ANGEWANDTE CHEMIE

V O L U M E 2 - N U M B E R7

JUI,Y 1963
P A G E S 333-404

Principles and Techniques of Diffuse-Reflectance Spectroscopy

BY PROF. DR. G. KORT~JM,DR. w. BRAUN, AND DIPL.-CHEM. G. HERZOG
INSTITUT FOR PHYSIKALISCHE CHEMIE DER UNIVERSITAT TU BINGEN (GERMANY)

The prerequisites for applicability of the Kubelka- Munk theor)) o f diffuse reflectance for
obtaining characteristic color curves qf powders are discussed, and some experimental
support .for the theory is given. Methods are described for eliminating surface reflection,
which is alwa.vs superimposed upon the diJuse reflection and which therefbre distorts the
spectrum. The effkct of interactions with the adsorbent on the rejlcction spectra of adsorbed
molecules is demonstrated by means of examples, and the speciul suitability qf the method
,for investigating chemisorption and for follo wing reactions at phase boundaries is pointed
out. By analogy to the Beer-Lambert law, the Kubelka-Mid ,function can be used for
quantitative photometric analysis. The reflectance curves of while standards are presented,
and measurements relating to the conditions fbr injinite1.v thich layers are reported. The
scattering coefficient of the Kubelka-Munk function has been es1imated.for several samples
of color-filters as a fiinction q f grain-size and wavelength. The c</fkctof moisture on difuse
reflectance spectra is discussed, and details of the measuring techniques are presented.

1. Test of the Kubelka-Munk Theory

standard, e . g . , MgO or NaCI, k is t h e molar absorption

coefficient of the sample, defincd by Lamberts law
4, cDo.exp(-k.d), a n d s is the scattering coefficient, which
will provisionally be assumed 1 0 be iirdependent of wavelength.
-2

The theory of diffuse reflection and of the transparency

of light-scattering and light-absorbing layers has been
investigated by numerous authors, since this problem is
of great practical significance in many areas (covering
power of paints, white content of papers, quality of
frosted and ground glass, etc.). The most general theory
was developed by Kubelka and Munk [1,2]. The equations derived by other authors proved as a rule to be
special cases of the Kubelka-Munk equation [*I. For
the special case of an infinitely thick, opaque layer (for
practical purposes, with fine powders, this is already
achieved at a layer depth of a few millimeters), the Kubelka-Munk equation reads

where R,
(I&ample/ostandard is t h e relative diffuse
reflectance of such a layer, referred t o a non-absorbing

According to Equation (I), a plot of F(R,) as a function of the extinction coefficient k, which is known from
transmission measurements, should yield a straight line.
This has been confirmed for weakly absorbing materials
[ 5 ] . Another example of the validity of the KubelkaMunk function is provided by Figure 1. Here the light
absorption of a didymium-glass filter [* *I was measured
in comparison to a quartz plate of equal thickness, in
order to eliminate losses due to surface reflection. The
glass was then triturated for several hours in an agate
mortar, and the reflectance of the resulting powder
measured. For comparison window glass ground in the
same manner was used, sincc the glass matrix of the
filter (without the coloring rare-earth ions) was unavailable. Figure 1 shows that the transmittance and reflectance curves can be brought more or less into coincidence by a shift parallel to the ordinate axis [?]. Thus,

[I] P. Kubelka and F. Munk, Z . techn. Physik 12, 593 (1931).

[2] P.Kubelka, J. opt. SOC.America 38, 448 (1948).
[ * ] For a brief summary of the theory and derivation of the
Kubelka-Munk function see for instance Kortunz [3] and Korturn and Vogel[4].
[ 3 ] G. Kortum, Rev. univ. Mines Ser. 9, Bd. 15, S. 495 (1959).
[4] G . Kortum and J . Vogel, Z. physik. Chem. N. F. 18,110 (1958).

[ 5 ] G. Kortum and H . Schottler, Z. Elektrochem. Ber. Bunsenges.

physik. Chem. 57, 353 (1953).
[**I Filter No. BG 36 by Schott and Gen., Mainz (Germany).
[t]Analogous measurements carricd out previously by Schreyer
[3, 61 with the similar color glass BG 20 allegedly yielded appreciably different transmittance and reflectance spectra. Reexamination of these measurements showed that they evidently
were incorrect.

Angew. Chem. internat. Edit.

Vol. 2 (1963) No. 7

333

measurement of the diffuse reflectance yields a "characteristic color curve" for the investigated material.

-0.5

-1.0

b) by diluting the sample with a white standard.

Method a) has already been described in detail [8].
Plane polarized light is used, and care must be taken to
ensure that the angle between the plane of incidence and
theplaneof vibration of the radiation is either 0" or90".
Under these conditions, regularly reflected light maintains its state of polarization and can be eliminated by
an analyzer crossed against the polarizer, while the diffusely reflectcd light changes its state of polarization and
is thus partly transmitted by the analyzer. The polarizing
sheets previously employed have the disadvantage of
being sufficiently transparent only in the visible region
of the spectrum [*I. For this reason, a reflection attachment for the Zeiss spectrophotometer PMQ I1 was
developed in which the polarizer consists of a Glan
prism (aperture 7 "), while the analyzer is a shortened
Glan-Thompson prism (aperture 30 ") [**I; the attachment extends the range down to about 250 mp.

