Вы находитесь на странице: 1из 11

Fuel 85 (2006) 1–11

www.fuelfirst.com

Molecular size and weight of asphaltene and asphaltene solubility


fractions from coals, crude oils and bitumen
Sophie Badrea, Cristiane Carla Goncalvesa, Koyo Norinagab,
Gale Gustavsona, Oliver. C. Mullinsa,*
a
Schlumberger-Doll Research, OFS, Old quarry Road, Ridgefield, CT 06877, USA
b
Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Katahira 2-1-1, Sendai 980-8577, Japan
Received 4 May 2005; received in revised form 31 May 2005; accepted 31 May 2005
Available online 28 June 2005

Abstract
The molecular weight of asphaltenes has been a controversy for several decades. In recent years, several techniques have converged on the
size of the fused ring system; indicating that chromophores in virgin crude oil asphaltenes typically have 4–10 fused rings. Consequently, the
molecular weight debate is equivalent to determining whether asphaltenes are monomeric (one fused-ring system per molecule) or whether
they are polymeric. Time-resolved fluorescence depolarization (FD) is employed here to interrogate the absolute size of asphaltene molecules
and to determine the relation of the size of the fused ring system to that of the corresponding molecule. Coal, petroleum and bitumen
asphaltenes are compared. Molecular size of coal asphaltenes obtained here by FD-determined rotational diffusion match closely with
Taylor-dispersion-derived translational diffusion measurements with UV absorption [1]. Coal asphaltenes are smaller than petroleum
asphaltenes. N-methyl pyrrolidinone (NMP) soluble and insoluble fractions are examined. NMP soluble and insoluble fractions of
asphaltenes are monomeric. It is suggested that the ‘giant’ asphaltene molecules reported from SEC studies using NMP as the eluting solvent
may actually be the expected flocs of asphaltene which are not soluble in NMP. Data is presented that intramolecular electronic relaxation in
asphaltenes does not perturb FD results.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Asphaltene; Molecular weight; bitumen

1. Introduction changes, viscosity, and interfacial properties of crude oils


are strongly affected by asphaltenes. The lack of proper
Asphaltenes are the most complex component of crude understanding of asphaltenes has limited the predictive
oil [2–5]. The term asphaltene is defined operationally; capabilities of petroleum science. Francis Crick advises “if
asphaltenes are defined to be toluene soluble and n-heptane you want to understand function, study structure” [6].
insoluble. Other similar asphaltene definitions exist impact- Correspondingly, if the structure of asphaltenes is unknown
ing derived properties somewhat. This definition is actually to within one or several orders of magnitude, then predictive
quite useful and not arbitrary in that it captures the most science is precluded and phenomenological approaches
aromatic component of crude oil. This definition does prevail. Such is the case particularly for properties of crude
correspond to a fraction of coal, but asphaltene is not the oils where asphaltenes play a key role. The new field of
most aromatic component of coal. Asphaltenes are a friable, petroleomics necessarily requires resolution of the asphalt-
infusible solid microcolloidally suspended in crude oil. ene molecular weight debate. The premise of petroleomics
They are characterized by complex chemistry and their is the prediction of structure based on function. First, one
presence in crude oils impacts the oil properties. Phase must have proper asphaltene molecular structures.
Many bulk properties of asphaltenes are known well.
The saturate to aromatic carbon is known from 13C NMR
* Corresponding author. Tel.: C1 20 3431 5572; fax: C1 20 3438 3819. studies [7]. X-ray Raman spectroscopy (XRRS) has shown
E-mail address: omullins@ridgefield.oilfield.slb.com (O.C. Mullins). that the overall geometry of the ring systems is largely
0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. pericyclic [8] as expected from chemical stability
doi:10.1016/j.fuel.2005.05.021 considerations. The sulfur [9] and nitrogen [10] chemical
2 S. Badre et al. / Fuel 85 (2006) 1–11

