeRnitions
1.2. Principal Methods
1.3. ClassiRcation According to the Shape of the Chromatographic Bed
1.4. ClassiRcation According to the Physical State of the Mobile Phase
1.5. ClassiRcation According to the Mechanism of Separation
1.6. Special Techniques
2. Terms Related to the Chromatographic System
2.1. Apparatus in Column Chromatography
2.2. Apparatus in Planar Chromatography
3. Terms Related to the Chromatographic Process and the Theory of Chromatography
3.1. The Chromatographic Medium
3.2. The Column
3.3. The Chromatogram
3.4. Diffusion
3.5. Temperatures
3.6. The Mobile Phase
3.7. Retention Parameters in Column Chromatography
3.8. Retention Parameters in Planar Chromatography
3.9. Distribution Constants
3.10. Terms Expressing the EfRciency of Separation
4. Terms Related to Detection
4.1. ClassiRcation of Detectors
4.2. Detector Response
 APPENDIX 12A / NOMENCLATURE / Chromatography 4715

4.3. Noise and Drift

4.4. Minimum Detectability
4.5. Linear and Dynamic Ranges

5. Special Terminology Used in Ion-exchange Chromatography

5.1. Basic DeRnitions
5.2. The Mobile Phase
5.3. The Chromatographic Medium
5.4. Capacity Values
5.5. Diffusion, Selectivity and Separation
5.6. Distribution Constants

6. Special Terminology Used in Exclusion Chromatography

6.1. The Column
6.2. Retention Parameters
6.3. EfRciency Terms

Tables
1. Index of Terms
2. List of Symbols
3. List of Acronyms Used in Chromatography

Figures
1. Typical Chromatograms
2. Typical Plane Chromatogram
3. Widths of the Gaussian Peak at Various Heights as a Function of the Standard Deviation of the Peak
4. Measurement of the Noise and Drift of a Chromatographic Detector
5. Linearity Plot of a Chromatographic Detector
6. Determination of the Linear and Dynamic Ranges of a Chromatographic Detector
7. Retention Characteristics in Exclusion Chromatography

1. General Terminology
1.1. Basic De\nitions
1.1.01. Chromatography Chromatography is a physical method of separation in which the components to
be separated are distributed between two phases, one of which is stationary (stationary phase) while the other
(the mobile phase) moves in a deRnite direction.

1.1.02. Chromatogram A graphical or other presentation of detector response, concentration of analyte in

the efSuent or other quantity used as a measure of efSuent concentration versus efSuent volume or time. In
planar chromatography `chromatograma may refer to the paper or layer with the separated zones.

1.1.03. Chromatograph (verb) To separate by chromatography.

1.1.04. Chromatograph (noun) The assembly of apparatus for carrying out chromatographic separation.

1.1.05. Stationary phase The stationary phase is one of the two phases forming a chromatographic system.
It may be a solid, a gel or a liquid. If a liquid, it may be distributed on a solid. This solid may or may not
contribute to the separation process. The liquid may also be chemically bonded to the solid (Bonded Phase) or
immobilized onto it (Immobilized Phase).
The expression Chromatographic Bed or Sorbent may be used as a general term to denote any of the
different forms in which the stationary phase is used.

Note: Particularly in gas chromatography where the stationary phase is most often a liquid, the term Liquid
Phase is used for it as compared to the Gas Phase, i.e. the mobile phase. However, particularly in the
4716 APPENDIX 12A / NOMENCLATURE / Chromatography

early development of liquid chromatography, the term liquid phase had also been used to characterize
the mobile phase as compared to the solid phase i.e. the stationary phase. Due to this ambiguity, the
use of the term liquid phase is discouraged. If the physical state of the stationary phase is to be
expressed, the use of the adjective forms such as Liquid Stationary Phase and Solid Stationary Phase,
Bonded Phase or Immobilized Phase is proposed.

1.1.05.1. Bonded phase A stationary phase which is covalently bonded to the support particles or to the
inside wall of the column tubing.

1.1.05.2. Immobilized phase A stationary phase which is immobilized on the support particles, or on the
inner wall of the column tubing, e.g. by in situ polymerization (cross-linking) after coating.

1.1.06. Mobile phase A Suid which percolates through or along the stationary bed, in a deRnite direction.
It may be a liquid (Liquid Chromatography) or a gas (Gas Chromatography) or a supercritical Suid
(Supercritical-Fluid Chromatography). In gas chromatography the expression Carrier Gas may be used for the
mobile phase. In elution chromatography the expression Eluent is also used for the mobile phase.

1.1.07. Elute (verb) To chromatograph by elution chromatograph. The process of elution may be stopped
while all the sample components are still on the chromatographic bed or continued until the components have
left the chromatographic bed.

Note: The term elute is preferred to the term Develop used in former nomenclatures of planar chromato-
graphy.

1.1.08. EfWuent The mobile phase leaving the column.

1.1.09. Sample The mixture consisting of a number of components the separation of which is attempted on
the chromatographic bed as they are carried or eluted by the mobile phase.

1.1.10. Sample components The chemically pure constituents of the sample. They may be unretained (i.e.
not delayed) by the stationary phase, partially retained (i.e. eluted at different times) or retained permanently.
The terms Elute or Analyte are also acceptable for a sample component.

1.1.11. Solute A term referring to the sample components in partition chromatography.

1.1.12. Solvent A term sometimes referring to the liquid stationary phase in partition chromatography.

Note: In liquid chromatography the term solvent has been often used for the mobile phase. This usage is not
recommended.

1.1.13. Zone A region in the chromatographic bed where one or more components of the sample are
located. The term Band may also be used for it.

1.2. Principal Methods

1.2.01. Frontal chromatography A procedure in which the sample (liquid or gas) is fed continuously into
the chromatographic bed. In frontal chromatography no additional mobile phase is used.

1.2.02. Displacement chromatography A procedure in which the mobile phase contains a compound (the
Displacer) more strongly retained than the components of the sample under examination. The sample is fed
into the system as a Rnite slug.

1.2.03. Elution chromatography A procedure in which the mobile phase is continuously passed through or
along the chromatographic bed and the sample is fed into the system as a Rnite slug.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4717

1.3. Classi\cation According to the Shape of the Chromatographic Bed

1.3.01. Column chromatography A separation technique in which the stationary bed is within a tube. The
particles of the solid stationary phase or the support coated with a liquid stationary phase may Rll the whole
inside volume of the tube (Packed Column) or be concentrated on or along the inside tube wall leaving an
open, unrestricted path for the mobile phase in the middle part of the tube (Open-Tubular Column).

1.3.02. Planar chromatography A separation technique in which the stationary phase is present as or on
a plane. The plane can be a paper, serving as such or impregnated by a substance as the stationary bed (Paper
Chromatography, PC) or a layer of solid particles spread on a support, e.g. a glass plate (Thin-Layer
Chromatography, TLC). Sometimes planar chromatography is also termed Open-Bed Chromatography.

1.4. Classi\cation According to the Physical State of the Mobile Phase

1.4.01. Chromatographic techniques are often classiRed by specifying the physical state of both phases used.
Accordingly, the following terms are in use:

Gas-liquid chromatography (GLC)

Gas-solid chromatography (GSC)
Liquid-Liquid chromatography (LLC)
Liquid-solid chromatography (LSC)

The term Gas-Liquid Partition Chromatography (GLPC) can also be found in the literature. However, often
distinction between these modes is not easy. For example, in GC, a liquid may be used to modify an
adsorbent-type solid stationary phase.

1.4.02. Gas chromatography (GC) A separation technique in which the mobile phase is a gas. Gas
chromatography is always carried out in a column.

1.4.03. Liquid chromatography (LC) A separation technique in which the mobile phase is a liquid. Liquid
chromatography can be carried out either in a column or on a plane.

Note: Present-day liquid chromatography generally utilizing very small particles and a relatively high inlet
pressure is often characterized by the term High-Performance (or High-Pressure) Liquid Chromatogra-
phy, and the acronym HPLC.

1.4.04. Supercritical-Wuid chromatography (SFC) A separation technique in which the mobile phase is
a Suid above and relatively close to its critical temperature and pressure.

Note: In general the terms and deRnitions used in gas or liquid chromatography are equally applicable to
supercritical-Suid chromatography.

1.5. Classi\cation According to the Mechanism of Separation

1.5.01. Adsorption chromatography Separation is based mainly on differences between the adsorption
afRnities of the sample components for the surface of an active solid.

1.5.02. Partition chromatography Separation is based mainly on differences between the solubilities of the
sample components in the stationary phase (gas chromatography), or on differences between the solubilities of
the components in the mobile and stationary phases (liquid chromatography).

1.5.03. Ion-exchange chromatography Separation is based mainly on differences in the ion exchange
afRnities of the sample components.

Note: Present day ion-exchange chromatography on small particle high efRciency columns and usually
utilising conductometric or spectroscopic detectors is often referred to as Ion Chromatography (IC).
4718 APPENDIX 12A / NOMENCLATURE / Chromatography

1.5.04. Exclusion chromatography Separation is based mainly on exclusion effects, such as differences in
molecular size and/or shape or in charge. The term Size-Exclusion Chromatography may also be used when
separation is based on molecular size. The terms Gel Filtration and Gel-Permeation Chromatography (GPC)
were used earlier to describe this process when the stationary phase is a swollen gel. The term Ion-Exclusion
Chromatography is speciRcally used for the separation of ions in an aqueous phase.

1.5.05. AfVnity chromatography This expression characterizes the particular variant of chromatography
in which the unique biological speciRcity of the analyte and ligand interaction is utilized for the separation.

1.6. Special Techniques

1.6.01. Reversed-phase chromatography An elution procedure used in liquid chromatography in which the
mobile phase is signiRcantly more polar then the stationary phase, e.g. a microporous silica-based material
with chemically bonded alkyl chains.
Note: The term `reverse phasea is an incorrect expression to be avoided.

1.6.02. Normal-phase chromatography An elution procedure in which the stationary phase is more polar
than the mobile phase. This term is used in liquid chromatography to emphasize the contrast to reversed-phase
chromatography.

1.6.03. Isocratic analysis The procedure in which the composition of the mobile phase remains constant
during the elution process.

1.6.04. Gradient elution The procedure in which the composition of the mobile phase is changed continu-
ously or stepwise during the elution process.

1.6.05. Stepwise elution The elution process in which the composition of the mobile phase is changed in
steps during a single chromatographic run.

1.6.06. Two-dimensional chromatography A procedure in which parts or all of the separated sample
components are subjected to additional separation steps. This can be done, e.g. by conducting a particular
fraction eluting from the column into another column (system) having different separation characteristics.
When combined with additional separation steps, this may be described as Multi-Dimensional Chromato-
graphy.
In planar chromatography two-dimensional chromatography refers to the chromatographic process in
which the components are caused to migrate Rrst in one direction and subsequently in a direction at right
angles to the Rrst one; the two elutions are carried out with different eluents.

1.6.07. Isothermal chromatography A procedure in which the temperature of the column is kept constant
during the separation.

1.6.08. Programmed-temperature chromatography (temperature programming) A procedure in which the

temperature of the column is changed systematically during a part or the whole of the separation.

1.6.09. Programmed-Wow chromatography (Wow programming) A procedure in which the rate of Sow of
the mobile phase is changed systematically during a part or the whole of the separation.

1.6.10. Programmed-pressure chromatography (pressure programming) A procedure in which the inlet

pressure of the mobile phase is changed systematically during a part or whole of the separation.

1.6.11. Reaction chromatography A technique in which the identities of the sample components are
intentionally changed between sample introduction and detection. The reaction can take place upstream of the
column when the chemical identity of the individual components passing through the column differs from that
of the original sample, or between the column and the detector when the original sample components are
separated in the column but their identity is changed prior to entering the detection device.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4719

1.6.11.1. Pyrolysis-gas chromatography A version of reaction chromatography in which a sample is

thermally decomposed to simpler fragments before entering the column.

1.6.11.2. Post-column derivatization A version of reaction chromatography in which the separated sample
components eluting from the column are derivatized prior to entering the detector. The derivatization process
is generally carried out `on-the-Sya, i.e. during transfer of the sample components from the column to the
detector. Derivatization may also be carried out before the sample enters the column or the planar medium;
this is pre-column (preliminary) derivatization.

2. Terms Related to the Chromatographic System

2.1. Apparatus in Column Chromatography
2.1.01. Pump A device designed to deliver the mobile phase at a controlled Sow-rate to the separation
system. Pumps are generally used in liquid chromatography.

2.1.01.1. Syringe pumps Pumps with a piston, which advances at a controlled rate within a smooth
cylinder to displace the mobile phase.

2.1.01.2. Reciprocating pumps Pumps with a single or multiple chamber, from which the mobile phase is
displaced by reciprocating piston(s) or diaphragm(s).

2.1.01.3. Pneumatic pumps Pumps which employ a gas to displace the liquid mobile phase either directly
or via a piston.

2.1.02. Sample injector A device by which a liquid, solid or gaseous sample is introduced into the mobile
phase of the chromatographic bed.

2.1.02.1. Direct injector A device which directly introduces the sample into the mobile-phase stream.

2.1.02.2. Bypass injector A device in which the sample is Rrst introduced into a chamber (loop), temporar-
ily isolated from the mobile phase system by valves, which can be switched to make an instantaneous diversion
of the mobile phase stream through the chamber to carry the sample to the column. A bypass injector may also
be known as a Valve Injector or Sampling Valve (see 2.1.02.7).

2.1.02.3. On-column injector A device in which the sample is directly introduced into the column. In gas
chromatography the on-column injector permits the introduction of the liquid sample into the column without
prior evaporation.

2.1.02.4. Flash vaporizer A heated device used in gas chromatography. Here the liquid sample is introduc-
ed into the carrier gas stream with simultaneous evaporation and mixing with the carrier gas prior to entering
the column.

2.1.02.5. Split injection A sample introduction technique used in gas chromatography. The sample is Sash
vaporized and after thorough mixing of the sample with carrier gas, the stream is split into two portions, one
being conducted to the column and the other being discarded.

2.1.02.6. Programmed temperature vaporizer (PTV) A sample introduction device used in gas chromatog-
raphy. The liquid sample is introduced, usually with a syringe, into a device similar to a Sash vaporizer, the
temperature of which is kept low, below the boiling point of the sample components. After withdrawal of the
syringe, the device is heated up very rapidly in a controlled fashion to evaporate the sample into the
continuously Sowing carrier gas stream. The PTV may also be used in the split mode: in this case, the carrier
gas stream containing the evaporated sample components is split into two portions, one of which is conducted
into the column while the other is discarded.
4720 APPENDIX 12A / NOMENCLATURE / Chromatography

2.1.02.7. Gas sampling valve A bypass injector permitting the introduction of a gaseous sample of a given
volume into a gas chromatograph.

2.1.03. Column oven A thermostatically controlled oven containing the column, the temperature of which
(Separation Temperature or Column Temperature) can be varied within a wide range.

2.1.04. Fraction collector A device for recovering fractional volumes of the column efSuent.

2.1.05. Detector A device that measures the change in the composition of the eluent by measuring physical
or chemical properties.

2.2. Apparatus in Planar Chromatography

2.2.01. Spotting device The syringe or micropipet used to deliver a Rxed volume of sample as a spot or
streak to the paper or thin-layer media at the origin.

2.2.02. Elution chamber (developing chamber) A closed container, the purpose of which is to enclose the
media used as well as the mobile phase to maintain a constant environment in the vapor phase.

2.2.02.1. Sandwich chamber A chamber in which the walls are close enough to the paper or plate to
provide a relatively fast equilibration.

2.2.02.2. Ascending elution (ascending development) A mode of operation in which the paper or plate is in
a vertical or slanted position and the mobile phase is supplied to its lower edge; the upward movement depends
on capillary action.

2.2.02.3. Horizontal elution (horizontal development) A mode of operation in which the paper or plate is
in a horizontal position and the mobile-phase movement along the plane depends on capillary action.

2.2.02.4. Descending elution (descending development) A mode of operation in which the mobile phase is
supplied to the upper edge of the paper or plate and the downward movement is governed mainly by gravity.

2.2.02.5. Radial elution (radial development) or circular elution (circular development) A mode of opera-
tion in which the sample is spotted at a point source at or near the middle of the plane and is carried outward in
a circle by the mobile phase, also applied at that place.

2.2.02.6. Anticircular elution (anticircular development) The opposite of 2.2.02.5. Here the sample as well
as the mobile phase is applied at the periphery of a circle and both move towards the center.

2.2.02.7. Chamber saturation (saturated development) This expression refers to the uniform distribution
of the mobile phase vapor through the elution chamber prior to chromatography.

2.2.02.8. Unsaturated elution (unsaturated development) This expression refers to chromatography in an

elution chamber without attaining chamber saturation.

2.2.02.9. Equilibration The expression refers to the level of saturation of the chromatographic bed by the
mobile-phase vapor prior to chromatography.

2.2.03. Visualization chamber A device in which the planar media may be viewed under controlled-
wavelength light, perhaps after spraying with chemical reagents to render the separated components as visible
spots under speciRed conditions.

2.2.04. Densitometer A device which allows portions of the developed paper or thin-layer media to be
scanned with a beam of light of a speciRed wavelength for measurements of UV or visible light absorption or
Suorescence, providing values which can be used for the quantisation of the separated compounds.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4721

3. Terms Related to the Chromatographic Process and

the Theory of Chromatography
3.1. The Chromatographic Medium
3.1.01. Active solid A solid with sorptive properties.

3.1.02. ModiVed active solid An active solid the sorptive properties of which have been changed by some
treatment.

3.1.03. Solid support A solid that holds the stationary phase but, ideally, does not contribute to the
separation process.

3.1.04. Binders Additives used to hold the solid stationary phase to the inactive plate or sheet in thin-layer
chromatography.

3.1.05. Gradient layer The chromatographic bed used in thin-layer chromatography in which there is
a gradual transition in some property.

3.1.06. Impregnation The modiRcation of the separation properties of the chromatographic bed used in
planar chromatography by appropriate additives.

3.1.07. Packing The active solid, stationary liquid plus solid support, or swollen gel contained in a tube.

3.1.07.1. Totally porous packing Here the stationary phase permeates each porous particle.

3.1.07.2. Pellicular packing In this case the stationary phase forms a porous outer shell on an impermeable
particle.

3.1.08. Particle diameter (dp) The average diameter of the solid particles.

3.1.09. Pore radius (rp) The average radius of the pores within the solid particles.

3.1.10. Liquid-phase loading A term used in partition chromatography to express the relative amount of
the liquid stationary phase in the column packing. It is equal to the mass fraction (%) of liquid stationary phase
in the total packing (liquid stationary phase plus support).

3.2. The Column

3.2.01. Column The tube and the stationary phase contained within, through which the mobile phase
passes.

3.2.02. Packed column A tube containing a solid packing.

3.2.03. Open-tubular column A column, usually having a small diameter in which either the inner tube
wall, or a liquid or active solid held stationary on the tube wall acts as the stationary phase and there is an
open, unrestricted path for the mobile phase.

3.2.03.1. Wall-coated open-tubular (WCOT) column In these columns the liquid stationary phase is
coated on the essentially unmodiRed smooth inner wall of the tube.

3.2.03.2. Porous-layer open-tubular (PLOT) column In these columns there is a porous layer on the inner
wall. Porosity can be achieved by either chemical means (e.g. etching) or by the deposition of porous particles
on the wall from a suspension. The porous layer may serve as a support for a liquid stationary phase or as the
stationary phase itself.
4722 APPENDIX 12A / NOMENCLATURE / Chromatography

3.2.03.3. Support-coated open-tubular (SCOT) column A version of a PLOT column in which the porous
layer consists of support particles and was deposited from a suspension.

3.2.04. Capillary column A general term for columns having a small diameter. A capillary column may
contain a packing or have the stationary phase supported on its inside wall. The former case corresponds to
a Packed Capillary Column while the latter case corresponds to an Open-Tubular Column. Due to the
ambiguity of this term its use without an adjective is discouraged.

3.2.05. Column volume (Vc) The geometric volume of the part of the tube that contains the packing:

Vc"AcL

where Ac is the internal cross-sectional area of the tube and L is the length of the packed part of the column.
In the case of wall-coated open-tubular columns the column volume corresponds to the geometric volume of
the whole tube having a liquid or a solid stationary phase on its wall.

3.2.06. Bed volume Synonymous with Column Volume for a packed column.

3.2.07. Column diameter (dc) The inner diameter of the tubing.

3.2.08. Column radius (rc) The inside radius of the tubing.

3.2.09. Column length (L) The length of that part of the tube which contains the stationary phase.

3.2.10. Cross-sectional area of the column (Ac) The cross-sectional area of the empty tube:

Ac"r2c"(dc/2)2

3.2.11. Interparticle volume of the column (Vo) The volume occupied by the mobile phase between the
particles in the packed section of a column. It is also called the Interstitial Volume or the Void Volume of the
column.

3.2.11.1. In liquid chromatography, the interparticle volume is equal to the mobile-phase hold-up volume
(VM) in the ideal case, neglecting any extra-column volume.

3.2.11.2. In gas chromatography, the symbol VG may be used for the interparticle volume of the column. In
the ideal case, neglecting any extra-column volume, VG is equal to the corrected gas hold-up volume (VoM) (see
3.6.03 and 3.7.04):
VG"V oM"VM ) j

3.2.12. Interparticle porosity (
) The interparticle volume of a packed column per unit column volume:

"Vo /Vc

It is also called the Interstitial Fraction of the column.

3.2.13. Extra-column volume The volume between the effective injection point and the effective detection
point, excluding the part of the column containing the stationary phase. It is composed of the volumes of the
injector, connecting lines and detector.

3.2.13.1. Dead-volume This term is also used to express the extra-column volume. Strictly speaking, the
term `dead-volumea refers to volumes within the chromatographic system which are not swept by the mobile
phase. On the other hand, mobile phase is Sowing through most of the extra-column volumes. Due to this
ambiguity the use of the term `dead-volumea is discouraged.

3.2.14. Liquid-phase Vlm thickness (df) A term used in connection with open-tubular columns to express
the average thickness of the liquid stationary phase Rlm coated on the inside wall of the tubing.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4723

3.2.15. Stationary-phase volume (VS) The volume of the liquid stationary phase or the active solid in the
column. The volume of any solid support is not included. In the case of partition chromatography with a liquid
stationary phase, it is identical to the Liquid-Phase Volume (VL).

3.2.16. Mass (weight) of the stationary phase (WS) The mass (weight) of the liquid stationary phase or the
active solid in the column. The mass (weight) of any solid support is not included. In the case of partition
chromatography with a liquid stationary phase it is identical to the Liquid Phase Mass (Weight) (WL).

3.2.17. Phase ratio () The ratio of the volume of the mobile phase to that of the stationary phase in
a column:
"Vo /VS

In the case of open-tubular columns the geometric internal volume of the tube (Vc) is to be substituted for Vo.

3.2.18. SpeciVc permeability (Bo) A term expressing the resistance of an empty tube or packed column to
the Sow of a Suid (the mobile phase). In the case of a packed column

d 2p3 d 2p
Bo" 2+
180(1!) 1000

In the case of an open-tubular column

r2c
Bo"
8

3.2.19. Flow resistance parameter (
) This term is used to compare packing density and permeability of
columns packed with different particles; it is dimensionless.

"d 2p/Bo

where dp is the average particle diameter. In open-tubular columns "32.

3.3. The Chromatogram

3.3.01. Differential chromatogram A chromatogram obtained with a differential detector (see Figure 1A).

3.3.02. Integral chromatogram A chromatogram obtained with an integral detector (see Figure 1B).

Figure 1 Typical chromatogram: A, differential record produced by differential detector; B, integral record produced by integral
detector.
4724 APPENDIX 12A / NOMENCLATURE / Chromatography

Figure 2 Typical planar chromatogram.

3.3.03. Starting point or line The point or line on a chromatographic paper or layer where the substance to
be chromatographed is applied (P in Figure 2).

3.3.04. Spot A zone in paper and thin-layer chromatography of approximately circular appearance.

3.3.04.1. Spot diameter (ST in Figure 2) The width of the sample component spot before or after
chromatography.

3.3.05. Baseline The portion of the chromatogram recording the detector response when only the mobile
phase emerges from the column.

3.3.06. Peak The portion of a differential chromatogram recording the detector response when a single
component is eluted from the column (see Figure 1A). If separation is incomplete, two or more components
may be eluted as one Unresolved Peak.

3.3.06.1. Peak base (CP in Figure 1A) The interpolation of the baseline between the extremities of the
peak.

3.3.06.2. Peak area (CHFEGJP in Figure 1A) The area enclosed between the peak and the peak base.

3.3.06.3. Peak maximum (E in Figure 1A) The point on the peak at which the distance to the peak base,
measured in a direction parallel to the axis representing detector response, is a maximum.

3.3.06.4. Peak height (EB in Figure 1A) The distance between the peak maximum and the peak base,
measured in a direction parallel to the axis representing detector response.

3.3.06.5. Standard deviation () The term in the exponent of the equation relating the width and height of
a Gaussian peak:

 
x2
y"yo . exp.!
22

where y is the peak height at any point on the peak, yo is the peak height at maximum, x is the distance from
the ordinate (i.e. half of the width at that point), and  is the standard deviation of the peak. In practice, the
standard deviation can be calculated from one of the peak-width values speciRed below.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4725

Figure 3 Widths of a Gaussian peak at various heights, as a function of the standard deviation of the peak.

3.3.06.6. Variance of the peak The square of the standard deviation (2).

3.3.07. Peak-widths Peak-widths represent retention dimensions (time or volume) parallel to the baseline.
If the baseline is not parallel to the axis representing time or volume, then the peak-widths are to be drawn
parallel to this axis. Three peak-width values are commonly used in chromatography (see Figure 1A and
Figure 3).

3.3.07.1. Peak-width at base (wb) (KL in Figure 1A and Figure 3) The segment of the peak base intercepted
by the tangents drawn to the inSection points on either side of the peak.

3.3.07.2. Peak-width at half height (wh) (HJ in Figure 1A and Figure 3) The length of the line parallel to
the peak base at 50% of the peak height that terminates at the intersection with the two limbs of the peak

Note: The peak-width at base (wb) may be called the base width. However, the peak width at half height (wh)
must never be called the half width because that has a completely different meaning. Also, the symbol
w1/2 should never be used instead of wh.

3.3.07.3. Peak-width at inUection points (wi) (FG in Figure 1A and Figure 3) The length of the line drawn
between the inSection points parallel to the peak base.

3.3.07.4. In the case of Gaussian (symmetrical) peaks, the peak-widths are related to the standard deviation
() of the peak according to the following equations:

wb"4

wh"2((2 ln 2)"2.355

wi"2

3.3.08. Tailing Asymmetry of a peak such that, relative to the baseline, the front is steeper than the rear. In
paper chromatography and thin-layer chromatography, it refers to the distortion of a spot showing a diffuse
region behind the spot in the direction of Sow.
4726 APPENDIX 12A / NOMENCLATURE / Chromatography

3.3.09. Fronting Asymmetry of a peak such that, relative to the baseline, the front is less steep than the
rear. In paper chromatography and thin-layer chromatography, it refers to the distortion of a spot, showing
a diffuse region in front of the spot in the direction of Sow.

3.3.10. Step The portion of an integral chromatogram recording the amount of a component, or the
corresponding change in the signal from the detector as the component emerges from the column (see
Figure 1B).

3.3.10.1. Step height (NM in Figure 1B) The distance, measured in the direction of detector response,
between straight-line extensions of the baselines on both sides of a step.

3.3.11. Internal standard A compound added to a sample in known concentration to facilitate the
qualitative identiRcation and/or quantitative determination of the sample components.

3.3.12. External standard A compound present in a standard sample of known concentration and volume
which is analysed separately from the unknown sample under identical conditions. It is used to facilitate the
qualitative identiRcation and/or quantitative determination of the sample components. The volume of the
external standard (standard sample) need not to be known if it is identical to that of the unknown sample.

3.3.13. Marker A reference substance chromatographed with the sample to assist in identifying the
components.

3.4. Diffusion
3.4.01. The diffusion coefRcient (D) is the amount of a particular substance that diffuses across a unit area in
1 s under the inSuence of a gradient of one unit.
It is usually expressed in the units cm2 s\1.

3.4.02. Diffusion coefVcient in the stationary phase (DS or DL) The diffusion coefRcient characterizing the
diffusion in the stationary phase. In partition chromatography with a liquid stationary phase, the symbol
DL may be used to express this term.

3.4.03. Diffusion coefVcient in the mobile phase (DM or DG) The diffusion coefRcient characterizing the
diffusion in the mobile phase. In gas chromatography where the mobile phase is a gas, the symbol DG may be
used to express this term.

3.4.04. Diffusion velocity (uD) This term is used in liquid chromatography in the expression of the reduced
mobile-phase velocity (see 3.6.05.3). The diffusion velocity expresses the speed of diffusion into the pores of
the particles:
uD"DM/dp

3.5. Temperatures
3.5.01. Ambient temperature (Ta) The temperature outside the chromatographic system.

3.5.02. Injection temperature The temperature within the injection device.

3.5.03. Separation temperature (Tc) The temperature of the chromatographic bed under isothermal opera-
tion. In column chromatography it is called the Column Temperature.

3.5.04. Temperatures during programmed-temperature analysis

3.5.04.1. Initial temperature The temperature of the chromatographic bed (column) at the start of the
analysis. Temperature programming might start immediately upon sample introduction or it can be preceded
by a short isothermal period (Initial Isothermal Temperature). In the case, the time of the Initial Isothermal
Period must also be speciRed.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4727

3.5.04.2. Program rate The rate of increase of column temperature. The rate of temperature increase is
usually linear (3C.min\1) but it may also be non-linear. During one analysis the temperature rate may be
changed and/or the temperature programming may be interrupted by an isothermal period. In this case one is
speaking about Multiple Programming. In multiple programming each program must be speciRed by its initial
and Rnal temperatures and program rate.

3.5.04.3. Mid-analysis isothermal temperature The temperature of the column in an isothermal period
during elution. The corresponding time (Mid-Analysis Isothermal Period) must also be speciRed.

3.5.04.4. Final temperature The highest temperature to which the column is programmed.

3.5.04.5. Final isothermal temperature The Rnal temperature of the program if it is followed by an
isothermal period. The time corresponding to the Final Isothermal Period must also be speciRed.

3.5.04.6. Retention temperature The column temperature corresponding to the peak maximum.

3.5.05. Detector temperature The temperature of the detector cell. In the case of a detector incorporating
a Same, it refers to the temperature of the detector base.

3.6. The Mobile Phase

3.6.01. Mobile phase viscosity () The viscosity of the mobile phase at the temperature of the chromato-
graphic bed.

3.6.02. Pressures
3.6.02.1. Inlet pressure (pi) The absolute pressure at the inlet of a chromatographic column.

3.6.02.2. Outlet pressure (po) The absolute pressure at the exit of a chromatographic column. It is usually
but not necessarily equal to the Ambient Pressure (pa), the atmospheric pressure outside the chromatographic
system.

3.6.02.3. Pressure drop (p) The difference between the inlet and outlet pressures:

p"pi!po

3.6.02.4. Relative pressure (P) The ratio of the inlet and outlet pressures:

P"pi /po

3.6.03. Mobile phase compressibility correction factor (j ) A factor, applying to a homogeneously Rlled
column of uniform diameter, that corrects for the compressibility of the mobile phase in the column. It is
also called the Compressibility Correction Factor. In gas chromatography, the correction factor can be
calculated as:

3 p2!1 3 (pi/po)2!1
j" 3 "
2 p !1 2 (pi/po)3!1

In liquid chromatography the compressibility of the mobile phase is negligible.

Note: In former nomenclatures the term pressure gradient correction factor was sometimes used to express
the same term. This is, however, an incorrect name, because it is not the pressure gradient but the
compression of the mobile phase which necessitates the use of this factor. In liquid chromatography,
where mobile phase compression is negligible, no correction factor has to be applied to the mobile
phase velocity; however, there is still a pressure gradient along the column.
4728 APPENDIX 12A / NOMENCLATURE / Chromatography

3.6.04. Flow rate The volume of mobile phase passing through the column in unit time.
3.6.04.1. The Sow rate is usually measured at the column outlet, at ambient pressure (pa) and temperature
(Ta, in K); this value is indicated with the symbol F. If a water-containing Sowmeter was used for the
measurement (e.g. the so-called soap bubble Sowmeter) then F must be corrected to dry gas conditions in order
to obtain the Mobile Phase Flow Rate at Ambient Temperature (Fa):

Fa"F(1!pw/pa)

where pw is the partial pressure of water vapor at ambient temperature.

3.6.04.2. In order to specify chromatographic conditions in column chromatography, the Sow-rate (Mobile
Phase Flow Rate at Column Temperature, Fc) must be expressed at Tc (kelvin), the column temperature:
Fc"Fa(Tc/Ta)

3.6.05. Velocities
3.6.05.1. Mobile-phase velocity (u) The linear velocity of the mobile phase across the average cross-section
of the chromatographic bed or column. It can be calculated from the column Sow-rate at column temperature
(Fc), the cross-sectional area of the column (Ac) and the interparticle porosity ():

u"Fc/(Ac)

In practice the mobile phase velocity is usually calculated by dividing the column length (L) by the retention
time of an unretained compound (tM; see 3.7.03):

u"L/tM

3.6.05.2. In gas chromatography, due to the compressibility of the carrier gas, the linear velocity will be
different at different longitudinal positions in the column. Therefore two terms must be distinguished:
The Carrier Gas Velocity at the Column Outlet (uo) can be obtained as above, from the carrier gas Sow rate
measured at column outlet:

uo"Fc/(Ac)

The Average Linear Carrier Gas Velocity (uN ) is obtained from uo, by correcting it for gas compressibility:

uN "uoj

The average linear carrier gas velocity can also be obtained by dividing the column length (L) by the retention
time of an unretained compound (tM):

uN "L/tM

In liquid chromatography where mobile phase compression is negligible, uN "u.