-1.5

15000

20000

25000

IA2901]
Fig. 1. Transmittance spectrum (Curve 1) and reflectance spectrum
(Curve 2 ) of a didymium-glass filter. The transmittance spectrum was
measured relative to q u a r t z glass 0.5 mm thick, the reflectance spectrum
relative to powdered colorless glass.

Ordinates: left (Curve 1): log E.

right (Curve 2 ) : log F(Rm)
Abscissa: Wave number icm 'I.

Particularly in the short-wave region of t h e spectrum, there

are small systematic deviations, due essentially to a slight
dependence of the scattering coefficient s on wavelength;
for further details, see Section 6. These deviations may also
be due in part to the fact that the glass matrix of the filter
was not available for comparison.
I t is important for these measurements that t h e coloring
components in the glass be present in highly dilute concentration. This ensures that, for equal grain-size of the
sample and standard, the scattering coefficients will likewise
be equal, so that these factors cancel o u t in comparison
measurements, if s is independent of A. On the other hand,
if the sample measured consists of pure absorbents, then
the scattering coefficients of sample and standard will
generally be different, even when the mean grain-sizes of
the two powders a r e equal, and thus the characteristic color
curve obtained will be distorted. T h e reflection spectrum
of the sample would then also become dependent upon the
white standard used. Moreover, with pure absorbents, even
for finest grain-sizes, a regular (surface) reflection is always
observed [ 7 ] , which adultelates the diffuse reflectance spectrum in proportion t o the absorbance of the material.

2. Elimination of Regular Reflection

The regular reflection arising from the plane surfaces of
crystallites can be eliminated in two ways:
a) by making measurements between crossed polarizing
sheets (Polaroids) or prisms.
[6] G. Kortiim and G. Schreyeu, Z. Naturforsch. I l a , 1018 (1956).
[7] G. Kortiim and J . Vogrl, Z. physik. Chem. N. F. 18,230 (1958).

334

15000

20000

25000

30000

35000

Fig. 2. Reflectance spectra of a c h r o m e alum sample (grain-size about

100 I*) measured relative to Gal'* with polarized light (Curve I ) and
natural light (Curve 2).
Ordinate: log F(Rm).
Abscissa: Wave number [cni-'I.

Figure 2 shows measurements carried out on a sample

of chrome alum (grain-size about 100 p) with this reflection attachment. These were made with CaF2 as
standard, with and without polarizing prisms. Curve 1
shows log F(R,) values obtained using polarized light
and lies above Curve 2, obtained with unpolarized light.
It can be seen that measurements performed without
polarizing prisms (Curve 2) give a spectrum which is depressed by the component of regular reflection the more
strongly, the more strongly the material absorbs, since
diffuse and regular reflection counteract each other.
Strong absorption causes strong regular reflection, but
weak diffuse reflection.
In method b) for eliminating regular reflection, the
sample is highly diluted with the non-absorbing standard
and then measured against the pure standard [9]. At
sufficiently high dilution ( I : l o 3 to 1 :lO5), the regular
reflections from the sample and from the standard are
approximately equal and, within the limits of experimental accuracy, do not affect the results. This is
the case with the color filter mentioned above. Figure 3

[*I Recently, polarizing sheets that are transparent in the ultraviolet have also been reported [8].
[**I Obtainable from B. Hallc, Berlin-Steglitz (Germany); cf.
G. Kortiim: Koiorimetrie, Photometrie und Spektrometrie. 4th
Edit., Springer, Heidelberg 1962, p. 356.
[8] A . S. Makas, J . opt. SOC.America 52, 43 (1962).
[9] G. Kortiim and G. Schreyrr, Angew. Chem. 67, 694 (1955).
Angew. Chem. internat. Edit.

Vol. 2 (1963) No. 7

provides an example of the effect of dilution on the reflectance spectrum. Pure anthraquinone was diluted in
stages by triturating it in a ball mill with thoroughly

45

40

3.5

sample and standard are practically eliminated. If it is

also ensured that the dilution I S sufficient (depending on
the specific surface area of the white standard, the mole
fraction should lie between 10 3 and l O - 5 ) , so th,tt no
complete monomolecular laycr of the materisl under investigation is present on the adsorbent, then the spectrum of the adsorbed single malecule is obtained, and,
because of interaction with the adsorbent, this is more
or less different from that of the free molecules (in the
vaporized state or in inert solvents).
If the adsorption forces are only of the vein dei Wnals
type (physical adsorption), thcn, as a rule, the following
differences between the reflectance and the transmittance
spectra of a diluted solution are observed: the reflectnnce
spectrum is broadened; the bands are displiced toward
longer wavelengths; the vibrational structure, if present
at ali, is strongly suppressed, frequently appearing only
as broad humps. Examples of this are the spectra of
anthracene (Fig. 4) and anthraquinone (Fig. 3).