moieties are known from X-ray absorption near edge alkane chains increasing solubility. Fourth, the trend towards
structure (XANES). Infrared spectroscopy (IR) and NMR smaller asphaltenes in mildly thermally cracked feedstock
has shown that (petroleum) asphaltene hydrogen resides results from the loss of alkane groups reducing steric
mostly on saturated carbon [7]. The size of the aromatic disruption thereby mandating smaller ring systems for the
ring systems has been imaged directly by scanning same solubility class [22]. Fifth, the FD analysis of solubility
tunneling microscopy (STM) [11]. High-resolution, trans- subfractions of asphaltenes shows that asphaltene properties
mission electron microscopy (HRTEM) studies yield the grade continuously from one subclass to the next; there is no
same size of the rings systems as STM; although of lower large discontinuity of asphaltene properties [21]. Monomeric
resolution, HRTEM can naturally sample a much larger asphaltenes, but not polymeric asphaltenes are consistent
sample size [12]. STM [11] and HRTEM [12] both with these extensive FD results. Furthermore, Taylor-
indicate that there are about 6–7 fused rings in petroleum dispersion measurements of translational diffusion [1]
asphaltenes on average. Optical absorption and fluor- performed on coal asphaltenes match the FD-derived
escence spectroscopy [13–15], when coupled with mol- rotational diffusion measurements. Agreement between
ecular orbital calculations [16] provide guidance as to the two very different techniques with very different detection
ring size within the standard quantum-particle-in-a-box methods confirms the validity of both sets of studies.
precepts. At present, there is little controversy surrounding Of course, mass spectroscopy (MS) can also be used
these bulk parameters. All techniques that speak to ring to obtain molecular weight. In an early study, field-
size support a conclusion of 4–10 rings in typical virgin ionization mass spectroscopy (FIMS) was used and an
crude oil asphaltene molecules. asphaltene molecular weight of w700 g/mol was obtained
The one area of asphaltene molecular structure that has [23]. Laser-desorption mass spectroscopy (LDMS) has
been the most controversial is that of asphaltene molecular resulted in various and inconsistent asphaltene molecular
weight. Results in the literature vary by significantly more weights [1,24–26]. One LDMS study obtained w500 g/mol
than one order of magnitude for asphaltene molecular for asphaltenes [24]. This study noted the extreme care that
weight. Because there is general agreement about the is required for baseline subtraction. Electrospray-ionization
constituent molecular fragments of asphaltenes (e.g. size of Fourier-transform ion-cyclotron resonance mass spec-
fused ring systems, size of alkane substituents, etc.), the troscopy (ESI-FT-ICR-MS) has been used on heavy oils
debate regarding asphaltene molecular weight reduces to and similar materials finding molecular mass consistent
whether asphaltenes are monomeric or polymeric. For with the monomeric description of asphaltenes [27,28].
instance, one structure proposed for asphaltenes from Atmospheric pressure chemical ionization mass spec-
athabasca bitumen showed a cross-linked network of troscopy (APCI-MS) has been used on asphaltenes yielding
aromatic ring systems—essentially a polymer, although molecular weights w700 g/mol [29]. The only mass
the authors of the report expressed significant uncertainty spectral technique that seems to provide self-contradictory
regarding their determination of the proper molecular results is LDMS; ironically, this technique remains popular
weight [17]. (We thank the authors for pointing this out.) for determination of asphaltene molecular weight in certain
The chemical and physical properties of monomers differ laboratories.
enormously from the corresponding polymers; it is essential One sustaining source for the molecular weight
for the advancement of petroleum science to resolve the controversy is that asphaltenes are known to aggregate at
debate on the molecular weight of asphaltenes. low concentrations. Surface tension measurements have
In a series of publications, our laboratory has shown that shown that asphaltenes exhibit a critical nanoaggregate
FD strongly supports the monomeric description of concentration (CNAC) at w400 mg/l in pyridine [30]. (A
asphaltenes. The FD technique works by measuring the terminology change is being attempted per reviewers’
rate of rotational diffusion of molecules in dilute solution. request; the term nanoaggregate is replacing micelle for
First, FD has shown that asphaltene molecules undergo rapid asphaltenes.) Microcalorimetry also shows asphaltene
rotational diffusion, approximately the same rate as aggregation (likely not primary aggregation) but at much
molecules in w750 g/mol range [18–22]. Second, the large higher concentrations and in the presence of some dissolved
wavelength dependence of the rotational diffusion times water in the toluene [31]. Recent high-Q, ultrasonic spectral
shows that for each asphaltene molecule, a single chromo- studies of asphaltenes clearly show asphaltene CNAC in the
phore is a large part of the molecule, thus, there must be one range of 50–150 mg/l [32,33]. Furthermore, it has been
or sometimes two chromophores (fused ring system) per shown that removing the more soluble component of a
molecule [18–22]. Third, FD has shown that large whole asphaltene leads to a remaining fraction that exhibits
chromophores require substantial alkane substitution while a much lower solubility [34]. The likely explanation is that
small chromophores mandate minimal alkane substitution above CNAC, a nanocolloidal dispersion is stable for the
[20]. This is interpretable in terms counterbalancing forces whole asphaltene—appearing to be a solution, but for the
controlling molecular solubility. In asphaltenes, there is a more monodisperse low solubility asphaltene component,
balance of opposing effects; p-bond stacking (via Van der initial aggregation above the CNAC, does not result in a
Waal’s force) reducing solubility vs steric repulsion of stable nanocolloid but proceeds to floc formation.
S. Badre et al. / Fuel 85 (2006) 1–11 3

Any techniques that report molecular weight by making asphaltene forms floc in NMP which then went rapidly
measurements on solutions such as vapor pressure osmo- through the column. This explanation should be tested by
metry (VPO) or Size Exclusion Chromatography (SEC) using an SEC solvent that asphaltenes are known to dissolve
need to account for aggregation. The large molecular in—say toluene—and comparing with the study cited using
weights reported for asphaltenes from these techniques are an SEC solvent that asphaltenes are known not to dissolve
most likely due in part to the measurement of aggregate in—NMP. The authors did note that this SEC peak for the
weight, not molecular weight. This problem is exacerbated very large asphaltene molecules was not obtained when
for asphaltene fractions whose preparation is based on low THF was used as the solvent (although other difficulties
solubility, such as for NMP-insoluble asphaltene fractions. were mentioned) [36].
Any description of asphaltenes must be consistent with Previously, a single coal asphaltene was compared to
the observation that properties of asphaltenes vs resins and several petroleum asphaltenes [20]. The principle finding
asphaltene subfractions vary continuously. The defining was that alkane chain length scaled with the size of the
characteristic of asphaltenes—solubility—necessarily asphaltene molecule and also with the asphaltene chromo-
grades continuously from resins—by definition. Further- phore (the aromatic ring system). The solubility classifi-
more, subfractions of asphaltenes obtained by changing cation of asphaltenes mandates a balance between
solvent ratios necessarily grade continuously in solubility. intermolecular attraction via van der Waals interaction vs
Asphaltene solubility subfractions have also been shown to steric hindrance disrupting binding. Direct observation of
grade continuously one into the next with respect to ring stacking via HRTEM supports the contention that
molecular size [21]. Asphaltene molecular properties have alkanes disrupt stacking [12]. It is conventional wisdom that
also been shown to grade continuously into resins, the next in general coal asphaltenes are smaller than petroleum
heaviest fraction of crude oil [19]. That is, there is no large asphaltenes; this must be tested.
discontinuity in asphaltene properties. One should not be In this report, we extend our FD results comparing
confused with the distinct rheological properties of several coal asphaltenes with petroleum asphaltenes; certain
asphaltenes and resins that most likely simply have to do systematics are obtained. FD interrogation of asphaltene
with intermolecular binding energy vs temperature. Con- solubility subfractions is extended to include N-methyl
sider the rheological difference of the oil n-hexadecane and pyrrolidinone (NMP) solubility as the solvent. We also
the waxy n-heptadecane, all at room temperature; their include Athabasca bitumen as the source material to test
chemistry is nearly identical. Furthermore, crude oils and systematics of asphaltene properties. The FD results
asphaltenes exhibit the Urbach tail phenomenon in the obtained here bolster conclusions drawn from previous FD
electronic absorption edge [15]—a result familiar in work.
condensed matter physics [35]. This observation indicates
large ring systems grow from small ring systems. Again, the
electronic absorption edge of crude oils grades continuously 2. Experimental section
into that of asphaltene; there is no large discontinuity.
Consequently, the view that asphaltenes are bimodal in The specifics of how we employ FD have been described
properties [26] is inconsistent with many observations and elsewhere [18–22]. Here, the fundamentals of FD are
with the solubility classification of asphaltenes. briefly described. FD employs a laser polarized in the lab
There is a study based on Size Exclusion Chromatog- frame to define a unique axis. The solvent used for FD was
raphy (SEC) that claims there is a large molecular weight toluene at room temperature unless otherwise specifically
fraction of asphaltenes [36]. In fact, this study finds that stated. Absorption of polarized photons gives an ensemble
asphaltenes and other comparable carbonaceous species are of polarized asphaltene molecules. Rotational diffusion
bimodal in their molecular weight distribution. In this SEC reduces the extent of molecular alignment; after a
study, the two very different ‘molecular’ weight components sufficiently long period of time, no polarization remains.
are separated rather cleanly by many SEC peak widths [36]. The time dependence of the polarization is measured using
We find such description of asphaltene molecular weight to a fast pulse on a photomultiplier tube (PMT) attached to the
be highly improbable and view one should search for emission monochromator of a fluorescence system. Our
alternate explanations. For some reason, this study measurements were made using a PTI time-resolved
employed NMP as the eluting solvent. As is established fluorescence spectrometer. The system contains a C-72
here and elsewhere [37], asphaltenes are not totally soluble fluorescence lifetime spectrometer with a PTI A-720
in NMP. There is an NMP-soluble fraction and an NMP- steady-state fluorescence spectrometer. The C-72 system
insoluble fraction in asphaltene. We view that the far more employs a nitrogen-pumped dye laser to excite the
likely explanation for a strongly bimodal distribution of fluorescence. Various Exciton dyes were used to cover
mass obtained in these NMP SEC studies is due to partial our excitation range. The dye laser output was sent through
dissolution of asphaltene in the NMP. Perhaps the a Glan–Thompson polarizer of known orientation and then
description is: the NMP-soluble component eluted slowly used to excite a dilute solution of asphaltenes. The
in accord with its low molecular weight. The NMP-insoluble fluorescence emission at 908 was imaged onto an entrance
4 S. Badre et al. / Fuel 85 (2006) 1–11