3.6.05.3. Reduced mobile phase velocity () A term used mainly in liquid chromatography. It compares the
mobile phase velocity with the velocity of diffusion into the pores of the particles (the so-called diffusion
velocity, uD: see 3.4.04):

"uN /uD"uN dp/DM

In open-tubular chromatography:

"uN dc/DM
 APPENDIX 12A / NOMENCLATURE / Chromatography 4729

3.7. Retention Parameters in Column Chromatography

3.7.01. Retention parameters may be measured in terms of chart distances or times, as well as mobile phase
volumes; e.g., tYR (time) is analogous to V R (volume). If recorder speed is constant, the chart distances are
directly proportional to the times; similarly if the Sow rate is constant, the volumes are directly proportional to
the times.

Note: In gas chromatography, or in any chromatography where the mobile phase expands in the column, VM,
VR and VYR represent volumes under column outlet pressure. If Fc, the carrier gas Sow rate at the column
outlet and corrected to column temperature (see 3.6.04.2), is used in calculating the retention volumes
from the retention time values, these correspond to volumes at column temperatures.

3.7.02. The various conditions under which retention volumes (times) are expressed are indicated by
superscripts: thus, a prime ( ; as in V R) refers to correction for the hold-up volume (and time) while a circle (3;
as in VR3) refers to correction for mobile-phase compression. In the case of the net retention volume (time)
both corrections should be applied: however, in order not to confuse the symbol by the use of a double
superscript, a new symbol (VN, tN) is used for the net retention volume (time).

3.7.03. Hold-up volume (time) (VM, tM) The volume of the mobile phase (or the corresponding time)
required to elute a component the concentration of which in the stationary phase is negligible compared to that
in the mobile phase. In other words, this component is not retained at all by the stationary phase. Thus, the
hold-up volume (time) is equal to the Retention Volume (Time) of an Unretained Compound. The hold-up
volume (time) corresponds to the distance OA in Figure 1A and it includes any volumes contributed by the
sample injector, the detector, and connectors.

tM"VM/Fc

In gas chromatography this term is also called the Gas Hold-up Volume (Time).

3.7.04. Corrected gas hold-up (volume (VM3) The gas hold-up volume multiplied by the compression
(compressibility) correction factor (j):

VM3"VM.j

Assuming that the inSuence of extracolumn volume on VM is negligible,

VM3"VG

(see 3.2.11.2)

3.7.05. Total retention volume (time) (VR, tR) The volume of mobile phase entering the column between
sample injection and the emergence of the peak maximum of the sample component of interest (OB in
Figure 1a), or the corresponding time. It includes the hold-up volume (time):

tR"VR/Fc

3.7.06. Peak elution volume (time) (V

R, tR
) The volume of mobile phase entering the column between the
start of the elution and the emergence of the peak maximum, or the corresponding time. In most of the cases,
this is equal to the total retention volume (time). There are, however, cases when the elution process does not
start immediately at sample introduction. For example, in liquid chromatography, sometimes the column is
washed with a liquid after the application of the sample to displace certain components which are of no
interest and during this treatment the sample does not move along the column. In gas chromatography, there
are also cases when a liquid sample is applied to the top of the column but its elution starts only after a given
period. This term is useful in such cases.
4730 APPENDIX 12A / NOMENCLATURE / Chromatography

3.7.07. Adjusted retention volume (time) (V R, tR ) The total elution volume (time) minus the hold-up
volume (time). It corresponds to the distance AB in Figure 1a:

V R"V R!VM

t R"tR!tM"(V R!VM)/Fc"V R/Fc

3.7.08. Corrected retention volume (time) (VR3, tR3) The total retention volume (time) multiplied by the
compression correction factor (j):

VR3"VR . j

tR3"VR . j/Fc"VR3/Fc

3.7.09. Net retention volume (time) (VN, tN) The adjustment retention volume (time) multiplied by the
compression correction factor (j):

VN"V R . j

tN"V R . j/Fc"VN/Fc

3.7.10. In liquid chromatography, the compression of the mobile phase is negligible and thus, the compres-
sion correction factor does not apply. For this reason, the total and corrected retention volumes (times) are
identical (VR"V 3R; tR"tN) and so are the adjusted and net retention volumes (times) (V R"VN; t R"tN).

3.7.11. SpeciVc retention volumes

3.7.11.1. The speciTc retention volume at column temperature (VFg) The net retention volume per gram of
stationary phase (stationary liquid, active solid or solvent-free gel (WS):

VFg"VN/WS

3.7.11.2. SpeciTc retention volume at 03C (Vg) The value of V
g corrected to 03C:

273.15 K VN 273.15 K
Vg"VFg "
Tc WS Tc

where Tc is the column temperature (in kelvin).

3.7.12. Retention factor (k) The retention factor is a measure of the time the sample component resides in
the stationary phase relative to the time it resides in the mobile phase: it expresses how much longer a sample
component is retarded by the stationary phase than it would take to travel through the column with the
velocity of the mobile phase. Mathematically, it is the ratio of the adjusted retention volume (time) and the
hold-up volume (time):

k"V R/VM"t R/tM

If the distribution constant (see 3.9) is independent of sample component concentration, then the retention
factor is also equal to the ratio of the amounts of a sample component in the stationary and mobile phases
respectively, at equilibrium:

amount of component in stationary phase

k"
amount of component in mobile phase
 APPENDIX 12A / NOMENCLATURE / Chromatography 4731

If the fraction of the sample component in the mobile phase is R (see 3.7.13), then the fraction in the stationary
phase is (1!R); thus

k"(1!R)/R

Note: In former nomenclatures and in the literature one may Rnd the expressions Partition Ratio, Capacity
Ratio, Capacity Factor or Mass Distribution Ratio to describe this term.
In the literature the symbol k is often used for the retention factor, particularly in liquid chromatogra-
phy. The original reason for this was to clearly distinguish it from the partition coefRcient (distribution
constant) for which the symbol K had been utilized. Since, however, the distribution constants are all
identiRed with a subscript, there is no reason to add the prime sign to this symbol. It should be
emphasized that all the recognized nomenclatures (IUPAC, BS, ASTM) have always clearly identiRed
the capacity factor with the symbol k and not k .

3.7.12.1. Logarithm of the retention factor This term is equivalent to the RM value used in planar
chromatography (see 3.8.05). The symbol  is suggested to express log k:

"log k"log[(1!R)/R].

3.7.13. Retardation factor (R) The fraction of the sample component in the mobile phase at equilibrium; it
is related to the retention factor and other fundamental chromatography terms:

R"1/(k#1)

3.7.14. Relative retention values

3.7.14.1. Relative retention (r) The ratio of the adjusted or net retention volume (time) or retention factor
of a component relative to that of a standard, obtained under identical conditions:

r"V Ri/V R(st)"VNi/VN(st)"t Ri/t R(st)"ki/kst

Depending on the relative position of the peak corresponding to the standard compound in the chromatogram,
the value of r may be smaller, larger or identical to unity.

3.7.14.2. Separation factor () The relative retention value calculated for two adjacent peaks (V R2'V R1):

"V R2/V R1"VN2/VN1"t R2/t R1"k2/k1

By deRnition, the value of the separation factor is always greater than unity.
The separation factor is also identical to the ratio of the corresponding distribution constants.

Note: The separation factor is sometimes also called the selectivity. The use of this expression is dis-
couraged.

3.7.14.3. Unadjusted relative retention (rG or G) Relative retention calculated by using the total retention
volumes (times) instead of the adjusted or net retention volumes (times):

ki#1
rG"VRi/VR(st)"tRi/tR(st)"
kst#1

Subscript G commemorates E. Glueckauf, who Rrst used this expression.

3.7.14.4. Relative retention (r) and separation factor () values must always be measured under isothermal
conditions. On the other hand, the unadjusted relative retention (rG or G) values may also be obtained in
4732 APPENDIX 12A / NOMENCLATURE / Chromatography

programmed-temperature or gradient-elution conditions. Under such conditions, the symbol RRT (for
Relative Retention Time) has also been used to describe the unadjusted relative retention values.
Using the same stationary and mobile phases and temperature, the relative retention and separation factor
values are reproducible between chromatographic systems. On the other hand, the unadjusted relative
retention (and relative retention time) values are only reproducible within a single chromatographic system.

3.7.15. Retention index; KovaH ts (retention) index (I ) The retention index of a sample component is
a number, obtained by interpolation (usually logarithmic), relating the adjusted retention volume (time) or the
retention factor of the sample component to the adjusted retention volumes (times) of two standards eluted
before and after the peak of the sample component.
In the Kova& ts Index or Kova& ts Retention Index used in gas chromatography, n-alkanes serve as the
standards and logarithmic interpolation is utilized:

 
log Xi!log Xz
I"100 #z
log X(z#1)!log Xz

where X refers to the adjusted retention volumes or times, z is the number of carbon atoms of the n-alkane
eluting before, and (z#1) is the number of carbon atoms of the n-alkane eluting after the peak of interest:

V Rz(V Ri(VR(z#1)

The Kova& ts (Retention) Index expresses the number of carbon atoms (multiplied by 100) of a hypothetical
normal alkane which would have an adjusted retention volume (time) identical to that of the peak of interest
when analyzed under identical conditions.
The Kova& ts Retention Index is always measured under isothermal conditions. In the case of temperature-
programmed gas chromatography a similar value can be calculated utilizing direct numbers instead of their
logarithm. Since both the numerator and denominator contain the difference of two values, here we can use
the total retention volumes (times). Sometimes this value is called the Linear Retention Index:

 
t !tTRz
T
Ri
IT"100 #z
t T
!tTRz
R(z#1)

where tTR refers to the total retention times (chart distances) measured under the conditions of temperature
programming. The value of IT will usually differ from the value of I measured for the same compound under
isothermal conditions, using the same two phases.

3.8. Retention Parameters in Planar Chromatography

3.8.01. Mobile-phase front The leading edge of the mobile phase as it traverses the planar media. In all
forms of development except radial, the mobile phase front is essentially a straight line parallel to the mobile
phase surface. It is also called the Liquid Front or Solvent Front.

3.8.02. Mobile-phase distance The distance travelled by the mobile phase travelling along the medium
from the starting (application) front or line to the mobile phase front. It is the distance a in Figure 2.

3.8.03. Solute distance The distance travelled by the solute along the medium from the starting (applica-
tion) point or line to the center of the solute spot. If the solute spot is not circular, an imaginary circle is used
whose diameter is the smallest axis of the spot. It is the distance b in Figure 2.

3.8.04. Retardation Factor (RF) Ratio of the distance travelled by the center of the spot to the distance
simultaneously travelled by the mobile phase. Using the symbols of Figure 2:

RF"b/a
 APPENDIX 12A / NOMENCLATURE / Chromatography 4733

By deRnition the RF values are always less than unity. They are usually given to two decimal places. In order to
simplify this presentation the hRF Values may be used: they correspond to the RF values multiplied by 100.
Ideally, the RF values are identical to the R values (see 3.7.13).

3.8.05. RM value A logarithmic function of the RF value:

 
1!RF 1
RM"log "log !1
RF RF

3.8.06. Relative retardation (Rrel) This term is equivalent to relative retention used in column chromatogra-
phy: it is the ratio of the RF value of a component to the RF value of a standard (reference) substance. Since the
mobile phase front is common for the two components, the Rrel value can be expressed directly as the ratio of
the distances travelled by the spot of the compound of interest (bi) and the reference substance (bst)
respectively:

Rrel"RF(i)/RF(st)"bi/bst

Note: In former nomenclatures the symbol Rs was used to express relative retardation in planar chromatogra-
phy. Because of its identity with the symbol for peak resolution (see 3.10.01) the symbol Rrel is
suggested for relative retardation in planar chromatography.

3.9. Distribution Constants

The distribution constant is the concentration of a component in or on the stationary phase divided by the
concentration of the component in the mobile phase. Since in chromatography a component may be present in
more than one form (e.g. associated and dissociated forms), the analytical condition used here refers to the
total amount present without regard to the existence of various forms.
There terms are also called the Distribution CoefTcients. However, the present term conforms more closely
to the general usage in science.
The concentration in the mobile phase is always calculated per unit volume of the phase. Depending on the
way the concentration in the stationary phase is expressed various forms of the distribution constants may
exist.

3.9.01. Distribution Constant (Kc) In the general case, the concentration in the stationary phase is
expressed per unit volume of the phase. This term is mainly applicable to partition chromatography with
a liquid stationary phase but can also be used with a solid stationary phase:

Wi(S)/VS
Kc"
Wi(M)/VM

where Wi(S) and Wi(M) are the amounts of component i in the stationary and mobile phases, while VS and VM are
the volumes of the stationary and mobile phases, respectively.
The term Distribution Constant and the symbol Kc are recommended in preference to the term Partition
CoefTcient which has been in use in partition chromatography with a liquid stationary phase.
The value of Kc is related to the retention volume (VR) of a sample component and the volumes of the
stationary (VS) and mobile phases (VM) in the column:

VR"VM#KcVS

In gas chromatography both VR and VM have to be corrected for gas compressibility: therefore V3R(see 3.7.08)
is to be used for VR, and VG"V3M (see 3.2.11.2) is to be used for VM.

V3R"VG#Kc VS
4734 APPENDIX 12A / NOMENCLATURE / Chromatography

3.9.02. Distribution constant (Kg) In the case of a solid stationary phase, the distribution constant may be
expressed per mass (weight) of the dry solid phase:

Wi(S)/WS
Kg"
Wi(M)/VM

where Wi(st) and Wi(M) are the amounts (masses) of the component i in the stationary and mobile phases,
respectively, Wst is the mass (weight) of the dry stationary phase, and VM is the volume of the mobile phase in
the column.

3.9.03. Distribution constant (Ks) In the case of adsorption chromatography with a well characterized
adsorbent of known surface area, the concentration in the stationary phase may be expressed per unit surface
area:

Wi(S)/AS
Ks"
Wi(M)/VM

where Wi(S) and Wi(M) are the amounts (masses) of the component i in the stationary and mobile phases,
respectively, AS is the surface area of the stationary phase, and VM is the volume of the mobile phase in the
column.

Note: The symbols used in 3.9.01 through 3.9.03 are generalized.

3.10. Terms Expressing the Ef\ciency of Separation
3.10.01. Peak resolution (Rs) The separation of two peaks in terms of their average peak width at base
(tR2'tR1):

(tR2!tR1) 2(tR2!tR1)
Rs" "
(wb1#wb2)/2 wb1#wb2

In the case of two adjacent peaks it may be assumed that wb1+wb2, and thus, the width of the second peak
may be substituted for the average value:

Rs+(tR2!tR1 )/wb2

3.10.02. Separation number (SN) This expresses the number of peaks which can be resolved in a given part
of the chromatogram between the peaks of two consecutive n-alkanes with z and (z#1) carbon atoms in their
molecules:

tR(z#1)!tRz
SN" !1
whz#wh(z#1)

In the German literature the symbol TZ (Trennzahl) is commonly used to express the separation number.
As the separation number depends on the n-alkanes used for the calculation, they always must be speciRed
with any given SN value.

3.10.03. Plate number (N) A number indicative of column performance, calculated from the following
equations which depend on the selection of the peak width expression (see 3.3.07):

N"(V R/)2"(tR/)2

N"16(V R/wb)2"16(tR/wb)2

N"5.545(V R/wh)2"5.545(tR/wh)2

The value of 5.545 stands for 8 ln 2 (see 3.3.07.4). These expressions assume a Gaussian (symmetrical) peak.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4735

In these expressions the units for the quantities inside the brackets must be consistent so that their ratio is
dimensionless: i.e., if the numerator is a volume, then peak width must also be expressed in terms of volume.

Note: In former nomenclatures the expressions Number of Theoretical Plates or Theoretical Plate Number
were used for the same term. For simpliRcation, the present name is suggested.

3.10.04. Effective plate number (Neff) A number indicative of column performance calculated by using the
adjusted retention volume (time) instead of the total retention volume (time). It is also called the Number of
Effective Plates:
Neff"(V R/)2"(t R/)2

Neff"16(V R/wb)2"16(t R/wb)2

Neff"5.545(V R/wh)2"5.545(t R/wh)2

The plate number and effective plate number are related to each other:

 
2
k#1
N"Neff
k

Where k is the retention factor (see 3.7.12).

Notes: In the former literature the expression number of effective theoretical plates had been used to express
this term. This is incorrect since the plate number is either theoretical or effective, but cannot be both.
In former nomenclatures the respective symbols n and N have been used for the plate number and the
effective plate number. However, there was often a confusion in the proper selection of lower case and
capital letters; therefore, the present usage, characterizing the effective plate number by a subscript, is
suggested.

3.10.05. Plate height (H) The column length (L) divided by the plate number:

H"L/N

It is also called the Height Equivalent to One Theoretical Plate (HETP).

3.10.06. Effective plate height (Heff) The column length divided by the effective plate number:

Heff"L/Neff

It is also called the Height Equivalent to One Effective Plate.

Notes: In the former literature the expression height equivalent to one effective theoretical plate had been
used to express this term. This is incorrect, since the plate height is either theoretical or effective (see
3.10.04), but cannot be both.
In former nomenclatures the respective symbols h and H have been used for the plate height and the
effective plate height, respectively. However, there was often a confusion in the proper selection of
lower case and capital letters and also due to the fact that h (lower case letter) is also used to express
the reduced plate height (see 3.10.07). The present usage is suggested in order to avoid any confusion.

3.10.07. Reduced plate height (h) A term used in liquid chromatography. It is the ratio of the plate height
to the average particle diameter:
h"H/dp
For open-tubular columns:
h"H/dc
4736 APPENDIX 12A / NOMENCLATURE / Chromatography

4. Terms Related to Detection

4.1. Classi\cation of Detectors
4.1.01. ClassiVcation according to the form of response
4.1.01.1. Differential detectors These measure the instantaneous difference in the composition of the
column efSuent.

4.1.01.2. Integral detectors These measure the accumulated quantity of sample component(s) reaching the
detector.

4.1.02. ClassiVcation according to the basis of response

4.1.02.1. Concentration-sensitive detector A device the response of which is proportional to the concentra-
tion of a sample component in the eluent.

4.1.02.2. Mass-Uow sensitive detector A device the response of which is proportional to the amount of
sample component reaching the detector in unit time.

4.1.03. ClassiVcation according to Detector selectivity

4.1.03.1. Universal detector A detector which responds to every component in the column efSuent except
the mobile phase.

4.1.03.2. Selective detector A detector which responds to a related group of sample components in the
column efSuent.

4.1.03.3. SpeciTc detector A detector which responds to a single sample component or to a limited number
of components having similar chemical characteristics.
4.2. Detector Response
4.2.01. Detector sensitivity (S) The signal output per unit concentration or unit mass of a substance in the
mobile phase entering the detector.

4.2.01.1. In the calculation of detector sensitivity the signal output of the detector is given as peak area in
mV.min, A.s or AU.min (AU"absorbance unit). These values are obtained from the integrated peak area
converted to the units speciRed.
Alternately, the peak area can also be obtained by multiplying the peak height at maximum (in mV, A or
AU) by the peak-width at half height (in time units). The peak area calculated in this way will be 6% less than
the true integrated peak area, assuming that peak is Gaussian.

4.2.01.2. In the case of concentration-sensitive detectors, sensitivity is calculated per unit concentration in
the mobile phase:
S"AiFc /Wi"E/Ci

where Ai is the integrated peak area (in mV.min or AU.min), E is the peak height (in mV or AU), Ci is the
concentration of the particular substance in the mobile phase at the detector (in g.cm\3), Fc is the mobile phase
Sow rate corrected to column temperature (in cm3.min\1) and Wi is the mass (amount) of the substance
present (in mg). The dimensions of detector sensitivity are mV.cm3.mg\1 or AU.cm3.mg\1.

4.2.01.3. In the case of thermal-conductivity detectors, this sensitivity value is also called the Dimbat-Porter-
Stross Sensitivity of the detector.
In the case of mass-Uow sensitive detectors, sensitivity is calculated per unit mass of the test substance in the
mobile phase entering the detector:
S"Ai /Wi"Ei /Mi
 APPENDIX 12A / NOMENCLATURE / Chromatography 4737

where Ai is the integrated peak area (in A.s), Ei is the peak height (in A), Mi is the mass rate of the test substance
entering the detector (in g.s\1), and Wi is the mass (amount) of test substance present (in g). The dimension of
detector sensitivity is A.s.g\1 or C.g\1.

4.2.02. Relative detector response factor (f ) The relative detector response factor expresses the sensitivity
of a detector relative to a standard substance. It can be expressed on an equal mole, equal volume or equal
mass (weight) basis:
fi"(Ai/Ast)fst

where A refers to the peak area of the compound of interest (subscript i) and standard (subscript st)
respectively, and fst is the response factor of the standard compound. Usually, an arbitrary value (e.g. 1
or 100) is assigned to fst. Expressing the relative molar responses and using n-alkanes as the standards, the
assigned value of fst is usually the number of carbon atoms of the n-alkanes multiplied by 100 (e.g. 600 for
n-hexane).
If the relative detector response factor is expressed on an equal mass (weight) basis, the determined
sensitivity values can be substituted for the peak area.

4.3. Noise and Drift

4.3.01. Noise (N) (see Figure 4) The amplitude expressed in volts, amperes, or absorbance units of the
envelope of the baseline which includes all random variations of the detector signal the frequency of which is in
the order of one or more cycles per minute. In the case of photometric detectors the amplitude may be
expressed in absorbance units per unit cell length.

4.3.02. Drift (see Figure 4) The average slope of the noise envelope, expressed in volts, amperes, or
absorbance units per hour. It may be actually measured for 0.5 hour and extrapolated to one hour.

4.4. Minimum Detectability

The concentration or mass Sow of a sample components in the mobile phase gives a detector signal equal to
twice the noise level. It can be calculated from the measured sensitivity (S) and noise (N):

D"2N/S

where D is the minimum detectability, expressed either as concentration or mass-Sow of the substance of
interest in the mobile phase at the detector. Both sensitivity and minimum detectability must be determined for
the same substance.

Figure 4 Measurement of the noise and drift of a chromatographic detector.

4738 APPENDIX 12A / NOMENCLATURE / Chromatography

4.5. Linear and Dynamic Ranges

4.5.01. Linear range The linear range of a chromatographic detector represents the range of concentrations
or mass Sows of a substance in the mobile phase at the detector over which the sensitivity of the detector is
constant within a speciRed variation, usually $5 percent.

4.5.01.1. The best way to present detector linear range is the Linearity Plot (see Figure 5) plotting detector
sensitivity against amount injected, concentration or mass Sow-rate. Here, the upper limit of linearity can be
graphically established as the amount, concentration, or mass Sow-rate) at which the deviation exceeds the
speciRed value ($x% window around the plot). The lower limit of linearity is always the minimum detectable
amount determined separately for the same compound.

4.5.01.2. Alternatively, the linear range of a detector may be presented as the plot of peak area (height)
against concentration or mass Sow-rate of the test substance in the column efSuent at the detector (see
Figure 6). This plot may be either linear or log/log. The upper limit of linearity is that concentration (mass
Sow-rate) at which the deviation from an ideal linearity plot is greater than the speciRed percentage deviation
($x% window).

4.5.01.3. Numerically, the linear range can be expressed as the ratio of the upper limit of linearity obtained
from the linearity plot and the minimum detectability, both measured for the same substance.

4.5.01.4. When presenting the linear range of a detector, either as a plot as a numerical value, the test
substance, the minimum detectability, and the speciRed deviation must be stated.

4.5.02.
4.5.02.1. Dynamic range The dynamic range of detector is that range of concentration or mass Sow-rates
of a substance over which an incremental change in concentration or mass Sow-rate produced an incremental
change in detector signal. Figure 6 Presents a plot used for the determination of the dynamic range of
a detector.

4.5.02.2. The lower limit of the dynamic range is the minimum detectability. The upper limit is the highest
concentration at which a further increase in concentration (mass Sow-rate) will still give an observable

Figure 5 Linearity plot of a chromatographic detector. The Figure 6 Determination of the linear and dynamic ranges of
scale of the ordinate is linear: the scale of the abscissa may be a chromatographic detector. Such a plot is usually in a log-log
either linear or logarithmic. scale.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4739

increase in detector signal, and the dynamic range is the ratio of the upper and lower limits. The dynamic range
is greater than the linear range.

4.5.02.3. Numerically the dynamic range can be expressed as the ratio of the upper limit of the dynamic
range obtained from the plot and the minimum detectability, both measured for the same substance.

4.5.02.4. When expressing the dynamic range of a detector, the test substance and the minimum detectability
must be stated.

Table 1 Index of terms

Compound terms are generally listed at two places: under the main term and also under the full name. Exception was made when both
the main term and its adjective start with same letter. For example, ion-exchange chromatography is listed as ion-exchange
chromatography and chromatography, ion-exchange; on the other hand, column chromatography is listed only as chromatography,
column. The numbers refer to the relevant sections. Terms specifically used in planar chromatography (PC), ion-exchange
chromatography (IEC) and exclusion chromatography (EC) are indicated by the corresponding acronyms.

A Chamber saturation (PC) 2.2.02.7

Active solid 3.1.01 Chromatogram 1.1.02
Adjusted retention time 3.7.07; (EC) 6.2.03 - differential 3.3.01
Adjusted retention volume 3.7.07; (EC) 6.2.03 - integral 3.3.02
Adsorption chromatography 1.5.01 Chromatograph 1.1.03; 1.1.04
Affinity chromatography 1.5.05 Chromatographic bed 1.1.05
Ambient pressure 3.6.02.2 Chromatography 1.1.01
Ambient temperature 3.5.01 - adsorption 1.5.01
Anion exchange (IEC) 5.1.10 - affinity 1.5.05
Anion-exchange membrane (IEC) 5.3.04 - column 1.3.01
Anion exchanger (IEC) 5.3.03 - displacement 1.2.02
- basic form 5.3.03.1 - elution 1.2.03
Anion-exchange resin (IEC) 5.3.03 - exclusion 1.5.04
Anticircular development (PC) 2.2.02.6 - frontal 1.2.01
Application point (line) (PC) 3.3.03 - gas 1.1.06; 1.4.02
Ascending elution (development) (PC) 2.2.02.2 - gas-liquid 1.4.01
- gas-solid 1.4.01
B - gel-permeation 1.5.04
Band 3.3.05 - ion 1.5.03
Baseline 3.3.05 - ion-exchange 1.5.03
Bed density (IEC) 5.6.03 - ion-exclusion 1.5.04
Bed volume 3.2.06 - isothermal 1.6.07
Bed volume capacity (IEC) 5.4.03 - liquid 1.1.06; 1.4.03
Bifunctional ion exchanger (IEC) 5.3.01.3 - liquid-liquid 1.4.01
Binder (PC) 3.1.04 - liquid-solid 1.4.01
Bonded phase 1.1.05.1 - normal-phase 1.6.02
Break-through capacity (IEC) 5.4.05 - open-bed 1.3.02
Bypass injector 2.1.02.2 - paper 1.3.02
- partition 1.5.02
C - planar 1.3.02
Capacity (IEC) - programmed-flow 1.6.09
- bed-volume 5.4.03 - programmed-temperature 1.6.08
- break-through 5.4.05 - pyrolysis-gas 1.6.11.1
- practical specific 5.4.04 - reaction 1.6.11
- theoretical specific 5.4.01 - reversed-phase 1.6.01
- volume 5.4.02 - supercritical-fluid 1.4.04
Capacity factor 3.7.12; (EC) 6.2.05 - thin-layer 1.3.02
Capacity ratio 3.7.12 - two-dimensional 1.6.06
Capillary column 3.2.04 Circular elution (development) (PC) 2.2.02.5
- open-tubular 3.2.03 Co-ions (IEC) 5.1.08
- packed 3.2.04 Column 3.2
Carrier gas 1.1.06 - capillary 3.2.04
- velocities 3.6.05.2 - cross-sectional area 3.2.10
Cation exchange (IEC) 5.1.09 - diameter 3.2.07
Cation-exchange membrane (IEC) 5.3.04 - interparticle porosity 3.2.12
Cation exchanger (IEC) 5.3.02 - interparticle volume 3.2.11; (EC) 6.1.01
- acid form 5.3.02.1 - interstitial fraction 3.2.12
4740 APPENDIX 12A / NOMENCLATURE / Chromatography

Table 1 Continued

Column 3.2 Distribution constants 3.9; (IEC) 5.6; (EC) 6.2.06

- interstitial volume 3.2.11; (EC) 6.1.01 Drift 4.3.02
- intraparticle volume (EC) 6.1.02 Dynamic range (of detector) 4.5.02
- intrastitial volume (EC) 6.1.02
- length 3.2.02 E
- open-tubular 3.2.03 Effective plate height 3.10.06; (EC) 6.3.04
- packed 3.2.02 Effective plate number 3.10.04; (EC) 6.3.04
- radius 3.2.08 Effluent 1.1.08
- temperature 2.1.03 Electron exchanger (IEC) 5.3.01.7
- void volume 3.2.11 Elute 1.1.07
- volume 3.2.05 Eluent 1.1.06
Column oven 2.1.03 Elution
Compression (Compressibility) - anticircular 2.2.02.6
correction factor 3.6.03 - ascending 2.2.02.2
Concentration-sensitive detectors 4.1.02.1 - chamber 2.2.02
Corrected gas hold-up volume (time) 3.7.04 - circular 2.2.02.5
Corrected retention volume (time) 3.7.08 - descending 2.2.02.4
Corrected selectivity coefficient (IEC) 5.5.03 - horizontal 2.2.02.3
Counter-ions (IEC) 5.1.02 - radial 2.2.02.5
- unsaturated 2.2.02.8
D Elution chromatography 1.2.03
Dead volume 3.2.13.1 Equilibration (PC) 2.2.02.9
Densitometer (PC) 2.2.04 Exclusion chromatography 1.5.04
Derivatisation External solution (IEC) 5.2.02
- post-column 1.6.11.2 External standard 3.3.12
- pre-column 1.6.11.2 Extracolumn volume 3.2.13
Descending elution (development) (PC) 2.2.02.4
F
Detector 2.1.05
Film thickness (of the liquid phase) 3.2.14
- concentration-sensitive 4.1.02.1
Final isothermal period 3.5.04.5
- differential 4.1.01.1
Final isothermal temperature 3.5.04.5
- dynamic range 4.5.02
Final temperature 3.5.04.4
- integral 4.1.01.2
Fixed ions (IEC) 5.1.03
- linearity plot 4.5.01.2
Flash vaporizer 2.1.02.4
- linear range 4.5.01
Flow programming 1.6.09
- mass-sensitive 4.1.02.2
Flow-rates 3.6.04
- minimum detectability 4.4
Flow-resistance parameter 3.2.19
- relative response factor 4.2.02
Fraction collector 2.1.04
- selective 4.1.03.2
Frontal chromatography 1.2.01
- sensitivity 4.2.01
Fronting 3.3.09
- specific 4.1.03.3
- universal 4.1.03.1 G
Develop (PC) 1.1.07 Gas chromatography 1.4.02
Developing chamber (PC) 2.2.02 Gas hold-up time 3.7.03
Development (PC) Gas hold-up volume 3.7.03
- anticircular 2.2.02.6 - corrected 3.7.04
- ascending 2.2.02.2 Gas-liquid chromatography 1.4.01
- circular 2.2.05.5 Gas phase 1.1.05
- descending 2.2.02.4 Gas sampling valve 2.1.02.7
- horizontal 2.2.02.3 Gas-solid chromatography 1.4.01
- radial 2.2.02.5 Gel-permeation chromatography 1.5.04
- saturated 2.2.02.7 Gradient elution 1.6.04
- unsaturated 2.2.02.8 Gradient layer 3.1.05
Differential chromatogram 3.3.01
Diffusion 3.4 H
Diffusion coefficient Height equivalent to one effective
- in ion exchanger (IEC) 5.5.01 plate 3.10.06; (EC) 6.3.04
- in mobile phase 3.4.03 Height equivalent to one theoretical
- in stationary phase 3.4.02 plate 3.10.05; (EC) 6.3.03
Diffusion velocity 3.4.04 Heterogeneous ion-exchange
Dimbat-Porter-Stross sensitivity 4.2.01.2 membrane (IEC) 5.3.04
Direct injector 2.1.02.1 High-performance liquid
Displacement chromatography 1.2.02 chromatography 1.4.03
Displacer 1.2.02 - - thin-layer chromatography 1.4.03
 APPENDIX 12A / NOMENCLATURE / Chromatography 4741