30

Fig. 3. Reflectance spectra of anthraquinone

(1) undiluted,
(2) adsorbed on NaCl (mole fraction x = 1.26X lo-*),
(3) adsorbed on NaCl (x = 5 . 0 x 10- 3).
(4) adsorbed on NaCl (x = 1.9x 10-4).
Curve 5 is the absorption spectrum of a dilute alcoholic solution of
anthraquinone.
Ordinates: left (Curves 1--4): log [F(R,)/cl; right (Curve 5): log i.
Abscissa: Wave number [ c w 11.

dried NaCl; after each dilution, the reflectance spectrum

was measured against pure NaCl as standard. With increasing dilution, the two characteristic bands of anthraquinone at 29600 and 39000 cm-1 become increasingly
evident. At mole fractions of less than 3 x 10-4, the
spectrum becomes independent of the degree of dilution
and is similar to the spectrum in solution. The spectrum
therefore approaches that of single molecules adsorbed
on the surface of the standard, as can be seen from the
fact that an identical spectrum is obtained when the
same material is adsorbed by the standard from a dilute
solution.
The fact that here the short-wavelength bands of anthraquinone appear so markedly depressed in the spectrum of
the pure material is due not only to regular reflection, but
probably also to coupling of the %-electron systems of
several molecules, if the latter lie in parallel layers along the
axis of greatest polarizability. Similar observations were
made by Perkampus [lo] during transmission measurements
of thin layers of aromatic materials. This interaction can be
eliminated by sufficient d il IIt ion.

3. Alteration of Reflectance Spectra of Adsorbed

Molecules by Interaction with the Adsorbent

The dilution method is particularly suitable for reflection-spectroscopical investigations, since all uncertainties due to regular reflection or differing grain-size in the
[lo] H . H . Perkampus, Z . physik. Chem. N.F. 13, 278 (1957);
19,206 (1959).

Angew. Chem. internut. Edit. 1 Vol. 2 (1963) No. 7

-0

-1

-1 5

-2 0
I

@gx

30000

40000

Fig. 4. Reflectance spectra of anthracene adsorbed on anhydrous NaCl

(Curve 1) and on NaCl after admission of moisture (Curve 2). Mole
fraction of the adsorbed anthracene: Y -= 1.42x 10 4 . Curves 3 and 4
show the absorption spectra of alcoholic solutions of anthracene and
9,lO-dimethylanthracene,respectively.
Ordinates: left (Curves IJ): log F(RCn);inside right (Curve 3): log
outside right (Curve 4): log E .
Abscissa: Wave number [cm-'].

E;

These effects correspond roughly to the differences

due to solvatochromism bctween spectra in solution
and vapor spectra, except that, in general, the
former are more pronounccd, It is very characteristic
that the development of the vibrational structure is
highly dependent on whether the surface of the adsorbent is completely dry or whether it contains adsorbed
water (Fig. 4). Anthracene adsorbed on absolutely dry
NaCl shows hardly any vibrational structure on the pband; this indicates that thc residence time in the excited state is strongly curtailed by the adsorption. When
moist air is admitted, the anthracene molecules are evidently displaced from the surface by water, and the vi-

335

brational structure emerges again. The p-band of adsorbed anthracene is displaced toward longer wavelengths by about 1200 cm'I, and, thus, has about the
same position as the p-band of 9,IO-dimethylanthracene
in solution; from this it can be concluded that the interaction between the adsorbate and adsorbent is effected
primarily at the 9- and 10-positions of the anthracene.
With dianthracene, too, the two molecules are likewise
attached at C-9 and C-10, and the reflectance spectrum
corresponds to the transmission spectrum of o-xylene.

linear region can extend u p t o x 1 lo--;. The deviation

of the curve indicates that the surface of the adsorbent
is approaching saturation by the first monomolecular
layer. Conversely, the specific surface area of the adsorbent can be determined from this saturation concetitration [17]. If the exjwiiiicnt:~l data lying in the linear

I n the case of strong interaction between the adsorbent a n d

t h e adsorbate (chernisorption), which as a rule is due t o
charge transfer, correspondingly strong changes also occur
in t h e reflectance spectrum compared t o that of the free
molecules. This has often been discussed in recent years
[I1-15], Chemical and photochemical reactions a t phase
boundaries can likewise be easily followed quantitatively
and evaluated kinetically using diIt'use-Yeflectance spectrosCOPY [ 161.

4. Photometric Evaluation of the Kubelka-Munk

Function

When applying the dilution method, if the molar absorption coefficient k in Equation ( I ) is replaced by the
product 2 . 3 0 c q where e denotes the extinction coefficient and c represents the molar concentration, then

Fig. 5 . I<ubelka-MunL limiting curve. Dependence of t h e b a n d intensity

o f pyrene a t 29500 cni I oil t h e iiiiiltl Ii-action. Pyrene in t h e f o r m of a
molecular c o m p o u n d with 1,3,S-lrinitrohen-lene was adsorbed o n
anhydrous NaCl (mean grain-siie i i h o u l I [L).

