AN AMERICAN CHEMICAL SOCIETY JOURNAL

VOLUME 2, NUMBER 5 SEPTEMBER/OCTOBER 1988

0 Copyright 1988 by the American Chemical Society

Articles
Composition of Heavy Petroleums. 2. Molecular
Characterization
Mieczyslaw M. Boduszynski
Chevron Research Company, P.O. Box 1627, Richmond, California 94802-0627
Received November 12, 1987, Revised Manuscript Received April 2, 1988

The objective of this paper is to illustrate the variation of the chemical composition of heavy
petroleums as a function of atmospheric equivalent boiling point (AEBP) up to approximately 1400
O F and further as a function of solubility down to the “bottom of the barrel”. The first paper in this
series described the variation of heteroatom (S, N, 0, V, Ni, Fe) concentrations, hydrogen deficiency
(H/C atomic ratio), and molecular weight distributi0n.l This paper describes the variation of the
chemical composition of heavy petroleums on the molecular level. Two molecular characterization
approaches are discussed. The first approach utilizes “average molecular parameters”. The data
illustrate the changes in relative concentrations of the “molecular building blocks” with increasing
AEBP and decreasing solubility of heavy crude components. The second approach involves the
extensive use of high-performance liquid chromatography (HPLC) to separate “compound-class”
fractions for further molecular characterization. The results are expressed in terms of the homologous
series of compounds and their carbon number distributions by using field-ionization mass spectrometry
(FIMS). The results help unravel the immense complexity of heavy petroleums. Data show the great
diversity of molecular types, involving numerous homologous series of compounds, each series having
a very wide carbon number distribution (extending to over Clm) and very low relative concentrations
of individual carbon number homologues (from a few tenths of 1wt 5% to less than 1ppm by weight).
Experimental data reveal homologous series of compounds, apparently of “biomarker” origin, greatly
exceeding the carbon number reported for “biomarkers”. The results demonstrate the need for and
the importance of heavy oil fractionations to produce operationally well-defined fractions having
progressively higher AEBP’s, which can be then analyzed and compared on a consistent basis. Data
also illustrate the importance of HPLC separations for preconcentrating and isolating compounds
of interest from a complex sample matrix.

Introduction any heavy petroleum into a set of operationally well-de-

In recent years, research was undertaken in this labo-
ratory to delineate the complexity of heavy crude oils and
to provide information on the variation of their compo-
sition as a function of atmospheric equivalent boiling point
(AEBP). The analytical approach involved the fraction-
ation of heavy oils by using short-path distillation
(DISTACT-volatility fractionation) followed by sequen-
tial elution fractionation (SEF-solubility fractionation)
of DISTACT “nondistillable” residues. DISTACT and
SEF, used together, provide a robust method for dividing
fined fractions having progressively higher AEBP’s.~?~
The DISTACT-SEF fractions were subjected to further
characterization by various analytical methods, providing
data which can be compared on a consistent basis. The
first paper in this series described the variation of het-
eroatom (S, N, 0, V, Ni, Fe) concentrations, hydrogen
(1)Boduszynski, M.M.Energy Fuels 1987,1 , 2-11.
(2)Schwartz, H.E.;Brownlee, R. G.; Boduszynski, M. M.; Su, F. Anal.
Chem. 1987,59, 1393-1401.