Table 1 Continued

Hold-up time 3.7.03 Liquid chromatography 1.4.03

Hold-up volume 3.7.03 Liquid-liquid chromatography 1.4.01
- corrected 3.7.04 Liquid phase 1.1.05
Homogeneous ion-exchange - film thickness 3.2.14
membrane (IEC) 5.3.04 - loading 3.1.10
Horizontal elution (development) (PC) 2.2.02.3 - mass (weight) 3.2.16
- volume 3.2.15
I Liquid-solid chromatography 1.4.01
Immobilized phase 1.1.05.2 Liquid stationary phase 1.1.05
Impregnation (PC) 3.1.06
Initial isothermal period 3.5.04.1 M
Initial isothermal temperature 3.5.04.1 Macroporous ion exchanger 5.3.01.5
Initial temperature 3.5.04.1 Marker 3.3.13
Injection temperature 3.5.02 Mass distribution ratio 3.7.12
Injector 2.1.02 Mass-flow sensitive detectors 4.1.02.2
- bypass 2.1.02.2 Mid-analysis isothermal period 3.5.04.3
- direct 2.1.02.1 Mid-analysis isothermal temperature 3.5.04.3
- flash vaporizer 2.1.02.4 Minimum detectability 4.4
- on-column 2.1.02.3 Mobile phase 1.1.06
- programmed-temperature - compression correction factor 3.6.03
vaporizer 2.1.02.6 - distance (PC) 3.8.02
- split 2.1.02.5 - flow-rates 3.6.04
- valve 2.1.02.2 - front (PC) 3.8.01
Inlet pressure 3.6.02.1 - reduced 3.6.05.3
Integral chromatogram 3.3.02 - velocities 3.6.05.1; 3.6.05.2
Integral detector 4.1.01.2 - viscosity 3.6.01
Internal standard 3.3.1 Modified active solid 3.1.02
Interparticle porosity 3.2.1.2 Monofunctional ion exchanger 5.3.01.2
Interparticle volume 3.2.11; (EC) 5.1.01 Multi-dimensional chromatography 1.6.06
Interstitial fraction 3.2.12 Multiple programming 3.5.04.2
Interstitial volume 3.2.11; (EC) 6.1.01
Intraparticle volume (EC) 6.1.02 N
Intrastitial volume (EC) 6.1.02 Net retention volume (time) 3.7.09
Ion chromatography 1.5.03 Noise 4.3.01
Ion-exchange chromatography 1.5.03 Normal-phase chromatography 1.6.02
Ion exchange (IEC) 5.1.01 Number of effective plates 3.10.04; (EC) 6.3.04
- isotherm 5.1.04 Number of theoretical plates 3.10.03; (EC) 6.3.03
Ion-exchange membrane (IEC) 5.3.04
Ion exchanger (IEC) 5.3.01 O
- anion 5.3.03 On-column injector 2.1.02.3
- bifunctional 5.3.01.3 Open-bed chromatography 1.3.02
- cation 5.3.02 Open-tubular column 3.2.03
- macroporous 5.3.01.5 - porous-layer 3.2.03.2
- monofunctional 5.3.01.2 - support-coated 3.2.03.3
- polyfunctional 5.3.01.4 - wall-coated 3.2.04.1
- redox 5.3.01.8 Outlet pressure 3.6.02.2
- salt form 5.3.01.6 Oven 2.1.03
Ion-exclusion chromatography 1.5.04
Ionogenic group 5.1.07 P
Isocratic analysis 1.6.03 Packing 3.1.07
Isothermal chromatography 1.6.07 - pellicular 3.1.07.2
Isothermal period - totally porous 3.1.07.1
- final 3.5.04.5 Packed capillary column 3.2.04
- initial 3.5.04.1 Packed column 3.2.02
- mid-analysis 3.5.04.3 Paper chromatography (PC) 1.3.02
Particle diameter 3.1.08
K Partition chromatography 1.5.02
KovaH ts Retention Index 3.7.15 Partition coefficient 3.9.01
Partition ratio 3.7.12
L Peak 3.3.06
Linearity plot (of detector) 4.5.01.2 - area 3.3.06.2
Linear range (of detector) 4.5.01 - base 3.3.06.1
Linear retention index 3.7.15 - fronting 3.3.09
4742 APPENDIX 12A / NOMENCLATURE / Chromatography

Table 1 Continued

Peak 3.3.06 Relative retention 3.7.14

- height 3.3.06.4 - unadjusted 3.7.14.3
- maximum 3.3.06.3 Relative retention time 3.7.14.4
- resolution 3.10.01; (EC) 6.3.01 RM value (PC) 3.8.05
- standard deviation 3.3.06.5 Resin matrix (IEC) 5.3.01.1
- tailing 3.3.08 Resolution 3.10.01; (EC) 6.3.01
- variance 3.3.06.6 Retardation factor 3.7.13; (PC) 3.8.04
- widths 3.3.07 Retention factor 3.7.12; (EC) 6.2.05
Peak elution volume (time) 3.7.06 - logarithm 3.7.12.1
Pellicular packing 3.1.07.2 Retention index 3.7.15
Permeability, specific 3.2.18 Retention parameters 3.7; (PC) 3.8
Perm-selectivity (IEC) 5.3.04.1 Retention temperature 3.5.04.6
Phase Retention times
- immobilized 1.6.02 - adjusted 3.7.07; (EC) 6.2.03
- mobile 1.1.06; 3.6 - corrected 3.7.08
- stationary 1.1.05 - net 3.7.09
Phase ratio 3.2.17 - peak elution 3.7.06
Planar chromatography 1.3.02 - total 3.7.05; (EC) 6.2.02
Plate height 3.10.05; (EC) 6.3.03 - total mobile-phase 6.2.04
- effective 3.10.06; (EC) 6.3.04 - unretained compound 3.7.03; (EC) 6.2.02
- reduced 3.10.07; (EC) 6.3.05 Retention volumes
- theoretical 3.10.05; (EC) 6.3.03 - adjusted 3.7.07; (EC) 6.2.03
Plate number 3.10.03; (EC) 6.3.03 - corrected 3.7.08
- effective 3.10.04; (EC) 6.3.04 - net 3.7.09
- theoretical 3.10.03; (EC) 6.3.03 - peak elution 3.7.06
Pneumatic pump 2.1.01.3 - specific 3.7.11
Pore radius 3.1.09 - total 3.7.05; (EC) 6.2.02
Porous-layer open-tubular column 3.2.03.2 - total mobile phase (EC) 6.2.04
Post-column derivatization 1.6.11.2 - unretained compound 3.7.03; (EC) 6.2.01
Practical specific capacity (IEC) 5.4.04 Reversed-phase chromatography 1.6.01
Pre-column derivatization 1.6.11.2
Pressure 3.6.02 S
- ambient 3.6.02.2 Salt form of ion exchanger (IEC) 5.3.01.6
- drop 3.6.02.3 Sample 1.1.09
- inlet 3.6.02.1 - component 1.1.10
- outlet 3.6.02.2 - injector 2.1.02
- programming 3.6.11 Sampling valve 2.1.02.2
- relative 3.6.02.4 Selective detectors 4.1.03.2
Programmed-flow chromatography 1.6.09 Selectivity coefficient (IEC) 5.5.02
Programmed pressure - corrected 5.5.03
chromatography 1.6.10 Sensitivity of detectors 4.2.01
Programmed-temperature - Dimbat-Porter-Stross 4.2.01.2
chromatography 1.6.08 Separation factor 3.7.14.2; (IEC) 5.5.04
Programmed-temperature vaporizer 2.1.02.6 Separation number 3.10.02
Separation temperature 2.1.03
Program rate 3.5.04.2
Solid stationary phase 1.1.05
Pumps 2.1.01
Solid support 3.1.03
- pneumatic 2.1.01.3
Solute 1.1.11
- reciprocating 2.1.01.2
Solute distance (PC) 3.8.03
- syringe 2.1.01.1
Solvent 1.1.12; (IEC) 5.2.01
Pyrolysis-gas chromatography 1.6.11.1
Sorption (IEC) 5.1.05
- isotherm (IEC) 5.1.06
R Specific capacity (IEC)
Radial elution (development) (PC) 2.2.02.5 - practical 5.4.04
Reaction chromatography 1.6.11 - theoretical 5.4.01
Reciprocating pump 2.1.01.2 Specific detectors 4.1.03.3
Redox ion exchanger (IEC) 5.3.01.8 Specific permeability 3.2.18
Redox polymers (IEC) 5.3.01.7 Specific resolution (EC) 6.3.02
Reduced plate height 3.10.07; (EC) 6.3.05 Specific retention volume 3.7.11
Reduced velocity (of mobile phase) 3.6.05.3 Split injection 2.1.02.5
Relative pressure 3.6.02.4 Spot (PC) 3.3.04
Relative response factor (of detector)4.2.02 - diameter 3.3.04.1
Relative retardation (PC) 3.8.06 Spotting device (PC) 2.2.01
 APPENDIX 12A / NOMENCLATURE / Chromatography 4743

Table 1 Continued

Standard deviation 3.3.06.5 Theoretical plate number 3.10.03; (EC) 6.3.03

Starting point (PC) 3.3.03 Theoretical specific capacity (IEC) 5.4.01
Stationary mobile-phase volume (EC) 6.1.03 Thin-layer chromatography 1.3.02
Stationary phase 1.1.05 Total mobile-phase time (EC) 6.2.04
- volume 3.2.15 Total mobile-phase volume (EC) 6.1.03; 6.2.04
- mass (weight) 3.2.16 Totally porous packing 3.1.07.1
Step 3.3.10 Trennzahl 3.10.02
- height 3.3.10.1 Two-dimensional chromatography 1.6.06
Stepwise elution 1.6.05
Supercritical-fluid chromatography 1.4.04 U
Support 3.1.03 Unadjusted relative retention 3.7.14.3
Support-coated open-tubular column 3.2.03.2 Universal detectors 4.1.03.1
Syringe pump 2.1.01.1 Unsaturated elution (development) (PC) 2.2.02.8

T V
Tailing 3.3.08 Valve injector 2.1.02.2
Temperature Variance 3.3.06.6
- ambient 3.5.01 Velocities (of mobile phase) 3.6.05.1
- column 2.1.03; 3.5.03 Viscosity (of mobile phase) 3.6.01
- detector 3.5.05 Visualization chamber (PC) 2.2.03
- final 3.5.04.4 Void volume 3.2.11
- final isothermal 3.5.04.5 Volume capacity (IEC) 5.4.02
- initial 3.5.04.1 Volume swelling ratio (IEC) 5.3.06
- initial isothermal 3.5.04.1
- injection 3.5.02 W
- mid-analysis isothermal 3.5.04.3 Wall-coated open tubular column 3.2.03.1
- program rate 3.5.04.2 Weight-swelling ratio in solvent (IEC) 5.3.05
- retention 3.5.04.6
- separation 2.1.03; 3.5.03 Z
Theoretical plate height 3.10.05; (EC) 6.3.03 Zone 1.1.13

Table 2 List of symbols

The numbers in parentheses refer to the relevant sections. Symbols used specifically in planar chromatography (PC), ion-exchange
chromatography (IEC) or exclusion-chromatography (EC) are indicated by the corresponding acronyms.

a Mobile phase-distance in PC (3.8.02)

A Peak area (4.2.01)
Ac Cross-sectional area of a column (3.2.10)
AS Surface area of stationary phase in column (3.9.03)
b Solute distance in PC (3.8.03)
Bo Specific permeability (3.2.18)
Ci Concentration of a test substance in the mobile phase at the detector (4.2.01.2)
dc Column inside diameter (3.2.07)
df Thickness of liquid phase film (3.2.14)
dp Particle diameter (3.1.08)
D Minimum detectability of a detector (4.4)
D Diffusion coefficient in general (3.4)
Dex Diffusion coefficient in an ion exchanger (5.5.01)
DG Diffusion coefficient in the gas phase (3.4.03)
DL Diffusion coefficient in the liquid stationary phase (3.4.02)
DM Diffusion coefficient in the mobile phase (3.4.03)
DS Diffusion coefficient in the stationary phase (3.4.02)
E Peak height (4.2.01.2)
f Relative detector response factor (4.2.02)
F Mobile-phase flow-rate, measured at column outlet under ambient conditions with a wet flowmeter (3.6.04.1)
Fa Mobile-phase flow-rate at ambient temperature (3.6.04.1)
Fc Mobile-phase flow-rate, corrected to column temperature (3.6.04.2)
h Reduced plate height (3.10.07)
hRF RF;100 (3.8.04)
H Plate height (height equivalent to one theoretical plate) (3.10.05)
Heff Effective plate height (height equivalent to one effective plate) (3.10.06; in EC: 6.3.04)
4744 APPENDIX 12A / NOMENCLATURE / Chromatography

Table 2 Continued

I Retention index; KovaH ts (retention) index (3.7.15)

lT Retention index obtained in programmed temperature analysis; Linear retention index (3.7.15)
j Mobile phase compression (compressibility) correction factor (3.6.03)
k Retention factor (capacity factor) (3.7.12)
ke Retention factor (capacity factor) in EC (6.2.05)
kA/B Selectivity coefficient in IEC (5.5.02)
kaA/B Corrected selectivity coefficient (IEC) (5.5.03)
K Distribution constants in general (3.9)
Kc Distribution constant in which the concentration in the stationary phase is expressed as mass of substance per volume of the
phase (3.9.01). In IEC, it refers to unit volume of the swollen ion exchanger (5.6.01)
Kg Distribution constant in which the concentration in the stationary phase is expressed as mass of substance per mass of the
solid phase (3.9.02). In IEC, it refers to unit mass of the dry ion exchanger (5.6.02)
Ks Distribution constant in which the concentration in the stationary phase is expressed as mass of substance per surface area
of the solid phase (3.9.03)
Kv Distribution constant used in IEC, in which the concentration in the stationary phase is expressed as volume of substance per
volume of the dry ion exchanger (5.6.03)
Ko Distribution constant in EC (6.2.06)
L Column length (3.2.09)
M in EC: molecular mass (6.3.01 & 6.3.02)
Mi Mass rate of the test substance entering the detector (4.2.01.3)
N Noise of a detector (4.3.01)
N Plate number (number of theoretical plates) (3.10.03)
Neff Effective plate number (number of effective plates) (3.10.04; in EC: 6.3.04)
p Pressure in general (3.6.02)
pa Ambient pressure (3.6.02.2)
pi Inlet pressure (3.6.02.1)
po Outlet pressure (3.6.02.2)
pw Partial pressure of water at ambient temperature (3.6.04.1)
p Pressure drop (3.6.02.3)
P Relative pressure (3.6.02.4)
QA Practical specific capacity of an ion exchanger (5.4.04)
QB Break-through capacity of an ion-exchange bed (5.4.05)
QV Volume capacity of an ion exchanger (5.4.02)
r Relative retention (3.7.14.1)
rc Inside column radius (3.2.08)
rG Unadjusted relative retention (3.7.14.3)
rp Pore radius (3.1.09)
R Retardation factor in column chromatography; fraction of a sample component in the mobile phase (3.7.12 & 3.7.13)
(R!1) Fraction of a sample component in the stationary phase in column chromatography (3.7.12)
RF Retardation factor in PC (3.8.04)
RM Logarithmic function of RF (PC) (3.8.05)
Rrel Relative retardation in PC (3.8.06)
Rs Peak resolution (3.10.01)
Rsp Specific resolution in EC (6.3.02)
R1/2 Peak resolution in EC (6.3.02)
S Detector sensitivity (4.2.01)
SN Separation number
t Time in general
ti Retention time corresponding to the interparticle volume (Vi) of the column (EC) (6.1.02)
tt Retention time corresponding to the total mobile phase volume (Vt) in the column (EC) (6.2.04)
to Retention time of an unretained compound in EC (6.2.01)
tM Mobile-phase hold-up time; except in EC (see 6.1.01) it is also equal to the retention time of an unretained compound (3.7.03)
tN Net retention time (3.7.09)
tR Total retention time (3.7.05; in EC: 6.2.02)
tTR Total retention time in temperature-programmed analysis (3.7.05 & 3.7.15)
tR Peak elution time (3.7.06)
t R Adjusted retention time (3.7.07; in EC: 6.2.03)
toR Corrected retention time (3.7.08)
T Temperature in general (always in kelvin) (3.5)
Ta Ambient temperature (3.5.01)
Tc Column temperature (3.5.03)
TZ Trennzahl number (separation number) (3.10.02)
u Mobile-phase velocity (3.6.05.1)
 APPENDIX 12A / NOMENCLATURE / Chromatography 4745

Table 2 Continued

uN Average linear carrier gas velocity (3.6.05.2)

uD Diffusion velocity (3.4.04)
uo Carrier gas velocity at column outlet (3.6.05.2)
V Volume in general
Vc Column volume (3.2.05)
V(DIE) Volume of dry ion exchanger (5.6)
Vext Extra-column volume (6.1.03)
Vg Specific retention volume at 03C (3.7.11.2)
V Fg Specific retention volume at column temperature (3.7.11.1)
Vi Intraparticle volume of column in EC (6.1.02)
VG Interparticle volume of column in GC (3.2.11.2)
VL Liquid-phase volume (3.2.15)
VM Mobile-phase hold-up volume; except in EC (see 6.1.01) it is also equal to the retention volume of an unretained compound
(3.7.03)
V oM Corrected gas hold-up volume (3.7.04)
VM Volume of mobile phase in column (3.9.01)
VN Net retention volume (3.7.09)
Vo Interparticle volume of column (3.2.11); in EC, it is also equal to the retention volume of an unretained compound (6.2.01)
VR Total retention volume (3.7.05; in EC: 6.2.02)
VR Peak elution volume (3.7.06)
V R Adjusted retention volume (3.7.07; in EC: 6.2.03)
V oR Corrected retention volume (3.7.08)
VS Volume of stationary phase in column (3.2.15, 3.9.01)
V(SIE) Volume of swollen ion-exchanger (5.6)
V (sol)
t Total mobile-phase volume in the column (the mobile phase hold-up volume in EC) (6.1.03 and 6.2.04)
wb Peak width at base (3.3.07.1)
wh Peak with at half height (3.3.07.2)
wi Peak width at the inflection points (3.3.07.3)
W Amount (mass) in general
Wi Amount (mass) of a test substance present (4.2.01.2)
Wi(IE) Amount of the component i in the ion exchanger (5.6)
Wi(M) Amount of component i in the mobile phase (3.9)
Wi(S) Amount of the component i in the stationary phase (3.9)
WL Amount (mass) of the liquid phase in the column (3.2.16)
WS Amount (mass) of the stationary phase in the column (3.2.16)
z Number of carbon atoms of a n-alkane eluted before the peak of interest (3.7.15)
(z#1) Number of carbon atoms of a n-alkane eluted after the peak of interest (3.7.15)

Greek symbols
 Separation factor (relative retardation) (3.7.4.2)
 Separation factor in IEC (5.5.04)
G Unadjusted separation factor (relative retention) (3.7.14.3)
 Phase ratio (3.2.17)
 Interparticle porosity (3.2.12)
 Mobile phase viscosity (3.6.01)
superscript in Vg
(3.7.11.1)
 log k (3.7.12.1)
 Reduced mobile phase velocity (3.6.05.3)

Bed density in IEC (5.6.03)
 Standard deviation of a Gaussian peak (3.3.06.5)
2 Variance of a Gaussian peak (3.3.06.5)

Flow resistance parameter (3.2.19)

Subscripts
The generally used subscripts are listed. There are a few specific subscripts not listed here.
a Ambient
c Column
eff Effective
f Film of liquid phase
i Compound of interest
o Outlet of column
p Particle
st Standard
4746 APPENDIX 12A / NOMENCLATURE / Chromatography

Table 2 Continued

Subscripts
G Gas phase
L Liquid stationary phase
M Mobile phase; also external solution in IEC
N Net (as in net retention time or volume; correction for both the holdup time (volume) and gas compressibility)
R Retention (as in retention time or volume)
S Stationary phase; in IEC: ion exchanger
1,2 Two adjacent (tR2'tR1 except in EC where M2'M1 and thus tR1'tR2

Superscripts
T
Indication that value was obtained in programmed-temperature analysis
Adjusted (as in adjusted retention time or volume)
o
Corrected (as in corrected retention time or volume)

Table 3 List of acronyms used in chromatography

EC Exclusion chromatography
GC Gas chromatography
GLC Gas-liquid chromatography
GLPC Gas-liquid partition chromatography
GPC Gel-permeation chromatography
GSC Gas-solid chromatography
HETP Height equivalent to one theoretical plate
HPLC High-performance liquid chromatography
IC Ion chromatography
IEC Ion-exchange chromatography
LC Liquid chromatography
LLC Liquid-liquid chromatography
LSC Liquid-solid chromatography
PC Paper chromatography or Planar chromatography
PLOT Porous-layer open-tubular (column)
PTV Programmed-temperature vaporizer
RRT Relative retention time
SCOT Support-coated open-tubular (column)
SFC Supercritical-fluid chromatography
TLC Thin-layer chromatography
WCOT Wall-coated open-tubular (column)

References
1. Preliminary Recommendations on Nomenclature and Presentation of Data in Gas Chromatography. Pure. Appl.
Chem. 1, 177}186 (1960).
2. Recommendations on Nomenclature and Presentation of Data in Gas Chromatography. Pure Appl. Chem. 8, 553}562
(1964).
3. Recommendations on Nomenclature for Ion Exchange. Information Bulletin Appendices on Tentative Nomenclature,
Symbols, Units and Standards, No. 5, IUPAC Secretariat, Oxford, January 1970.
4. Recommendations on Ion-Exchange Nomenclature. Pure Appl. Chem. 29, 619}624 (1972).
5. Recommendations on Nomenclature for Chromatography. Information Bulletin Appendices on Tentative Nomencla-
ture, Symbols, Units and Standards, No. 15, IUPAC Secretariat, Oxford, February 1972.
6. Recommendations on Nomenclature for Chromatography. Pure Appl. Chem. 37, 447}462 (1974).
7. Glossary of Terms of Gas Chromatography. British Standard 3382. British Standards Institution, London. First
published: 1963; latest revision 1969.
8. Gas Chromatography Terms and Relationships. ASTM E 355. American Society for Testing & Materials, Philadel-
phia, PA; originally published in 1968, latest revision: 1989.
9. Packed Column Gas Chromatography. ASTM E 260. American Society for Testing & Materials, Philadelphia, PA;
originally published in 1965, latest revision: 1991.
10. Calculation of Gas Chromatography Response Factors. ASTM D 4626. American Society for Testing & Materials,
Philadelphia, PA; originally published in 1986, latest revision: 1990.
11. Thermal Conductivity Detectors Used in Gas Chromatography. ASTM E 516. American Society for Testing & Mater-
ials, Philadelphia, PA; originally published in 1974, latest revision: 1991.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4747

12. Flame Ionization Detectors Used in Gas Chromatography. ASTM E 594. American Society for Testing & Materials,
Philadelphia, PA; originally published in 1977.
13. Electron Capture Detectors Used in gas Chromatography. ASTM E 697. American Society for Testing & Materials,
Philadelphia, PA; originally published in 1979, latest revision: 1991.
14. Nitrogen/Phosphorus Thermionic Ionization Detectors for Use in Gas Chromatography. ASTM E 1140. American
Society for Testing & Materials, Philadelphia, PA; originally published in 1986.
15. Flame Photometric Detectors Used in Gas Chromatography, ASTM E 840. American Society for Testing & Materials,
Philadelphia, PA; originally published in 1981, latest revision: 1991.
16. Supercritical-Fluid Chromatography Terms and Relationships. ASTM E 1449. American Society for Testing & Mater-
ials, Philadelphia, PA; originally published in 1992.
17. Liquid Chromatography Terms and Relationships. ASTM E 682. American Society for Testing & Materials,
Philadelphia, PA; Originally published in 1979.
18. Refractive Index Detectors Used in Liquid Chromatography. ASTM E 1303. American Society for Testing & Mater-
ials, Philadelphia, PA; originally published in 1989.
19. Fixed-Wavelength Photometric Detectors Used in Liquid Chromatography. ASTM E 685. American Society for
Testing & Materials, Philadelphia, PA; originally published in 1979.
20. Ion Chromatography Terms and relationships. ASTM 1151. American Society for Testing & Materials, Philadelphia,
PA; originally published in 1987.
21. Use of Liquid Exclusion Chromatography Terms and Relationships. ASTM D 3536. American Society for Testing
& Materials, Philadelphia, PA; originally published in 1978, latest revision: 1986.
22. Molecular Weight Averages and Molecular Weight Distribution by Liquid Exclusion Chromatography (Gel-Per-
meation Chromatography GPC) ASTM D 3536. American Society for Testing & Materials, Philadelphia, PA;
originally published in 1976, latest revision: 1991.
23. Molecular Weight Averages and Molecular Weight Distribution by Certain Polymers by Liquid Size Exclusion
Chromatography (Gel-Permeation Chromatography GPC) Using Universal Calibration. ASTM D 3593. American
Society for Testing & Materials, Philadelphia, PA; originally published in 1977, latest revision: 1986.
24. E. Stahl: Nomenclature in Chromatography. Chromatographia 1, 338}342 (1968).
25. L. S. Ettre: The Nomenclature of Chromatography. I. Gas Chromatography. J. Chromatogr. 165, 235}256 (1979).
26. L. S. Ettre: The Nomenclature of Chromatography. II. Liquid Chromatography. J. Chromatogr. 220, 29}63 (1981).
27. L. S. Ettre: The Nomenclature of Chromatography, III. General Rules for Future Revisions. J. Chromatogr. 220,
65}69 (1981).
28. I. Mills, T. Cvitas\ , K. Homann, N. Kalley, and K. Kuchitsu, Quantities, Units and Symbols in Physical Chemistry,
Blackwell ScientiRc Publications, Oxford, UK, 1988.

5. Special Terminology Used in Ion-exchange Chromatography

The general terms and deRnitions discussed in the previous chapters are also valid in ion-exchange chromatog-
raphy. In addition, the following terms and deRnitions refer speciRcally to this variant of the technique.

5.1. Basic De\nitions

5.1.01. Ion exchange The process of exchanging ions between a solution and an ion exchanger.

5.1.02. Counter-ions In an ion exchanger, the mobile exchangeable ions.

5.1.03. Fixed ions In an ion exchanger, the non-exchangeable ions which have a charge opposite to that of
the counter-ions.

5.1.04. Ion-exchange isotherm The concentration of a counter-ion in the ion exchanger expressed
as a function of its concentration in the external solution under speciRed conditions and at constant
temperature.

5.1.05. Sorption Uptake of electrolytes or non-electrolytes by the ion exchanger through mechanisms other
than pure ion exchange.

5.1.06. Sorption isotherm The concentration of a sorbed species in the ion exchanger, expressed as
a function of its concentration in the external solution under speciRed conditions and at constant temperature.
4748 APPENDIX 12A / NOMENCLATURE / Chromatography

5.1.07. Ionogenic groups The Rxed groupings in an ion exchanger which are either ionized or capable of
dissociation into Rxed ions and mobile counter-ions

5.1.08. Co-ions The mobile ionic species in an ion exchanger with a charge of the same sign as the Rxed
ions.

5.1.09. Cation exchange The process of exchanging cations between a solution and a cation exchanger.

5.1.10. Anion exchange The process of exchanging anions between a solution and an anion exchanger.

5.2. The Mobile Phase

5.2.01. Solvent The term used in classical ion exchange to express the mobile phase.

5.2.02. External solution The solution in contact with the ion exchanger which contains the ionized species
before and after exchange with the ion exchanger.

5.3. The Chromatographic Medium

5.3.01. Ion exchangers A solid or liquid, inorganic or organic substance containing ions exchangeable with
others of the same charge, present in a solution in which the ion exchanger is considered to be insoluble.

Note: It is recognized that there are cases where liquid exchangers are employed and where it may be difRcult
to distinguish between the separation process as belonging to ion exchange or liquid-liquid distribution,
but the broad deRnition given here is regarded as that which is most appropriate.

5.3.01.1. Resin matrix The molecular network of an ion exchanger which carries the ionogenic groups.

5.3.01.2. Monofunctional ion exchanger An ion exchanger containing only one type of ionogenic group.

5.3.01.3. Bifunctional ion exchanger An ion exchanger containing two types of ionogenic groups.

5.3.01.4. Polyfunctional ion exchanger An ion exchanger containing more than one type of ionogenic
groups.

5.3.01.5. Macroporous ion exchanger An ion exchanger with pores that are large compared to atomic
dimensions.

5.3.01.6. Salt form of an ion exchanger The ionic form of an ion exchanger in which the counter-ions are
neither hydrogen nor hydroxide ions. When only one valence is possible for the counter-ion, or its exact form
or charge is not known, the symbol or the name of the counter-ion without charge is used, e.g., sodium-form or
Na-form, tetramethylammonium-form, orthophosphate-form. When one of two or more possible forms is
exclusively present, the oxidation state may be indicated by a Roman numeral, e.g. FeII-form, FeIII-form.

5.3.01.7. Redox polymers Polymers containing functional groups which can be reversibly reduced or
oxidized. Electron Exchanger may be used as a synonym.

5.3.01.8. Redox ion exchangers Conventional ion exchangers in which reversible redox couples have been
introduced as counter-ions either by sorption or complex formation. They closely resemble redox polymers in
their behavior.

5.3.02. Cation exchanger Ion exchanger with cations as counter-ions. The term Cation-Exchange Resin
may be used in the case of solid organic polymers.

5.3.02.1. Acid form of a cation exchanger The ionic form of a cation exchanger in which the counter-ions
are hydrogen ions (H-form) or the ionogenic groups have added a proton forming an undissociated acid.
 APPENDIX 12A / NOMENCLATURE / Chromatography 4749

5.3.03. Anion exchanger Ion exchanger with anions as counter-ions. The term Anion-Exchange Resin may
be used in the case of solid organic polymers.

5.3.03.1. Base form of an anion exchanger The ionic form of an anion exchanger in which the counter-ions
are hydroxide groups (OH-form) or the ionogenic groups form an uncharged base, e.g. }NH2.

5.3.04. Ion-exchange membrane A thin sheet or Rlm of ion-exchange material which may be used to
separate ions by allowing the preferential transport of either cations (in the case of a Cation-Exchange
Membrane) or anions (in the case of an Anion-Exchange Membrane). If the membrane material is made from
only ion-exchanging material, it is called a Homogeneous Ion-Exchange Membrane. If the ion-exchange
material is embedded in an inert binder, it is called a Heterogeneous Ion-Exchange Membrane.

5.3.04.1. Perm-selectivity A term used to deRne the preferential permeation of certain ionic species
through ion-exchange membranes.

5.3.05. Weight-swelling ratio in solvent Mass of solvent taken up by unit mass of the dry ion exchanger.
The solvent must always be speciRed.

5.3.06. Volume-swelling ratio Ratio of the dry swollen volume to the true dry volume of the ion exchanger.

5.4. Capacity Values

5.4.01. Theoretical speciVc capacity Amount (mmol) of ionogenic group per mass (g) of dry ion exchanger.
If not otherwise stated, the capacity should be reported per mass (g) of the H-form of a cation exchanger and of
the Cl-form of an anion exchanger.

5.4.02. Volume capacity (Qv) Amount (mmol) or ionogenic group per volume (cm3) of swollen ion
exchanger. The ionic form of the ion exchanger and the medium should be stated.

5.4.03. Bed volume capacity Amount (mmol) of ionogenic group per bed volume (cm3) (see 3.2.06)
determined under speciRed conditions. The conditions should always be speciRed.

5.4.04. Practical speciVc capacity (QA) Total amount of ions (mmole) taken up per mass (g) of dry ion
exchanger under speciRed conditions. The conditions should always be speciRed.

5.4.05. Break-through capacity of ion-exchange bed (QB) The practical capacity of an ion exchanger bed,
obtained experimentally by passing a solution containing a particular ionic or molecular species through
a column containing the ion exchanger. This is under speciRed conditions and is determined by measuring the
amount of species which has been taken up when the species is Rrst detected in the efSuent or when the
concentration in the efSuent reaches some arbitrarily deRned value. The break-through capacity may be
expressed in millimoles or milligrams taken up per gram of dry ion exchanger of per cm3 of bed volume.

5.5. Diffusion, Selectivity and Separation

5.5.01. Diffusion coefVcient in the ion exchanger (Dex) The meaning of this term is the same as the speciRed
in 3.4.01-3.4.02.

5.5.02. Selectivity coefVcient (kA/B) The equilibrium coefRcient obtained by application of the law of mass
action to ion exchange and characterizing quantitatively the ability of an ion exchanger to select one of two
ions present in the same solution. The ions involved in the exchange should be speciRed as subscripts.
Examples:
Exchange: Mg2#!Ca2#

[Mg]S/[Ca]S
kMg/Ca"
[Mg]S/[Ca]S
4750 APPENDIX 12A / NOMENCLATURE / Chromatography

Exchange: SO2#!Cl\

[SO4]S/[Cl]2S
kSO4/Cl"
[SO4]M/[Cl]2M

In the above equations subscript S refers to the ion exchanger (stationary phase) and M to the external
solution (mobile phase). For exchanges involving counter-ions differing in their charges, the numerical value
of kA/B depends on the choice of the concentration scales in the ion exchanger and the external solution (molal
scale, molar scale, mole fraction scale, etc.). Concentration units must be clearly stated for an exchange of ions
of differing charges.

5.5.03. Corrected selectivity coefVcient (kaA/B) This is calculated in a way identical to the selectivity
coefRcient except that the concentrations in the external solutions are replaced by activities.

5.5.04. Separation factor (A/B) The deRnition of this term is identical to the deRnition given in 3.7.14.2. In
an exchange of counter-ions of equal charge the separation factor is equal to the selectivity coefRcient (see
5.5.01), provided that only one type of ion represents the analytical concentration (e.g. in exchanges of
K# and Na#) but not in systems where several individual species are included in the analytical concentrations.

5.6. Distribution constants

A Distribution Constant is the concentration of a component in the ion exchanger (the stationary phase)
divided by its concentration in the external solution (the mobile phase). The concentration in the external
solution is always calculated per unit volume. Depending on the way the concentration in the ion exchanger is
expressed three forms of the distribution constant may exist.
In 5.6.01-5.6.03, Wi(IE) and Wi(sol) are the amounts of the component i in the ion exchanger and in the
external solution; VSIE and DIE are the volumes of the swollen and dry ion exchanger, respectively; V(sol) is the
volume of the external solution.

5.6.01. Distribution constant (Kc) In this case, the concentration in the ion exchanger is calculated as mass
(weight)/volume and it refers to the swollen ion exchanger:

Wi(IE)/V(SIE)
Kc"
Wi(sol)/V(sol)

5.6.02. Distribution constant (Kg) In this case, the concentration in the ion exchanger is calculated as
mass/mass (weight/weight) and it refers to dry ion exchanger:

Wi(IE)/V(DIE)
Kg"
Wi(sol)/V(sol)

5.6.03. Distribution constant (Kv) In this case, the concentration in the ion exchanger is calculated as
volume/volume and it refers to the dry ion exchanger:

Vi(IE)/V(DIE)
Kv"
Wi(sol)/V(sol)

If the Bed Density is
, expressed in grams of dry resin per cm3 of bed, then

Kv"Kg

6. Special Terminology Used in Exclusion Chromatography

Besides the terms and deRnitions used in general in chromatography, a number of special terms exist in
exclusion chromatography. In addition, due to the different nature of the chromatographic separation, some
 APPENDIX 12A / NOMENCLATURE / Chromatography 4751

Figure 7 Retention characteristics in exclusion chromatography. A standard sample in analysed (top); subsequently, the retention
volumes (times) are plotted against the logarithms of the corresponding molecular weights. Peak A corresponds to a non-retained
sample component the molecules of which are larger than the largest pores in the gel particles (total exclusion); peak D corresponds to
a sample component the molecules of which are smaller than the smallest pores in the gel particles (total penetration).

of the general chromatographic terms have a different meaning here. For further explanation of some of the
terms, see Figure 7.
Below, only the chromatography terms are listed. For a discussion of the molecular weight terms calculated
from the chromatographic data see the specialized nomenclatures (e.g. refs. 15}18).