O r d i n a t e : F(R

,)

Abscissa: Mole fraction

[pyrencl

[pyrenel I INnCIl

i. e. at constant external conditions (grain-size, wavelength, temperatut e, c f c . ) ,the Kubelka-Munk function

is proportional to the molar concentration c, or
at
also to the mole fracfion x
sufficiently high dilution
of the absorbate; if F(R ) is plotted as a function of c
or x, it should yield a straight line passing through the
origin. It should therefore be possible to employ Equation (2) for quantitative photometric analysis in the
same way as the Beer-Lambert law, remembering that
both are liniiting laws for sufficientlylow concentrations.
As an example of the validity of Equation (2) and of the
accuracy attainable in photometric determinations of
concentration, the function F(R,) of the pyrene band
at 29500 cm I is shown in Figure 5 as a function of the
mole fraction x, using dry NaCl as adsorbent. The mean
grain-diameter of the NaCI, determined according to the
method of Brunauer, Emnwtt, and Teller (BET method)
was about I p. The curve remains linear up to
x 2 x 10-4 and then falls off. Just at what point it
bends away depends critically on the grain-size of the
adsorbent; for very fine grains (SiOz, MgO, etc.), the

of the measurements was 3.2

the average deviation
of the mean error rs,, ajn was approximately 1.3 %.
The mean error depends strongly on the moisture-content of the samples and on the uniformity of distribution
of the adsorbate over the itdsorbent (duration of grinding). Under favorable conditions, the mean error of
photometric determinations may be reduced to 1-2 %
[*I. If, even at very low concentrations, deviations from
the linear relationship given by Equation (2) are observed, then, as in the case o i true deviations from the
Beer-Lambert law, this indicates the presence of association or dissociation processes in the surface-boundary
phase. Such deviations can be used, for example, to
deter mine the dissociation constan ts of adsor bed molecular compounds [20].

[ I I] G. Kortiinr, J . Vogel, and W . Brnun, Angew. Chern. 70, 651

(1958).
[I21 G. Kurtiiwi and J . Vogel, Chem. Ber. 93, 706 (1960).
[I31 G. Korfiinr, Trans. Faraday Soc. 58, 1624 (1962).
[I41 G. Kortzin7 and G'. Ra,ver, Z. physik. Chem. N.F. 33, 254
(1962).
[IS] G. Korturn, and W. Rr.aun. Liebigs Ann. Chem. 632, 104
(1960).
[I61 G. Kortun?, M . Kortlirn-Seller. and S . D. Builey, J. physic.
Chem. 66,2439 (1962).

[ I 7 1 G. Kortiim and D. Oe//mx, 7,.

physik. Chem. N. F. 34, 58
( I 962).
[*I For exarnplc, photomctric determination of concentrations
by measurement o f diffuse rcllcctance has been employed for
evaluating paper chromatograms [IS], and for rutile-anatase
analyses [191.
1181 G. Kortiim and J . Vogel, Angcw. Chern. 71, 451 (1959).
[I91 G. Kortiiin and G. Herrog. %. analyt. Chern. 190, 239 (1962).
[20] G. Kurriin7 and W. Brtruii. %. physik. Chern. N.F. 18, 242
(1958); 28, 362 (IYhl).

region are reduced u p to the concentration x =- 2 x 10-4,

i.e. if, in view of the proportionality of F(R,)
and x,
the function is multiplied hy the appropriate concentration ratio, then the values ot' F(R,,) obtained lie distrib-

336

uted between 0.592 and 0.544. The mean error 0 =

'x;

Angew. Chem. infc~rirtrl.Edit.1 Vol. 2 (1963)

??
n 1
.-

No. 7

5. Reflectance of Various White Standards

For purposes of investigating chemical problems by

means of reflectance measurements, the white standard
must frequently be chosen in such a mrinner that certain
interactions between the adsorbate and adsorbent, e . g .
acid-base reactions, complex formation, ligand exchange, etc., are either ensured or avoided. It is therefore
desirable to measure the reflectance of as inany white
standards as possible ovei the whole presently accessible
spectrum range, since any significant self-absorption of
the white standard will adulterate the reflection spectrum
of a sample in the same way as transmission spectra are
distorted by seif-absorption or" the solvent.

prolonged triturating and statistical averaging o f the grainsize were deemed unnecessary, !,ince the experimental curves
were only to servc for a qualitative appraisal of t h e reflectance of t h e white standard i n any case. Since these curves
still depend on the grain-six, thc grain-size distribution, the
moisture-cgntent, as well a s oil the purity and origin of the
samples, they generally cannot lie reproduced quantitatively.

There is still no standard available that would provide

readily reproducible diffuse reflecrance that is greater
than about 95 for the spectral region between 2.5 p
and 200 mp. Generally, the white standard of best reproducibility is considered to be freshly prepared MgO;
its absolute diffuse reflectance has been measured by
several authors t o an accuracy of about , 0.5 % [21].
K'-f

'Dincidentl'"reflected

(3)