0887-0624/88/2502-0597$01.50/0 0 1988 American Chemical Society

598 Energy & Fuels, Vol. 2, No. 5, 1988
deficiency (H/C atomic ratio), and molecular weight dis-
tribution, all as a function of AEBP up t o about 1400 O F
and further as a function of solubility down to the “bottom
of the barrel”.’
The objective of this paper is to illustrate the variation
of the chemical composition of heavy petroleums on the
molecular level. Two analytical approaches are discussed
in this paper. The first approach utilizes elemental
analysis results, 13Cnuclear magnetic resonance (NMR)
data, and average molecular weight measurements and
describes heavy oil composition in terms of “average mo-
lecular parameters”. The second approach involves ex-
tensive use of high-performance liquid chromatography
(HPLC) to separate “compound-class”fractions for further
characterization in terms of the homologous series of
coFpounds and their carbon number distributions. The
compositional changes are presented as a function of
volatility (AEBP) and solubility (SEF) of heavy-oil com-
ponents. The advantages and limitations of the two
characterization approaches and the inherent ambiguity
of the compositional data for heavy oils are discussed.
Experimental Section
Materials Studied. Atmospheric residues (ARs; -650 OF+
AEBP) derived from different crude oils have been the focus of
this study. The first paper in this series describes six different
feedstocks.’ In this paper, compositional data are presented for
only selected samples that are identified in the text. Occasionally,
for proprietary reasons, the origin of samples has not been re-
vealed.
Volatility and Solubility Fractionations. The A R s were
first fractionated each into several distillate ”cuts” having pro-
gressively higher AEBP and one “nondistillable” residue (- 1300
OF+ AEBP) by using a DISTACT short-path distillation appa-
ratus (Leybold-Heraeus, GmBH). The “nondistillable” residues
were further fractionated with the SEF method, each into the
following four solubility fractions: (1)n-pentane soluble (SEF-l),
(2) cyclohexane solubleln-pentane insoluble (SEF-2), (3) toluene
soluble/cyclohexane insoluble (SEF-3), and (4) methylene chlo-
ridemethanol(41 v/v) soluble/toluene insoluble (SEF-4). The
details of the DISTACT-SEF fractionations are given in the
previous paper.l
Preparative Chromatographic Separations. Three prep-
arative HPLC methods have been developed as a part of this
research effort. The first separation method involves HPLC on
two alumina columns. The first column is packed with basic
alumina, and the second contains acidic alumina. This separation
is referred to as the HPLC-BA/AA method. The objective of this
separation step is to separate a sample into the following oper-
ationally defiied fractions: (a) “acidic” compounds, (b) “pyrrolic”
N-compounds, (c) “basic” compounds, and (d) “neutral” com-
pounds. The second method, HPLC-NH2,separates the “neutral”
compounds into the following fractions: (a) saturates and mo-
noaromatics, (b) diaromatics, (c) triaromatics, (d) tetraaromatics,
and (e) pentaaromatics and greater. Finally, the third method,
HPLC-SIL, produces fractions of saturates and monoaromatics.
HPLC-BA/AA Method. The HPLC-BA/AA system consists
of a programmable pump (Model 590, Waters) with an automatic
solvent selection valve, a sample loop injector (Model 210, Altex),
a series of two 9-mm-i.d. X 25-cm precision-bore glass columns
(Altex) connected through a switching valve, and a programmable
fraction collector (Foxy, ISCO). The first column is packed with
basic alumina (AG-10, BioRad), and the second column is packed
with acidic alumina (AG-4, BioRad). Fresh packing is used for
each separation. The separation conditions are as follows: sample
size, 200-400 mg dissolved in cyclohexane; flow rate, 10 mL/min.
The elution sequences are as follows: (a) “neutral” compounds,
a fraction eluted from the BA/AA columns with 110 mL of cy-
clohexane followed by 40 mL of toluene combined with a fraction
eluted from the AA column with 200 mL of methylene chloride;
(b) “basic”compounds, a fraction backflushed from the AA column
with 100 mL of a mixture of methylene chloride and methanol
(4:l v/v); (c) “pyrrolic” N compounds, a fraction eluted from the
Boduszynski