6.1. The Column

6.1.01. Interparticle volume of the column (Vo) The volume of the mobile phase in the interstices between
the gel particles. It is also called the Interstitial Volume of the column.
In exclusion chromatography, the interparticle volume of the column is equal to the retention volume of an
unretained compound; however, it is not equal to the mobile phase hold-up volume (Vt). The reason for this is
that is practice the mobile phase molecules are always smaller than the smallest pores of the column packing.
Thus, they will enter all the pores available in the packing and therefore, will be eluted last. As a contrast, in
general liquid chromatography, the mobile-phase hold-up volume (see 3.7.03) and the retention volume of
a non-retained compound are practically equal.

6.1.02. Intraparticle volume of the column (Vi) The volume of the mobile phase within the pores of the gel
particles. It is also called the Intrastitial Volume of the column or the Stationary Mobile-Phase Volume.
The retention time equivalent to Vi is ti:

ti"Vi/Fc

6.1.03. Total mobile-phase volume in column (Vt) The sum of the interparticle and intraparticle volumes:

Vt"Vo#Vi

In the deRnition of Vt the extra-column volume of the system (Vext; see 3.2.13) is neglected. If it is not
negligible, it must also be added:

Vt"Vo#Vi#Vext
4752 APPENDIX 12A / NOMENCLATURE / Chromatography

6.2. Retention Parameters

6.2.01. Retention volume (time) of an unretained compound (Vo, to) The retention volume of a sample
component the molecules of which are larger than the largest pores of the gel particles. These will be eluted
Rrst from the column. The corresponding retention time is to:

to"Vo /Fc

Ignoring any extra-column volume, Vo is equal to the interparticle volume of the column (see 6.1.01).

6.2.02. Retention volume (time) (VR, tR) The retention volume (time) of a sample component the molecules
of which are smaller than the largest pores of the gel particles but larger than the smallest pores. The
corresponding retention time is tR:

tR"VR /Fc

6.2.03. Adjusted retention volume (time) (V R, t R) The total retention volume less the retention volume of
an unretained compound:

V R"VR!Vo

The corresponding retention time is t R:

t R"tR!to"V R /Fc"(VR!Vo)/Fc

6.2.04. Total mobile phase volume (time) (Vt, tt) The retention volume (time) of a sample component
the molecules of which are smaller than the smallest pores of the gel particles. The corresponding retention
times is tt:

tt"Vt /Fc

6.2.05. Retention factor (ke) The ratio of the adjusted retention volume (time) and the retention volume
(time) of an unretained compound:

VR!Vo tR!to
ke" "
Vo to

It may also be called the Capacity Factor. However, the suggested expression better deRnes its real meaning
(see also 3.7.12).

6.2.06. Distribution constant in exclusion chromatography (ko) The fraction of the intraparticle volume
(the volume of the pores) available to the molecules of a particular sample component for diffusion:

VR!Vo
Ko"
Vi

For an unretained compound, VR"Vo and thus, Ko"0. On the other hand, for a compound the molecules of
which are smaller than the smallest pores, VR"Vt and thus, Ko"1. In other words, the value of Ko varies
between zero and unity.
In exclusion chromatography, Ko is related to the retention volume of a sample component and the inter-
and intraparticle volumes of the column (Vo and Vi, respectively) in a manner analogous to the relationship in
general liquid chromatography:

VR"Vo#KoVi
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4753

6.3. Ef\ciency Terms

6.3.01. Peak resolution (R1/2) The deRnition of this term is identical to that given in 3.10.01.

VR1!VR2
R1/2"
(wb1#wb2)/2

Here VR1 and VR2 represent the peaks corresponding to compounds with molecular masses M1 and M2 respec-
tively: by deRnition M2'M1. In exclusion chromatography, larger molecules are eluted Rrst, therefore,
VR1'VR2.
Because of the addition of a new term, the speciRc resolution (see 6.3.02), the symbol R1/2 is suggested for
peak resolution in exclusion chromatography.

6.3.02. SpeciVc resolution (Rsp) Peak resolution also considering the molecular masses of the two test
compounds:

VR1!VR2 1
Rsp"
(wb1#wb2)/2 log(M2/M1)

The test compounds used for the determination of the speciRc resolution should have a narrow molecular-mass
distribution (the ratio of the mass-average and number-average molecular masses should be equal to or less
than about 1.1) and differ by a factor of about 10 in their molecular masses.

Note: In some nomenclatures, the symbol Rs is used for the speciRc resolution. Due to the possibility of
confusing it with the general resolution term (see 3.10.01), the symbol Rsp is suggested here.

6.3.03. Plate number and plate height (N, H ) The deRnitions of these terms are identical to those given in
3.10.03 and 3.10.05.

6.3.04. Effective plate number and effective plate height (Neff, Heff) The deRnitions of these terms are
identical to those given in 3.10.04 and 3.10.06, except that the retention volume of a non-retained compound
(Vo; see 6.2.01) is used in the calculations:

   
VR!Vo 2
(VR!Vo) 2
Neff"16 "5.545
wb wh

   
tR!to 2
(tR!to) 2
Neff"16 "5.545
wb wh

Heff"L/Neff

6.3.05. Reduced plate height (h) The deRnition of this term is identical to that given in 3.10.07.

12B. Liquid-Liquid Distribution (Solvent Extraction)

(IUPAC Recommendations 1993)
Prepared for publication by
N. M. Rice, H. M. N. H. Irving (1980) and M. A. Leonard* (1987)
*The Queens University of Belfast, Belfast, UK
^ 1993 IUPAC
4754 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

Abstract
The widespread use of liquid-liquid distribution, ranging from an analytical chemical technique to a unit
operation in various Relds of chemical technology (e.g. petroleum reRning, nuclear fuel reprocessing, hydro-
metallurgy, food technology, biochemistry) has led to a proliferation of terminology. This paper extends the
scope of the deRnitions beyond those previously recommended (Pure Appl. Chem. 1970, 21, 111}113) and
a list of terms, including those previously published, is presented under the following general headings: general
deRnitions of phenomena, operations and relationships; components of the solvent phase; fundamental
parameters for quantitative description of liquid-liquid distribution systems; process terminology applicable to
large scale continuous operations.

Introduction
In 1970 IUPAC published Recommended Nomenclature for Liquid-Liquid Distribution [1] which represent-
ed the work of several generations of the Nomenclature Commission V.3 of the International Union of
Pure and Applied Chemistry, assisted by the work of an ad hoc working party which considered the whole
of the nomenclature of separation processes. The choice of terms selected for deRnition was somewhat
arbitrarily restricted to those commonly used in what might be termed small scale batch or laboratory
analytical procedures. Many requests were received to extend the range of terms deRned and the need
for a complete revision was emphasized by the publication independently of other nomenclature proposals
[2}5].
The wide variation [6] in the choice of symbols and in nomenclature adopted by authors of papers published
in the Proceedings of the International Conferences on Solvent Extraction in Gothenburg (1966) and in
Scheveningen (1971) convinced the organizing committee of the next conference at Lyons (ISEC-1974) of the
desirability of providing authors with an extensive set of recommendations in order to aim at consistency in
nomenclature and symbols in the published proceedings. These were drawn up by a working party of the
Solvent Extraction and Ion Exchange Group of the Society of Chemical Industry (SCI), after meetings at
Bradford and Birmingham (England) attended by representatives of all aspects of the Reld. Their report, drawn
up by Dr Rice (as Secretary) formed the basis of discussions at the IUPAC General Assembly held in Munich
during August 1973. This led to a new set of recommendations which were made available for comment at
ISEC-74. After further discussions at a meeting of Commission V.3 of IUPAC in London in November 1974
and at the IUPAC General Assembly in Madrid in 1975, a tentative document was issued as an Information
Bulletin [7] by IUPAC in July 1977 and further discussed at the IUPAC General Assembly in Warsaw in August
1977. At almost the same time the Solvent Extraction and Ion Exchange Group of the SCI authorized
publication of their considered proposals for Recommended Nomenclature for Solvent Extraction in Chem-
istry in Industry to coincide with the International Solvent Extraction Conference (ISEC-77) in Toronto [8].
A synthesis of all these proposals including the comments received on the tentative IUPAC proposals was
discussed at the IUPAC General Assembly in Davos in September 1979, and circulated in draft form at
ISEC-80 in Liege in September 1980, where some minor amendments to process terminology were proposed.
Further discussion took place at Commission meetings in London in December 1980 and Leuven in September
1981.
These revised recommendations have been drawn up in collaboration with the Committee of the Solvent
Extraction and Ion Exchange Group of the Society of Chemical Industry and have been discussed at several
International Solvent Extraction Conferences.
Following the rearrangement of Commissions at the 35th IUPAC General Assembly in Lund in 1989, this
project became the responsibility of the Limited Life Time Commission on Chromatography and Other
Analytical Separations.
Since automated methods of analysis frequently simulate many of the features of large-scale industrial
practice } not least in that the attainment of distribution equilibrium or quantitative extraction is not always
achieved or even necessary } it seemed important in revising the original nomenclature to increase the scope of
the terms previously deRned so as to provide chemical engineers, clinical biochemists, food technologists,
hydrometallurgists, nuclear technologists, petrochemists and physical as well as analytical chemists with
a comprehensive set of symbols and nomenclature. In order to distinguish what might be termed processing
nomenclature, from that of a more fundamental nature, the recommendations have been placed in separate
sections. However, the importance of achieving consistent usage and a common language for all users of
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4755

liquid-liquid extraction cannot be overstressed. The distinction between the two sets of terms is merely one of
convenience and does not imply any real difference in their importance. In many instances, it is
difRcult to decide the appropriate category for a certain term so that the classiRcation is somewhat
arbitrary. Furthermore, some additional terms commonly used in the metallurgical industry have been added
since publication of the tentative proposals [7].
In order to keep the deRnitions as general as possible the liquid phases involved have been designated as
extract or solvent rather than organic phase and other phase or feed rather than aqueous phase. The
need clearly to label and specify the relevant phase in any equations or graphs is emphasized. Symbols are
recommended for a few of the more important parameters. The clear designation of the extract phase
components is stressed; the term solvent should be reserved for the composite phase rather than any
individual component although it may be the only component in that phase. Terms which have become jargon
in particular industrial situations or which may be confusing because of ambiguity have been listed as `not
recommendeda. No attempt has been made to deRne standard chemical engineering terms, e.g. mass transfer
coefRcient, but certain terms, e.g. equilibrium constant are discussed in relation to their usage in
liquid-liquid distribution.
When the nomenclature is applied to other types of extraction systems (e.g. two immiscible aqueous
phases, such as a concentrated salt solution and a concentrated aqueous solution of poly(ethylene glycol),
two immiscible non-aqueous liquids or two immiscible molten salts), the two immiscible phases should
be clearly speciRed and can equally well be distinguished and denoted by the general terms, Phase I and
Phase II. It is also recognized that in the petrochemical and food-processing industries the extractant
phase may well be an inorganic liquid (e.g. liquid sulfur dioxide, supercritical carbon dioxide). This situ-
ation can be described by using the terms `epi-phasea and `hypo-phasea for the less dense and more
dense phases, respectively, recognizing that mass transfer could be in either direction. Occasionally,
cases arise where metals are partitioned among three liquid phases, such as concentrated KCl, aceto-
nitrile and hexane. The application of the recommended nomenclature to any situation is clearly a
matter of common sense. What is important is that in a given situation the phases used should be completely
speciRed.
Since the organic phase is commonly quite a complex solution of one or more organic liquids containing
one or more extractants and possibly modiRers of various sorts as well as a diluent, particular care has been
taken over the deRnitions applicable to these various components.
Concentration is frequently used in the list of deRnitions. In general any suitable concentration units may
be employed but the same units should be used for each phase and should be clearly speciRed in the text.
Following normal convention, activities rather than concentrations should be used where thermodynamic
equilibrium quantities are implied. However, use of the appropriate concentration quotient as an approxima-
tion for the equilibrium constant in the limiting case of dilute solutions or in conjunction with the appropriate
activity coefRcients follows logically.
One semantic problem is the occurrence of three English terms } extraction, distribution and partition
} to describe the transfer of a solute between phases whereas in many other languages only one suitable term
exists (e.g. Verteilung). An attempt has been made to obviate this difRculty in the present work but its
complete avoidance is not possible owing to established usage. The terms and symbols, which have been
deRned are listed in Tables 1 and 2, respectively. In some cases these represent changes, clariRcations, or
speciRc usages of previously deRned terms [9], in particular those related to chromatographic separations [10],
and the differences are not in Appendix 1.
The current reviser (MAL) suggests use of the term distribution when referring to the total concentration of
related species and partition when referring to a single species. The term constant should be reserved for
Rxed thermodynamic true constants, otherwise ratio should be used. A survey of the literature shows
reasonable agreement over the symbol and title of partition constant and distribution ratio but nomenclature
for the partition ratio is a nightmare. Appendix 2 summarizes past usage.

1. General De\nitions
1.1 Antagonism
The converse to synergism (1.23).
Note: The terms anti-synergism, antisynergic and anti-synergistic should not be used.
4756 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

1.2 Coextraction
Formation of mixed-species aggregates in a low-polarity organic phase.

1.3 Conditioning
A synonym for pre-equilibration (1.16).

1.4 Distribution
The apportionment of a solute between two phases.
Note: the term partition (1.15) or extraction (1.9) may also be used in this sense where appropriate.

1.5 Distribution Isotherm

The relationship (algebraic or graphical) between the concentration of a solute in the extract (2.7) and the
corresponding concentration of the same solute in the other phase at equilibrium at a speciRed temperature.
Note: Alternative terms in common use are equilibrium line (1.7) and in the appropriate contexts: extraction
isotherm, scrubbing isotherm and stripping isotherm. Partition isotherm is not normal usage.

1.6 Equilibration
The operation by which a system of two or more phases is brought to a condition where further changes with
time do not occur.
Note: This term is not synonymous with pre-equilibration (1.16) and should not be used in that sense.

1.7 Equilibrium Line

A plot of the distribution isotherm (1.5).

1.8 Extract (Verb)

To transfer a solute from a liquid phase to another immiscible or partially miscible liquid phase in contact
with it.
Notes:
(i) The term is also applied to the dissolution of material from a solid phase with a liquid in which it is not
wholly soluble (i.e. leaching). See solvent extraction (1.19).
(ii) For usage as a noun see under 2 Components of the Solvent Phase.

1.9 Extraction (in Liquid-liquid Distribution)

See liquid-liquid extraction.
Notes:
(i) See under 4 Process Terminology for a more speciRc usage of extraction.
(ii) Distribution (1.4) and partition (1.15) are often used as synonyms for the general phenomenon of
extraction where appropriate.

1.10 Liquid Ion Exchange

A term used to describe a liquiddliquid extraction process that involves a transfer of ionic species from the
extractant to the aqueous phase in exchange for ions from the aqueous phase.
Notes:
(i) The term does not imply anything concerning the nature of the bonding in the extracted complex.
(ii) The term `Solvent Ion Exchangea (SIX) is not recommended.

1.11 Liquid-liquid Distribution (Extraction) (Partition)

The process of transferring a dissolved substance from one liquid phase to another (immiscible or partially
miscible) liquid phase in contact with it.
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4757

Note: Although extraction, partition and distribution are not synonymous, extraction may replace distribu-
tion where appropriate.

1.12 Macro-element See Main Solute.

Notes:
(i) This term is vague and is not recommended.
(ii) Macroelement has a different meaning in analytical chemistry and the term major component, the
meaning of which is obvious, is preferable.

1.13 Main (extractable) solute.

That (or those) species transferred which is of greatest economic or chemical interest.

Note: It is not necessarily the species present at greatest concentration.

1.14 Micro-element
This term should not be used in the sense of a minor component or a contaminant in the feed to a liquid-liquid
distribution system.

Note: Microelement has a different meaning in analytical chemistry and the terms minor component,
impurity or contaminant the meaning of which are obvious, are preferable.

1.15 Partition
This term is often used as a synonym for distribution (1.4) and extraction (1.9). However, an essential
difference exists by deRnition between distribution constant or partition ratio (3.17) and partition
constant (3.16).

Note: This term should be, but is not invariably, applied to the distribution of a single deRnite chemical species
between the two phases.

1.16 Pre-equilibration
(i) Preliminary treatment of a solvent in order to convert the extractants into a suitable chemical form.
(ii) Preliminary treatment of either phase with a suitable solution of the other phase (in the absence of main
extractable solute(s) (1.13)) so that when the subsequent equilibration (1.6)) is carried out changes in the
(volume) phase ratio (3.19) or in the concentrations of other components are minimised.

Notes:
(i) The use of equilibration (1.6) in this sense is confusing and should be avoided.
(ii) The term conditioning may be used as a synonym for pre-equilibration.

1.17 Re-extraction
Since the preRx re- can signify back as well as again this term is ambiguous and should be avoided, except
where the process of extraction (e.g. from aqueous solution to an organic phase) in a single direction is
repeated (following stripping). It should not be used as a synonym for stripping (4.3) or back-extraction (4.1).

1.18 Salting Out

The addition of particular electrolytes to an aqueous phase in order to increase the distribution ratio (3.5) of
a particular solute.

Notes:
(i) The addition of electrolytes to improve phase separation behaviour should not be referred to as salting
out.
(ii) The term is also used for the addition of electrolytes to reduce the mutual partial miscibility of two
liquids.
(iii) It has no connection with synergism (1.23).
4758 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

1.19 Solvent Extraction

The process of transferring a substance from any matrix to an appropriate liquid phase. If the substance is
initially present as a solute in an immiscible liquid phase the process is synonymous with liquid-liquid
extraction (1.11).
Notes:
(i) If the extractable material is present in a solid (such as a crushed mineral or an ore) the term leaching may
be more appropriate. The extractable material may also be a liquid entrapped within or adsorbed on
a solid phase.
(ii) Common usage has established this term as a synonym for liquid}liquid distribution (1.11). This is
acceptable provided that no danger of confusion with extraction from solid phases exists in a given context.
1.20 Solvent Ion Exchange (SIX)
This term is not recommended (see liquid ion exchange) (1.10).
1.21 Sublation
A Sotation process in which the material of interest, adsorbed on the surface of gas bubbles in a liquid, is
collected on an upper layer of immiscible liquid.
Notes: There is no liquid-phase mixing in the bulk of the system; as a result recoveries can approach 100%.
1.22 Substoichiometric Extraction
Here the amount of reagent used is lower than that dictated by stoichiometry. If the constants of formation
and extraction of the complexes are high, the amount of extracted metal is dictated by the amount of
extractant introduced.
1.23 Synergism
A term describing the co-operative effect of two (or more) extractants (2.8) where the distribution ratio
(3.5) for the combination is greater than the largest individual distribution ratio (measured under comparable
conditions)
Notes:
(i) The corresponding adjective is synergic and the term synergistic should not be used.
(ii) No standard method for quantiRcation of the phenomenon has been agreed and any approach should be
clearly deRned in a given situation.

2. Components of the Solvent Phase

2.1 Accelerator See Catalyst (2.3), Kinetic Synergist (2.10), Modi\er (2.11)
Note: This term may be used as a synonym for catalyst.
2.2 Carrier See Diluent (2.5)
This term is not recommended.
2.3 Catalyst (in Liquid-liquid Distribution)
A substance included in the solvent (2.12) to increase the rate of transfer without affecting the position of
equilibrium.
Notes: The term accelerator may also be used but kinetic synergist is not recommended.
2.4 Cosolvent See Diluent (2.5)

2.5 Diluent
The liquid or homogeneous mixture of liquids in which extractant(s) (2.8) and possible modiTer(s) (2.11) may
be dissolved to form the solvent (2.12) phase.
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4759

Notes:
(i) The term carrier, which implied an inert diluent is not recommended.
(ii) Although the diluent may well be a single liquid or even the major portion of the extracting phase, the
term solvent (2.12) should not be used in this sense as it has a much wider meaning in the context of
liquid-liquid extraction, although the term cosolvent may be used in certain circumstances.
(iii) The diluent by itself does not extract the main (extractable) solute appreciably

2.6 Epi-Phase
The less dense phase in a distribution system.
Note: The term is often used when two non-aqueous phases are present or when the solvent (2.12) is an
aqueous solution. See also hypo-phase (2.9).

2.7 Extract (Noun)

The separated phase (often but not necessarily organic) that contains the material extracted from the other
phase.
Notes:
(i) Where appropriate the term `loaded solventa (4.15) may be used, but is not recommended.
(ii) For usage as a verb see 1.8.

2.8 Extractant
The active component(s) primarily responsible for transfer of a solute from one phase to the other.
Notes:
(i) The term extracting agent is a synonym but solvent (2.12) and ligand should not be used in this context.
(ii) Certain extractants that consist of liquids immiscible with water (e.g. Tributyl phosphate or certain
ketones) might comprise the only component of the initial organic phase but extractant(s) can also be
dissolved in diluent (2.5).

2.9 Hypo-Phase
The denser phase in an extraction system.
Note: The term is often used when two non-aqueous phases are present or when the solvent (2.12) is an
aqueous phase. See also epi-phase (2.6).

2.10 Kinetic Synergist

This term is not recommended as a synonym for catalyst (2.3) or accelerator (2.1).

2.11 Modi\er
A substance added to a solvent (2.12) to improve its properties e.g. by increasing the solubility of an extractant
(2.8), changing interfacial parameters, or reducing adsorption losses.
Note: Additives used to enhance extraction rates should be called accelerators (2.1) or catalysts (2.3).

2.12 Solvent (in Liquid-liquid Distribution)

The term applied to the whole initial liquid phase containing the extractant (2.8).
Notes:
(i) The solvent may contain only extractant or it may be a composite homogeneous mixture of extractant(s)
(2.8) with diluent(s) (2.5) and also sometimes modiTers (2.11) and accelerators (2.1).
(ii) The term solvent must not be used as a synonym for any of the individual components of a composite
liquid phase even where, in the case of a single component (e.g. 3-methylbutan-2-one or tributyl
phosphate), it becomes identical with the extractant.
(iii) The term may be qualiRed to denote the extract from a given processing step (4.41), e.g. loaded solvent
(4.15).
4760 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

3. Fundamental Parameters
3.1 Concentration Factor
Not recommended. See extraction factor (3.10).

3.2 Decontamination Factor

The ratio of the proportion of contaminant to product before treatment to the proportion after treatment. It is
the reciprocal of the enrichment factor (3.6).

3.3 Distribution Coef\cient

This term is not recommended as a synonym for distribution ratio (3.5).

3.4 Distribution Constant

A synonym for partition ratio (3.17).

3.5 Distribution Ratio (in Liquid-liquid Distribution) (D)

The ratio of the total analytical concentration of a solute in the extract (2.7) (regardless of its chemical form) to
its total analytical concentration in the other phase.
Notes:
(i) If there is possible confusion with the extraction factor or (mass) distribution ratio (3.13), the term
concentration distribution ratio (symbol DC) should be used, but this is not common usage. This is
reasonably compatible with chromatographic nomenclature.
(ii) The terms distribution coefTcient, extraction coefTcient and, where appropriate, scrubbing
coefTcient, stripping coefTcient are widely used alternatives but are not recommended. If they
must be used in a given situation the term ratio is preferable to coefTcient.
(iii) In equations relating to aqueous/organic systems the organic phase concentration is, by convention, the
numerator and the aqueous phase concentration the denominator. In the case of stripping ratio the
opposite convention is sometimes used but should then be clearly speciRed.
(iv) In the past there has been much confusion between the distribution ratio as deRned above, the value of
which varies with experimental conditions, e.g. pH, presence of complexing agents, extent of achieve-
ment of equilibrium, etc. and the true partition constant (3.16) which is by deRnition invariable or the
partition coefTcient or distribution constant which apply to a particular chemical species under
speciRed conditions. For this reason the terms distribution constant (3.4), partition constant (3.16),
partition coefTcient (3.15), partition ratio (3.17) and extraction constant (3.9) should not be used in
this context.
(v) The use of the ratio: light phase concentration to heavy phase concentration is ambiguous and is not
recommended.
(vi) The distribution ratio is an experimental parameter and its value does not necessarily imply that
distribution equilibrium between the phases has been achieved.

3.6 Enrichment Factor (in Liquid-liquid Distribution) (S )

The factor by which the ratio of the amounts of two substances in the feed (4.11) must be multiplied to give
their ratio after treatment.
QA/QB"SA,B(Q
A/Q
B) where QA and Q
A are the Rnal and initial amounts of species A and QB and Q
B are the
Rnal initial amounts of species B. Hence SA,B"EA/EB where E is the fraction extracted (3.11). In terms of D, n,
r (where n is the number of stages and r the phase ratio (3.19))

1!(1#rDA)\n
SA,B"
1!(1#rDB)\n

3.7 Extractability
A property which qualitatively indicates the degree to which a substance is extracted.
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4761

Note: This term is imprecise and generally used in a qualitative sense. It is not a synonym for fraction extracted
(3.11).

3.8 Extraction Coef\cient

This term is not recommended as a synonym for distribution ratio (3.5).

3.9 Extraction (Equilibrium) Constant at Zero Ionic Strength (K oex )

The equilibrium constant of the distribution reaction expressed in terms of the reacting species. Thus, for the
gross reaction:
#
aq #n HLorg & MLn,org#n Haq
Mn#

in which the reagent HL initially dissolved in an organic phase reacts with a metal ion Mn# in aqueous
solution to form a product MLn which is more soluble in the organic phase than in water,

aMLn,org;anH#, aq
Koex"
aMn#, aq;anHL,org

Notes:
(i) When concentrations are used instead of activities or mixed terms are employed as when H# and/or
Mn# are measured with an electrode, the appropriate name is extraction constant, symbol Kex,
accompanied by a careful deRnition. Koex may be termed the thermodynamic extraction constant.
(ii) The extraction constant is related to other terms relevant to such systems by:
DMLn n Kna
Kex"
DnHL
where n is the overall formation constant of MLn and Ka is the dissociation constant of HL. When the
reagent HL is more soluble in water than the other immiscible phase it may be more convenient to deRne
a special equilibrium constant in terms of HLaq:

Kex"DML,n n Kna

(iii) In distribution equilibria involving non-aqueous systems, e.g. liquid SO2, molten salts and metals, the
mass action equilibrium constant for the relevant extraction process can be identiRed with Kex which
should be explicitly deRned in this context.
(iv) In actual practice, it may be necessary to include other terms to take into account other complexes
formed by auxiliary reagents and the solvation and/or polymerization of the various species. In such
cases, Kex must be deRned with reference to the relevant explicit chemical equation. An example is
complex formation between the metal ion and an uncharged crown ether or cryptand molecule followed
by ion-pair extraction:
aq #Lorg#nA\
Mn# aq"(ML
n#
) Ann\)org

[MLn#Ann\]org
Kex" n#
[M ]aq[L]org[A\]naq
(v) Use of Ringboms `conditional extraction constanta,
anH# ) [ML
n]org
ex "
Keff
[M
]aq[HL
]norg
in conjunction with alpha coefRcients is useful [11].
(vi) The phases can also be speciRed by the formula of the solvent or by other symbols (preferably Roman
numerals) or by overlining formulae referring to one phase, usually the less polar one. The subscript aq
(or w) is often omitted; aq is preferable to w as the latter is appropriate only in English and German.
(vii) The qualiRcation `Equilibriuma is often omitted.
(viii) The terms partition constant and distribution constant must not be used in this sense.
4762 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

3.10 Extraction Factor (Dm)

The ratio of the total mass of a solute in the extract to that in the other phase.
Notes:
(i) It is the product of the (concentration) distribution ratio and the appropriate phase ratio.
(ii) It is synonymous with the concentration factor or mass distribution ratio, this latter term being
particularly apt.
(iii) The term concentration factor is often employed for the overall extraction factor in a process or process
step.

3.11 Fraction Extracted (E )

The fraction of the total quantity of a substance extracted (usually by the solvent) under speciRed conditions,
i.e. EA"QA/Q
A where QA is the mass of A extracted and Q
A is the total mass of A present at the start.
Notes:
(i) E may be expressed as a percentage, %E.
(ii) The term extractability is qualitative and should not be used as a synonym for fraction extracted.
(iii) If the aqueous phase is extracted with n successive portions of solvent, the phase volume ratio
(solvent/feed) being r each time, the fraction extracted is given by:

En"1!(rD#1)\n

If n"r"1, E1"D/(1#D) this expression is a concept of value in chromatography theory.

(iv) The fraction extracted is also known as the recovery factor, especially for a multistage process.

3.12 Loading Capacity

The maximum concentration of solute(s) that a solvent (2.12) can contain under speciRed conditions.

Notes:
(i) The terms maximum loading, saturation capacity and saturation loading are synonymous.
(ii) All the above terms should clearly be distinguished from ultimate capacity (3.29)

3.13 Mass Distribution Ratio See Extraction Factor (3.10)

3.14 Maximum Loading See Loading Capacity (3.12)

3.15 Partition Coef\cient

This term is not recommended and should not be used as a synonym for partition constant (3.16), partition
ratio or distribution ratio (3.5).

3.16 Partition Constant (K oD )

The ratio of activity of a given species A in the extract to its activity in the other phase with which it is in
equilibrium, thus

(KoD)A"aA,org/aA,aq

Its value should not vary with composition but depends on the choice of standard states and on the
temperature (and eventually the pressure).
Note: See transfer activity coefTcient (3.28).

3.17 Partition Ratio (KD)

The ratio of the concentration of a substance in a single deRnite form, A, in the extract (4.8) to its
concentration in the same form in the other phase at equilibrium, e.g. for an aqueous/organic system

(KD)A"[A]org/[A]aq
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4763

Notes:
(i) KD is sometimes called the distribution constant; this is a good synonym. The terms distribution
coefTcient, distribution ratio (3.5), partition constant (3.16) and extraction constant (3.9) should
not be used as synonyms for partition ratio.
(ii) The use of the inverse ratio (aqueous/organic) may be appropriate in certain cases, e.g. where the organic
phase forms the feed (4.11) but its use in such cases should be clearly speciRed. The ratio of the
concentration in the denser phase to the less dense phase is not recommended as it can be ambiguous.
(iii) If the pure solvent and inRnitely dilute feed are taken as the standard state, KDPKoD as the total
concentration of dissolved materials decreases.

3.18 pH0.5 or pH1/2

That value of pH in an aqueous phase at which the distribution ratio (3.5) is unity at equilibrium.

Note: 50% of the solute is extracted (E"0.5) only when the phase ratio (3.19) is unity.

3.19 Phase Ratio (in Liquid-liquid Distribution) (r )

The ratio of the quantity of the solvent (2.12) to that of the other phase.

Notes:
(i) Unless otherwise speciRed the phase ratio refers to the phase volume ratio.
(ii) If other aspects of the phase ratio are employed viz. phase mass ratio, phase Uow ratio, these should be
speciRed.

3.20 Recovery Factor

This term is not recommended. Fraction extracted (3.11) should be used.

3.21 Saturation Capacity See Loading Capacity (3.12)

3.22 Saturation Loading See Loading Capacity (3.12)

3.23 Selectivity Coef\cient

This term should not be used as a synonym for separation factor (3.26).

Note: This term has a speciRc meaning in relation to ion exchange by solid exchangers.

3.24 Selectivity Ratio

Synonym for selectivity coefTcient (3.23). It should not be used as a synonym for separation factor (3.26).

3.25 Separation Coef\cient

This term is not recommended. A synonym for separation factor (3.26).

3.26 Separation Factor (in Liquid-liquid Distribution) (
A,B)

The ratio of the respective distribution ratios (3.5) of two extractable solutes measured under the same
conditions.

A,B"DA/DB

Notes:
(i) By convention the solutes designated as A and B in the above are chosen so as to make
'1.
(ii) The term separation coefTcient is not recommended.
(iii) The terms selectivity coefTcient (3.23) and selectivity ratio (3.24) are not synonymous and should
not be used.
4764 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

3.27 Stoichiometric Capacity See Ultimate Capacity (3.29)

3.28 Transfer Activity Coef\cient (  t)

A term used to quantify the difference in the free energy of a solute ion in two different standard
states often in two different liquid phases. The relationship is tG3"v RT ln t where tG3 is the transfer
Gibbs energy and v is the number of ions in the solute. See partition constant.
Notes:
(i) See IUPAC Information Bulletin No. 34 (1974) [12] for full details.
(ii) It should not be confused with the mass transfer coefTcient which represents the speciRc rate of
transfer of a species from one phase to another.
(iii) It does not necessarily imply the physical transfer of a solute between two liquid phases.

3.29 Ultimate Capacity

The theoretical maximum capacity of a solvent (2.12) containing a given concentration of extractant (2.8) for
a solute under any conditions.
Note: Where appropriate the term stoichiometric capacity can be used.

4. Process Terminology
4.1 Back Extraction
A synonym for stripping (by extraction) (4.43).

4.2 Back Washing

Often used as a synonym for stripping (4.43). This term is not recommended.

4.3 Continuity Inversion

A change in the mutual dispersion of two phases in contact. See inversion (4.13).

4.4 Crowding
The displacement of an impurity from an extract phase by contact with a solution containing the main
extractable solute. See scrubbing (4.23), exchange extraction (4.8).
Note: The main solute need not be present in a pure solution but should have a higher distribution ratio (3.5)
than the impurities present.

4.5 Crud
A deposit or emulsion at the interface between two partially settled phases.
Notes:
(i) The phenomenon of crud formation arises from many causes and this deRnition does not imply any
single one.
(ii) Other terms } some unprintable } have been used but crud is the generally accepted term.