However, particularly in the ultraviolet region, its reflectance varies with time. The frosted glass standards
that are commercially available are c'onstant, but their
reflectances decrease sharply above 30000 cm-1. Hence,
for our standard, we employed a quirtz cuvette with
matted windows made of Suprasil@, which was filled
with Suprasil powder (grain-diameter = 1 p) and fused
off in an atmosphere of nitrogen [*I. This standard is
easy to clean, and, when placed in the same position in
the sample holder of the reflection attachment of a
Beckman D K 2 spectrophotometer, it always gives a
constant reflectance. It was standardized once against
freshly prepared MgO; the absolute reflectance of the
latter (R'm) obtained by averaging; the values given in
the literature is reproduced in Fi2ure6. For waveriumbers
above 30000 cm-1, for which no values are given i n the
literature, we arbitrarily assumed :I h e a r decrease from
0.950 a t 30000 cm-1 to 0.800 at 45000 c m 1.~ The latter
value was chosen such that R', did not exceed unity
for the materials which reflect best in the ultraviolet
region of the spectrum (c'.g. MgS04). The values measured against the Suprasil standard were then converted
into absolute diffuse reflectances by calculation on the
basis of the R', values of freshly prepared MgO.
T h e materials t o be measured were generally dehydrated for
two hours a t t h e temperatures given in Fig. 6, cooled over
PzO5 in a desiccator, triturated for two hours in a micro
agate ball mill (capacity 50 nil) with a grinding ball o f hard
metal (tungsten carbide + cobalt) [21a], and then measurcd
immediately. We always employed t h e purest materials
commercially available. T h e grain-size was determined by
means of a measuring microscope. Finer pulveriration by
[21] cf.,for example, F. Benfbrd et at., J. opt. Soc. America 38,
445 (1948); W. E. K . Middletun and C. L . Sanders, ibid. 4 / , 419
(1951); C . L. Sunders and W. E. K . Middleton, ibid. 43, 58 (1953);
J . T . Gier et al., ibid. 44, 558 (1954); P. A . T d k x and J . R . Wrrldrun, ibid. 45, 19 (1955).
[ * ] Produced by Heraeus Quarzschmelze, Hanau (Germany).
[21 a] Mfd. by Alfred Fritsch, Idar-Oberstein (Germany).

Angew. Chem. internat. Edit.

1 Vol. 2 (1963) 1 No. 7

i j l l ~ ~ . O ~ , i ?1:11"1.
p
L J - l - -

i o o o 6000
A! 8
U-E!

8000

-~

moon

~J

7001in

I+ *:

,--

....

3OOOO

40000

1+6000

Fig. 6. Dependence o n wavelength 01 the a b s o l u t e diHuse reflectance of

several white s t a n d a r d s (maximum rrain-sizr i n ir) referred t o t h a t of
freshly prepared M y 0 (uppermosf curve). T h e temperatures given a r e
drying temperatures.
[ * ] Silica gel.

[**I

Grain-size n o t measured.

Ordinate: Absolute diffuse reflectan<e R

Abscissa: W a v e n u m b e r [cni '1.

The curves for the most important materials are presented in Figure 6. The dat;i are absolute values calculated by means of Equition (4). The experimental error
is I 2

'x.

KCI, KBr, and CaFz behave i n a manner similar t o NaCI.

K I frequently contains traces of elementary iodine, a n d must
therefore always be rccrystalli/cd with complete exclusion of
air. T h e reflectance o f freshly precipitated B a s 0 4 in t h e
short - wave ultraviolet region is apprcciably better than
that shown in Figure 6. T h e low reflectance of silica gel and
A1203 above 30000 cm I is reinarkable and is probably d u e
to impurities, since Aerosil ['"I, for example, possesses excellent reflectance. Aerosil is ohlainable in various grain-sizes,
and thus may well serve as a standard for certain purposes
[22]. Since t h e curves obtaincd a r e rcfexed to t h e measured
absolute reflectance of freshly prepared MgO, they may, in
some measure, deviate from the true curves in t h e shortwave ultraviolet region.

[*I Aerosil'"

chemically purc silica gel. Supplier: Degussa,

Rhcinfeldcn (Germany).
[ 2 2 ] Cf. G. Kurtiirn and H . Ko//i,f., Bcr. Bunscngcs. physik. Chem.

67, 6 1 (1963).

331

Following these measurenients, we investigated how

thick the sample Must be in order to satisfy the condition
of infinite layer thickness which entered into the derivation of Equation ( I ) . This can be tested easily by
measuring the reflectance R of the layer of desired thickness of the sample in question with a strongly absorbing
background (soot, black velvet) and comparing it to
that of the same powder with a non-absorbing background (e.g. metallic mirror). I r R is appreciably different from unity, the layer still transmits radiation. This
test was carried out with Aerosil powders of various
grain-size, which had a rather narrow grain-size distribution [23]. The surface areas of three samples as determined by the BET method were 38, 106, and 376 mz/g.
Since Aerosils have no internal surfaces (pores, e f c . ) ,
the mean grain-size is inversely proportional to the surface area determined by the BET method. A layer thickness of I cm was chosen. The measurements established
that the samples of coarsest grain (surface 38 mz/g, grainsize 500-2000 A) are opaque up to a wavelength of
2.4 p, while finer samples with grain-sizes of 140 to
560 8, and 50 to 200 A begin to transmit at 1 and 0.7 p,
respectively. At 2.4 p, R is only 92 % and 70 %, respectively. This corresponds to the well known greater penetrability of long-wave radiation through scattering
media (infrared photography). Thus, for measurements
in the short-wave infrared region and for very fine-grain
powders, the layer thickness must be increased to 1 cm
or more if the Kubelka-Munk function is to be applied.