BA column with 200 mL of methylene chloride; (d) “acidic”
compounds, a fraction backflushed from the BA column with 100
mL of a mixture of methylene chloride and methanol (4:l v/v).
Solvents are removed from the fractions by using standard
procedures. Weight percent yields of the recovered material are
determined gravimetrically. Fractions from duplicate separations
are combined to produce sufficient quantities of material for
further characterization work. The reproducibility between du-
plicate separations is about 5% or better for “neutral” fractions,
and about 5-15% for “basic”, “pyrrolic”, and “acidic” fractions.
HPLC-NH2Method. The HPLC system consists of a gradient
pump (Model 8800, Du Pont Instruments) equipped with prep-
arative heads, a preparative ZORBAX-NH2column (21.2 mm i.d.
X 25 cm, 8-pm particle size, Du Pont), a UV/vis diode-array
spectrophotometer (Model 8451A, Hewlett-Packard) as a detector,
a Nelson Analytical 760 Series interface connected to a Hew-
lett-Packard 9920 computer system, and a plotter (Model 7470A,
Hewlett-Packard). The separation conditions are as follows:
sample size, 100-200 mg dissolved in heptane; flow rate, 20
mL/min. Elution sequence: heptane, 0-29.5 Min; isocratic,
29.5-54.5 min; heptane and methylene chloride (0-50%) gradient.
Cut points between the ring-type subfractions are determined
by using HPLC-UV profiles generated by the UV diode-array
detector. Weight percent yields of the fractions are determined
gravimetrically. The reproducibility between duplicate separations
is about 5-15%.
HPLC-SIL Method. The HPLC system involves a preparative
isocratic pump (Model 8800, Du Pont Instruments), a preparative
ZORBAX-SIL column (21.2 mm i.d. X 25 cm, 8-pm particle size,
Du Pont), a differential refractometer detector (Model R401,
Waters), a variable wavelength UV detector (Model 450, Waters),
and a Nelson Analytical 760 Series interface connected to a
Hewlett-Packard 9920 computer system. The separation con-
ditions are as follows: sample size, 100-200 mg dissolved in
heptane; mobile phase, heptane, isocratic, 10 mL/min flow rate.
Weight percent yields of the recovered material are determined
gravimetrically. The reproducibility between duplicate separations
is better than 5%.
Molecular Weight Measurements. Vapor pressure osmo-
metry (VPO) was used to determine number-average molecular
weights of the DISTACT-SEF fractions. The VPO measurements
were conducted in a toluene solution at a temperature of 50 “C
and in a pyridine solution a t a temperature of 90 “C on a vapor
pressure osmometer (Model 232A, Wescan Instruments). Mea-
surements were performed at three or four different concentra-
tions, and the results were calculated by extrapolation to infinite
dilution. It is important to note that the calibration constant a
in
lime- ( A T / C ) = a / M ,
(where AT = temperature difference, C = concentration, and M ,
= number-average molecular weight) is solute independent for
this i n ~ t r u m e n t . ~
In addition, field-ionization (FI) and field-desorption (FD) mass
spectrometry (MS) have been used to determine molecular weight
distributions of the DISTACTSEF fractions. Details were given
in the earlier paper.’
I3C NMR Analysis. 13C NMR spectrometry was used to
determine concentrations of aromatic and aliphatic carbons in
the DISTACT-SEF fractions. The 13C NMR spectra were ob-
tained on 40% (v/v) solutions in CDC13. The final solution
contained 0.04-0.05 M Cr(acac), according to the procedure of
Shoolery and B ~ d d e .Pulse~ parameters were applied every 2.4
s. Data were acquired for 0.4 s, which required 16 384 points at
the spectral width of 20 kHz. Line broadening of 3 Hz was applied
prior to Fourier transformation. Each spectrum was the result
of 500-2000 coadded free-induction decays.
FIMS Analysis. The FIMS measurements for HPLC fractions
were obtained a t SRI International, Menlo Park, CA, by using
procedures previously d e ~ c r i b e d . ~The FIMS data processing
(3)Burge, D.E.J. Appl. Polym. Sci. 1979,24,293-299.
(4)Schoolery, J. N.;Budde, W. L. Anal. Chem. 1976,48,1458-1461.
(5)Buttrill, S.E., Jr. Final Technical Report, SRI Project PYU 8903,
1981;SRI International: Menlo Park, CA.
Composition of Heavy Petroleums Energy &Fuels, Vol. 2, No. 5, 1988 599
50% AEBP,
Gravity, Fraction “F
“API
0 *15000~~** 451