4.6 Density Inversion

The interchange of the denser and less dense phases due to changes in solute concentration. See inversion
(4.13).
Note: Phase inversion (4.20) is often used in this context but is ambiguous.

4.7 Differential Contactor

A type of continuous multistage extraction equipment in which there is only one interface at which phase
separation by settling occurs. See theoretical stage (4.52).
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4765

4.8 Exchange Extraction

An extraction operation or process in which a metal from one phase is exchanged with the equivalent amount
of a second metal from the other phase. See crowding (4.4).
Note:
(i) This term may be used in connection with any step (e.g. loading, (4.16), scrubbing (4.23) or stripping
(4.43) in a process).
(ii) This applies also to organic or molecular species.
4.9 Extraction (in Process Liquid-liquid Distribution)
In connection with processes, this term often refers to the initial transfer step whereby the main solute (1.13),
often together with impurities, is transferred from feed to solvent (2.12). See loading (4.16).
Notes:
(i) Partition and distribution (1.4) are not synonyms in this speciRc instance.
(ii) The term extraction may be used in a more general sense. See under `General DeRnitionsa (1.9).
4.10 Extraction Isotherm See Distribution Isotherm (1.5)

4.11 Feed
A solution introduced into an extraction system.
Note: It should be clearly identiRed (e.g. scrub feed) but, if used without qualiRcation, it may be taken to
designate the initial liquid phase containing the main solute to be transferred.
4.12 Height Equivalent to a Theoretical Stage (HETS)
See explanation of Theoretical Stage (4.52).
4.13 Inversion (or Phase Inversion)
This term is used in two senses which should be speciRed.
(i) density inversion (4.6)
(ii) continuity inversion (4.3)
4.14 Load (in Liquid-liquid Distribution) (Verb)
To transfer solute from a feed (4.11) to another liquid phase.
4.15 Loaded Solvent See Extract (2.7)
Note: This term is usually used to denote the extract (2.7) after completion of a particular step, e.g. extraction
or scrubbing (4.23)
4.16 Loading (Noun)
The concentration of an extracted solute in the extract (2.7).
4.17 Loading (Verb) See Load (4.14)
Note: Used in this sense the term normally refers to the operation of transferring the main solute (1.13)), often
with impurities from the feed to the solvent (2.12).
4.18 O.K. Liquor
Sometimes used as a synonym for strip product solution (4.48) or strip liquor (4.42)
Note: This term is confusing and should not be used.
4.19 Operating Line
A graphical representation of the mass balance relationship of a solute across an extraction process step (4.41)
or stage (4.38).
4766 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

4.20 Phase Inversion See Density Inversion (4.6)

4.21 Raf\nate
The phase remaining after extraction of some speciRed solute(s). When necessary it should be further speciRed,
e.g. scrub rafTnate (4.30).
Note: The original meaning of rafTnate as a `reRned producta has become extended and changed by
common usage.
The term should normally be applied only to waste streams but the latter may form the feed to a further
extraction process for another solute.

4.22 Regeneration See Solvent Regeneration (4.37)

4.23 Scrubbing See Crowding (4.2) and Selective Stripping (4.33)

The process of selectively removing contaminating solutes (impurities) from an extract (2.7) that contains
these as well as the main extractable solute (1.13) by treatment with a new immiscible liquid phase.
Note: The term stripping (4.43) has a different meaning and should not be used in this sense although this
usage has been customary in certain industries.

4.24 Scrubbing Agent

The chemical reagent used to effect scrubbing (4.23).
Note: Often used as a synonym for its solution.

4.25 Scrubbing Agent Solution

The solution used to effect scrubbing (4.23)
Note: The term scrub solution is ambiguous and is not recommended.

4.26 Scrubbing Isotherm See Distribution Isotherm (1.5)

4.27 Scrub Feed

The extract (2.7) to be scrubbed.

4.28 Scrub Liquor See Scrub Raf\nate (4.30)

Note: This term is ambiguous and is not recommended.

4.29 Scrub Product Solution

The solution that results from the scrubbing of impurities from an extract phase.
Note: The term scrub liquor is also used but can be confused with the scrubbing agent solution (4.25) and is
not recommended. See scrub rafTnate (4.30).

4.30 Scrub Raf\nate

This term should only be used where the product solution from scrubbing is discharged to waste. Scrub
product solution (4.29) is better where this stream is combined with feed (4.11) to the loading section.

4.31 Scrubbing Ratio See Distribution Ratio (3.5)

Note: The term scrubbing coefTcient is not recommended. This term is not common.

4.32 Scrub Solution See Scrubbing Agent Solution (4.25)

Note: This term should not be used as it is ambiguous and can be confused with scrub rafTnate (4.30) or
scrub product solution (4.29).
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4767

4.33 Selective Scrubbing See stripping (4.43)

4.34 Solvent Inventory

The total quantity of solvent present in the process.

4.35 Solvent Loss

The total quantity of solvent lost during the operation of a process.
Note: There are a number of ways currently in use to express both solvent inventory and solvent loss and
authors should carefully deRne how they are using the terms until a generally agreed procedure can be
recommended.

4.36 Solvent Puri\cation

See solvent regeneration. The description solvent puriTcation naturally applies also to the puriRcation of fresh
solvent (2.12).

4.37 Solvent Regeneration

Treatment of the solvent for re-cycling, e.g. by removal of degradation products or non-strippable solutes.
Note: The term solvent puriTcation is synonymous, but the terms scrubbing (4.23), stripping (4.31) and
washing should not be used in this context.

4.38 Stage
That physically distinct part of an extraction process in which transfer of solute(s) occurs, followed by phase
separation. See theoretical stage (4.52).
Notes:
(i) For certain types of equipment with a single phase separation interface, the term theoretical stage (4.52)
is more appropriate.
(ii) Equilibrium need not necessarily be established in a stage.

4.39 Stagewise Contactor

A type of continuous multi-stage liquiddliquid contactor in which each stage has a physically distinct cycle of
interphase contact and separation.
Note: There will be the same number of phase separation interfaces as there are stages.

4.40 Steady State (in Liquid-liquid Distribution)

The state of a continuous process when it is operating in such a way that the concentration of solutes in exit
streams remains constants with respect to time for constant feed concentrations, even though the two phases
are not necessarily in thermodynamic equilibrium in any part of the process.
Note: The term equilibrium should not be used to describe this situation.

4.41 Step (in Liquid-liquid Distribution)

That operation in an overall extraction process in which transfer of solute(s) occurs in a particular direction,
e.g. Loading (4.16), stripping (4.43), scrubbing (4.23).

4.42 Strip Liquor

A liquid phase resulting from the operation of stripping (4.43). See strip solution (4.50) and strip rafTnate
(4.49).
Notes:
(i) This term is ambiguous and should be used carefully. Strip rafTnate (4.49) is more appropriate.
(ii) The term O.K. Liquor (4.18) is not recommended.
4768 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

4.43 Stripping
The process of removing solute(s) from a loaded solvent or extract (2.7). Generally this refers to the main
solute(s) present.
Notes:
(i) Where appropriate, e.g. when liquid-liquid distribution is used for stripping, the term back-extraction
can be used. The terms back-washing and re-extraction (1.17) are not recommended.
(ii) The recent application of selective stripping of solutes as a separation method leads to some confusion
between the terms stripping and scrubbing (4.23). It is recommended that the term scrubbing be reserved
for the operation of removing contaminants (impurities) from an extract (2.7) (where the scrub
rafTnate (4.30) is often recycled to the loading step) and the term selective stripping be used where
two or more main solutes are stripped successively from an extract, usually with different stripping
agents (4.44), with a view to their subsequent separate recovery from solution for analysis.
4.44 Stripping Agent
The active substance effective in stripping (4.43).
4.45 Stripping Agent Solution
The liquid phase used to accomplish stripping (4.43).
4.46 Stripping Ratio See Distribution Ratio (3.5)
Notes:
(i) This term is usually deRned as the inverse ratio to the distribution ratio (3.5, comment iii), i.e. in
aqueous-organic systems the aqueous phase concentration of solute is the numerator and the organic
phase concentration the denominator. Their usage should be clearly deRned.
(ii) The term stripping coefTcient is not recommended.
4.47 Stripping Ratio See Distribution Isotherm (1.5), Equilibrium Line (1.7)
Note: In the graphical representation of stripping isotherms, the axes are often interchanged from those used
to represent the phases for extraction isotherms. It is essential that the axes be clearly labelled.
4.48 Strip Product Solution
The liquid phase resulting from stripping (4.43) of a solvent (2.12). See stripping liquor (4.42), strip solution
(4.50), strip rafTnate (4.30), O.K. liquor (4.18).
Note: The last four terms are not recommended.
4.49 Strip Raf\nate
This term is not recommended. RafTnate (4.21) should be reserved for waste streams and the liquid phase
resulting from stripping normally contains the desired product.
4.50 Strip Solution
The liquid phase used for stripping (4.43).
Note: There is some ambiguity between the terms strip liquor and strip solution. Perhaps strip product
solution (4.48) would be more appropriate to the former and stripping agent solution (4.45) for the
latter. See stripping agent (4.44).
4.51 Tenor
A term sometimes used to denote the concentration levels of various solutes in the feed (4.11). It is not
recommended.
4.52 Theoretical Stage
That part of a continuous multi-stage contactor in which the amount of solute transferred from one phase to
the other is equivalent to that which would be transferred in an actual stage at equilibrium under comparable
conditions of solute concentration in each phase as determined from the distribution isotherm (1.5) and
operating line (4.19) for the system.
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4769

Note: Thus from the number of theoretical stages so determined and the height of the contactor the height
equivalent to a theoretical stage (HETS) may be calculated.
4.53 Washing See solvent regeneration (4.37)
Note: This term is vague and is not recommended.

Table 1 Index of terms

Accelerator 2.1 Partition coefficient 3.15

Antagonism 1.1 Partition constant 3.16
Back extraction 4.1 Partition ratio 3.17
Back washing 4.2 pH0.5 or pH1/2 3.18
Carrier 2.2 Phase inversion 4.13, 4.20
Catalyst (in liquid-liquid distribution) 2.3 Phase ratio (in liquid-liquid distribution) 3.19
Coextraction 1.2 Pre-equilibration 1.16
Concentration factor 3.1 Raffinate 4.21
Conditioning 1.3 Regeneration 4.22
Continuity inversion 4.3 Recovery factor 3.20
Cosolvent 2.4 Re-extraction 1.17
Crowding 4.4 Salting out 1.18
Crud 4.5 Saturation capacity 3.21
Decontamination factor 3.2 Saturation loading 3.22
Density inversion 4.6 Scrubbing 4.23
Differential contactor 4.7 Scrubbing agent 4.24
Diluent 2.5 Scrubbing agent solution 4.25
Distribution 1.4 Scrubbing isotherm 4.26
Distribution coefficient 3.3 Scrub feed 4.27
Distribution constant 3.4 Scrub liquor 4.28
Distribution isotherm 1.5 Scrub product solution 4.29
Distribution ratio (in liquid-liquid distribution) 3.5 Scrub raffinate 4.30
Enrichment factor (in liquid-liquid distribution) 3.6 Scrubbing ratio 4.31
Epi-phase 2.6 Scrub solution 4.32
Equilibration 1.6 Selectivity coefficient 3.23
Equilibrium line 1.7 Selectivity ratio 3.24
Exchange extraction 4.8 Selective stripping 4.33
Extract (noun) 2.7 Separation coefficient 3.25
Extract (verb) 1.8 Separation factor (in liquid-liquid distribution) 3.26
Extractability 3.7 Solvent 2.12
Extractant 2.8 Solvent extraction 1.19
Extraction 1.9 Solvent inventory 4.34
Extraction (in process liquid-liquid distribution) 4.9 Solvent ion exchange (six) 1.20
Extraction coefficient 3.8 Solvent loss 4.35
Extraction factor 3.10 Solvent purification 4.36
Extraction isotherm 4.10 Solvent regeneration 4.37
Feed 4.11 Stage 4.38
Fraction extracted 3.11 Stagewise contactor 4.39
Height equivalent to a theoretical stage (HETS) 4.12 Steady state (in liquid-liquid distribution) 4.40
Hypo-phase 2.9 Step (in liquid-liquid distribution) 4.41
Inversion 4.13 Stoichiometric capacity 3.27
Kinetic synergist 2.10 Strip liquor 4.42
Liquid ion exchange 1.10 Stripping 4.43
Liquid-liquid distribution (extraction) (partition) 1.11 Stripping agent 4.44
Load (in liquid-liquid distribution) (verb) 4.14 Stripping agent solution 4.45
Loaded solvent 4.15 Stripping ratio 4.46
Loading (noun) 4.16 Stripping isotherm 4.47
Loading (verb) 4.17 Strip product solution 4.48
Loading capacity 3.12 Strip raffinate 4.49
Macro-element 1.12 Strip solution 4.50
Main (extractable) solute 1.13 Sublation 1.21
Mass distribution ratio 3.13 Substoichiometric extraction 1.22
Maximum loading 3.14 Synergism 1.23
Micro-element 1.14 Tenor 4.51
Modifier 2.11 Theoretical stage 4.52
O.K. Liquor 4.18 Transfer activity coefficient 3.28
Operating line 4.19 Ultimate capacity 3.29
Partition 1.15 Washing 4.53
4770 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution

Table 2 Index of symbols

D Distribution ratio (in liquid-liquid distribution) 3.5

S Enrichment factor (in liquid-liquid distribution) 3.6
Dm Extraction factor 3.10
K oex Extraction (equilibrium) constant at zero 3.9
ionic strength
E Fraction extracted 3.11
K od Partition constant 3.16
Kd Partition ratio 3.17
r Phase ratio (in liquid-liquid distribution) 3.19
A,B Separation factor (in liquid-liquid distribution) 3.26
t Transfer activity coefficient 3.28

Appendix 1. Comparison of Nomenclature With Previously De\ned Terms

(OB, Orange Book [9]; Chrom, Nomenclature for Chromatography [10]

Carrier (2.12) This term is not recommended in this area (OB p54)
Catalyst (in liquid-liquid distribution) (2.3) DeRned as a speciRc use of catalyst (OB p56).
Diluent (2.5) RedeRned from OB 9.2.4 and 9.2.10 in a more general sense
Distribution (1.4) Now deRned } only mentioned in OB 9.2.6
Distribution coefTcient (3.3) Not recommended in this area (OB 9.4.10)
Distribution constant (3.4) Matches uses in Chrom 3.9 and 5.6.
Distribution ratio (in liquid-liquid distribution) (3.5) Slight clariRcation of usage from OB 9.2.6
Enrichment factor (in liquid-liquid distribution) (3.6) Rewording of OB 9.2.8 to make more general.
Extractant (2.8) RedeRned compared to OB 9.2.11
Extraction (in liquid-liquid distribution) (1.9) See more precise term OB 9.2.4
Extraction coefTcient (3.8) Not recommended (OB 9.2.6)
Extraction constant (3.9) Slightly amended from OB 9.2.5
Liquid-liquid distribution (1.11) RedeRned from OB 9.2.4
Mass distribution ratio (3.13) Unchanged from OB 9.4.10. Not recommended in Chrom 3.7.12.
ModiTer (2.11) Term now deRned (see OB 9.2.4)
Partition (1.15) Term now deRned (OB 9.2.1)
Partition coefTcient (3.15) and partition constant (3.16) Not recommended (OB 9.2.6) agrees 3.9.01
Recovery factor (3.20) Now not recommended (OB 9.2.7)
Salting out (1.18) DeRnition broadened from OB 9.2.15
Selectivity coefTcient (3.23) Not recommended in this area (used in Chrom 5.5.02)
Separation factor (in liquid-liquid distribution) (3.25) SpeciRcally deRned for this area to distinguish from
Chrom 3.7.14.2 and 5.5.04
Solvent (in liquid-liquid distribution) (2.12) SpeciRc deRnition provided for this area more limited than OB
9.1.2 and redeRnes 9.2.9. Differs from Chrom 1.1.11 and 5.2.01
 APPENDIX 12B / NOMENCLATURE / Liquid-Liquid Distribution 4771

Appendix 2. Survey of Partition Terminology Used by Authors of Books

aA,org (CA)org
Term: Activities of a single species Term: Total concentration of related species
aA,aq (CA)aq)

Symbol Name Reference Symbol Name Reference

K3D Partition constant 9 Dc Concentration distribution

P3 Activity partition constant 13 ratio 9, 16, 31
P Partition coefRcient or D Distribution ratio 14, 19, 20,
Distribution coefRcient 14, 15 22, 23, 25,
Kp Thermodynamic partition 26, 27, 31,
constant 16 33, 34, 37
p3 or K3p Thermodynamic partition D Distribution coefRcient 17, 18, 37,
constant 17 38
P Thermodynamic partition DTotal Distribution coefRcient 34
coefRcient or q Distribution coefRcient 11
Partition coefRcient 18 Q Extraction coefRcient 39
KD Distribution coefRcient 19 E Extraction coefRcient 36
K Distribution coefRcient 15 E Distribution ratio 17, 35

[A]org
Term: Concentration distribution of a single Term: Equilibrium constant for:
[A]aq
species #
aq #nHLorg&MLn,org#nHaq
Mn#
Symbol Name Reference
[MLn]org;[H#]naq
i.e."
KD Distribution constant 9, 20, 21 [Mn]#
aq ;[HL]org
n

KD Distribution coefRcient 21, 22, 23,

24 Symbol Name Reference
KD Distribution or partition
coefRcient 25 K3ex Extraction constant at
KD Partition coefRcient 26 zero ionic strength (activities
KD Partition constant 27 in above) 9
K Partition coefRcient 14, 28, 29 Kex Extraction constant or
K or  Partition ratio 30 Overall extraction constant 9, 19, 20
K Partition or Distribution Kext Extraction equilibrium
coefRcient 31 constant 32
K Distribution coefRcient 15 K Extraction constant 39
Kp Partition coefRcient 16 K
ex Conditional extraction
P Partition constant or constant 11
coefRcient 32
P Partition coefRcient or
distribution constant 33
P Partition ratio 17,18
D Distribution ratio 11,34
Q Partition ratio 35
Qc Molar partition constant 36
4772 APPENDIX 12C / NOMENCLATURE / Non-Linear Chromatography

References
1. Recommended Nomenclature for Liquid-Liquid Distribution, Pure Appl. Chem., 21, 111}113 (1970).
2. Y. Marcus, Rev. Pure Appl. Chem., 18, 460}464 (1969).
3. A. W. Ashbrook and G. M. Ritcey, Canad. Mining J, 70}72 (May 1972).
4. D. W. Bridges and J. B. Rosenbaum, U.S. Bureau of Mines Information Circular, IC 7139, (1962).
5. W. Fischer, K. Biesenberger, J. Happner and U. Noltzol, `Old and New Processes for Multiplicative Distribution
(liquid-liquid extraction)a Angew. Chem. Internat. Edn., 3, 791}800 (1964).
6. Proceedings ISEC-71, Society of Chemical Industry, London, 2, 25}27 (1971).
7. H. M. N. H. Irving and N. M. Rice, IUPAC Inform. Bull. No. 63. (July 1977).
8. N. M. Rice, Chem. Ind., 718}723 (1977).
9. H. Freiser and G. H. Nancollas, Compendium of Analytical Nomenclature, Blackwell ScientiRc Publications, Oxford,
2nd Ed., (1987).
10. Recommendations on Nomenclature for Chromatography, Pure Appl. Chem., 65, 819}872 (1993).
11. A. Ringbom, Complexation in analytical chemistry, Interscience, New York, 1963.
12. IUPAC Inform. Bull. 34, (1974).
13. I. M. Kolthoff, E. B. Sandell, E. J. Meehan and S. Bruckenstein, Quantitative chemical analysis, 4th Ed.,
Macmillan, London, 1969.
14. M. S. Cresser, Solvent extraction in Uame spectroscopic analysis, Butterworths, London, 1978.
15. E. W. Berg, Physical and chemical methods of separation, McGraw Hill, New York, 1963.
16. D. G. Peters, J. M. Hayes and G. M. Hieftje, Chemical Separations and measurements, theory and practice of
analytical chemistry, Saunders, New York, 1974.
17. E. B. Sandell and H. Onishi, Photometric determination of traces of metals (General aspects), 4th Ed., Part 1., Wiley
Interscience, New York, 1978.
18. Y. Marcus and A. S. Kertes, Ion-exchange and solvent extraction of metal complexes, Wiley, Chichester, 1969.
19. R. A. Day and A. L. Underwood, Quantitative analysis, Prentice-Hall, Engelwood Cliffs, NJ, 1980.
20. H. A. Laitinen and W. E. Harris, Chemical analysis, 2nd Ed., McGraw Hill, New York, 1975.
21. J. S. Fritz and G. H. Shenk, Quantitative analytical chemistry, Allyn and Bacon, Boston, 1969.
22. G. H. Morrison and H. Freiser, Solvent extraction in analytical chemistry, Wiley, Chichester, 1957.
23. G. D. Christian and J. E. O'Reilly, Instrumental analysis, Allyn and Bacon, Boston, 1986.
24. D. A. Skoog and D. M. West, Fundamentals of analytical chemistry, Holt, Rinehart and Winston, New York, 1976.
25. A. I. Vogel, Quantitative inorganic analysis, 3rd Ed., Longmans, London, 1961.
26. F. W. FiReld and D. Kealy, Principles and practice of analytical chemistry, International Textbook, London, 1983.
27. A. S. Kertes and Y. Marcus (Eds), Solvent extraction chemistry 1968, Wiley InterScience, New York, 1969.
28. Cumming and Kay, Revised by R. A. Chalmers, Quantitative chemical analysis, 11th Ed. Oliver and Boyd, Edinburgh,
1956.
29. R. U. Brumblay, A Trst course in quantitative analysis, Addison Welsey, Reading, MA, 1970.
30. H. F. Walton, Principles and methods of chemical analysis, 2nd Ed., Prentice Hall, London, 1964.
31. D. J. Pietrzyk and C. W. Frank, Analytical chemistry, Academic, New York, 1979.
32. I. M. Kolthoff and E. B. Sandell, Textbook of quantitative inorganic analysis, Macmillan, London, 1950.
33. H. A. Flaschka, A. J. Barnard, and P. E. Sturrock, Quantitative analytical chemistry, 2nd Ed., Willard
Grant/Wadsworth, Belmont CA 1980.
34. G. H. Brown and E. M. Sallee, Quantitative chemistry, Prentice Hall, London, 1963.
35. R. A. Chalmers, Aspects of analytical chemistry, Oliver and Boyd, Edinburgh, 1968.
36. L. Sucha and S. Kotryl. Solution equilibria in analytical chemistry, Van Nostrand/Reinhold, New York, 1972.
37. H. A. C. McKay, T. V. Healy, I. L. Jenkins and A. Naylor, Solvent extraction of metals, Macmillan, London, 1966.
38. Z. Marczenko, Separation and spectrophotometric determination of elements, Ellis Horwood, Chichester, 1986.
39. J. Stary, Solvent extraction of metal chelates, Pergamon, Oxford, 1974.

12C. Non-Linear Chromatography

(IUPAC Recommendations 1996)
Prepared for publication by
J. A> . Jo] nsson, University of Lund, Lund, Sweden
^ 1996 IUPAC
4772 APPENDIX 12C / NOMENCLATURE / Non-Linear Chromatography

References
1. Recommended Nomenclature for Liquid-Liquid Distribution, Pure Appl. Chem., 21, 111}113 (1970).
2. Y. Marcus, Rev. Pure Appl. Chem., 18, 460}464 (1969).
3. A. W. Ashbrook and G. M. Ritcey, Canad. Mining J, 70}72 (May 1972).
4. D. W. Bridges and J. B. Rosenbaum, U.S. Bureau of Mines Information Circular, IC 7139, (1962).
5. W. Fischer, K. Biesenberger, J. Happner and U. Noltzol, `Old and New Processes for Multiplicative Distribution
(liquid-liquid extraction)a Angew. Chem. Internat. Edn., 3, 791}800 (1964).
6. Proceedings ISEC-71, Society of Chemical Industry, London, 2, 25}27 (1971).
7. H. M. N. H. Irving and N. M. Rice, IUPAC Inform. Bull. No. 63. (July 1977).
8. N. M. Rice, Chem. Ind., 718}723 (1977).
9. H. Freiser and G. H. Nancollas, Compendium of Analytical Nomenclature, Blackwell ScientiRc Publications, Oxford,
2nd Ed., (1987).
10. Recommendations on Nomenclature for Chromatography, Pure Appl. Chem., 65, 819}872 (1993).
11. A. Ringbom, Complexation in analytical chemistry, Interscience, New York, 1963.
12. IUPAC Inform. Bull. 34, (1974).
13. I. M. Kolthoff, E. B. Sandell, E. J. Meehan and S. Bruckenstein, Quantitative chemical analysis, 4th Ed.,
Macmillan, London, 1969.
14. M. S. Cresser, Solvent extraction in Uame spectroscopic analysis, Butterworths, London, 1978.
15. E. W. Berg, Physical and chemical methods of separation, McGraw Hill, New York, 1963.
16. D. G. Peters, J. M. Hayes and G. M. Hieftje, Chemical Separations and measurements, theory and practice of
analytical chemistry, Saunders, New York, 1974.
17. E. B. Sandell and H. Onishi, Photometric determination of traces of metals (General aspects), 4th Ed., Part 1., Wiley
Interscience, New York, 1978.
18. Y. Marcus and A. S. Kertes, Ion-exchange and solvent extraction of metal complexes, Wiley, Chichester, 1969.
19. R. A. Day and A. L. Underwood, Quantitative analysis, Prentice-Hall, Engelwood Cliffs, NJ, 1980.
20. H. A. Laitinen and W. E. Harris, Chemical analysis, 2nd Ed., McGraw Hill, New York, 1975.
21. J. S. Fritz and G. H. Shenk, Quantitative analytical chemistry, Allyn and Bacon, Boston, 1969.
22. G. H. Morrison and H. Freiser, Solvent extraction in analytical chemistry, Wiley, Chichester, 1957.
23. G. D. Christian and J. E. O'Reilly, Instrumental analysis, Allyn and Bacon, Boston, 1986.
24. D. A. Skoog and D. M. West, Fundamentals of analytical chemistry, Holt, Rinehart and Winston, New York, 1976.
25. A. I. Vogel, Quantitative inorganic analysis, 3rd Ed., Longmans, London, 1961.
26. F. W. FiReld and D. Kealy, Principles and practice of analytical chemistry, International Textbook, London, 1983.
27. A. S. Kertes and Y. Marcus (Eds), Solvent extraction chemistry 1968, Wiley InterScience, New York, 1969.
28. Cumming and Kay, Revised by R. A. Chalmers, Quantitative chemical analysis, 11th Ed. Oliver and Boyd, Edinburgh,
1956.
29. R. U. Brumblay, A Trst course in quantitative analysis, Addison Welsey, Reading, MA, 1970.
30. H. F. Walton, Principles and methods of chemical analysis, 2nd Ed., Prentice Hall, London, 1964.
31. D. J. Pietrzyk and C. W. Frank, Analytical chemistry, Academic, New York, 1979.
32. I. M. Kolthoff and E. B. Sandell, Textbook of quantitative inorganic analysis, Macmillan, London, 1950.
33. H. A. Flaschka, A. J. Barnard, and P. E. Sturrock, Quantitative analytical chemistry, 2nd Ed., Willard
Grant/Wadsworth, Belmont CA 1980.
34. G. H. Brown and E. M. Sallee, Quantitative chemistry, Prentice Hall, London, 1963.
35. R. A. Chalmers, Aspects of analytical chemistry, Oliver and Boyd, Edinburgh, 1968.
36. L. Sucha and S. Kotryl. Solution equilibria in analytical chemistry, Van Nostrand/Reinhold, New York, 1972.
37. H. A. C. McKay, T. V. Healy, I. L. Jenkins and A. Naylor, Solvent extraction of metals, Macmillan, London, 1966.
38. Z. Marczenko, Separation and spectrophotometric determination of elements, Ellis Horwood, Chichester, 1986.
39. J. Stary, Solvent extraction of metal chelates, Pergamon, Oxford, 1974.

12C. Non-Linear Chromatography

(IUPAC Recommendations 1996)
Prepared for publication by
J. A> . Jo] nsson, University of Lund, Lund, Sweden
^ 1996 IUPAC
 APPENDIX 12C / NOMENCLATURE / Non-Linear Chromatography 4773

Synopsis
This report summarizes and comments on terms and symbols used for the description of non-linear chromato-
graphy.

Introduction
In the IUPAC recommendations Nomenclature for Chromatography [1], the conditions of linear chromatog-
raphy are tacitly assumed. In all versions of chromatography, however, non-linear effects are common.
These are seen as concentration-dependent retention times and asymmetric (e.g. tailing or fronting) peaks.
Asymmetric peaks can result from a number of other causes as well, i.e. large extra-column volumes. In many
applications, non-linear effects are disadvantageous as they decrease peak resolution and disturb quantit-
ative evaluation. However, in preparative chromatography, heavy overloading is employed in order to
increase material throughput, leading to prominent non-linear effects. A comprehensive text on non-
linear chromatography has recently been published [2].
In this paper, some of the concepts and terms used for non-linear chromatography are described. It is to read
as a complement to the Nomenclature for Chromatography (CN) [1], to which numerous references are given.

1. Terms Related to Isotherms

1.1 Distribution Isotherm (in Chromatography)
The equilibrium relation between the concentration of a sample component in the stationary phase cS, and in
the mobile phase cM, expressed as a function cS"f(cM).

Note: The relation can be inSuenced also by concentrations of other sample components. cS and cM are usually
expressed per unit volume of the phase; cS may also be expressed per mass of the dry solid phase or per
unit surface area. This is discussed in CN, section 3.9.

In some versions of chromatography, a distribution isotherm can be seen as a partition isothem, an adsorption
isotherm, or a combination of these, depending on the mechanism of separation (cf. CN 1.5).

1.1.1 Partition isotherm (in chromatography) Isotherm describing partition of the sample component
between the bulk of a liquid stationary phase and a liquid, gaseous or supercritical mobile phase.

1.1.2 Adsorption isotherm Isotherm describing adsorption of the sample component on the surface of the
stationary phase from the mobile phase.

Note: Adsorption isotherms can be described by Langmuir, Freundlich and other adsorption isotherm
equations. See [3], p. 13.

1.2 Linear Distribution Isotherm

A distribution isotherm which can be approximated as cS"KC cM, where KC is a constant.

Note: At low concentrations, all distribution isotherms tend towards being linear. KC is the distribution
constant (cf. CN 3.9 and 3.4 in ref. 4).

1.3 Non-linear Distribution Isotherm

A distribution isotherm which is not linear.

Note: A non-linear isotherm can have several shapes, as classiRed by Brunauer et al. [5]. In chromatography
convex or concave shapes are common, as well as combinations.

1.3.1 Convex isotherm Distribution isotherm, the slope of which is continuously decreasing (see Figure 1A).
4774 APPENDIX 12C / NOMENCLATURE / Non-Linear Chromatography

Figure 1 Different types of distribution isotherms for the concentrations of a compound in the stationary (cS) and mobile (cM) phases:
(A) convex isotherm, (B) concave isotherm.

Note: The resulting chromatographic peak is tailing (CN 3.3.08). Adsorption isotherms are often of this type.
A special case is the Langmuir adsorption isotherm.

1.3.2 Concave isotherm Distribution isotherm, the slope of which is continuously increasing (see
Figure 1B).

Note: The resulting chromatographic peak is fronting (CN 3.3.09). In gas-liquid chromatography, overload-
ing often results in a concave isotherm.

2. Types of Chromatographic Processes

2.1 Linear Chromatography
Chromatographic process, where the retention is governed by a linear distribution isotherm.

2.2 Non-linear Chromatography

Chromatographic process, where the retention is governed by a non-linear distribution isotherm.

2.3 Ideal Chromatography

Chromatographic process, where no peak-broadening effects (such as diffusion, slow mass transfer,
etc.) operate.

Note: This is a hypothetical case, implying that the plate number (CN 3.10.03) is inRnite.

2.4 Non-ideal Chromatography

Chromatographic process with normal peak-broadening effects.

2.5 Non-ideal, Linear Chromatography

Chromatographic process, where the retention is governed by a linear distribution isotherm and normal
peak-broadening take place.

Note: This case is commonly assumed in analytical chromatography, as described in ref. 1.

2.6 Ideal, Non-linear Chromatography

Chromatographic process, where only the curvature of the distribution isotherm determines the shape of the
peaks while other peak-broadening processes are neglected.

Note: The assumption of ideal, non-linear (INL) chromatography is often made in order to facilitate
theoretical treatments. It can be justiRed in cases of efRcient columns and distribution isotherms
with prominent non-linearity.
 APPENDIX 12C / NOMENCLATURE / Non-Linear Chromatography 4775

2.7 Non-ideal, Non-linear Chromatography

Chromatographic process, where both isotherm curvature and other peak-broadening processes (such as
diffusion) contribute to the peak shape.

Note: This case comprises most peaks in common practice that are characterized as tailing or fronting.

3. Retention Parameters in Non-Linear Chromatography

3.1 Total Retention Volume (Time) in Ideal, Non-linear Chromatography (VR(INL) , t R(INL) )
The volume of mobile phase entering the column between sample introduction and the emergence of a certain
concentration of the sample component at the column outlet; or the corresponding time.