6. The Scattering Coefficient

were rneisured. From the data, the mean grain-size

was calculated using Equation (5) [26]:

[*I

where ni = t h e number and d i the diameter of the

particles. The following are the mean grain-sizes obtained for the four samples:
Sample I :

l/~

Sample3:

4 . 2 b ; Sample 2 :

1 . 5 1 ~ ;S a m p l e 4 :

4-

dz = 2.41-L;

l/$

1.1 b.

Figure 6a shows the distribution of grain-size of the

most finely divided sample 4. The other more coarse
samples had similar distribution curves with maxima
being lower and broader.

Fig. 6a. Distribution of grain-size of triturated didymium-glass filter

ds

= I . I w). Below 1 : L the microscopic determination of

(sample 4,
size-distribution is n o longer possible.

Ordinate: (Ni/N).lOO [ 73.

In the derivation of the Kubelka-Munk function [Equation (I)], monochromatic radiation is assumed. The
theory therefore says nothing concerning the dependence
on wavelength of the scattering coefficient s defined by
Equation (I). Furthermore, since we are dealing with
multiple scattering from closely packed particles of different forms and sizes, no theoretical predictions concerning this dependence are possible. The theory of Mie
[24] concerning the scattering of a planar wave from isotropic spheres at great distances apart is restricted to
single scattering and cannot be applied directly to the
present problem [25]. However, the wavelength dependence can be estimated experimentally from the KubelkaMunk function, if the related values of k are available
from transmission measurements: Here it is essential
that the spectra be not altered by interactions due to
solvation or adsorption. This is for instance the case for
the glass filter BG 36 mentioned above. In order to obtain the dependence of the scattering coefficient on
mean grain-size simultaneously, we prepared four samples by varying the duration of triturating for 1, 2, 4, and
14 hours, respectively. These were then placed in the
thinnest possible layers under the measuring microscope,
photographed, and at least 500 particles in each sample
[23] E. Wagner and H. Briinner, Angew. Chem. 72, 744 (1960).
[24] G. Mie, Ann. Physik 25, 377 (1908).
[25] In this connection, cf. W . Jaenicke, Z . Elektrochem., Ber.
Bunsenges. physik. Chem. 60, 163 (1956).

338

Abscissa:

ddz[wl.

The reflectance spectra of the four samples, measured

against the same standard (MgO) under identical conditions, are shown in Figure I.

-05

-151

L
Fig. 7. Effect of grain-size (root-mean-souare grain-diameter) on the
reflectance spectrum of a pulverized didymium-glass filter, measured
against MgO.
Ordinate: log F(R,).
Abscissa: Wave number [cm 1.

[*I For simplicity, the expression mean grain-size will be used

instead of the more cumbersome but correct expression rootmean-square grain-diameter.
[26] Cf. G. Kortiim and P. Hrrug. Z. Naturforsch. 8 a , 372 (1953).
Angew. Chem. interntct. Edit.

1 V d . 2 (1963)

No. 7

The spectral bandwidth was 18 cm-1 for the transmission

measurements; by amplifying the photocurrent, this same
bandwidth was maintained i n the reflectance measurements,
and, between 14000 and 28000 cm-1, it did not fluctuate.
The measurements were made with a Zeiss P M Q I1 spectrophotometer equipped with an RA 2 reflection attachment [27].
The scattering coefficient s calculated from these reflectance readings and the k-values obtained from transmission measurements was found to be inversely proportional to the mean grain-size (Fig. X), confirming
previous observations [26].

mean error of the data is quitc naturally much greater.

This dependence of the scattering coefficient on wavelength evidently implies a distortion of the short-wavelength spectra obtained from the Kubelka-Munk function on assumption of constant s ; however, this distortion is not very great, as can be seen from Fig. 1 .

7. Effect of Moisture on Reflectance Spectra

As is evident from Figure 4, the reflectance spectrum of

600

300

200

100

05

10

Fig. 8. Dependence of the scattering coefficient on grain-size, measured

o n the four samples described in Figure 7.
Ordinate: Scattering coefficient, s [cm-11.
Abscissa: 1/d/a
[d?
in' w ] .

an adsorbed material (e.g. anthracene on NaCL) is

strongly altered by moisture, not only in its vibrational
structure, but also in intensity. In the example illustrated in Figure 4, this is due t o displacement of the adsorbate from the surface by water molecules. This phenomenon was also observed i n other cases [ l l , 121; dipole forces and hydrogen bonds are probably involved
here. However, even when such displacements are excluded o r are very unlikely, it is frequently noted that on
admission of moisture, the samples appear darker ; this
is also true with pure substances. Since it was to be expected that such effects should depend on whether the
material is sparingly o r highly soluble in water, we carried out several experiments with K2Cr04 and Cr203 as
samples and NaCl and CaF2 as standards.
The samples and standards werc carefully dried and measured with complete exclusion of air. The quartz covering
plate was then carefully withdrawn, the sample exposed to an
atmosphere of constant humidity, covered again after a while,

The dependence on wavelength of the scattering coefficient s is shown in Figure 9 for various mean grain-sizes.
It can be seen that, for coarse-grain powders, s is practically independent of wavelength in the visible region,
I

60C

50C

40C

15000

20000

25000

30000

300
Fig. 10. Reflectance spectra of C r ? 0 3 measured relative to NaC1.
Curve 1 : anhydrous material; Curve 2 : CrzO3, exposed for 24 hours to
an atmosphere saturated with water wpor at room temperature.