slit of the emission monochromator fitted with a second alone. We have used two candidate references, deionized
Glan–Thompson polarizer. The corresponding photomulti- water and ludox (colloidal suspension of latex spheres).
plier was pulsed so as to acquire signal only over a short They give slightly different results for several wavelengths;
time window. With the polarization decay from a standard the ludox scatterer often shows slightly better performance
to characterize the laser pulse, deconvolution software from at several wavelengths, but much worse results at one
PTI was employed to obtain the rotational correlation. wavelength. Consequently, we have chosen to keep water as
We utilize several wavelengths in order to interrogate our reference because the dispersion is a very important part
different asphaltene fractions in a single solution. A red shift of the results. The standard is not that important in our
of 40 nm (increase in emission wavelength over excitation experiments because (1) this choice has relatively small
wavelength) is typically employed. This is essentially the effect on absolute rotational correlation times; and (2) the
Stokes shift—the energy difference between excitation and depolarization curves for the model compound used for
emission between two electronic states of a molecule in comparison are influenced in exactly the same way as the
solution. A larger red shift can lead to electronic thus, corresponding curves for asphaltenes—thus, the differential
polarization relaxation in the excited state manifold. A comparison does not change at all.
smaller red shift can give rise to improper detection of The asphaltene samples were prepared using 40cc
scattered excitation light in the emission channel altering n-heptane added per gram of crude oil (or bitumen). After
the polarization curve. In these experiments, the asphaltenes 24 h with occasional stirring, the resulting asphaltene
precipitate was washed with n-heptane until it was colorless.
are typically in a concentration range of 10 mg/l. Using
The resulting asphaltene was dissolved and the same
750 g/mol (our result) for asphaltene molecular weight, this
precipitation process was followed a second time. We
gives a solution of w13 mM. Recent ultrasonic experiments
refer to these samples as n-heptane asphaltenes. We have
show that the CNAC of asphaltenes is significantly higher in
performed various FD and sulfur XANES studies on the first
concentration than this (CNACw100 mg/l) [32,33].
and second precipitations but we have never observed any
Fig. 1 shows a typical set of decay curves that are acquired
effect from this second precipitation process. Nevertheless,
in order to determine fluorescence depolarization. The four
we do so for completeness.
possible (908) combinations of the two polarizers, one on the The N-methyl pyrrolidinone (NMP) samples (all from
excitation side, the other on the emission side, are employed. UG8 asphaltene) were prepared by dissolving 25 mg of
The horizontal excitation, vertical emission (designated as n-heptane UG8 asphaltenes in 50 ml of toluene. 100 ml of
H–V) and the H–H configurations produce curves that NMP was then added. After 18 h, the solution was filtered.
overlay. The initial polarization that decays away with time is The precipitated solid was then filtered. After drying, this
shown by the initial separation in the V–H and V–V curves. solid was dissolved in minimum toluene and precipitated
This polarization decay and the fitted curve are also shown in with excess NMP. The resulting solid was labeled as NMP
Fig. 1. insoluble. The solution from this second process was labeled
To measure the decay curve, it is always necessary to as NMP soluble. The soluble portion from the first solution
measure the response from a system that yields scattering is comparable in spectra and in tr, consequently, we list
parameters and spectra for the second solution only.
100 25
VV
Difference
Coal asphaltenes were prepared from Illinois (IL) and
HV Least SquareFit Pocahantas (POC)coals (Argonne Premium Coal Samples,
80 VH 20
http://www.anl.gov/PCS/) and liquefaction products from
HH
Banko coal. Extraction and fractionation procedures were
followed by Iino et al. [38]. Dried Illinois and Pocahantas
coals were extracted with pyridine. The pyridine soluble
Counts (/103)