Note: This volume (time) can be measured to the peak maximum or to other points on the peak. Under the
conditions of ideal, non-linear chromatography, the total retention volume is given by:

*cS
VR(INL)"VM# ) VS (1)
*cM

With a constant Sow rate Fc through the column, the total retention time in ideal, non-linear chromatography
is given by tR(INL)"VR(INL)/Fc as in CN 37.05. If appropriate, VS in equation (1) may be exchanged for the
surface area of the stationary phase or the mass of the stationary phase, depending on the deRnition of cS (cf.
1.1 and CN 3.9). In the case of a linear distribution isotherm, equation (1) is in agreement with corresponding
equation in CN 3.9.01. Note that the retention is determined by the slope of the isotherm, not by the ratio
cS/cM. This particular point was discussed by Helfferich [6].
Typical peaks in ideal, non-linear chromatography are shown in Figure 2. The curved (diffuse) Sanks
are described by equation (1) and the area of the peak (determined by the total amount of the sample
component) gives the position of the vertical Sank.
The retention volume in ideal, non-linear chromatography is thus a function of the mobile phase concentra-
tion of the sample component. The retention volume to the maximum of the peak (cf. CN 3.7.05) is related the
value of the slope of the distribution isotherm at the maximum value of the mobile phase concentration of the
sample component at the column outlet.
The broadening of the peaks in Figure 2 is totally caused by the isotherm non-linearity. As the derivation of
equation (1) implies that the plate number N is inRnite, it is obviously meaningless to apply equations such as
those described in CN 3.10.03 and 3.10.04 to characterize peaks of this kind.

3.2 Total Retention Volume (Time) in Non-ideal, Non-linear Chromatography (VR(NINL), tR(NINL) )
The deRnition is analogous to that in 3.1 above.

Figure 2 Typical peak shapes in ideal, non-linear chromatography. Peaks A and B are generated by equation (1) from the distri-
bution isotherms in Figures 1A and 1B, respectively. The numbers 1, 2, 4 signify the relative amounts of the sample component. The
retention time at low sample concentration, i.e. in the case that the curvature of the distribution isotherm is negligible, is indicated with
an arrow.
4776 APPENDIX 12D / NOMENCLATURE / Supercritical Fluid Chromatography and Extraction

Note: In the general case of non-ideal, non-linear (NINL) chromatography, only numerical solutions to the
applicable non-linear partial differential equations involved can be found. Several examples are
found ref 2, where simulated NINL peaks are compared with INL peaks with the same parameters,
except for the diffusion term. It is seen that the NINL peaks are lower, wider and more tailing than
the INL peak. With a reasonably efRcient column (N'5000 for symmetric peaks), the differ-
ence might be neglected for practical purposes.
Thus, even if no explicit equation for the retention volume in the NINL case can be given, Equation (1)
is approximately valid also for a NINL peak, the discrepancy depending on the column efRciency.
To measure distribution isotherms by chromatography, the so-called Elution by Characteristic Point
(ECP) method has been suggested. Retention volumes to several points on the curved Sank of an
experimental peak are measured and related to the solute concentration at those points. The distribu-
tion isotherm can then be calculated using Equation (1). The validity of the method depends on the
efRciency of the column used for these measurements.
There is no known general way of calculating meaningful peak broadening parameters, such as plate
numbers from NINL peaks. As the NINL case in practice is common, this observation is important: The
usual equations for the calculation of plate numbers (such as those described in CN 3.10.03) should
only be applied to effectively symmetrical peaks.

References
1. Recommendations for Nomenclature for Chromatography. Pure Appl. Chem. 65, 819}872 (1993).
2. G. Guiochon, S. Golshan Shirazi and A. M. Katti, Fundamentals of Preparative and Non-linear Chromatography.
Academic Press. Inc. Boston (1994).
3. V. Gold, K. L. Loeming, A. D. McNaught and P. Sehml. Compendium of Chemical Terminology. Blackwell Science
Publishers. Oxford, UK. 1987.
4. Recommendations for Nomenclature for Liquid-Liquid Distribution (Solvent Extraction). Pure. Appl. Chem. 65,
2372}2396 (1993).
5. S. Brunauer, L. S. Deming, W. E. Deming and E. Teller, J. Amer. Chem. Soc. 62, 1723 (1940)
6. F. Helfferich. J. Chem. Educ. 41, 410 (1964).

12D. Supercritical Fluid Chromatography and Extraction

(IUPAC Recommendations 1993)

Prepared for publication by

R. M. Smith, Loughborough University of Technology, Loughborough, Leicestershire, UK
^ 1993 IUPAC

Abstract
The report present deRnitions for the terms and symbols used when supercritical Suids are employed as the
liquid phase in chromatography and allied areas including sample extraction. The terms supplement those in
the general Nomenclature for Chromatography and includes additional more speciRc terms.

Introduction
Following the General Assembly Meeting in 1989 the Limited Life Time Commission for Chromatography
and Other Analytical Separations took over the work on the nomenclature for chromatography that had
previously been undertaken by the Commission for Analytical Nomenclature. A major part of the work was
the Nomenclature for Chromatography which had been developed over a number of years by L. S. Ettre and
has recently been published [1]. This work was comprehensive and included all the major areas of chromato-
graphy. Specialist chapters covered the speciRc areas of size exclusion chromatography and ion-exchange
 APPENDIX 12D / NOMENCLATURE / Supercritical Fluid Chromatography and Extraction 4777

chromatography. However, it was clear that as further new areas of separation science were developed speciRc
terminologies of further additional terms and deRnitions would be needed.
Over the last few years the use of a supercritical Suid as the mobile phase in chromatography has become an
accepted routine method. In the Nomenclature for Chromatography Ettre noted (1.4.04) `In general, the
terms and deRnitions used for gas or liquid chromatography are equally applicable to supercritical-Suid
chromatographya. However, supercritical Suid chromatography has also lead to the use in the literature of
a number of new terms whose meanings have been generally adopted by workers in the Reld. These new terms
are formalized in the present Supplement to the general nomenclature of chromatography. Supercritical Suids
have also been used for the extraction of samples and frequently similar equipment and operating conditions
have been employed and many of the terms are also applicable in this Reld.
This nomenclature is designed to be used as a supplement to the principal nomenclature paper and is written
as Section 7 of that paper [1]. It therefore omits any terms which have already been deRned unless a new or
additional deRnition has been necessary. The paper is also complementary to the deRnitions and terms for
supercritical Suid chromatography recently published by ASTM [2].

*7. Special Terminology Used in Supercritical-Fluid Chromatography and

Extraction
7.1 Basic De\nitions
7.1.1 Critical temperature (Tc) The maximum temperature at which a gas can be converted into a liquid by
an increase in pressure.

7.1.2 Critical pressure (pc) The minimum pressure which would sufRce to liquefy a substance at its
critical temperature. Above the critical pressure, increasing, the temperature will not cause a Suid to vaporize
to give a two-phase system.

7.1.3 Critical point The characteristic temperature (Tc) and pressure (pc) above which a gas cannot be
liqueRed.

7.1.4 Supercritical Wuid The deRned state of a compound, mixture or element above its critical pressure (pc)
and critical temperature (Tc).

7.1.5 Reduced temperature (Tr) The ratio of the temperature (T) in the system to the critical temperature
(Tc)

Tr"T/Tc

7.1.6 Reduced pressure (pr) The ratio of the pressure in the system (p) to the critical pressure (pc).

pr"p/pc

7.2 The Mobile Phase

7.2.1 The mobile phase was deVned previously in 1.1.06
7.2.2 Mobile-phase pressure
7.2.2.1 Outlet pressure (po) DeRned as in 3.6.02.2. However, unlike gas and liquid chromatography the
outlet pressure in supercritical-Suid chromatography has to be maintained above ambient pressure by a Sow
restrictor (7.3.1) or back-pressure regulator (7.3.2).

7.2.2.2 Pressure drop across the column (
p) DeRned as 3.6.02.3.

* For Sections 1}6 see Pure & Appl. Chem., Vol. 65, No. 4, pp. 819}872, 1993.
4778 APPENDIX 12D / NOMENCLATURE / Supercritical Fluid Chromatography and Extraction

7.2.3 Mobile-phase volume Wow rate. DeRned as 3.6.04. In supercritical-Suid chromatography this is
usually quoted as the rate of delivery of the pumping system.

7.2.4 Mobile-phase mass Wow rate The rate of mass Sow through the column. It is usually determined by
measuring the gas-Sow rate (or liquid-Sow rate) at ambient conditions after the mobile phase has been
depressurized. If liquid modiRers are present in the mobile phase, corrections will be needed.

7.2.5 Mobile-phase composition The composition of the mobile-phase which is delivered to the column.
This should be described in such a way that it can be reproduced in different laboratories. It can be
expressed on a mass, volume, or mole fraction basis but in each case the temperature and pressure must also be
deRned.
If the individual components are pumped separately, the relative delivery Sow rates should be deRned.
Premixed eluents are often used and can be deRned by their mass composition as recorded by the
manufacturer. However, the delivered composition may depend on the relative volatility of the components
and can change as a function of syringe pump volume and time.

7.2.5.1 Mobile-phase modiTer ModiRers are materials (usually organic compounds such as methanol or
acetonitrile) added to the supercritical Suid being used as the mobile phase to alter the elution properties.
7.3 Instrumentation
Most of the components of the instrumentation for supercritical-Suid chromatography are in common with
liquid and gas chromatography and are deRned in Section 2.1 Apparatus for Column Chromatography.

7.3.1 Flow restrictor This is a device which restricts the Sow of the mobile phase leaving the column and is
used to maintain the pressure in the chromatographic column.

7.3.1.1 Capillary restrictor This is a capillary tube which may be tapered or constricted and acts as
a mass-Sow controller. The column pressure is controlled by adjusting the pump Sow rate.

7.3.1.2 Frit restrictor A frit placed at the end of an open-tubular column to act as a Sow restrictor.
Sometimes referred to as an integral frit restrictor.

7.3.2 Back-pressure regulator This is a device which is placed after the column and is used to regulate the
pressure in the column by a pressure-adjustable diaphragm or controlled nozzle so that the same column-outlet
pressure is maintained irrespective of the mobile-phase pump Sow rate.

7.3.3 Sample injector as deVned in 2.1.02. The most common form in supercritical-Suid chromatography
is the bypass injector (see 2.1.02.2). In capillary supercritical-Suid chromatography a timed injector is often
used.

7.3.3.1 Timed injector This is a form of bypass injector in which the rotation of the valve is timed so that
only a portion of the contents of the sample loop can pass to the column.

7.3.4 High-pressure Wow cell A Sow-through cell (usually spectroscopic) designed for use at high pressures
so that the sample remains dissolved in the mobile phase during detection.
7.4 The Chromatographic Medium
Supercritical-Suid chromatographic separations are carried out using capillary columns or packed columns
similar to those used in gas or liquid chromatography (see Section 3.1). Stationary phases are usually
chemically bonded to the support. Non-chemically bonded phases are often unsuitable as the stationary phase
may be soluble in the mobile phase.
7.5 Terms Related to the Chromatographic Process
7.5.1 Isobaric separation Chromatographic separation carried out using constant inlet and outlet pressure
conditions.
 APPENDIX 12D / NOMENCLATURE / Supercritical Fluid Chromatography and Extraction 4779

7.5.2 Isopycnic separation Chromatographic separations carried out using constant density conditions.
The temperature and pressure may be altered during the run (originally the term isoconfertic separation was
used but this term is not recommended).

7.5.3 Programmed elution A procedure is which the conditions of the separation are changed in a pro-
grammed manner. Unlike gas or liquid chromatography both the pressure and temperature can be pro-
grammed.
The term `gradient elutiona should be restricted to changes in composition of the mobile phase with time
(see 1.6.04).

7.5.3.1 Density-programmed elution A separation carried out using a pressure and/or temperature pro-
gramme so that the density of the mobile phase changes with time in a pre-determined manner during the
separation.

7.5.3.2 Pressure-programmed elution A separation carried out using a programmed increasing pressure
with time.

7.5.3.3 Pressure/temperature-programmed elution A separation carried out using conditions where the
pressure and temperature are programmed simultaneously. The temperature may be programmed to increase
or decrease.

7.6 Coupled-systems
As well as discrete chromatographic detectors, supercritical-Suid chromatography has been coupled to more
complex detectors and to other separation techniques and the most widely used are listed here.

7.6.1 Coupled supercritical-Wuid chromatography-mass spectrometry (SFC-MS) Separation system in

which the column efSuent from a supercritical-Suid chromatograph is passed directly to the inlet chamber
of a mass spectrometer.

7.6.2 Coupled supercritical-Wuid chromatography-Fourier-transform infrared spectrometry (SFC-FTIR)

Separation system in which the column efSuent from a supercritical-Suid chromatograph is passed directly
through a Fourier-transform infrared specrometer.

7.6.3 Coupled supercritical-Wuid chromatography-gas chromatography (SFC-GC) Separation system in

which a fraction from the supercritical-Suid chromatograph efSuent is transferred directly to the inlet port
or column of a gas chromatograph system.

7.7 Supercritical-]uid Extraction

7.7.1 Supercritical-Wuid extraction (SFE) Extraction of a material using a supercritical Suid. The extracted
material is usually recovered by reducing the temperature or pressure of the extraction Suid and allowing the
volatile components of the mobile phase to evaporate. Instrumentally supercritical-Suid extraction can use
many of the components of a supercritical-Suid chromatographic system. It can used either as an on-line
sample introduction method for a chromatographic separation or as an off-line sample preparation
method.

7.7.2 Coupled supercritical-Wuid extraction-supercritical-Wuid chromatography (SFE-SFC) System in

which a sample is extracted with a supercritical-Suid which then places the extracted material in the inlet port
of a supercritical-Suid-chromatographic system. The extract is then chromatographed directly using a super-
critical Suid.

7.7.3 Coupled supercritical-Wuid extraction-gas chromatography (SFE-GC) and Coupled supercritical-Wuid

chromatography-liquid chromatography (SFE-LC) System in which a sample is extracted using a supercriti-
cal Suid which is then depressurized to deposit the extracted material in the inlet port or column of a gas or
liquid chromatographic system, respectively. The extract is then chromatograped directly.
4780 APPENDIX 12D / NOMENCLATURE / Supercritical Fluid Chromatography and Extraction

Table 1 Index of additional terms Table 2 List of symbols

Back-pressure regulators 7.3.2 pc Critical pressure 7.1.2

Capillary restrictors 7.3.1.1 Tc Critical temperature 7.1.1
Coupled supercritical-fluid chromatography-mass 7.6.1 pr Reduced pressure 7.1.6
spectrometry Tr Reduced temperature 7.1.5
Coupled supercritical-fluid extraction-Fourier- 7.6.2
transform infrared spectroscopy
Coupled supercritical-fluid chromatography-gas 7.6.3
chromatography Table 3 List of acronyms
Coupled supercritical-fluid extraction-gas 7.7.3
chromatography SFE Supercritical-fluid extraction
Coupled supercritical-fluid extraction-liquid 7.7.3 SFC-FTIR Supercritical-fluid chromatography-Fourier
chromatography transform infrared spectroscopy
Coupled supercritical-fluid extraction-supercritical- 7.7.2 SFC-GC Supercritical-fluid chromatography-gas
fluid chromatography chromatography
Critical point 7.1.3 SFC-LC Supercritical-fluid chromatography-liquid
Critical pressure 7.1.2 chromatography
Critical temperature 7.1.1 SFC-MS Supercritical-fluid chromatography-mass
Density-programmed elution 7.5.3.1 spectrometry
Flow restrictors 7.3.1 SFE-GC Supercritical-fluid extraction-gas chromatography
Frit restrictor 7.3.1.2 SFE-LC Supercritical-fluid extraction-liquid chromatography
High-pressure flow cell 7.3.4 SFE-SFC Supercritical-fluid extraction-supercritical-fluid
Integral frit restrictor 7.3.1.2 chromatography
Isobaric separation 7.5.1
Isopycnic separation 7.5.2
Mobile-phase composition 7.2.5
Mobile-phase modifiers 7.2.5.1
Mobile-phase mass flow rate 7.2.4
Mobile-phase volume flow rate 7.2.3
Pressure programmed elution 7.5.3.2
Pressure/temperature-programmed elution 7.5.3.3
Programmed elution 7.5.3
Reduced temperature 7.1.5
Reduced pressure 7.1.6
Supercritical fluid 7.1.4
Supercritical-fluid extraction (SFE) 7.7.1
Timed injector 7.3.3.1

References
1. Recommendations for Nomenclature for Chromatography, Pure and Applied Chemistry, 65, 819}872 (1993).
2. Standard Guide for Supercritical Fluid Chromatography terms and relationships, ASTM E 1449, American Society
for Testing and Materials, Philadelphia, PA, 1992.
 APPENDIX 13 / pH SCALE FOR AQUEOUS SOLUTIONS 4781

13. pH SCALE FOR AQUEOUS SOLUTIONS

Values of pH For Primary Standard Reference Solutions

Primary ref. Temperature (3C)

standard
0 5 10 15 20 25 30 35 37 40 50 60 70 80 90 95

Saturated * * * * * 3.557 3.552 3.549 3.548 3.547 3.549 3.560 3.580 3.610 3.650 3.674
(at 253C)
potassium
hydrogentartrate
0.1 mol kg\1 3.863 3.840 3.820 3.802 3.788 3.776 3.766 3.759 3.756 3.754 3.749 * * * * *
Potassium
dihydrogencitrate
0.025 mol kg\1 6.984 6.951 6.923 6.900 6.881 6.865 6.853 6.844 6.841 6.838 6.833 6.836 6.845 6.859 6.876 6.886
Disodium hydro-
genphosphate
#0.025 mol kg\1
potassium dihydro-
gen phosphate
0.03043 mol kg\1 7.534 7.500 7.472 7.448 7.429 7.413 7.400 7.389 7.386 7.380 7.367 * * * * *
Disodium hydro-
genphosphate #
0.008695 mol kg\1
potassium
dihydrogen
phosphate
0.01 mol kg\1 9.464 9.395 9.332 9.276 9.225 9.180 9.139 9.102 9.088 9.068 9.011 8.962 8.921 8.884 8.850 8.833
Disodium
tetraborate
0.025 mol kg\1 10.317 10.245 10.179 10.118 10.062 10.012 9.966 9.926 9.910 9.889 9.828 * * * * *
Sodium hydro-
gencarbonate
#0.025 mol kg\1
sodium carbonate

Note: Based on an uncertainty of $0.2 mV in determined (E!E
), the uncertainty is $0.003 in pH in the range 0}503C.
4782 APPENDIX 13 / pH SCALE FOR AQUEOUS SOLUTIONS

pH Values of Operational Reference Solutions

Operational Temperature (3C)

standard
ref. solution 0 5 10 15 20 25 30 37 40 50 60 70 80 90 95

0.1 mol kg\1 Potassium * * * * 1.475 1.479 1.483 1.490 1.493 1.503 1.513 1.52 1.53 1.53 1.53
tetroxalatea
0.05 mol kg\1 Potassium * * 1.638 1.642 1.644 1.646 1.648 1.649 1.650 1.653 1.660 1.671 1.689 1.72 1.73
tetroxalatea
0.05 mol kg\1 Sodium * 3.466 3.470 3.476 3.484 3.492 3.502 3.519 3.527 3.558 3.595 * * * *
hydrogendiglycolateb
Saturated (at 253C) * * * * * 3.556 3.549 3.544 3.542 3.544 3.553 3.570 3.596 3.627 3.649
Potassium hydrogen-
tartrate
0.05 mol kg\1 Potassium 4.000 3.998 3.997 3.998 4.000 4.005 4.011 4.022 4.027 4.050 4.080 4.115 4.159 4.21 4.24
hydrogenphthalate
(RVS)
0.1 mol dm\3 Acetic 4.664 4.657 4.652 4.647 4.645 4.644 4.643 4.647 4.650 4.663 4.684 4.713 4.75 4.80 4.83
acid#0.1 mol dm\3
sodium acetate
0.01 mol dm\3 Acetic 4.729 4.722 4.717 4.714 4.712 4.713 4.715 4.722 4.726 4.743 4.768 4.800 4.839 4.88 4.91
acid#0.1 mol dm\3
sodium acetate
0.02 mol kg\1 * 6.477 6.419 6.364 6.310 6.259 6.209 6.143 6.116 6.030 5.952 * * * *
Piperazine phosphatec
0.025 mol kg\1 6.961 6.935 6.912 6.891 6.873 6.857 6.843 6.828 6.823 6.814 6.817 6.830 6.85 6.90 6.92
Disodium hydrogen-
phosphate#
0.025 mol kg\1
potassium dihydrogen
phosphate
0.03043 mol kg\1 7.506 7.482 7.460 7.441 7.423 7.406 7.390 7.369 * * * * * * *
Disodium hydrogen-
phosphate#
0.008695 mol kg\1
potassium disodium
phosphate
0.04 mol kg\1 * 7.512 7.488 7.466 7.445 7.428 7.414 7.404 * * * * * * *
Disodium hydrogen-
phosphate#
0.01 mol kg\1
potassium dihydrogen
phosphate
0.05 mol kg\1 Tris 8.399 8.238 8.083 7.933 7.788 7.648 7.513 7.332 7.257 7.018 6.794 * * * *
hydrochloride#
0.01667 mol kg\1
Trisd
0.05 mol kg\1 9.475 9.409 9.347 9.288 9.233 9.182 9.134 9.074 9.051 8.983 8.932 8.898 8.88 8.84 8.89
Disodium tetraborate
(Na2B4O7)
0.01 mol kg\1 9.451 9.388 9.329 9.275 9.225 9.179 9.138 9.086 9.066 9.009 8.965 8.932 8.91 8.90 8.89
Disodium tetraborate
(Na2B4O7)
0.025 mol kg\1 Sodium 10.273 10.212 10.154 10.098 10.045 9.995 9.948 9.889 9.866 9.800 9.753 9.728 9.725 9.75 9.77
hydrogencarbonate#
0.025 mol kg\1
sodium carbonate
Saturated (at 203C) 13.360 13.159 12.965 12.780 12.602 12.431 12.267 12.049 11.959 11.678 11.423 11.192 10.984 10.80 10.71
calcium hydroxide

Note: Uncertainty is$0.003 in pH between 0 and 603C rising to $0.01 above 703C.
a
Potassium trihydrogen dioxalate (KH3C4O8).
b
Sodium hydrogen 2,2
-oxydiethanoate.
c
C4H10N2 ) H3PO4.
d
2-Amino-2(hydroxymethyl)-1,3 propanediol or tris(hydroxymethyl)aminomethane.
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4783

Useful Data on Some Standard Buffer Solutions

Molecular Molarity Relative Density Molarity Mass of 1 l Mass Mass tolerance

formula (mol kg\1) molar at 203C at 203C at 203C (g) tolerance expressed as a
mass (g cm\3) (mol l\1) for $0.001 percentage (%)
pHa (g)

Potassium tetraoxalate KH3C4O8 ) 0.1 254.1913 1.0091 0.09875 25.1017 0.07 0.27
2H2O
Potassium tetraoxalate KH3C4O8 ) 0.05 254.1913 1.0038 0.04965 12.6202 0.034 0.26
2H2O


Disodium hydrogen Na2HPO4 0.025 141.9588 3.5379 0.02 0.56
orthophosphate
1.0038 0.02492
Potassium dihydrogen KH2PO4 0.025 136.0852 3.3912 0.02 0.58
orthophosphate
Disodium tetraborate Na2B4O7 ) 0.05 381.367 1.0075 0.04985 19.0117 0.9 4.73
10H2O
Disodium tetraborate Na2B4O7 ) 0.01 381.367 1.0001 0.009981 3.8064 0.19 0.49
10H2O


Sodium carbonate Na2CO3 0.025 105.9887 2.6428 0.017 0.064
1.0021 0.02494
Sodium hydrogencarbonate NaHCO3 0.025 84.0069 2.0947 0.013 0.62

a
Calculated from known dilution value of solution.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd edn. Oxford: Blackwell ScientiRc
Publications.)

14. PROPERTIES OF PARTICLES, ELEMENTS

AND NUCLIDES
Properties of Some Particles
Name Symbola Spin I Charge Rest mass Magnetic moment /N Meanlife
/s
number z
m/u mc2/MeV

Photon  1 0 0 0
Neutrino e 1/2 0 0 0
Electronb e 1/2 !1 5.485 799 03 (13);10\4 0.510 999 06 (15) 1.001 159 652 193 (10)c
Muon ! 1/2 $1 0.113 428 913 (17) 105.658 389 (34) 1.001 165 923 (8)d 2.197 3 (4);10\6
Pion ! 1 $1 0.149 832 3 (8) 139.5679 (7) 2.6030 (24);10\8
Pion 0 1 0 0.144 9008 (9) 134.9743 (8) 8.4 (6);10\17
Proton p 1/2 1 1.007 276 470 (12) 938.272 31 (28) 2.792 847 386 (63)
Neutron n 1/2 0 1.008 664 904 (14) 939.565 63 (28) !1.913 042 75 (45) 889.1 (21)
Deuteron d 1 1 2.013 553 214 (24) 1875.613 39 (53) 0.857 437 6 (1)
Triton t 1/2 1 3.015 500 71 (4) 2808.921 78 (85) 2.978 960 (1)
Helion h 1/2 2 3.014 932 23 (4) 2808.392 25 (85) !2.127 624 (1)
-Particle  0 2 4.001 506 170 (50) 3727.380 3 (11) 0

a
The Particle Data Group recommends the use of italic symbols for particles and this has been adopted by many physicists.
b
The electron as -particle is sometimes denoted by .
c
The value is given in Bohr magnetons /B, B"e
/2me.
d
The value is given as / , where  "e
/2m .
I I I
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd edn. Oxford: Blackwell ScientiRc
Publications.)

In nuclear physics and chemistry the masses of particles are often quoted as their energy equivalents (usually in
mega electronvolts). The uniRed atomic mass unit corresponds to 931.494 32 (28) MeV.
Atom-like pairs of a positive particle and an electron are sometimes sufRciently stable to be treated as
individual entities with special names.
4784 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Examples
positronium (e#e\) m(e#e\)"1.097 152 503(26);10\3u
muonium (
#e\; Mu) m(Mu)"0.113 977 478(17)u

The positive or negative sign for the magnetic moment of a particle implies that the orientation of the
magnetic dipole with respect to the angular momentum corresponds to the rotation of a positive or negative
charge respectively.

Standard Atomic Weights of the Elements 1991

As agreed by the IUPAC Commission on Atomic Weights and Isotopic Abundances in 1979 the relative atomic
mass (atomic weight) of an element, E, can be deRned for any speciRed sample. It is the average mass of its
atoms in the sample divided by the uniRed atomic mass unitH or alternatively the molar mass of its atoms
divided by the standard molar mass MF"Lmu"1 g mol\1:

Ar(E)"mN a(E)/u"M(E)MF

The variations in isotopic composition of many elements in samples of different origin limit the
precision to which a relative atomic mas can be given. The standard atomic weights revised biennially by the
IUPAC Commission on Atomic Weights and Isotopic Abundances are meant to be applicable for normal
materials. This means that to a high level of conRdence the relative atomic mass of an element in any normal
sample will be within the uncertainty limits of the tabulated value. By normal it is meant here that the
material is a reasonably possible source of the element or its compounds in commerce for industry and science
and that it has not been subject to signiRcant modiRcation of isotopic composition within a geologically brief
period. This, of course, excludes materials studied themselves for very anomalous isotopic composition.
The relative atomic masses of many elements depend on the origin and treatment of the materials. The notes
to this table explain the types of variation to be expected for individual elements.
A value in brackets denotes the mass number of the most stable isotope.  denotes density, C,m denotes
melting temperature, C,b denotes boiling temperature and cp denotes speciRc heat capacity. subl. denotes
sublimes.

Element Symbol Atomic Relative  (g cm\3) C,m (3C) C,b (3C) cp (J kg\1 K\1) Oxidation Note
number atomic mass states

Actinium Ac 89 (227) 10.1 1050 3200 3 A

Aluminium Al 13 26.9815 2.70 660 2470 900 3
Americium Am 95 (243) 11.7 (1200) (2600) 140 3,4,5,6 A
Antimony Sb 51 121.75 6.62 630 1380 209 3,5 g
Argon Ar 18 39.948 1.40(87 K) !189 !186 519 g,r
Arsenic (, grey) As 33 74.9216 5.72 613 subl. 326 3,5
Astatine At 85 (210) (302) A
Barium Ba 56 137.34 3.51 714 1640 2
Berkelium Bk 97 (247) 3,4 A
Beryllium Be 4 9.01218 1.85 1280 2477 1.82;103 2
Bismuth Bi 83 208.9806 9.80 271 1560 121 3,5
Boron B 5 10.81 2.34 2300 3930 1.03;103 3 g,m,r
Bromine Br 35 79.904 3.12 !7.2 58.8 448 1,3,4,5,6
Cadmium Cd 48 112.40 8.64 321 765 230 2 g
Caesium Cs 55 132.9055 1.90 28.7 690 234 1
Calcium Ca 20 40.08 1.54 850 1487 653 2 g
Californium Cf 98 (251) 3 A
Carbon (graphite) C 6 12.011 2.25 (graphite) 3730 subl. 4830 711 (graphite) 2,4 r
3.51 (diamond) 519 (diamond)

*Note that the atomic mass constant, mu , is equal to the uniRed atomic mass unit, u, and is deRned in terms of the mass of the carbon-12
atom: mu"1u"ma (12C)/12.
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4785

Element Symbol Atomic Relative  (g cm\3) C,m (3C) C,b (3C) cp (J kg\1 K\1) Oxidation Note
number atomic mass states

Cerium Ce 58 140.12 6.78 795 3470 184 3,4 g

Chlorine Cl 17 35.453 1.56(238 K) !101 34.7 477 1,3,4,5,6,7 m
Chromium Cr 24 51.996 7.19 1890 2482 448 2,3,6
Cobalt Co 27 58.9332 8.90 1492 2900 435 2,3
Copper Cu 29 63.546 8.92 1083 2595 385 1,2 r
Curium Cm 96 (247) 3 A
Dysprosium Dy 66 162.50 8.56 1410 2600 172 3 g
Einsteinium Es 99 (254) 3 A
Erbium Er 68 167.26 9.16 1500 2900 167 3 g
Europium Eu 63 151.96 5.24 826 1440 138 2,3 g
Fermium Fm 100 (253) 3 A
Fluorine F 9 18.9984 1.11 (85 K) !220 !188 824 1
Francium Fr 87 (223) (27) 1 A
Gadolinium Gd 64 157.25 7.95 1310 3000 234 3 g
Gallium Ga 31 69.72 5.91 29.8 2400 381 3
Germanium Ge 32 72.59 5.35 937 2830 322 4
Gold Au 79 196.9665 19.3 1063 2970 130 1,3
Hafnium Hf 72 178.49 13.3 2220 5400 146 4
Helium He 2 4.00260 0.147 (4 K) !270 !269 5.19;103 g,r
Holmium Ho 67 164.9303 8.80 1460 2600 163 3
Hydrogen H 1 1.0080 0.070 (20 K) !259 !252 1.43;104 1 g,m,r
Indium In 49 114.82 7.30 157 2000 238 1,3
Iodine I 53 126.9045 4.93 114 184 218 1,3,5,7
Iridium Ir 77 192.22 22.5 2440 5300 134 2,3,4,6
Iron Fe 26 55.847 7.86 1535 3000 448 2,3,6
Krypton Kr 36 83.80 2.16 (121 K) !157 !152 247 2 g,m
Lanthanum La 57 138.9055 6.19 920 3470 201 3 g
Lawrencium Lr 103 (257) A
Lead Pb 82 207.2 11.3 327 1744 130 2,4 g,r
Lithium Li 3 6.941 0.53 180 1330 3.39;103 1 g,m,r,
Lutetium Lu 71 174.97 9.84 1650 3330 155 3 g
Magnesium Mg 12 24.305 1.74 650 1110 1.03;103 2
Manganese Mn 25 54.9380 7.20 1240 2100 477 2,3,4,6,7
Mendelevium Md 101 (256) 3 A
Mercury Hg 80 200.59 13.6 !38.9 357 138 1,2
Molybdenum Mo 42 95.94 10.2 2610 5560 251 2,3,4,5,6 g
Neodymium Nd 60 144.24 7.00 1020 3030 188 3 g
Neon Ne 10 20.179 1.20 (27 K) !249 !246 1.03;103 g,m
Neptunium Np 93 (237) 20.4 640 3,4,5,6 A
Nickel Ni 28 58.71 8.90 1453 2730 439 2,3
Niobium Nb 41 92.9064 8.57 2470 3300 264 3,5
Nitrogen N 7 14.0067 0.808 (77 K) !210 !196 1.04;103 1,2,3,4,5 g,r
Nobelium No 102 (254) A
Osmium Os 76 190.2 22.5 3000 5000 130 2,3,4,6,8 g
Oxygen O 8 15.9994 1.15 (90 K) !218 !183 916 2 g,r
Palladium Pd 46 106.4 12.0 1550 3980 243 2,4 g
Phosphorus P 15 30.9738 1.82 (white) 44.2 (white) 280 (white) 757 (white) 3,5
2.34 (red) 590 (red) 670 (red)
Platinum Pt 78 195.09 21.4 1769 4530 134 2,4,6
Plutonium Pu 94 (242) 19.8 640 3240 3,4,5,6 A
Polonium Po 84 (210) 9.4 254 960 126 2,4 A
Potassium K 19 39.102 0.86 63.7 774 753 1
Praseodymium Pr 59 140.9077 6.78 935 3130 192 3,4
Promethium Pm 61 (147) 1030 2730 184 3 A
Protoactinium Pa 91 (231) 15.4 1230 121 4,5 Z
Radium Ra 88 (226) 5.0 700 1140 121 2 A
Radon Rn 86 (222) 4.4 (211 K) !71 !61.8 92 A
Rhenium Re 75 186.2 20.5 3180 5630 138 2,4,5,6,7
Rhodium Rh 45 102.9055 12.4 1970 4500 243 2,3,4
Rubidium Rb 37 85.4678 1.53 38.9 688 360 1 g
Ruthenium Ru 44 101.07 12.3 2500 4900 238 3,4,5,6,8 g
4786 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Element Symbol Atomic Relative  (g cm\3) C,m (3C) C,b (3C) cp (J kg\1 K\1) Oxidation Note
number atomic mass states