200

Ordinate: log F(R,).

Abscissa: Wave number [cm- 'I.

1rJc

15000

20000

25000

Fig. 9. Dependence of the scattering coefficient on wavelength, for

various grain-sizes. Measured o n the four samples described in Figure 7.
Ordinate: Scattering coefficient, s [cm-11.
Abscissa: Wave number [cm- 'I.

as was provisionally assumed above. With the finer powders, a distinct increase in the scattering coefficient is
observed at higher wave numbers, although here the
[27] For further information, see G. Korfiimt Kolorimetrie,
Photometrie und Spektrometrie. 4th Edit., Springer-Verlag,
Heidelberg 1962.

Angew. Chem. internat. Edit. / Vol. 2 (1963) [ No. 7

and measured once more. It wa\ found that the spectrum of

pure Cr203, measured against dry NaCl as standard, remains
virtually unchanged when the sample is exposed for 24 hours
to an atmosphere saturated with water vapor at room temperature (Fig. 10). On the othcr hand, the absorption of pure
KzCr04 is greatly increased under the same conditions
(Fig. 11).
When mixed with the sparingly water - soluble standard
CaF2 and subjected to prolonged treatment with watersaturated air Crz03 shows a detectable change in its
spectrum. This can be practically reversed by storing the
sample over P205 in VLICUO (Fit:. 12). The mixture K2Cr04/
CaF2 also shows an increase in absorption if the dry sample
is exposed to an atmosphere containing water vapor; this
change, however, is irreversible.

339

Fig. I I .

Reflectance spectra of K z C r 0 4 measured relative t o NaCI.

Curve 1 : anhydrous material; Curve 2 : KzCr04, exposed for 24 hours

t o a n atmosphere saturated with water vapor at room temperature.
Ordinate: log F(R,).
Abscissa: Wave number [cm 1.

When mixed with the highly soluble standard NaCI, both

Cr203 and K2Cr04 show appreciable elevation of the bands
even after being kept for only an hour over 40% H2SO4
( ~ H ~
= o1 I m m Hg); this can be reversed only to a slight
extent by subsequent storage over P205 in vucuo.

With water-soluble powders such as K2Cr04 and

K2Cr04 + NaCl, additional recrystallization effects are
encountered which lead to an increase in grain-size and
thus to increased absorption as a result of the greater
depth of penetration of the radiation [4].These effects
naturally cannot be revcrsed by subsequent drying;
moreover, they should be much larger than the effects
mentioned before. Superficially, this recrystallization
can sometimes be recognized merely from the observation that a powder that was originally packed loosely
into the measuring cell contracts t o a hard cake under
the influence of moist air. Independent of this, the addition of water vapor can naturally lead to reactions at the
phase boundary (hydration of ions, shifts of chemisorption equilibria, erc.),which may lead to additional sharp
changes in the spectrum, as already indicated. In general, the use of water-insoluble standards is recommended, if this is not ruled out for other reasons. It is generally preferable to work with complete exclusion of
moisture; this, however, calls for appreciably greater
experimental expenditure.

8. Details of Measuring Techniques

,o-o-o

-101

1500U

20000

25000

Fig. 12. Reflectance spectra of a mixture of C r 2 0 3 a n d CaFz

( c c r , 0 3 = 0.902 moleil) measured relative t o CaF2. Curve 1: anhydrous
material; Curve 2 : after exposure f o r 72 hours t o air saturated with
water vapor; Curve 3 : After 7 days over P 2 0 5 in vacuo.

Ordinate: log F(Rm,).

Abscissa: Wave number [ c n - I.

These observations can be interpreted as follows. If the

powders under examination are difficultly soluble in
water, as Cr2O3 and CaF2 are, the only effect of the
water is evidently condensation in the pores of the powder and partial displacement of the air present there.
Since the scattering coefficient depends on the ratio of
the refractive indices of the sample and the surrounding
medium, and since this ratio is decreased by the penetration of the water, the scattering coefficient is decreased; thus, according to Equation (l), F(R,) is increased. This effect must be largely reversible, as is in
fact observed with pure 0 2 0 3 and Cr2O3 + CaF2. The
observation that the change in the spectrum cannot be
completely reversed may be due t o the fact that the
condensed water is extremely difficult t o remove from
the finest pores.