60 15
portions were subsequently fractionated into toluene
solubles and insolubles. The toluene solubles were further
40 10 fractionated into n-hexane solubles and insolubles (asphalt-
ene). The yields of asphaltenes were 8 and !1 wt% (daf)
from IL and POC coals, respectively. Banko (BA) coal was
20 5 liquefied with tetralin at 400 8C, under 9 MPa hydrogen
pressure for 30 min [39]. The liquefaction products were
recovered with THF. The THF soluble portions were
0 0
64 66 68 70 72 74 76 extracted with toluene and further fractionated with
Time (ns) n-hexane to get asphaltene (toluene soluble/n-hexane
insoluble). The asphaltene yield from Banko coal was
Fig. 1. The raw data curves utilized in FD. By measuring the time
dependence of the four 908 linear polarization combinations for incident 30 wt% (daf).
and fluorescent photons, one can measure the rate of polarization decay Table 1 lists the elemental weight percent for the various
(also shown) due to molecular rotation. asphaltenes and Athabasca bitumen. These data were
S. Badre et al. / Fuel 85 (2006) 1–11 5

Table 1
The elemental mass fractions various petroleum asphaltenes, coal asphaltenes and a bitumen

Asph sample C H N O S C:H


UG8 81.07 7.11 1.02 1.6 8.94 1:1.045
Athabasca 77.03 8.01 1.27 3.00 8.18 1:1.239
Ven20 84.75 7.81 1.75 1.72 4.57 1:1.098
Denver Citya 72.63 6.58 0.75 3.19 5.32 1:1.1080
Iino 90.35 5.53 2.23 1.9 !0.23 1:0.729
Ill #8 80.98 6.52 1.33 8.80 0.39 1:0.960
BA 79.23 6.19 1.49 1.36 0.39 1:0.931
Athasaca Bitumenb 83.72 10.39 !0.5 15.03 4.73 1:1.479
a
Pipeline deposit.
b
Bitumen sample, not an asphaltene.

obtained from Galbraith Laboratories, Knoxville, TN. The experiment, can now be written as:
petroleum asphaltenes; UG8, Athabasca, Ven20 and Denver Vh
City have H–C atomic ratios exceeding one, while the coal tr Z (6)
kT
asphaltenes; Ill#8, BA and Iino have H–C ratios less than
one. The Iino coal sample (from Tanito Harum coal) has an It is straightforward to check this analysis by measuring
H–C ratio much less than one. known dye compounds and comparing the resulting
Some fluorescence lifetime data shown herein which will molecular size with calculated results. These known
be identified was acquired on Beamline U9B at the National references allow the molecular weight to molecular size
Synchrotron Light Source at Brookhaven National Labs. correlation to be made.
Finally, the impact on the rotational correlation time of
having asymmetric tops has been discussed in detail. (cf
Ref. [18, 21] and references therein). From Eq. (6) one can
3. Data interpretation obtain a molecular radius from the rotational correlation
time assuming the molecule is a sphere. The derived radius
As described elsewhere [18 and references therein], the increases by 12.4% for the long axis if assumes an oblate
following definitions are used spheroid shape with an aspect ratio of 2 [21]. We presume
asphaltenes are oblate spheroids with an aspect ratio of 2 for
TðtÞ Z Is ðtÞ K It ðtÞ (1) our analysis.

SðtÞ Z Is ðtÞ C 2It ðtÞ (2)


4. Results and discussion
and
4.1. Asphaltenes; coal vs petroleum
TðtÞ
rðtÞ Z (3)
SðtÞ Fig. 2 shows the FD data for several coal asphaltenes vs
where Ijj ðtÞ and It ðtÞ denote the detection of intensity of UG8 (petroleum) asphaltene; the coal asphaltenes exhibit
light linearly polarized parallel and perpendicular to the trends similar to the petroleum asphaltenes. The trs are not
polarization axis of the excitation beam and r(t) represents large when compared to those of model compounds and
the anisotropy of the fluorescence emission. For spherical increase as the emission wavelength increases. Coal
rotors, the following equation applies: asphaltenes have tr smaller than those for petroleum
asphaltenes meaning that at a given interrogation wave-
2 length, coal asphaltene molecules are smaller than those of
rðtÞ Z eK6Dt : (4)
5 petroleum asphaltenes. Furthermore, the coal fluorescence
spectra peak at shorter wavelengths than the petroleum
The impact of molecular asymmetry has been checked asphaltenes as seen in Fig. 3. Consequently, the coal
[18,21]; it is not large. For a sphere, the following equation centroid for molecular size is shifted to shorter wavelength
applies: and smaller tr compared to petroleum asphaltenes. (More
kT extensive fluorescence spectra of petroleum asphaltenes are
6D Z ; (5) published elsewhere [13]; UG8 asphaltene exhibits a typical
Vh
petroleum asphaltene fluorescence spectrum.) Conse-
where V is volume of the molecule, h is viscosity of the quently, coal asphaltenes are characterized by a much
solvent, and D is solute rotational diffusion coefficient. The smaller tr, roughly comparable to that of OEP which has a
decay time of the anisotropy, tr, the parameter of our molecular weight of 535 g/mol. It is possible the coal
6 S. Badre et al. / Fuel 85 (2006) 1–11