Samarium Sm 62 150.4 7.54 1070 1900 197 2,3 g

Scandium Sc 21 44.9559 2.99 1540 2730 556 3
Selenium Se 34 78.96 4.81 217 685 322 2,4,6
Silicon Si 14 28.086 2.33 1410 2360 711 4 r
Silver Ag 47 107.868 10.5 961 2210 234 1 g
Sodium Na 11 22.9898 0.97 97.8 890 1.23;103 1
Strontium Sr 38 87.62 2.62 768 1380 284 2 g,r
Sulphur (, rhombic)S 16 32.06 2.07 () 113 () 445 732 2,4,6 g,r
1.96 () 119 ()
Tantalum Ta 73 180.9479 16.6 3000 5420 138 5
Technetium Tc 43 (99) 11.5 2200 3500 243 7 A
Tellurium Te 52 127.60 6.25 450 990 201 2,4,6 g
Terbium Tb 65 158.9254 8.27 1360 2800 184 3,4
Thallium Tl 81 204.37 11.8 304 1460 130 1,3
Thorium Th 90 232.0381 11.7 1750 3850 113 3,4 g,Z
Thulium Tm 69 168.9342 9.33 1540 1730 159 2,3
Tin (white) Sn 50 118.69 7.28 (white) 232 2270 218 2,4 g
5.75 (grey)
Titanium Ti 22 47.90 4.54 1675 3260 523 2,3,4
Tungsten W 74 183.85 19.4 3410 5930 134 2,4,5,6
Unnilennium Une 109 A,U
Unnilhexium Unh 106 A,U
Unniloctium Uno 108 A,U
Unnilpentium Unp 105 A,U
Unnilquadium Unq 104 A,U
Unnilseptium Uns 107 A,U
Uranium U 92 238.029 19.1 1130 3820 117 3,4,5,6 g,m,Z
Vanadium V 23 50.9414 5.96 1900 3000 481 2,3,4,5
Xenon Xe 54 131.30 3.52 (165 K) !112 !108 159 2,4,6,8 g,m
Ytterbium Yb 70 173.04 6.98 824 1430 146 2,3 g
Yttrium Y 39 88.9059 4.34 1500 2930 297 3
Zinc Zn 30 65.37 7.14 420 907 385 2
Zirconium Zr 40 91.22 6.49 1850 3580 276 2,3,4 g

(g) geologically exceptional specimens are known in which the element has an isotopic composition outside the limits for normal material. The
difference between the average relative atomic mass of the element in such specimens and that given in the table may exceed considerably the
implied uncertainty.
(m) modiRed isotopic compositions may be found in commercially available material because it has been subjected to an undisclosed or inadvertent
isotopic separation. Substantial deviations in relative atomic mass of the element from that given in the table can occur.
(r) range in isotopic composition of normal terrestrial material prevents a more precise relative atomic mass being given; the tabulated Ar(E) value
should be applicable to any normal material.
(A) Radioactive element that lacks a characteristic terrestrial isotopic composition.
(Z) An element without stable nuclide(s), exhibiting a range of characteristic terrestrial compositions of long-lived radionuclide(s) such that
a meaningful relative atomic mass can be given.
(U) The names and symbols given here are systematic and based on the atomic numbers of the elements as recommended by the IUPAC Commission
on the Nomenclature of Inorganic Chemistry. The names are composed of the following roots representing digits of the atomic number:

1 un, 2 bi, 3 tri, 4 quad, 5 pent,

6 hex, 7 sept, 8 oct, 9 enn, 0 nil

The ending -ium is then added to the three roots. The three-letter symbols are derived from the Rrst letters of the corresponding roots.
(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd edn. Oxford ScientiRc
Publications.)
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4787

Electronic Con\gurations of the Elements (Ground States)

Atomic Element Shell

number
K L M N

1s 2s 2p 3s 3p 3d 4s 4p 4d 4f

1 Hydrogen 1
2 Helium 2

3 Lithium 2 1
4 Beryllium 2 2
5 Boron 2 2 1
6 Carbon 2 2 2
7 Nitrogen 2 2 3
8 Oxygen 2 2 4
9 Fluorine 2 2 5
10 Neon 2 2 6

11 Sodium 2 2 6 1
12 Magnesium 2 2 6 2
13 Aluminium 2 2 6 2 1
14 Silicon 2 2 6 2 2
15 Phosphorus 2 2 6 2 3
16 Sulphur 2 2 6 2 4
17 Chlorine 2 2 6 2 5
18 Argon 2 2 6 2 6

19 Potassium 2 2 6 2 6 1
20 Calcium 2 2 6 2 6 2
21 Scandium 2 2 6 2 6 1 2
22 Titanium 2 2 6 2 6 2 2
23 Vanadium 2 2 6 2 6 3 2
24 Chromium 2 2 6 2 6 5 1
25 Manganese 2 2 6 2 6 5 2
26 Iron 2 2 6 2 6 6 2
27 Cobalt 2 2 6 2 6 7 2
28 Nickel 2 6 2 6 8 2
29 Copper 2 2 6 2 6 10 1
30 Zinc 2 2 6 2 6 10 2
31 Gallium 2 2 6 2 6 10 2 1
32 Germanium 2 2 6 2 6 10 2 2
33 Arsenic 2 2 6 2 6 10 2 3
34 Selenium 2 2 6 2 6 10 2 4
35 Bromine 2 2 6 2 6 10 2 5
36 Krypton 2 2 6 2 6 10 2 6

Atomic Element Shell

number
K L M N O P

4s 4p 4d 4f 5s 5p 5d 5f 6s 6p 6d

37 Rubidium 2 8 18 2 6 1
38 Strontium 2 8 18 2 6 2
39 Yttrium 2 8 18 2 6 1 2
40 Zirconium 2 8 18 2 6 2 2
41 Niobium 2 8 18 2 6 4 1
42 Molybdenum 2 8 18 2 6 5 1
43 Technetium 2 8 18 2 6 6 1
44 Ruthenium 2 8 18 2 6 7 1
45 Rhodium 2 8 18 2 6 8 1
46 Palladium 2 8 18 2 6 10
4788 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Atomic Element Shell

number
K L M N O P

4s 4p 4d 4f 5s 5p 5d 5f 6s 6p 6d

47 Silver 2 8 18 2 6 10 1
48 Cadmium 2 8 18 2 6 10 2
49 Indium 2 8 18 2 6 10 2 1
50 Tin 2 8 18 2 6 10 2 2
51 Antimony 2 8 18 2 6 10 2 3
52 Tellurium 2 8 18 2 6 10 2 4
53 Iodine 2 8 18 2 6 10 2 5
54 Xenon 2 8 18 2 6 10 2 6

55 Caesium 2 8 18 2 6 10 2 6 1
56 Barium 2 8 18 2 6 10 2 6 2
57 Lanthanum 2 8 18 2 6 10 2 6 1 2
58 Cerium 2 8 18 2 6 10 2 2 6 2
59 Praseodymium 2 8 18 2 6 10 3 2 6 2
60 Neodymium 2 8 18 2 6 10 4 2 6 2
61 Promethium 2 8 18 2 6 10 5 2 6 2
62 Samarium 2 8 18 2 6 10 6 2 6 2
63 Europium 2 8 18 2 6 10 7 2 6 2
64 Gadolinium 2 8 18 2 6 10 7 2 6 1 2
65 Terbium 2 8 18 2 6 10 9 2 6 2
66 Dysprosium 2 8 18 2 6 10 10 2 6 2
67 Holmium 2 8 18 2 6 10 11 2 6 2
68 Erbium 2 8 18 2 6 10 12 2 6 2
69 Thulium 2 8 18 2 6 10 13 2 6 2
70 Ytterbium 2 8 18 2 6 10 14 2 6 2
71 Lutetium 2 8 18 2 6 10 14 2 6 1 2
72 Hafnium 2 8 18 2 6 10 14 2 6 2 2
73 Tantalum 2 8 18 2 6 10 14 2 6 3 2
74 Tungsten 2 8 18 2 6 10 14 2 6 4 2
75 Rhenium 2 8 18 2 6 10 14 2 6 5 2
76 Osmium 2 8 18 2 6 10 14 2 6 6 2
77 Iridium 2 8 18 2 6 10 14 2 6 9
78 Platinum 2 8 18 2 6 10 14 2 6 9 1
79 Gold 2 8 18 2 6 10 14 2 6 10 1
80 Mercury 2 8 18 2 6 10 14 2 6 10 2
81 Thallium 2 8 18 2 6 10 14 2 6 10 2 1
82 Lead 2 8 18 2 6 10 14 2 6 10 2 2
83 Bismuth 2 8 18 2 6 10 14 2 6 10 2 3
84 Polonium 2 8 18 2 6 10 14 2 6 10 2 4
85 Astatine 2 8 18 2 6 10 14 2 6 10 2 5
86 Radon 2 8 18 2 6 10 14 2 6 10 2 6

Atomic Element Shell

number
K L M N O p Q

5s 5p 5d 5f 6s 6p 6d 7s

87 Francium 2 8 18 32 2 6 10 2 6 1
88 Radium 2 8 18 32 2 6 10 2 6 2
89 Actinium 2 8 18 32 2 6 10 2 6 1 2
90 Thorium 2 8 18 32 2 6 10 2 6 2 2
91 Protoactinium 2 8 18 32 2 6 10 2 2 6 1 2
92 Uranium 2 8 18 32 2 6 10 3 2 6 1 2
93 Neptunium 2 8 18 32 2 6 10 4 2 6 1 2
94 Plutonium 2 8 18 32 2 6 10 6 2 6 2
95 Americium 2 8 18 32 2 6 10 7 2 6 2
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4789

Atomic Element Shell

number
K L M N O p Q

5s 5p 5d 5f 6s 6p 6d 7s

96 Curium 2 8 18 32 2 6 10 7 2 6 1 2
97 Berkelium 2 8 18 32 2 6 10 8 2 6 1 2
98 Californium 2 8 18 32 2 6 10 10 2 6 2
99 Einsteinium 2 8 18 32 2 6 10 11 2 6 2
100 Fermium 2 8 18 32 2 6 10 12 2 6 2
101 Mendelevium 2 8 18 32 2 6 10 13 2 6 2
102 Nobelium 2 8 18 32 2 6 10 14 2 6 2
103 Lawrencium 2 8 18 32 2 6 10 14 2 6 1 2

(Reprinted with permission from Mills I et al. (1993) Quantities, Units and Symbols in Physical Chemistry, 2nd edn. Oxford: Blackwell ScientiRc
Publications.)

Properties of Nuclides
The table contains the following properties of naturally occurring and some unstable nuclides:

Column
1. Z is the atomic number (number of protons) of the nuclide.
2. Symbol of the element.
3. A is the mass number of the nuclide. The H sign denotes an unstable nuclide (for elements without naturally
occurring isotopes it is the most stable nuclide) and the C sign a nuclide of sufRciently long lifetime to
enable the determination of its isotopic abundance.
4. The atomic mass is given in uniRed atomic mass units, u"ma(12C)/12, together with the standard errors in
parentheses and applicable to the last digit quoted.
5. Isotopic abundances are given as mole fractions, x, of the corresponding atoms in percents. They were
recommended in 1989 by the IUPAC Commission on Atomic Weights and Isotopic Abundances. The
uncertainties given in parentheses are applicable to the last digits quoted and cover the range of probable
variations in the materials as well as experimental errors.
6. I is the nuclear spin quantum number.
7. Under magnetic moment the maximum z-component expectation value of the magnetic dipole moment, m,
in nuclear magnetons is given. The positive or negative sign implies that the orientation of the magnetic
dipole with respect to the angular momentum corresponds to the rotation of a positive or negative charge,
respectively. An asterisk H indicates that more than one value is given in the original compilation. The value
of highest precision or most recent data is given here.
8. Under quadrupole moment, the electric quadrupole moment area is given in units of square femtometres,
fm2"10\30 m2, although most of the tables quote them in barns (1 barn"10\28 m2"100 fm2). The
positive sign implies a prolate nucleus, the negative sign an oblate nucleus. The data for Z420 were taken
from the compilation by P. PyykkoK with values for Cl and Ca corrected by D. Sundholm (private
communication), and the others from P. Raghavan. An asteriskH indicates that more than one value is given
in the original compilation.

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

1 H 1 1.007 825 035 (12) 99.985 (1) 1/2 #2.792 847 386 (63)
(D) 2 2.014 101 779 (24) 0.015 (1) 1 #0.857 438 230 (24) #0.2860 (15)
(T) 3H 3.016 049 27 (4) 1/2 #2.978 962 479 (68)

2 He 3 3.016 029 31 (4) 0.000 137 (3) 1/2 !2.127 624 848 (66)
4 4.002 603 24 (5) 99.999 863 (3) 0 0

3 Li 6 6.015 1214 (7) 7.5 (2) 1 #0.822 056 67 (26)H !0.082 (4)
7 7.016 0030 (9) 92.5 (2) 3/2 #3.256 462 53 (40)H !4.01
4790 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

4 Be 9 9.012 1822 (4) 100 3/2 !1.177 492 (17)H #5.288 (38)

5 B 10 10.012 936 9 (3) 19.9 (2) 3 #1.800 644 75 (57) #8.459 (24)
11 11.009 3054 (4) 80.1 (2) 3/2 #2.688 6489 (10) #4.059 (10)

6 C 12 12 (by deRnition) 98.90 (3) 0 0

13 13.003 354 826 (17) 1.10 (3) 1/2 #0.702 4118 (14)
14H 14.003 241 982 (27) 0 0

7 N 14 14.003 074 002 (26) 99.634 (9) 1 #0.403 761 00 (6) #2.01 (2)
15 15.000 108 97 (4) 0.366 (9) 1/2 !0.283 188 842 (45)

8 O 16 15.994 914 63 (5) 99.762 (15) 0 0

17 16.999 1312 (4) 0.038 (3) 5/2 !1.893 80 !2.558 (22)
18 17.999 1603 (9) 0.200 (12) 0 0

9 F 19 18.998 403 22 (15) 100 1/2 #2.628 868 (8)

10 Ne 20 19.992 4356 (22) 90.48 (3) 0 0

21 20.993 8428 (21) 0.27 (1) 3/2 !0.661 797 (5) #10.155 (75)
22 21.991 3831 (18) 9.25 (3) 0 0

11 Na 23 22.989 7677 (10) 100 3/2 #2.217 6556 (6)H #10.06 (20)

12 Mg 24 23.985 0423 (8) 78.99 (3) 0 0

25 24.985 8374 (8) 10.00 (1) 5/2 !0.855 465 (8) #19.94 (20)
26 25.982 5937 (8) 11.01 (2) 0 0

13 Al 27 26.981 5386 (8) 100 5/2 #3.641 504 687 (65) #14.03 (10)

14 Si 28 27.976 9271 (7) 92.23 (1) 0 0

29 28.976 4949 (7) 4.67 (1) 1/2 !0.555 29 (3)
30 29.973 7707 (7) 3.10 (1) 0 0

15 P 31 30.973 7620 (6) 100 1/2 #1.131 60 (3)

16 S 32 31.972 070 70 (25) 95.02 (9) 0 0

33 32.971 458 43 (23) 0.75 (1) 3/2 #0.643 8212 (14) !6.78 (13)
34 33.967 866 65 (22) 4.21 (8) 0 0
36 35.967 080 62 (27) 0.02 (1) 0 0

17 Cl 35 34.968 852 721 (69) 75.77 (5) 3/2 #0.821 8743 (4) !8.11 (8)
37 36.965 902 62 (11) 24.23 (5) 3/2 #0.684 1236 (4) !6.39 (6)

18 Ar 36 35.967 545 52 (29) 0.337 (3) 0 0

38 37.962 7325 (9) 0.063 (1) 0 0
40 39.962 3837 (14) 99.600 (3) 0 0

19 K 39 38.963 7074 (12) 93.2581 (44) 3/2 #0.391 507 31 (12)H #5.9 (6)
40 39.963 9992 (12) 0.0117 (1) 4 !1.298 1003 (34) !7.3 (7)
41 40.961 8254 (12) 6.7302 (44) 3/2 #0.214 870 09 (22) #7.2 (7)

20 Ca 40 39.962 5906 (13) 96.941 (18) 0 0

42 41.958 6176 (13) 0.647 (9) 0 0
43 42.958 7662 (13) 0.135 (6) 7/2 !1.317 643 (7) !4.09 (8)
44 43.955 4806 (14) 2.086 (12) 0 0
46 45.953 689 (4) 0.004 (4) 0 0
48 47.952 533 (4) 0.187 (4) 0 0

21 Sc 45 44.955 9100 (14) 100 7/2 #4.756 4866 (18) !22 (1)H
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4791

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

22 Ti 46 45.952 6294 (14) 8.0 (1) 0 0

47 46.951 7640 (11) 7.3 (1) 5/2 !0.788 48 (1) #29 (1)
48 47.947 9473 (11) 73.8 (1) 0 0
49 48.947 8711 (11) 5.5 (1) 7/2 !1.104 17 (1) #24 (1)
50 49.944 7921 (12) 5.4 (1) 0 0

23 V 50C 49.947 1609 (17) 0.250 (2) 6 #3.345 6889 (14) 20.9 (40)H
51 50.943 9617 (17) 99.750 (2) 7/2 #5.148 705 73 (18) !5.2 (10)H

24 Cr 50 49.946 0464 (17) 4.345 (13) 0 0

52 51.940 5098 (17) 83.789 (18) 0 0
53 52.940 6513 (17) 9.501 (17) 3/2 !0.474 54 (3) !15 (5)H
54 53.938 8825 (17) 2.365 (7) 0 0

25 Mn 55 54.938 047 1 (16) 100 5/2 #3.468 7190 (9) #33 (1)H

26 Fe 54 53.939 6127 (15) 5.8 (1) 0 0

56 55.934 9393 (16) 91.72 (30) 0 0
57 56.935 3958 (16) 2.2 (1) 1/2 #0.090 623 00 (9)H
58 57.933 2773 (16) 0.28 (1) 0 0

27 Co 59 58.933 1976 (16) 100 7/2 #4.627 (9) #40.4 (40)H

28 Ni 58 57.935 3462 (16) 68.077 (9) 0 0

60 59.930 7884 (16) 26.223 (8) 0 0
61 60.931 0579 (16) 1.140 (1) 3/2 !0.750 02 (4) #16.2 (15)
62 61.928 3461 (16) 3.634 (2) 0 0
64 63.927 9679 (17) 0.926 (1) 0 0

29 Cu 63 62.929 5989 (17) 69.17 (3) 3/2 #2.2227 3456 (14)H !21.1 (4)H
65 64.927 7959 (20) 30.83 (3) 3/2 #2.381 61 (19)H !19.5 (4)

30 Zn 64 63.929 1448 (19) 48.6 (3) 0 0

66 65.926 0347 (17) 27.9 (2) 0 0
67 66.927 1291 (17) 4.1 (1) 5/2 #0.875 2049 (11)H #15.0 (15)
68 67.924 8459 (18) 18.8 (4) 0 0
70 69.925 325 (4) 0.6 (1) 0 0

31 Ga 69 68.925 580 (3) 60.108 (9) 3/2 #2.016 589 (44) #16.8H
71 70.924 7005 (25) 39.892 (9) 3/2 #2.562 266 (18) #10.6H

32 Ge 70 69.924 2497 (16) 21.23 (4) 0 0

72 71.992 0789 (16) 27.66 (3) 0 0
73 72.923 4626 (16) 7.73 (1) 9/2 !0.879 4677 (2) !17.3 (26)
74 73.921 1774 (15) 35.94 (2) 0 0
76 75.921 4016 (17) 7.44 (2) 0 0

33 As 75 74.921 5942 (17) 100 3/2 #1.439 475 (65) #31.4 (6)H

34 Se 74 73.922 4746 (16) 0.89 (2) 0 0

76 75.919 2120 (16) 9.36 (1) 0 0
77 76.919 9125 (16) 7.63 (6) 1/2 #0.535 074 24 (28)H
78 77.917 3076 (16) 23.78 (9) 0 0
80 79.916 5196 (19) 49.61 (10) 0 0
82 81.916 6978 (23) 8.73 (6) 0 0

35 Br 79 78.918 3361 (26) 50.69 (7) 3/2 #2.106 400 (4) #33.1 (4)
81 80.916 289 (6) 49.31 (7) 3/2 #2.270 562 (4) #27.6 (4)

36 Kr 78 77.920 396 (9) 0.35 (2) 0 0

80 79.916 380 (9) 2.25 (2) 0 0
82 81.913 482 (6) 11.6 (1) 0 0
83 82.914 135 (4) 11.5 (1) 9/2 !0.970 669 (3) #25.3 (5)
84 83.911 507 (4) 57.0 (3) 0 0
86 85.910 616 (5) 17.3 (2) 0 0
4792 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

37 Rb 85 84.911 794 (3) 72.165 (20) 5/2 #1.353 3515 (8)H #22.8 (43)H
87C 86.909 187 (3) 27.835 (20) 3/2 #2.751 818 (2) #13.2 (1)

38 Sr 84 83.913 430 (4) 0.56 (1) 0 0

86 85.909 2672 (28) 9.86 (1) 0 0
87 86.908 8841 (28) 7.00 (1) 9/2 !1.093 6030 (13)H #33.5 (20)
88 87.905 6188 (28) 82.58 (1) 0 0

39 Y 89 88.905 849 (3) 100 1/2 !0.137 415 42 (34)H

40 Zr 90 89.904 7026 (26) 51.45 (3) 0 0

91 90.905 6439 (26) 11.22 (4) 5/2 !1.303 62 (2) !20.6 (10)
92 91.905 0386 (26) 17.15 (2) 0 0
94 93.906 3148 (28) 17.38 (4) 0 0
96 95.908 275 (4) 2.80 (2) 0 0

41 Nb 93 92.906 3772 (27) 100 9/2 #6.1705 (3) !32 (2)H

42 Mo 92 91.906 809 (4) 14.84 (4) 0 0

94 93.905 0853 (26) 9.25 (3) 0 0
95 94.905 8411 (22) 15.92 (5) 5/2 !0.9142 (1) !2.2 (1)H
96 95.904 6785 (22) 16.68 (5) 0 0
97 96.906 0205 (22) 9.55 (3) 5/2 !0.9335 (1) #25.5 (13)H
98 97.905 4073 (22) 24.13 (7) 0 0
100 99.907 477 (6) 9.63 (3) 0 0

43 Tc 98H 97.907 215 (4) 6

44 Ru 96 95.907 599 (8) 5.52 (6) 0 0

98 97.905 287 (7) 1.88 (6) 0 0
99 98.905 9389 (23) 12.7 (1) 5/2 !0.6413 (51)H #7.9 (4)
100 99.904 2192 (24) 12.6 (1) 0 0
101 100.905 5819 (24) 17.0 (1) 5/2 !0.7188 (60)H #45.7 (23)
102 101.904 3485 (25) 31.6 (2) 0 0
104 103.905 424 (6) 18.7 (2) 0 0

45 Rh 103 102.905 500 (4) 100 1/2 !0.088 40 (2)

46 Pd 102 101.905 634 (5) 1.02 (1) 0 0

104 103.904 029 (6) 11.14 (8) 0 0
105 104.905 079 (6) 22.33 (8) 5/2 !0.642 (3) #66.0 (11)H
106 105.903 478 (6) 27.33 (3) 0 0
108 107.903 895 (4) 26.46 (9) 0 0
110 109.905 167 (20) 11.72 (9) 0 0

47 Ag 107 106.905 092 (6) 51.839 (7) 1/2 !0.113 679 65 (15)H
109 108.904 756 (4) 48.161 (7) 1/2 !0.130 690 62 (22)H

48 Cd 106 105.906 461 (7) 1.25 (4) 0 0

108 107.904 176 (6) 0.89 (2) 0 0
110 109.903 005 (4) 12.49 (12) 0 0
111 110.904 182 (3) 12.80 (8) 1/2 !0.594 886 07 (84)H
112 111.902 757 (3) 24.13 (28) 0 0
113C 112.904 400 (3) 12.22 (8) 1/2 !0.622 300 92 (87)
114 113.903 357 (3) 28.73 (28) 0 0
116 115.904 755 (4) 7.49 (12) 0 0

49 In 113 112.904 061 (4) 4.3 (2) 9/2 #5.5289 (2) #79.9
115C 114.903 882 (4) 95.7 (2) 9/2 #5.5408 (2) #81.0H

50 Sn 112 111.904 826 (5) 0.97 (1) 0 0

114 113.902 784 (4) 0.65 (1) 0 0
115 114.903 348 (3) 0.34 (1) 1/2 !0.918 83 (7)
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4793

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

50 Sn 116 115.901 747 (3) 14.53 (11) 0 0

117 116.902 956 (3) 7.68 (7) 1/2 !1.001 04 (7)
118 117.901 609 (3) 24.23 (11) 0 0
119 118.903 311 (3) 8.59 (4) 1/2 !1.047 28 (7)
120 119.902 1991 (29) 32.59 (10) 0 0
122 121.903 4404 (30) 4.63 (3) 0 0
124 123.905 2743 (17) 5.79 (5) 0 0

51 Sb 121 120.903 8212 (29) 57.36 (8) 5/2 #3.3634 (3) !36 (4)H
123 122.904 2160 (24) 42.64 (8) 7/2 #2.5498 (2) !49 (5)

52 Te 120 119.904 048 (21) 0.096 (2) 0 0

122 121.903 050 (3) 2.603 (4) 0 0
123 122.904 2710 (22) 0.908 (2) 1/2 !0.736 9478 (8)
124 123.902 8180 (18) 4.816 (6) 0 0
125 124.904 4285 (25) 7.139 (6) 1/2 !0.888 505 13 (43)H
126 125. 903 3095 (25) 18.95 (1) 0 0
128 127.904 463 (4) 31.69 (1) 0 0
130 129.906 229 (5) 33.80 (1) 0 0

53 I 127 126.904 473 (5) 100 5/2 #2.813 273 (84) !78.9

54 Xe 124 123.905 8942 (22) 0.10 (1) 0 0

126 125.904 281 (8) 0.09 (1) 0 0
128 127.903 5312 (17) 1.91 (3) 0 0
129 128.904 7801 (21) 26.4 (6) 1/2 !0.777 9763 (84)
130 129.903 5094 (17) 4.1 (1) 0 0
131 130.905 072 (5) 21.2 (4) 3/2 #0.691 8619 (39) !12.0 (12)
132 131.904 144 (5) 26.9 (5) 0 0
134 133.905 395 (8) 10.4 (2) 0 0
136 135.907 214 (8) 8.9 (1) 0 0

55 Cs 133 132.905 429 (7) 100 7/2 #2.582 0246 (34)H !0.371 (14)H

56 Ba 130 129.906 282 (8) 0.106 (2) 0 0

132 131.905 042 (9) 0.101 (2) 0 0
134 133.904 486 (7) 2.417 (27) 0 0
135 134.905 665 (7) 6.592 (18) 3/2 #0.837 943 (17)H #16.0 (3)H
136 135.904 553 (7) 7.854 (36) 0 0
137 136.905 812 (6) 11.23 (4) 3/2 #0.937 365 (20)H #24.5 (4)H
138 137.905 232 (6) 71.70 (7) 0 0

57 La 138C 137.907 105 (6) 0.0902 (2) 5 #3.713 646 (7) #45 (2)H
139 138.906 347 (5) 99.9098 (2) 7/2 #2.783 0455 (9) #20 (1)

58 Ce 136 135.907 140 (50) 0.19 (1) 0 0

138 137.905 985 (12) 0.25 (1) 0 0
140 139.905 433 (4) 88.48 (10) 0 0
142 141.909 241 (4) 11.08 (10) 0 0

59 Pr 141 140.907 647 (4) 100 5/2 #4.2754 (5) !5.89 (42)

60 Nd 142 141.907 719 (4) 27.13 (12) 0 0

143 142.909 810 (4) 12.18 (6) 7/2 !1.065 (5) !63 (6)
144 143.910 083 (4) 23.80 (12) 0 0
145 144.912 570 (4) 8.30 (6) 7/2 !0.656 (4) !33 (3)
146 145.013 113 (4) 17.19 (9) 0 0
148 147.916 889 (4) 5.76 (3) 0 0
150 149.920 887 (4) 5.64 (3) 0 0

61 Pm 145H 144.912 743 (4) 5/2

4794 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

62 Sm 144 143.911 998 (4) 3.1 (1) 0 0

147C 146.914 894 (4) 15.0 (2) 7/2 !0.8148 (7) !25.9 (26)
148 147.914 819 (4) 11.3 (1) 0 0
149 148.917 180 (4) 13.8 (1) 7/2 !0.6717 (7)H #7.5 (8)H
150 149.917 273 (4) 7.4 (1) 0 0
152 151.919 728 (4) 26.7 (2) 0 0
154 153.922 205 (4) 22.7 (2) 0 0

63 Eu 151 150.919 702 (8) 47.8 (15) 5/2 #3.4717 (6) #90.3 (10)H
153 152.921 225 (4) 52.2 (15) 5/2 #1.5330 (8)H #241.2 (21)H

64 Gd 152 151.919 786 (4) 0.20 (1) 0 0

154 153.920 861 (4) 2.18 (3) 0 0
155 154.922 618 (4) 14.80 (5) 3/2 !0.257 23 (35)H #130 (2)H
156 155.922 118 (4) 20.47 (4) 0 0
157 156.923 956 (4) 15.65 (3) 3/2 !0.337 26 (55)H #136 (2)H
158 157.924 019 (4) 24.84 (12) 0 0
160 159.927 049 (4) 21.86 (4) 0 0

65 Tb 159 158.925 342 (4) 100 3/2 #2.014 (4) #143.2 (8)

66 Dy 156 155.924 277 (8) 0.06 (1) 0 0

158 157.924 403 (5) 0.10 (1) 0 0
160 159.925 193 (4) 2.34 (6) 0 0
161 160.926 930 (4) 18.9 (2) 5/2 !0.4803 (25)H #250.7 (20)H
162 161.926 795 (4) 25.5 (2) 0 0
163 162.928 728 (4) 24.9 (2) 5/2 #0.6726 (35) #264.8 (21)
164 163.929 171 (4) 28.2 (2) 0 0

67 Ho 165 164.930 319 (4) 100 7/2 #4.173 (27) #349 (3)H

68 Er 162 161.928 775 (4) 0.14 (1) 0 0

164 163.929 198 (4) 1.61 (1) 0 0
166 165.930 290 (4) 33.6 (2) 0 0
167 166.932 046 (4) 22.95 (15) 7/2 !0.563 85 (12) #356.5 (29)
168 167.932 368 (4) 26.8 (2) 0 0
170 169.935 461 (4) 14.9 (2) 0 0

69 Tm 169 168.934 212 (4) 100 1/2 !0.2316 (15)

70 Yb 168 167.933 894 (5) 0.13 (1) 0 0

170 169.934 759 (4) 3.05 (6) 0 0
171 170.936 323 (3) 14.3 (2) 1/2 #0.493 67 (1)H
172 171.936 378 (3) 21.9 (3) 0 0
173 172.938 208 (3) 16.12 (21) 5/2 !0.679 89 (3)H #280 (4)
174 173.938 859 (3) 31.8 (4) 0 0
176 175.942 564 (4) 12.7 (2) 0 0

71 Lu 175 174.940 770 (3) 97.41 (2) 7/2 #2.2327 (11)H #349 (2)H
176C 175.942 679 (3) 2.59 (2) 7 #3.1692 (45)H #492 (3)H

72 Hf 174 173.940 044 (4) 0.162 (3) 0 0

176 175.941 406 (4) 5.206 (5) 0 0
177 176.943 217 (3) 18.606 (4) 7/2 #0.7935 (6) #336.5 (29)H
178 177.943 696 (3) 27.297 (4) 0 0
179 178.945 8122 (29) 13.629 (6) 9/2 !0.6409 (13) #379.3 (33)H
180 179.946 5457 (30) 35.100 (7) 0 0

73 Ta 180 179.947 462 (4) 0.012 (2) 8

181 180.947 992 (3) 99.988 (2) 7/2 #2.3705 (7) #328 (6)H
 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES 4795

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

74 W 180 179.946 701 (5) 0.13 (4) 0 0

182 181.948 202 (3) 26.3 (2) 0 0
183 182.950 220 (3) 14.3 (1) 1/2 #0.117 784 76 (9)
184 183.950 928 (3) 30.67 (15) 0 0
186 185.954 357 (4) 28.6 (2) 0 0

75 Re 185 184.952 951 (3) 37.40 (2) 5/2 #3.1871 (3) #218 (2)H
187C 186.955 744 (3) 62.60 (2) 5/2 #3.2197 (3) #207 (2)H

76 Os 184 183.952 488 (4) 0.02 (1) 0 0

186 185.953 830 (4) 1.58 (30) 0 0
187 186.955 741 (3) 1.6 (3) 1/2 #0.064 651 89 (6)
188 187.955 830 (3) 13.3 (7) 0 0
189 188.958 137 (4) 16.1 (8) 3/2 #0.659 933 (4) #85.6 (28)
190 189.958 436 (4) 26.4 (12) 0 0
192 191.961 467 (4) 41.0 (8) 0 0

77 Ir 191 190.960 584 (4) 37.3 (5) 3/2 #0.1507 (6)H #81.6 (9)H
193 192.962 917 (4) 62.7 (5) 3/2 #0.1637 (6)H #75.1 (9)H

78 Pt 190 189.959 917 (7) 0.01 (1) 0 0

192 191.961 019 (5) 0.79 (6) 0 0
194 193.962 655 (4) 32.9 (6) 0 0
195 194.964 766 (4) 33.8 (6) 1/2 #0.609 52 (6)
196 195.964 926 (4) 25.3 (6) 0 0
198 197.967 869 (6) 7.2 (2) 0 0

79 Au 197 196.966 543 (4) 100 3/2 #0.148 158 (8)H #54.7 (16)H

80 Hg 196 195.965 807 (5) 0.15 (1) 0 0

198 197.966 743 (4) 9.97 (8) 0 0
199 198.968 254 (4) 16.87 (10) 1/2 #0.505 885 49 (85)
200 199.968 300 (4) 23.10 (16) 0 0
201 200.970 277 (4) 13.18 (8) 3/2 !0.560 2257 (14)H #38.5 (40)H
202 201.970 617 (4) 29.86 (20) 0 0
204 203.973 467 (5) 6.87 (4) 0 0