340

Commercially available spectrophotometers are generally provided with a special attachment for making
diffuse reflection measurements [27],either for direct or
for diffuse irradiation over a n integrating sphere. In the
latter case, the conditions of irradiation must be the
same for both sample and standard, since otherwise a
spherical error is introduced, which can become very
appreciable [27].
The essential difficulty involved in the measurements consists
in the preparation of reproducible samples with uniformly
flat surfaces and identical packing densities. To achieve this,
the powder to be measured is loosely heaped into the sample
cell and then compressed with a polished, carefully dried
metal stamp that fits exactly into the cell. Alternatively, the
powder can also be covered with a piece of glazed paper and
rolled flat with a glass rod.
The preparation of absolutely anhydrous samples has already
been described in detuil [I I]. It is an elaborate process that
must be carried out with extreme care, since moisture can
strongly affect the results of the measurements, as noted
earlier. If we are dealing with comparative or quantitative
photometric measurements, the process of triturating in a
ball mill that serves to dilute the samples with the white
standard should also be standardized, if possible; porcelain
mortars of 250 nil content with two large or six small porcelain balls have been found suitable for this purpose. If the
material to be ground i s hard, agate mortars with agate or
hard-metal balls (tungsten carbide + cobalt) are to be preferred. Within a given series of measurements, the same kind of
mortars with the same balls should always be employed if
the measurements are to be quantitatively reproducible. The
grain-size of the ground material depends essentially on the
kind and number of grinding halls. Even the number of
revolutions of t h e ball mill should be kept carefully constant
within a given series of measurements; the same holds true
for the duration of triturating. The standard must be prepared
and handled i n the same manner as the sample. The duration
of grinding differs from case to case, depending on the
hardness and the desired purpose of the ground material.
To obtain uniform adsorption of the material under investigation on the surface of the standard, grinding should
Angew. Chem. internut. Edit. 1 Vol. 2 (1963) I No. 7

be continued for at least six hours, preferably overnight. In

order to achieve the finest possible grain, 12-14 hours are
required; further grinding is usually of no advantage, since the
finest grains ultimately recoalesce, and an equilibrium in
respect to grain-size and grain-size distribution is attained.
During long grinding periods, slight abrasion of the mortars
and balls cannot be avoided. Systematic investigation has
shown that this abrasion does not affect the measurements
within the accuracy of the method; the F(R,) values remain
constant to within I -2'x if the grinding period is prolonged
beyond the time required to attain equilibrium.
I n choosing the standard for diluting the sample, care must
be taken that the latter does not react with the standard.
Thus, acids adsorbed on MgO or carbonates yielded the
spectrum of the corresponding salts; conversely, any salts
adsorbed on acidic A1203showed the superimposed spectrum
of salt and acid. The presence of moisture has various
effects, since hydration of ions on the surface can also
occasion additional changes in the spectrum.
Certain standards cannot be used in a highly dried state for
the adsorption of organic materials. Thus, for example, even
after drying for only an hour at 600 "C, silica gel becomes so
active that, during the grinding process, some organic
materials, especially aromatic materials, are largely de0
\

-0-SiO

-0

organic materials can react [28.29]. The action of the mixtures of Si02 and A1203 used industrially as cracking
catalysts can perhaps be understood in this way (cf. Fig. 13).

I
30000

LOO00

;A 290.131

Fig. 13. Reflectance spectra of anthrdcene adsorbed on SiOz (x =

5.OXlO-4). Curve 1 : SiOa not pre-dried; Curve 2: S i 0 2 dried for 1 hour
at 600C.
Ordinate: log F(R,).
Abscissa: Wave number [cnr '1

/O+ E!O-.Sj.-O-

20000

0 --

Received, October ZOth, 1962

I A 290/98 IEI
German version: Angew. Chem. 75, 653 (1963).

composed. Apparently, the Si - 0 - S i bridges formed during

the heating of the gel are mechanically destroyed in the
process of triturating and form active centers with which

[28] W. A . Weyl, Research 3, 230 ( I 950).

[29] R . E. Benson and J. E. Cas/le,I . physic. Chem. 62,840 (1958).

The Biosynthesis of Alkaloids I [*]

BY PROFESSOR DR. K. MOTHES AND DOZENT DR. H. R. SCHUTTE
DEUTSCHE AKADEMIE DER WISSENSCHAFTEN ZU BERLIN, INSTITUT FOR BIOCHEMIE
DER PFLANZEN, HALLE/SAALE (GERMANY)
Dedicated to Professor Dr. Adolf Birtenandt on thc occusiorr of his 60th birthday

1. lntroduction
2. Tropan alkaloids and related compounds
a) Hyoscyamine
b) Hygrine, cuscohygrine, and cocaine
c) Stachydrine
d) Hyoscine
e) Tropic acid
f ) Participation of Cl-metabolism
g) Site of biosynthesis in plants
3. Pyridine and piperidine alkaloids
a) The pyrrolidine ring of nicotine
b) The piperidine ring of anabasine

1. Introduction
The rapid development of present concepts concerning
the mechanisms of biosynthesis of secondary substances
in plants will soon cause a large number of classical
[*I Part I1 of this review will appear in a subsequent issue of
this journal.
Angew. Chern. internat. Edit.

Vol. 2 (1963) No. 7

1
I

c) The pyridine ring of nicotine and anabasine

d) The methyl group of nicoline
e) Ricinine
f ) Isopelletierine, methylisopelletierine and pseudopelletierine
g) Coniine
h) Lobeline, lobelanidine, and lobelanine

4. Pyrrolizidine and quinolizidine alkaloids

a)
b)
c)
d)

Pyrrolizidine alkaloids of the laburnine type

Retronecine and platynecine bases
Necinic acids
Lupin alkaloids

hypotheses, no matter how ingenious some of these undoubtedly are, to pass into oblivion, because the accumulating number of experimental facts permit definite conclusions to be drawn. These i n turn may lead on the one
hand to entirely new theories concerning the mode of
action of organisms, and on the other they may inosculate and confirm existing ideas established iong before
the development of modern biochemistry.

34 1