Asphaltene w0.7 nm. Fig. 2 confirms that the coal asphaltene molecules
0.8 are smaller. Fig. 3 shows that the coal asphaltenes have
UG8 CrudeOil
0.7 BA Coal shorter wavelength emission indicating that coal asphaltene
POC Coal ring systems are smaller than petroleum ring systems. Fig. 4
IL Coal
0.6 solar dye shows the coal asphaltenes are much lower in coloration
OEP (optical absorption in the visible) than petroleum asphalt-
− τr 0.5
enes, even though coal source materials for these
(ns)
0.4 asphaltenes are black. Thus, petroleum asphaltenes have
larger fused ring systems than coal asphaltenes. As shown
0.3 previously, coal asphaltenes lack alkane substitution
0.2 relative to petroleum asphaltenes [20]. The universal
asphaltene solubility definition mandates a balance of
0.1 attractive and repulsive molecular forces. Coal asphaltenes,
0
lacking alkanes, lack intermolecular repulsion. Conse-
400 450 500 550 600 quently, they must exhibit weaker intermolecular attractive
Emission Wavelength forces; this mandates smaller fused ring systems. The lack
Fig. 2. Rotational correlation times tr for several coal asphaltenes, one of long wavelength absorption in coal asphaltenes relative to
crude oil asphaltene and two known compounds. petroleum asphaltenes in Fig. 4 shows the lack of large
fused ring systems in coal asphaltenes relative to petroleum
asphaltenes are smaller than OEP; the limits of our time- asphaltenes.
resolved fluorescence depolarization system is approxi- The molecular size obtained here for coal asphaltenes
mately 100 ps, which is our measured correlation time for matches exactly the results obtained previously using a very
OEP. Thus, we lose resolving power for molecules smaller different technique. Taylor-dispersion measurements with
than OEP. The molecular size of OEP obtained by FD is UV absorption have been performed on dilute solutions of
very close to that obtained for OEP and other similar coal asphaltenes in a variety of solvents [1]. Taylor diffusion
porphyrins by a very different technique; perturbed angular determines the translational diffusion constant, FD deter-
correlation of gamma rays [40]. Thus, we believe we have mines the rotational diffusion constant. The Taylor diffusion
accurately measured the correlation time for OEP here. experiments used UV absorption for detection of asphalt-
Table 2 lists the parameters that are measured and enes; FD relies on fluorescence emission. The fact that these
derived for the coal asphaltenes, UG8 petroleum asphaltene very different techniques obtain the same centroid and range
and the model compounds. These values for both the coal of asphaltene molecular size is very encouraging. Table 1
and the crude oil asphaltene molecular size are consistent shows that coal asphaltene molecular size is w12 Å in
with corresponding HRTEM results [12]. Direct imaging of diameter (at the fluorescence maxima); Taylor-dispersion
the fused aromatic ring systems by HRTEM shows that the finds coal asphaltene molecular diameters to be w11 Å [1].
petroleum asphaltene ring systems are w1 nm in linear In addition, LDMS was used on the coal asphaltenes to
dimension while the coal asphaltene fused ring systems are obtain the molecular weight.
Petroleum asphaltenes have a correlation time compar-
able to the solar dye which has a molecular weight of
1
755 g/mol. These results support what was observed for one
coal asphaltene previously [20]. As shown previously, the
smaller fraction of alkane carbon in coal shown by 13C
Fluorescence Intensity (norm)

0.8
NMR yields less steric repulsion [20]. The smaller steric
repulsion must be balanced by smaller intermolecular
0.6 attraction to maintain constant solubility; thus, coal
aromatic ring systems are smaller. That is, alkane steric
UG8 repulsion balances p–p intermolecular attractive inter-
0.4
POC actions to maintain the same solubility—which is the
BA property that defines asphaltenes. At a given interrogation
0.2 wavelength in Fig. 2, the coal asphaltenes are seen to be
smaller than the petroleum asphaltenes. This is due in part to
the smaller alkane fraction of coal asphaltenes. The C:H
0 ratios for coal asphaltenes are much higher than for
350 400 450 500 550 600
petroleum asphaltenes (cf. Table 1) reflecting the lack of
Fluorescence Emission Wavelength (nm)
alkane substituents in coals.
Fig. 3. The fluorescence spectra of different asphaltenes, UG8 petroleum By spanning a large range in wavelength in Fig. 1, we can
asphaltene and POC and BA coal asphaltenes. interrogate different chromophores. Small chromophores
S. Badre et al. / Fuel 85 (2006) 1–11 7

Table 2
tr values and corresponding molecular diametersa or known compounds and for different coal and oil asphaltene samples at different wavelengths

l Emission OEP (ns) (Å) Solar Dye (ns) (Å) UG8 (ns) (Å) BA (ns) (Å) POC (ns) (Å) IL (ns) (Å)
410 0.18 ns 14.9 Å
450 0.088 ns 11.7 Å 0.32 ns 18.1 Å 0.13 ns 13.4 Å 0.11 ns 12.7 Å 0.092 ns 11.9 Å
480 0.4 ns 19.5 Å 0.24 ns 16.4 Å 0.21 ns 15.7 Å 0.165 ns 14.5 Å
520 0.56 ns 21.8 Å 0.36 ns 18.8 Å 0.35 ns 18.6 Å 0.22 ns 16.0 Å
570 0.7 ns 23.5 Å
530 0.48 ns 20.7 Å
a
Long axis assuming oblate spheroid, aspect ratioZ2.

are investigated with blue light and large chromophores 4.2. NMP
with red light. But Fig. 2 also shows that there is roughly an
order of magnitude increase of tr, thus large increase in There has been the suggestion that NMP insoluble
molecular size, in going from short wavelength to long asphaltene fractions are inaccessible to fluorescence
wavelength for the asphaltenes. This indicates that the small techniques [37]; and that possibly the NMP insolubles are
and large chromophores are not linked to each other. If they of a very different molecular weight than the NMP solubles.
were, then (1) large tr would result and (2) little dependence This suggestion would be that NMP mysteriously separates
of tr on wavelength would be observed. Thus, both the the presumed bimodal asphaltene molecular distribution
zeroth moment (small magnitude of tr) and first moment into low and high molecular weight fractions. This all seems
(large wavelength dependence of tr) in Fig. 2 indicate improbable to us; we have previously obtained solubility
asphaltenes are monomeric, not polymeric. subfractions of asphaltenes from various solvent systems
Comparison of Table 2 to previous work on the Iino coal and found no large mass fraction [21,22]. Only subtle
asphaltene sample (Tanito Harum) shows that the samples differences were found in molecular weight among the
that have smaller H–C ratios are characterized by smaller different solubility fractions independent of solvents used.
molecules. That is, the tr for the Iino coal asphaltene sample Again, all mass spectral techniques except laser desorption
[20] are less than for the coal asphaltenes presented here. ionization exhibit only small molecular masses for
Table 1 shows that this Iino coal sample has a much smaller asphaltenes [23,27–29]. And laser desorption techniques
H–C ratio (0.729) than the other coal asphaltenes measured in the literature are conflicting most likely due to the
here. And all coal asphaltenes we have measured have both enormous baseline subtraction issue [24–26]. In any event,
smaller tr and smaller H–C ratios than all petroleum we have obtained rotational correlation data for NMP
asphaltenes we have measured. Less alkane substituents solubility fractions.
produces smaller asphaltene molecules. Table 3 lists this data; molecular diameters for NMP
soluble and NMP insoluble fractions are listed; the
diameters for the NMP insoluble component is somewhat
2 1 bigger than the NMP soluble component but the difference
is not nearly an order of magnitude. Note that the tr of the
NMP soluble component is obtained in NMP solvent. We
0.8
have previously shown that, as expected, Eq. (6) properly
1.5 accounts for viscosity variations [20]. All measured tr are
all quite small and comparable indicating molecules in the
500–1000 g/mol range. Table 3 shows that, as expected,
Transmission