81 Tl 203 202.972 320 (5) 29.524 (14) 1/2 #1.622 257 87 (12)
205 204.974 401 (5) 70.476 (14) 1/2 #1.638 214 61 (12)

82 Pb 204 203.973 020 (5) 1.4 (1) 0 0

206 205.974 440 (4) 24.1 (1) 0 0
207 206.975 872 (4) 22.1 (1) 1/2 #0.582 583 (9)H
208 207.976 627 (4) 52.4 (1) 0 0

83 Bi 209 208.980 374 (5) 100 9/2 #4.1106 (2) !37.0 (26)H

84 Po 209H 208.982 404 (5) 1/2

85 At 210H 209.987 126 (12)

86 Rn 222H 222.017 571 (3) 0 0

87 Fr 223H 223.019 733 (4) 3/2 #1.17 (2) #117 (1)

88 Ra 226H 226.025 403 (3) 0 0

89 Ac 227H 227.027 750 (3) 3/2 #1.1 (1) #170 (20)

90 Th 232C 232.038 0508 (23) 100 0 0

91 Pa 231H 231.035 880 (3) 3/2 2.01 (2) !172 (5)

4796 APPENDIX 14 / PROPERTIES OF PARTICLES, ELEMENTS AND NUCLIDES

Z Symbol A Atomic mass, ma (u) Isotopic abundance, Nuclear Magnetic moment, Quadrupole
100 x spin, I m (
N) moment, Q (fm2)

92 U 233H 233.039 628 (3) 5/2 0.59 (5) #366.3 (8)

234C 234.040 9468 (24) 0.0055 (5) 0 0
235C 235.043 9242 (24) 0.7200 (12) 7/2 !0.38 (3)H #455 (9)H
238C 238.050 7847 (23) 99.2745 (60) 0 0

93 Np 237H 237.048 1678 (23) 5/2 #3.14 (4) #388.6 (6)

94 Pu 244H 244.064 199 (5) 0

95 Am 243H 243.061 375 (3) 5/2 #1.61 (4) #420 (130)

96 Cm 247* 247.070 347 (5)

97 Bk 247* 247.070 300 (6)

98 Cf 251* 251.079 580 (5)

99 Es 252* 252.082 944 (23)

100 Fm 257* 257.095 099 (8)

101 Md 258* 258.098 57 (22)

102 No 259* 259.100 931 (12)

103 Lr 260* 260.105 320 (60)

104 Unq 261* 261.108 69 (22)

105 Unp 262* 262.113 76 (16)

106 Unh 263* 263.118 22 (13)

107 Uns 262* 263.122 93 (45)

108 Uno 265* 265.130 16 (99)

109 Une 266* 266.137 64 (45)

(Reprinted with permission from Mills I et al. (1993) Quantities, Uses and Symbols in Physical Chemistry, 2nd edn. Oxford: Blackwell
ScientiRc Publications.)
 APPENDIX 15 / SOLVENTS FOR ULTRAVIOLET SPECTROPHOTOMETRY 4797

15. SOLVENTS FOR ULTRAVIOLET

SPECTROPHOTOMETRY
Solvent Cutoff wavelength Dielectric constant
(nm) (203C)

Acetic acid 260 6.15

Acetone 330 20.7 (253C)
Acetonitrile 190 37.5
Benzene 280 2.284
2-Butanol 260 15.8 (253C)
n-Butyl acetate 254
Carbon disulphide 380 2.641
Carbon tetrachloride 265 2.238
1-Chlorobutane 220 7.39 (253C)
Chloroforma 245 4.806
Cyclohexane 210 2.023
1,2-Dichloroethane 226 10.19 (253C)
1,2-Dimethoxyethane 240
N,N-Dimethylacetamide 268 59 (833C)
N,N-Dimethylformamide 270 36.7
Dimethyl sulphoxide 265 4.7
1,4-Dioxane 215 2.209 (253C)
Diethyl ether 218 4.335
Ethanol 210 24.30 (253C)
2-Ethoxyethanol 210
Ethyl acetate 255 6.02 (253C)
Glycerol 207 42.5 (253C)
n-Hexadecane 200 2.06 (253C)
n-Hexane 210 1.890
Methanol 210 32.63 (253C)
2-Methoxyethanol 210 16.9
Methyl cyclohexane 210 2.02 (253C)
Methyl ethyl ketone 330 18.5
Methyl isobutyl ketone 335
2-Methyl-1-propanol 230 1
N-Methyl-2-pyrrolidone 285 32.0
Pentane 210 1.844
n-Pentyl acetate 212
1-Propanol 210 20.1 (253C)
2-Propanol 210 18.3 (253C)
Pyridine 330 12.3 (253C)
Tetracholoroethyleneb 290
Tetrahydrofuran 220 7.6
Toluene 286 2.379 (253C)
1,1,2-Tricholro-1,2,2-triSuroethane 231
2,2,4-Trimethylpentane 215 1.936 (253C)
o-Xylene 290 2.568
m-Xylene 290 2.374
p-Xylene 290 2.270
Water 78.54 (253C)
a
Stabilized with ethanol to avoid phosgene formation.
b
Stabilized with thymol (isopropyl meta-cresol).
Reprinted from T. J. Bruno and P. D. N. Svoronos, CRC Handbook of Basic Tables for Chemical Analysis, CRC Press, Boca
Raton, FL, 1989, p. 212.
4798 APPENDIX 16 / STATISTICAL TABLES

16. STATISTICAL TABLES

The following tables are presented in a format that is compatible with the needs of analytical chemists: the
signiRcance level P"0.05 has been used in most cases, and it has been assumed that the number of
measurements available is fairly small. Except where stated otherwise, these abbreviated tables have been
taken, with permission, from Elementary Statistics Tables by Henry R. Neave, published by George Allen
& Unwin Ltd. (Tables 1}3, 5}6, and 7}11). The reader requiring statistical data corresponding to signiRcance
levels and/or numbers of measurements not covered in the tables is referred to these sources.

Table 1 The t-distribution

Value of
t
for a confidence interval of: 90% 95% 98% 99%

Critical value of t for P values of: 0.10 0.05 0.02 0.01
Number of degrees of freedom

1 6.31 12.71 31.82 63.66

2 2.92 4.30 6.96 9.92
3 2.35 3.18 4.54 5.84
4 2.13 2.78 3.75 4.60
5 2.02 2.57 3.36 4.03
6 1.94 2.45 3.14 3.71
7 1.89 2.36 3.00 3.50
8 1.86 2.31 2.90 3.36
9 1.83 2.26 2.82 3.25
10 1.81 2.23 2.76 3.17
12 1.78 2.18 2.68 3.05
14 1.76 2.14 2.62 2.98
16 1.75 2.12 2.58 2.92
18 1.73 2.10 2.55 2.88
20 1.72 2.09 2.53 2.85
30 1.70 2.04 2.46 2.75
50 1.68 2.01 2.40 2.68
R 1.64 1.96 2.33 2.58

The critical values of
t
are appropriate for a two-tailed test. For a one-tailed test the value is taken from the column for twice the desired
P-value, e.g. for a one-tailed test, P"0.05, 5 degrees of freedom, the critical value is read from the P"0.10 column and is equal
to 2.02.
 APPENDIX 16 / STATISTICAL TABLES 4799

Table 2 Critical values of F for a one-tailed test (P"0.05)

v1: 1 2 3 4 5 6 7 8 9 10 12 15 20
v2

1 161.4 199.5 215.7 224.6 230.2 234.0 236.8 238.9 240.5 241.9 243.9 245.9 248.0
2 18.51 19.00 19.16 19.25 19.30 19.33 19.35 19.37 19.38 19.40 19.41 19.43 19.45
3 10.13 9.552 9.277 9.117 9.013 8.941 8.887 8.845 8.812 8.786 8.745 8.703 8.660
4 7.709 6.944 6.591 6.388 6.256 6.163 6.094 6.041 5.999 5.964 5.912 5.858 5.803
5 6.608 5.786 5.409 5.192 5.050 4.950 4.876 4.818 4.772 4.735 4.678 4.619 4.558

6 5.987 5.143 4.757 4.534 4.387 4.284 4.207 4.147 4.099 4.060 4.000 3.938 3.874
7 5.591 4.737 4.347 4.120 3.972 3.866 3.787 3.726 3.677 3.637 3.575 3.511 3.445
8 5.318 4.459 4.066 3.838 3.687 3.581 3.500 3.438 3.388 3.347 3.284 3.218 3.150
9 5.117 4.256 3.863 3.633 3.482 3.374 3.293 3.230 3.179 3.137 3.073 3.006 2.936
10 4.965 4.103 3.708 3.478 3.326 3.217 3.135 3.072 3.020 2.978 2.913 2.845 2.774

11 4.844 3.982 3.587 3.357 3.204 3.095 3.012 2.948 2.896 2.854 2.788 2.719 2.646
12 4.747 3.885 3.490 3.259 3.106 2.996 2.913 2.849 2.796 2.753 2.687 2.617 2.544
13 4.667 3.806 3.411 3.179 3.025 2.915 2.832 2.767 2.714 2.671 2.604 2.533 2.459
14 4.600 3.739 3.344 3.112 2.958 2.848 2.764 2.699 2.646 2.602 2.534 2.463 2.388
15 4.543 3.682 3.287 3.056 2.901 2.790 2.707 2.641 2.588 2.544 2.475 2.403 2.328

16 4.494 3.634 3.239 3.007 2.852 2.741 2.657 2.591 2.538 2.494 2.425 2.352 2.276
17 4.451 3.592 3.197 2.965 2.810 2.699 3.614 2.548 2.494 2.450 2.381 2.308 2.230
18 4.414 3.555 3.160 2.928 2.773 2.661 2.577 2.510 2.456 2.412 2.342 2.269 2.191
19 4.381 3.522 3.127 2.895 2.740 2.628 2.544 2.477 2.423 2.378 2.308 2.234 2.155
20 4.351 3.493 3.098 2.866 2.711 2.599 2.514 2.447 2.393 2.348 2.278 2.203 2.124

v1"number of degrees of freedom of the numerator and v2"number of degrees of freedom of the denominator.

Table 3 Critical values of F for a two-tailed test (P"0.05)

v1: 1 2 3 4 5 6 7 8 9 10 12 15 20
v2

1 647.8 799.5 864.2 899.6 921.8 937.1 948.2 956.7 963.3 968.6 976.7 984.9 993.1
2 38.51 39.00 39.17 39.25 39.30 39.33. 39.36 39.37 39.39 39.40 39.41 39.43 39.45
3 17.44 16.04 15.44 15.10 14.88 14.73 14.62 14.54 14.47 14.42 14.34 14.25 14.17
4 12.22 10.65 9.979 9.605 9.364 9.197 9.074 8.980 8.905 8.844 8.751 8.657 8.560
5 10.01 8.434 7.764 7.388 7.146 6.978 6.853 6.757 6.681 6.619 6.525 6.428 6.329

6 8.813 7.260 6.599 6.227 5.988 5.820 5.695 5.600 5.523 5.461 5.366 5.269 5.168
7 8.073 6.542 5.890 5.523 5.285 5.119 4.995 4.899 4.823 4.761 4.666 4.568 4.467
8 7.571 6.059 5.416 5.053 4.817 4.652 4.529 4.433 4.357 4.295 4.200 4.101 3.999
9 7.209 5.715 5.078 4.718 4.484 4.320 4.197 4.102 4.026 3.964 3.868 3.769 3.667
10 6.937 5.456 4.826 4.468 4.236 4.072 3.950 3.855 3.779 3.717 3.621 3.522 3.419

11 6.724 5.256 4.630 4.275 4.044 3.881 3.759 3.664 3.588 3.526 3.430 3.330 3.226
12 6.554 5.096 4.474 4.121 3.891 3.728 3.607 3.512 3.436 3.374 3.277 3.177 3.073
13 6.414 4.965 4.347 3.996 3.767 3.604 3.483 3.388 3.312 3.250 3.153 3.053 2.948
14 6.298 4.857 4.242 3.892 3.663 3.501 3.380 3.285 2.209 3.147 3.050 2.949 2.844
15 6.200 4.765 4.153 3.804 3.576 3.415 3.293 3.199 3.123 3.060 2.963 2.862 2.756

16 6.115 4.687 4.077 3.729 3.502 3.341 3.219 3.125 3.049 2.986 2.889 2.788 2.681
17 6.042 4.619 4.011 3.665 3.438 3.277 3.156 3.061 2.985 2.922 2.825 2.723 2.616
18 5.978 4.560 3.954 3.608 3.382 3.221 3.100 3.005 2.929 2.866 2.769 2.667 2.559
19 5.922 4.508 3.903 3.559 3.333 3.172 3.051 2.956 2.880 2.817 2.720 2.617 2.509
20 5.871 4.461 3.859 3.515 3.289 3.128 3.007 2.913 2.837 2.774 2.676 2.573 2.464

v1"number of degrees of freedom of the numerator and v2"number of degrees of freedom of the denominator.
4800 APPENDIX 16 / STATISTICAL TABLES

Table 4 Critical values of Q (P"0.05) Table 7 Wilcoxon signed rank test. Critical values for the test
statistic at P"0.05
Sample size Critical value
n One-tailed test Two-tailed test
4 0.831
5 0.717 5 0 NA
6 0.621 6 2 0
7 0.570 7 3 2
8 0.524 8 5 3
9 0.492 9 8 5
10 0.464 10 10 8
11 13 10
Taken from E.P. King, J. Am. Statist. Assoc., 1958, 48, 531, by 12 17 13
permission of the American Statistical Association. 13 21 17
14 25 21
15 30 25

Table 5 Critical values of 2 (P"0.05) The null hypothesis can be rejected when the test statistic is4the
tabulated value. NA indicates that the test cannot be applied.
Number of degrees of freedom Critical value

1 3.84
2 5.99 Table 8 Wilcoxon rank sum test; Mann-Whitney U-test. Critical
3 7.81 values for U or the lower of T1 and T2 at P"0.05
4 9.49
5 11.07 n1 n2 One-tailed test Two-tailed test
6 12.59
7 14.07 3 3 0 NA
8 15.51 3 4 0 NA
9 16.92 3 5 1 0
10 18.31 3 6 2 1
4 4 1 0
4 5 2 1
4 6 3 2
4 7 4 3
Table 6 The sign test 5 5 4 2
5 6 5 3
n r"0 1 2 3 4 5 6 7 5 7 6 5
6 6 7 5
4 0.063 0.313 0.688 6 7 8 6
5 0.031 0.188 0.500 7 7 11 8
6 0.016 0.109 0.344 0.656
7 0.008 0.063 0.227 0.500 The null hypothesis can be rejected when U or the lower T value
8 0.004 0.035 0.144 0.363 0.637 is4the tabulated value. NA indicates that the test cannot be
9 0.002 0.020 0.090 0.254 0.500 applied.
10 0.001 0.011 0.055 0.172 0.377 0.623
11 0.001 0.006 0.033 0.113 0.274 0.500
12 0.000 0.003 0.019 0.073 0.194 0.387 0.613
13 0.000 0.002 0.011 0.046 0.133 0.290 0.500
14 0.000 0.001 0.006 0.029 0.090 0.212 0.395 0.605
15 0.000 0.000 0.004 0.018 0.059 0.151 0.304 0.500

The table uses the binomial distribution with P"0.5 to give the
probabilities of r or less successes for n"4}15. These values
correspond to a one-tailed sign test and should be doubled for
a two-tailed test.
 APPENDIX 16 / STATISTICAL TABLES 4801

Table 9 The Spearman rank correlation coefficient. Critical Table 11 The Kolomogorov test for normality
values for  at P"0.05
n One-tailed test Two-tailed test
n One-tailed test Two-tailed test
3 0.367 0.376
5 0.900 1.000 4 0.345 0.375
6 0.829 0.886 5 0.319 0.343
7 0.714 0.786 6 0.297 0.323
8 0.643 0.738 7 0.280 0.304
9 0.600 0.700 8 0.265 0.288
10 0.564 0.649 9 0.252 0.274
11 0.536 0.618 10 0.241 0.262
12 0.504 0.587 11 0.231 0.251
13 0.483 0.560 12 0.222 0.242
14 0.464 0.538 13 0.215 0.234
15 0.446 0.521 14 0.208 0.226
16 0.429 0.503 15 0.201 0.219
17 0.414 0.488 16 0.195 0.213
18 0.401 0.472 17 0.190 0.207
19 0.391 0.460 18 0.185 0.202
20 0.380 0.447 19 0.181 0.197
20 0.176 0.192

Critical values for one-tailed and two-tailed tests at P"0.05. The

approapriate value is compared with the maximum difference
Table 10 The Kolmogorov goodness of fit test between the experimental and theoretical cumulative frequency
curves.
n One-tailed test Two-tailed test

1 0.950 0.975
2 0.776 0.842
3 0.636 0.708
4 0.565 0.624
5 0.509 0.563
6 0.468 0.519
7 0.436 0.483
8 0.410 0.454
9 0.388 0.430
10 0.369 0.409
11 0.352 0.392
12 0.338 0.375
13 0.326 0.361
14 0.314 0.349
15 0.304 0.338
16 0.295 0.327
17 0.286 0.318
18 0.278 0.309
19 0.271 0.301
20 0.265 0.294

Critical values for one-tailed and two-tailed tests at P"0.05. The

appropriate value is compared with the maximum difference be-
tween the experimental and theoretical cumulative frequency
curves.
4802 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION

17. THIN LAYER (PLANAR)

CHROMATOGRAPHY: DETECTION
A. Misra, Bareilly, India
Copyright ^ 2000 Academic Press

The following tables list detection methods and reagents suitable for detecting and identifying substances
separated by thin-layer (planar) chromatography.

Table 1.1 Methods of detection on TLC plates (aluminium oxide) by heating (Types 150/T or 60/E)

Substances Temperature/time Remarks

Pesticides, e.g. aminocarb, captan, difolatan,
landrin, rotenone

4-3-Ketosteroids, e.g. testosterone and epi-
testosterone in urine

4-3-Ketosteroids, e.g. trimethylsilyl-testosterone
Testosterone
Testosterone

4-3-Ketosteroids, e.g. progesterone in plasma
2003C, 45 min Induction of fluorescence in weakly fluorescent
or nonfluorescent pesticides and amplification of
natural fluorescence. There are some
differences between basic and acidic aluminium
oxide layers
1803C, 20 min Pale blue induced fluorescence (fl"440 nm)
for
4-3-ketosteroids, detection limit: 5 ng
1803C, 20 min or 1503C, Conversion of
4-3-ketosteroids or their
20 min trimethylsilyl or acetyl derivatives in fluorescent
components, whereby the detection limits were
improved by 65% for the acetates.
5-3-keto-
and
5-3-OH-steroids also react with the same
sensitivity
1803C, 20 min Induced fluorescence (fl'430 nm, cut off filter)
by thermal treatment of the chromatogram, the
fluorescence increased by a factor of 2.5 by
dipping in a solution of Triton X-100
!chloroform (1#4). Working range: 2}50 ng
substance per chromatogram zone.
Prewashing the layers with methanol-ammonia
solution (25%) (50#50) increased the
precision
1803C, 20 min Induced fluorescence and fluorescence
amplification by a factor of 25 by dipping the
chromatogram in a solution of Triton X-100
!chloroform (1#4)
1503C, 20 min Conversion of
4-3-ketosteroids into fluorescent
derivatives (fl"440 nm). Relatively selective
for progesterone at 1503C detection limit: 2}5 ng
 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION 4803

Table 1.2 Methods of fluorimetric detection on TLC plates (silica gel) by heating

Substances Temperature/time Remarks

Essential oil components 800}9003C Induction of fluorescence in a special apparatus

Steroids, e.g. cholesterol, triolein, androsterone; 110}1503C, 2}12 h Conversion to fluorescent derivatives by heating
sugars, e.g. fructose, glucose, ribose; amino acids,
pyrimidines, purines, alkaloids

Alkaloids, e.g. raubasine and its metabolites in 1203C, 1 h Amplification of the natural fluorescence of
plasma, urine and bile raubasine (fl"482 nm), detection limit 20 ng

Alkaloids, e.g. reserpine, rescinnamine 1053C, 2 h Induced fluorescence (fl'500 nm, cut off
filter). Possibly formation of 3-dehydro
derivatives

Alkaloids, e.g. reserpine, ajmaline, rescinnamine 1053C, 2 h or 1053C, 15 h Induction of stable fluorescence (fl'480 nm,
cut off filter), detection limits 5}20 ng

Alkaloids, e.g. cocaine, ecgonine, 2803C, 8 min or 2603C, Pale blue induced fluorescence (fl'390 nm,
benzoylecgonine, ecgonine methyl ester 10}30 min cut off filter), fluorescence amplification by
a factor of 2 on dipping in liquid paraffin solution;
detection limits: (10 ng

Alkaloids, e.g. lupanine, angustifoline, sparteine, 1303C, 17}35 h Induced blue fluorescence (fl"400 nm),
lupinine, hydroxylupanine detection limits: 10 ng

Pesticides, e.g. dursban, azinphos-methyl, 200}2253C, 20}120 min Induced fluorescence or amplification of natural
menazon, imidan, phosalone, zinophos fluorescence; detection limits: 10}300 ng

Organophosphorus pesticides, e.g. coumaphos, 2003C, 45 min Induced fluorescence or amplification of natural
menazon, maretin, dursban fluorescence, detection limits: 1}80 ng

Pesticides, e.g. fuberidazol 2003C, 45 min Amplification of the natural fluorescence of

some pesticides and bathochromic shift of the
excitation and emission maxima; detection
limits: 5}100 ng

Pesticides, e.g. coumatetralyl, 2003C, 45 min Induced fluorescence (fl'430 nm, cut off
methabenzthiazuron, propylisom, naptalam, filter); detection limits: 6}600 ng
thioquinox, warfarin etc.

Coumaphos 2003C, 20 min Residue analysis; induced fluorescence on

heating (fl'400 nm); detection limit: 1 ng

Potasan, coumaphos, coroxon 2003C, 20 min Induced blue fluorescence (fl"430 nm or

450 nm), idenfication of the fluorescent
derivatives as chlorferon or 4-
methylumbelliferone

Coumaphos 2003C, 20 min Residue determination in honey, induced

fluorescence (fl'400 nm, cut off filter);
detection limit: 0.5 ng

Rubratoxin B 2003C, 10 min Induced fluorescence that can be intensified by

gassing the previously heated chromatogram
plates with ammonia vapours (10 min). This also
alters the colour of the emitted light to pale blue

Glucose or methyglucosides 1353C, 3 min or 1403C, Induced yellow fluorescence

10 min

Sugar derivatives Mild heating over a Bunsen No details of whether fluorescence was
burner produced or if a carbonization reaction occurred

Sugars, e.g. glucose, fructose, galactose, 1603C, 10 min Production of fluorescence by heating the
mannose etc. chromatogram after covering it with a glass plate.
Sugar alcohols and C1}C1 bonded oligosacchar-
ides do not react; detection limit: 10 ng
4804 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION

Table 1.2 Continued

Substances Temperature/time Remarks

Sugars, e.g. glucose, glucosamine, fucose,
raffinose, cellobiose, methylated sugars
80P2603C, gradient or Production of fluorescence by temperature gradi-
2003C, 5 min ents (103C/30 s) to determine the optimum heat-
ing temperature for the individual substances.
Oligosaccharides require higher temperatures
than monosaccharides. Detection limit: 1 nMol.
The fluorescence colours are characteristic
particularly for the methylated sugars

Lipids, e.g. -sitosterol, geraniol, dolichol, 2003C, 15 min Induced fluorescence; detection limits: (1
g
squalene, cholesterol cholesterol

C-Nucleosides Moderate heating on a hot No details of whether fluorescence or

plate carbonization was produced

Nomifensine and metabolites 703C, 2 h#UV254 Heating and simultaneous UV irradiation

produced intense yellow fluorescence
(fl'460 nm, cut off filter)

Reproduced with permission from, Jork H, Funk W, Fisher W and Wimmer H (1994). Thin Layer Chromatography : Regents and
Detection Methods, volume 1B. Weinheim: Wiley VCH.

Table 1.3 Examples of fluorimetric detection after thermal treatment of layer after chromatography

Substances Temperature/time Remarks

Sugars, e.g. lactose, glucose, fructose 1203C, 15 min Violet fluorescence on a dark blue background

Sugars, e.g. lactose, glucose, fructose 1203C, 15 min Induced fluorescence; detection limits in nano-
gram range

Glucose, fructose Infrared lamp or 1703C Heating produced stable bluish-white fluores-
each for 3 min cence (exe"365 nm and fl'400 nm, cut off
filter K 400), detection limits; 5}10 ng

Sugars, e.g. glucose, rhamnose, xylose etc. 1603C, 3}4 min or infrared Induction of brilliant stable fluorescence
lamp exe"365 nm and fl'400 nm (cut off filter
K 400), sugar alcohols do not fluoresce; detec-
tion limits; 5}10 ng

Creatine, creatinine, uric acid in urine and serum 1503C, 3}4 min Stable fluorescence exe"365 nm and fl'
400 nm (cut off filter K 400)

Sugars, e.g. sucrose, ribose, xylose 1503C, 3}4 min Induced fluorescence exe"365 nm and fl'
400 nm (cut off filter K 400)

Reproduced with permission from, Jork H, Funk W, Fisher W and Wimmer H (1994). Thin Layer Chromatography : Regents and
Detection Methods, volume 1B. Weinheim: Wiley VCH.
 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION 4805

Table 2 Some substances that produce intense fluorescence when treated with ionized nitrogen after they have been chromato-
graphed

Substance Exposure time [s] Substance Exposure time [s]

Cholesterol 60 Oleic acid 180

Cholesteryl pelargonate 60 Morphine 180
Progesterone 60 Codeine 180
Testosterone 60 Cocaine 180
Dieldrin 60 Dimerol 180
Tetrahydrocannabinol 60 Phenobarbital 180
Inositol 60 Chlorpromazine 180
Lauryl alcohol 180 d-Amphetamine sulfate 180
n-C22H46 180 Methadone 180
Phenol 180

Reproduced with permission from Jork H, Funk W, Fisher W and Wimmer H (1994). Thin Layer Chromatography : Regents and
Detection Methods, volume 1B. Weinheim: Wiley VCH.

Table 3 Reagents suitable for the recognition of functional groups

Functional group Reagent Remarks

Acetylene compounds Dicobaltoctacarbonyl Formation of coloured complexes. After the

Aldehydes 4-Amino-3-hydrazino-5-mercapto-
1,2,4-triazole (Purpald reagent)
Aldehydes 2,4-Dinitrophenylhydrazine
Aldehydes Hydrazine sulfate#hydrochloric acid
Alcohols 4-(4-Nitrobenzyl)pyridine
Alcohols (diols, polyols, sugars) Lead(IV) acetate
} dichlorofluorescein
Amines (primary) Ninhydrin
Amines (primary aliphatic and aromatic) Diphenylboric
anhydride#salicylaldehyde (DOOB)
reagent excess has been washed out,
reaction with bromine vapour yields cobalt
bromide, which reacts with -nitroso--
naphthol to yield red chromatogram zones
on an almost colourless background
Aldehydes yield violet chromatogram zones
on a whitish-yellow background. Some
alcohols form yellow to orange-coloured
chromatogram zones
Formation of coloured hydrazones or
osazones. It is possible to distinguish
between saturated and unsaturated
hydrazones using potassium
hexacyanoferrate (III)
Aromatic aldehydes yield coloured
hydrazones
Amino compounds, esters and ethers do not
interfere, but phenols and acids as well as
epoxides, olefins and substances containing
labile halogen probably do
Diol cleavage of vicinal diols, e.g. sugars,
sugar alcohols. The lead tetraacetate
consumed is no longer available to
decompose the fluorescent
dichlorofluorescein
Reddish or bluish chromatogram zones are
produced, amino sugars and amino acids
also react. Unexpectedly ascorbic acid also
reacts
Fluorescent reaction products are produced

Amines (primary) o-Phthalaldehyde (OPA) In the presence of mercaptoethanol

o-phthalaldehyde reacts with primary
amines and amino acids to yield
fluorescent isoindole derivatives
4806 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION

Table 3 Continued

Functional group Reagent Remarks

Amines (primary)
Amines (primary)
Amines (primary aromatic)
Amines (primary aromatic)
Amines (capable of coupling)
Amines (primary and secondary)
Amines (primary and
secondary aromatic)
Amines (secondary aliphatic
and alicyclic)
Amines (long-chain primary, secondary
and tertiary plus quaternary ammonium
salts)
Carboxyl groups (carboxylic acids)
Carboxyl groups (carboxylic acids)
Carboxyl groups (carboxylic acids)
Halogen derivatives
Ketones
Trinitrobenzenesulfonic acid (TNBS)
Fluorescamine
Sodium nitrite#-naphthol or
Bratton-Marshall reagent
4-(Dimethylamino)benzaldehyde
#acid
Fast blue salt B, fast blue salt BB, fast
black salt K, diazotized sulfanilic acid
(Paulys reagent), diazotized
sulfanilamide or 4-nitroaniline
7-Chloro-4-nitrobenzo-2-oxa-1,3-
diazole (NBD chloride)
p-Chloranil
Sodium nitroprusside#
acetaldehyde
Cobalt(II) thiocyanate
Indicators, e.g. bromocresol green,
bromocresol green#bromophenol
blue#potassium permanganate,
bromocresol purple, methyl
red#bromothymol blue
2,6-Dichlorophenol-indophenol
(Tillmanns reagent)
Aniline#aldose (e.g. glucose)
Silver nitrate, ammoniacal
(Dedonders, Tollens or Zaffaronis
reagent)
2,4-Dinitrophenylhydrazine
On heating primary amines react with TNBA
to yield intensely coloured Meisenheimer
complexes. Amino acids also react
Primary aliphatic and aromatic amines yield
fluorescent derivatives. Primary aromatic
amines yield stable yellow-coloured
derivatives that can be eluted from the TLC
layer
Diazotization of the primary amine followed
by coupling with -naphthol or N-(L-
naphthyl)ethylenediamine. Sulfonamides
also react
Alkaloids and indole derivatives also react
Intensely coloured azo dyes are produced.
Catecholamines, imidazoles and phenols
also react
Fluorescent 4-nitrobenzofurazan derivatives
are produced. Phenols and thiols also react
The reaction depends on the catalytic effect
of silica gel. Monochlorobenzene as solvent
for the reagent, also contributes. There is no
reaction on cellulose layers
Secondary aliphatic and alicyclic amines
yield blue-coloured chromatogram zones
(e.g. morpholine, diethanol amine)
Long-chain primary, secondary and tertiary
amines and long-chain quaternary
ammonium salts yield blue chromatogram
zones on a pink background
Detection depends on the colour change of
the indicator in acid medium. Quaternary
ammonium salts give a colour change in
some cases
Organic acids release the red undissociated
acid from the blue mesomerically stabilized
phenolate anion. Reductones reduce the
reagent to a colourless compound
The action of acid causes glucose to be
converted to furfural which reacts with aniline
to yield a coloured product
Halogen compounds yield black
chromatogram zones on a pale grey
background
Formation of coloured hydrazones or
osazones. It is possible to distinguish
between saturated and unsaturated
hydrazones using potassium
hexacyanoferrate (III)
 APPENDIX 17 / THIN LAYER (PLANAR) CHROMATOGRAPHY: DETECTION
Table 3 Continued
Functional group Reagent
Nitro derivatives Benzylcyanide#benzyl-
trimethylammonium hydroxide
Peroxides 1-Naphthol#N4-ethyl-N4\ (2-
methanesulfonamidoethyl)-2-methyl-
1,4-phenylenediamine (peroxide
reagent)
Peroxides Iron(II) sulfate#ammonium
thiocyanate
Peroxides Potassium iodide#starch
Peroxides N,N-Dimethyl-1,4-phenylenediamine
(N,N-DPDD), N,N,N,N-tetra-methyl-
1,4-phenylene-diamine (TPDD)
Phenols 7-Chloro-4-nitrobenzo-2-oxa-1,3-
diazole (NBD chloride)
Phenols (capable of coupling) Fast blue salt B, fast blue salt BB, fast
black salt K, diazotized sulfanilic acid
(Paulys reagent) diazotized
sulfanilamide or 4-nitroaniline
Thiols, thioethers, disulfides Sodium metaperiodate#benzidine
Thiols 7-Chloro-4-nitobenzo-2-oxa-1,3-
diazole (NBD chloride)
Reproduced with permission from Jork H, Funk W, Fisher W and Wimmer H (1994). Thin Layer Chromatography: Regents and
Detection Methods, volume 1B. Weinheim: Wiley VCH.
See also: II/Chromatography: Thin-Layer (Planar): Densitometry and Image Analysis; Spray Reagents.
4807
Remarks
Nitro compounds, e.g. explosives, or
pesticides containing nitro groups yield gray
to bluish-green chromatogram zones on
a brownish background
A quinonimine dyestuff is produced on
reaction with peroxides
Peroxides rapidly oxidize iron(II) to iron(III)
ions which react to yield brown-red iron(III)
thiocyanate complexes
Peroxides release free iodine which forms
a blue complex with the starch
Peroxides, e.g. alkyl hydroperoxides, oxidize
N,N-DPDD to Wursters red and TPDD to
Wursters blue
Fluorescent 4-nitrobenzofuran derivatives
are produced. Primary and secondary
aromatic amines and thiols also react
Intensely coloured azo dyes are formed.
Catecholamines, imidazoles and amines
capable of coupling also react
Substances with divalent sulfur yield white
chromatogram zones on a blue background
Fluorescent 4-nitrobenzofuran derivatives
are formed. Primary and secondary aromatic
amines and phenols also react
4808 APPENDIX 18 / WAVELENGTH SCALE

18. WAVELENGTH SCALE