0.6
Absorption

1
Table 3
tr values for 410 nm excitation, 450 nm emission for UG8 asphaltene and
UG8 Petroleum 0.4 solubility fractions
illinois #8
Iino Coal
Sample tr (ns) Solvent vis- Diameter Diameter
0.5
cosity (cp) (sphere) (Å) (oblate
0.2 spheroidc
Toluene 0.32 0.59 16.1 18.1
soluble
0 NMP 0.65 1.67 14.4 16.2
0
300 400 500 600 700 800 900 1000 soluble
Wavelength (nm) NMP 0.47 0.59 18.3 20.6
insoluble
Fig. 4. The optical absorption and transmission spectra of various petroleum
a
and coal asphaltenes. Long axis, aspect ratioZ2.
8 S. Badre et al. / Fuel 85 (2006) 1–11

the whole asphaltene lies in the middle of the NMP soluble acquired in the same conditions. It is evident the NMP
and NMP insoluble fractions in terms of molecular size. insoluble fraction has a lower fluorescence intensity. This
The NMP insoluble asphaltenes have a much smaller issue needs to be addressed. Some reduction in quantum
fluorescence quantum yield in the concentration range yield is observed when isolating the less soluble asphaltene
where we perform our measurements. Fig. 5 shows the subfractions from toluene–n-heptane mixtures. The larger
fluorescence spectra of the whole asphaltene, the NMP aromatic ring systems are known to be of lower quantum
soluble fraction and the NMP insoluble fraction; all yield due to the Energy Gap Law [13,41]. This basic
photophysical phenomenon does not impede our measure-
ments for interrogating different asphaltene solubility
5
fractions [21,22]. In our experiments, we utilize concen-
λex = 370 nm
trations 10 times below aggregation limits of whole
Fluorescence Instensity (arb.units)

4 NMP Soluble asphaltenes [32,33]. However, if one were to use higher


Whole Asphaltene concentrations, the fluorescence yield can be reduced to
NMP Insoluble
very low values due to collisional quenching [13]. This
3
photophysical effect has been studied in detail using
measurements of quantum yields, fluorescence lifetimes
2 and fluorescence spectra as a function of concentration, oil
type, and excitation wavelength [13]. If experiments do not
prove use of low concentrations, then the corresponding
1
results are in doubt. Furthermore, it has been shown that
asphaltene subfractions can have very different solubility
0 characteristics than the whole asphaltene [34]. In other
350 400 450 500 550 600 650 words, the aggregation and flocculation phenomena can
Wavelength (nm)
4 change depending on the asphaltene fractionation. It is
plausible that the concepts stating NMP insoluble asphalt-
λex = 410 nm enes are not fluorescent and polymeric are incorrect because
Fluorescence Instensity (arb.units)

the corresponding concentrations are too high, the asphalt-


3 NMP Soluble
Whole Asphaltene
enes are aggregated resulting in (1) no fluorescence and (2)
NMP Insoluble large apparent mass (aggregate mass). Recent high-Q
ultrasonic determination of aggregation limits for asphalt-
2 enes in toluene show the petroleum [32] and coal
asphaltenes [33] aggregate at w100 mg/l in toluene, much
lower than previously thought.
1
4.3. Bitumen

0 Fig. 6 compares athabasca bitumen asphaltenes to UG8


400 450 500 550 600 650 asphaltenes; this somewhat different source material gives
Wavelength (nm)
3 asphaltenes of essentially the same size as those of UG8
crude oil. There has been the suggestion that athabasca
λex = 440 nm
asphaltenes consist of aromatic rings systems that are
Fluorescence Instensity (arb.units)

covalently cross-linked [17]. Our data directly refute this


NMP Soluble
2 Whole Asphaltene model. Athabasca bitumen asphaltenes are similar to all
NMP Insoluble other crude oil asphaltenes we have measured; they are
monomeric and are small. Again the wavelength depen-
dence of tr indicates that the single chromophore (aromatic
ring system) interrogated is a large part of the entire
1
asphaltene molecule; there is one chromophore per
molecule.
The prevailing concept is that the solubility classification
of asphaltenes is fairly restrictive chemically thereby
0 preventing wild variations in corresponding molecular
450 500 550 600 650
structure. For grossly different source materials, coal vs
Wavelength (nm)
crude oil, some differences in asphaltene molecular
Fig. 5. The fluorescence spectra for asphaltene fractions; NMP soluble, structure exist, but still the differences are not large. For
NMP insoluble and whole UG8 asphaltene at three excitation wavelengths. asphaltene science, this result is good news. Asphaltenes,
S. Badre et al. / Fuel 85 (2006) 1–11 9

0.8
UG8 Asphaltene λex=316 nm, λem = 370 nm
0.7
Athabasca
Bitumen UG8 Asphaltene

Fluorescence Intensity
Asphaltene Concentration
0.6
Dilute
1%
τr 0.5
2.5%
(ns)
10%
0.4

0.3

0.2
0 10 20 30 40 50
0.1 Time (nano seconds)
400 450 500 550 600
Fluorescence Emission Wavelength (nm) Fig. 7. The fluorescence decay curves for asphaltene toluene solutions at
various concentrations. Intermolecular energy transfer and quenching at
Fig. 6. Comparison of Athabasca bitumen asphaltene and UG8 crude high concentrations cause a dramatic reduction in fluorescence lifetimes.
oil asphaltene.
By measuring lifetimes of dilute asphaltene solutions and
comparing those to deasphaltened crude oil (maltene)
while complex, have underlying simple governing
solutions, we can determine if we see a reduction of
principles.
lifetimes due to intramolecular relaxation. Thus, we can test
It is known that fluorescence quantum yields vary as a
the speculation regarding intramolecular relaxation in
function of various parameters in crude oils and asphaltenes.
asphaltenes [43]. Fig. 8 shows the fluorescence decay
There has been speculation in the literature [43], rather
curves for very dilute solutions of UG8 asphaltenes and the
extensive speculation at that, as to whether large ‘archipe-
corresponding UG8 crude oil without asphaltenes (UG8
lago type’ molecular structures could have excessive
maltenes). The lifetimes of the two solutions are very
intramolecular quenching of fluorescence. These structures
similar and much larger than the fluorescence lifetimes of
have been proposed for asphaltenes where there are various
concentrated asphaltene solutions [42]. Thus, there is little
islands of fused ring systems cross-linked by alkane chains.
or no measurable intramolecular relaxation in asphaltenes.
It is known that individual fluorophores can interact in
First, long fluorescence lifetimes as listed in Fig. 8 for dilute
solution if they become physically close enough—say by
asphaltene solutions are not compatible with extensive
diffusion. Electronic excitation energy of fluorophores can
intramolecular relaxation. Second, nobody is (yet) propos-
be transferred to a fluorophore of smaller energy excited
ing extra large molecular weight components in maltenes.
states resulting in red shifted fluorescence emission [14]. In
addition, fluorophore or chromophore interaction in solution λ ex=390 ns,
b) λ em = 430 ns
can result in quenching, thus reduced quantum yields [41].
Both processes of energy transfer and electronic energy UG8 De-Asphaltened Crude Oil
quenching are additional decay processes of the electronic τ=1.8 ns, 10.1 ns
Fluorescence Intensity

excited state thus necessarily give rise to a reduction of


fluorescence lifetime [44]. Fig. 7 shows what happens to the
fluorescence lifetimes of asphaltenes as relaxation processes UG8 Asphaltene (Dilute Soln.)
increase. By increasing asphaltene concentrations in τ=2.0 ns, 9.4 ns
toluene, relaxation processes of excited electronic states
increase. Fig. 7 shows the fluorescence decay curves for a
series of asphaltene solutions at different concentrations. A
UG8 Asphaltene (10% Soln.)
large reduction of asphaltene fluorescence lifetimes with
τ=0.7 ns
higher concentrations is observed. At the highest concen-
trations, the asphaltene fluorescence decay curve is 5 15 25 35 45 55 65
essentially the lamp function, the lifetime is below Time (nanoseconds)
measurement limit of the system. This occurs because of
the rapid concentration-dependent relaxation mechanisms Fig. 8. The fluorescence decay curves for (top) a dilute toluene solution of
deasphaltened crude oil UG8 (UG8 maltene); (middle) a dilute solution of
that speed up the decay of the fluorescent state. The ‘lamp’ UG8 asphaltenes in toluene; and (bottom) UG8 asphaltene in high
here is Beamline U9B at the National Synchrotron Light concentration (10%) in toluene. Intramolecular quenching effects are
Source at Brookhaven National Labs. absent from the dilute solution lifetime data.
10 S. Badre et al. / Fuel 85 (2006) 1–11

a) λex =290 nm, λem= 330 nm within the radiative lifetime. This exact chain length would
104 not be generally expected in an asphaltene molecule that
NS De-Asphaltened Crude Oil possesses two chromophores. Moreover, asphaltenes exhibit
τ =3.6 ns, 13.5 ns
steric interactions further impeding an eximer equivalent.
Fluorescence Intensity

103 And asphaltenes would contain multiple alkane links in the


NS Asphaltene (Dilute Soln.)
archipelago model further reducing flexibility. Our rather
τ = 3.5 ns, 12.1 ns large anisotropies also indicate that intramolecular energy
transfer is not so large as energy transfer with subsequent
102 fluorescence greatly reduces anisotropy [18,19]. The
NS Asphaltene (10% Soln.) archipelago model for asphaltene molecular structure is
τ =0.6 ns inconsistent with our data.
101
5 15 25 35 45 55 65
Time (nanoseconds)
5. Conclusions
Fig. 9. The fluorescence decay curves for (top) a dilute toluene solution of
deasphaltened crude oil NS (NS maltene); (middle) a dilute solution of NS
asphaltenes in toluene and; (bottom) NS asphaltene in high concentration
Fluorescence depolarization studies have proven to be a
(10%) in toluene. Intramolecular quenching effects are absent from the powerful and consistent tool to unravel asphaltene complex-
dilute solution lifetime data. ities. Asphaltene molecules are monomeric with relatively
small molecular weight. Independent of which source
The maltene chromophores are presumed not to be material or which solubility fraction, asphaltenes are
crosslinked. By comparison, the asphaltene chromophores consistently found by FD to be relative small molecules.
are not crosslinked. Monomeric asphaltene molecules Both the zeroth moment (absolute tr) and the first moment
prevail. (wavelength dispersion of tr) provide independent, compel-
Fig. 9 shows the fluorescence decay curves for dilute ling data. These conclusions are in accord with mass
solutions of NS asphaltene and for the NS crude oil without spectral techniques, FI-MS, ESI-FT-ICR-MS, APCI-MS.
asphaltenes (NS maltenes). This data is for a different Only LD-MS yields self-contradictory results in the
excitation and emission wavelengths than Fig. 8. Again, the literature most likely over the issue of proper baseline
asphaltene and maltene fluorescence decay curves are very subtraction. The coal asphaltene results presented here
similar and exhibit much longer lifetimes than the match closely those previously reported which were
concentrated solution of asphaltene. These data show obtained by Taylor-dispersion (coupled with UV absorption
there is no intramolecular relaxation in asphaltenes. The detection). Attempts to disqualify time-resolved fluor-
speculations in the literature that intramolecular relaxation escence depolarization have the additional problem to
in asphaltenes is dominant are contradicted by this data. The similarly disqualify the Taylor-dispersion studies as well
likely reason is that asphaltenes are primarily monomeric as as most mass spec studies. In addition, both direct molecular
all FD results suggests. In addition, other considerations imaging techniques, HRTEM and STM, provide corrobora-
argue against intramolecular relaxation. tive results consistent with the FD results. Asphaltenes are
Certainly, energy transfer can take place from blue- monomeric, not polymeric.
emitting chromophores (large HO–LU gap) [45] to small
HO–LU gap fluorophores. However, red-emitting chromo-
phores cannot transfer energy to large HO–LU gap Acknowledgements
chromophores. Toluene is an example of a large HO–LU
gap chromophore (and possible quencher). But energy We are deeply indebted to Prof. Masashi Iino for
transfer to toluene from visible fluorophores does not occur; providing us with one of the coal asphaltene samples.
in accord with energy conservation. For example, toluene
solutions of fluorophores are highly fluorescent. Thus, our
longest wavelength FD results would be unaffected in any
event by other blue-emitting asphaltene fluorophores. In References
addition, the study of bichromophoric molecules connected
[1] Wargadalam VJ, Norinaga K, Iino M. Fuel 2002;81:1403.
by a single alkane chain exhibit the ‘Hirayama’ rule [46]. [2] Chilingarian GV, Yen TF, editors. Bitumens, asphalts and tar sands.
The Hirayama rule which is generally applicable states that New York: Elsevier; 1978.
upon electronic excitation of the bichromophoric molecules, [3] Bunger JW, Li NC, editors. Chemistry of asphaltenes. Washington
eximers form only for the connecting chain length of three DC: American Chemical Society; 1981.
[4] Sheu EY, Mullins OC, editors. Asphaltenes: fundamentals and
carbons, for example, 1,3-diphenyl propane. Other connect- applications. New York: Plenum Publishing Co.; 1995.
ing chain lengths preclude eximer formation due to the [5] Mullins OC, Sheu EY, editors. Structure and dynamics of asphaltenes.
difficulty of the linked chromophores to find each other New York: Plenum Press; 1998.
S. Badre et al. / Fuel 85 (2006) 1–11 11

[6] Crick F. What mad pursuit, a personal view of scientific discovery. [26] Suelvas I, Islas CA, Millan M, Galmes C, Carter JF, Herod AA, et al.
New York: Basic Books; 1988. Fuel 2003;82:1.
[7] Scotti, R., Montanari, L. [Chapter 3. in Ref. [4]]. [27] Qian K, Rodgers RP, Hendrickson CL, Emmett MR, Marshall AG.
[8] Bergmann U, Groenzin H, Mullins OC, Glatzel P, Fetzer J, Cramer S Energy Fuels 2001;15:492.
P. Chem Phys Lett 2003;369:184. [28] Hughey CA, Rodgers RP, Marshall AG. Anal Chem 2002;74:4145.
[9] George GN, Gorbaty LL. J Am Chem Soc 1989;111:3182. [29] Cunico RI, Sheu EY, Mullins OC. Petroleum Sci Technol 2004;
[10] Mitra-Kirtley S, Mullins OC, Chen J, van Elp J, George SJ, Cramer S 22(7/8).
P. J Am Chem Soc 1993;115:252. [30] Sheu EY. J Phys Condens Matter 1996;8:A125.
[11] Zajac GW, Sethi NK, Joseph JT. Scan Microsc 1994;8:463. [31] Anderson SI, Christensen SD. Energy Fuels 2000;14:38.
[12] Sharma A, Groenzin H, Tomita A, Mullins OC. Energy Fuels 2002; [32] Andreatta G, Bostrom N, Mullins OC. Langmuir 2005;21:2728.
16:490. [33] Andreatta G, Goncalves CC, Buffin G, Bostrom N, Quintella CM,
[13] Mullins, O.C. [Chapter 2, in Ref. [4]]. Arteaga-Larios F, et al. Energy Fuels, in press.
[14] Downare TD, Mullins OC. Appl Spectrosc 1995;49:754.
[34] Acevedo S, Escobar O, Echevarria L, Gutierrez LB, Mendez B,
[15] Mullins OC, Mitra-Kirtley S, Zhu Y. Appl Spectrosc 1992;46:1405.
Energy Fuels 2004; 18:305.
[16] Ruiz-Morales Y. J Chem Phys A 2002;106:11283.
[35] Urbach F. Phys Rev 1953;92:1324.
[17] Strausz OP, Mojelsky TW, Lown EM. Fuel 1992;71:1355.
[36] Trevor TJ, Milan M, Behrouzi M, Herod AA, Kandiyoti R. Energy
[18] Groenzin H, Mullins OC. J Phys Chem A 1999;103:11237.
Fuels 2005;19:164.
[19] Groenzin H, Mullins OC. Energy Fuels 2000;14:677.
[37] Ascanius BE, Garcia DM, Andersen SI. Energy Fuels 2004;18:1827.
[20] Buenrostro-Gonzalez E, Groenzin H, Lira-Galeana C, Mullins OC.
Energy Fuels 2001;15:972. [38] Iino M, Takanohashi T, Ohsuga H, Toda K. Fuel 1988;67:1639.
[21] Groenzin H, Mullins OC, Eser S, Mathews J, Yang M-G, Jones D. [39] Arso A, Iino M. Fuel Process Techmol 2004;85:325.
Energy Fuels 2003;17:498. [40] Mullins OC, Kaplan M. J Chem Phys 1983;79:4475.
[22] Buch L, Groenzin H, Buenrostro-Gonzalez E, Andersen SI, Lira- [41] Ralston CY, Wu X, Mullins OC. Appl Spectrosc 1996;50:1563.
Galeana C, Mullins OC. Fuel 2003;82:1075. [42] Ralston CY, Mitra-Kirtley S, Mullins OC. Energy Fuels 1996;10:623.
[23] Boduszynski, M.W. [Chapter 7, Ref. [2]]. [43] Strausz OP, Peng P, Mugich J. Energy Fuels 2002;16:809.
[24] Miller JT, Fisher RB, Thiyagarajan P, Winans RE, Hunt JE. Energy [44] Wang X, Mullins OC. Appl Spectrosc 1994;48:977.
Fuels 1998;12:1290. [45] HO-LU gap is the energy gap between the highest occupied molecualr
[25] Yang M-G, Eser S. ACS reprints. ACS new orleans meeting 1999 orbital and the lowest unoccupied molecular orbital.
pp. 768. [46] De Schryver FC, Boens N, Put J. Adv Photochem 1977;10:359.

Вам также может понравиться