10 “500-650°F” 592

- 20
is cut 1 722
30
2
u Cut 2A 825
-e
E 40

’-
Cut 2 8 887
50
Cut 3A 955
60
.-L Cut 3 8 1023
m Cut 4A 1091
3 70
s
0 80
Cut 4 8
cut 5
1158
1239

SEF-1 (1365)
90
SEF-2
SEF-3
100 ccc 1

Table I. Carbon Number, Atmospheric Equivalent Boiling Point (AEBP), and the Number of Acyclic Alkane Isomers
carbon no. n-alkane AEBP,” O F no. of isomers* examples of Detroleum distillation cuts
5 97 3 1
8
10
12
258
345
42 1
75t
355)
gasoline

15
20
25
30
35
40
45 1022 82.2 x 1014
.-*- n-- --- I vacuum residue, asphalt

+ 1+ +
60 1139 221.5 X lozo
80 1242 1056.4 X lo2*
100 1306 5920 X 1 “nondistillable” residue
OFrom ref 64. bFrom ref 6.

and interpretation were performed in this laboratory.
Results and Discussion
General Considerations. Petroleum components can
be viewed as two major groups of compounds, namely,
hydrocarbons and nonhydrocarbons. The term hydro-
carbons is used to describe molecules that comprise only
carbon and hydrogen atoms. Hydrocarbons include acyclic
alkanes (normal and isoparaffins), cycloalkanes (naphth-
enes), and aromatic hydrocarbons. Most naphthenes
contain both saturated rings and side chains and are de-
fined by the number of naphthenic rings, i.e., monocyclic,
dicyclic, tricyclic, etc. Similarly, most of the aromatic
hydrocarbons bear normal or branched chains and na-
phthenic rings. A molecule containing one aromatic ring
is regarded a monoaromatic, a molecule with two aromatic
rings diaromatic, with three aromatic rings as triaromatic,
etc., even if several naphthenic rings and side chains are
attached to the aromatic ring. Non-hydrocarbons are
compounds that, in addition to carbon and hydrogen at-
oms, also include one or more heteroatoms such as sulfur,
nitrogen, oxygen, vanadium, nickel, or iron.
The complexity of petroleum increases rapidly with in-
creasing boiling point as a result of the increasing number
of atoms in a molecule and the immense number of their
possible structural arrangements. This makes the com-
positional analysis of high boiling petroleum fractions an
extremely difficult task. Figure 1 illustrates the carbon
number distribution (the total and aromatic carbon atoms)
as a function of AEBP for heavy (13.6 OAPI) Kern River
petroleum. (For more details see further discussion on
“average molecular parameters”.) The curve representing
the total number of carbon atoms covers a range from
about 10 to almost 300 carbon atoms per molecule. The
number of aromatic carbon atoms per molecule ranges
from about 1.3 a t the “top of the barrel” to almost 150 a t
the “bottom of the barrel”.
The immense complexity of petroleum can be illustrated
by using acyclic alkanes as an example. Table I shows the
number of possible acyclic alkane isomers at a given carbon
number.6 At Czo,a boiling point borderline for petroleum
(6) McClellan, A. L., Chevron Research Co., unpublished data.
600 Energy & Fuels, Vol. 2, No. 5, 1988
“heavy ends” (-650 O F + AEBP), the possible number of
alkane isomers is already tens of thousands. The numbers
become astronomical as the boiling point and the carbon
number increase.
Although alkanes are not a major component of heavy
petroleums, a large percentage of the carbon atoms in other
types of molecules is represented by paraffiiic chains. The
molecules having naphthenic and aromatic rings also have
a huge number of possible structural arrangements. The
problem is further complicated by the presence of het-
eroatoms involving a variety of functional groups at an
immense number of possible locations within a molecule.
Even if only a very small percentage of the thermody-
namically favored isomers were actually present in crude
oil, the number of compounds comprising high boiling
fractions of petroleum would be immense. This is one of
the key factors limiting our ability to characterize those
complex mixtures.
It was recognized long ago that the characterization of
high-boiling petroleum fractions in terms of the individual
components is impossible. Even a combination of gas
chromatography and mass spectrometry (GC/MS), one of
the most powerful analytical tools, is limited to relatively
low-boiling fractions. This limitation is due to the im-
mense number of isomers rather than to the limited sample
volatility. For example, petroleum middle distillates (-
350-650 O F AEBP) are already complex enough to result
in chromatograms consisting of hundreds of poorly re-
solved peaks. This makes the GC/MS analysis, particu-
larly the interpretation of the data, an extremely tedious
and difficult task. Analysis of vacuum gas oils (VGO’s),
which are also completely volatile (-650-1000 O F AEBP)
under GC/MS conditions but which contain significantly
more compounds that are more difficult to resolve, is a
practical impossibility.
It is obvious from the above discussion that the molec-
ular characterization of heavy oils must involve some
“grouping” of compounds, which inevitably leads to am-
biguity of the compositional data.
“Average Molecular Parameters”. The concept of an
“average molecule” has been widely used to characterize
fossil fuel liquids. It typically utilizes elemental analysis
results and number-average molecular weight measure-
ments to calculate the formulae of an “average molecule”.
This approach is an extreme example of “grouping”.
This analytical approach can be expanded by utilizing
results generated by ‘H and 13CNMR spectrometry, and
it allows for describing a complex mixture in terms of the
“average molecular parameter^".^ When it is combined
with high-resolution mass spectrometry (HRMS), it can
lead to significant insights into molecular structure.E
In this work, the “average molecular parameters” have
been obtained by utilizing elemental analysis (C, H, N, S,
0, V, Ni, and Fe) results, percent aliphatic and aromatic
carbon from 13C NMR, and number-average molecular
weight measurements. This approach provides a conven-
ient means for comparing complex mixtures without re-
vealing their actual components. The “average molecular
parameters” are being used here only to illustrate the
changes in relative concentrations of the “molecular
building blocks” with increasing AEBP. No attempt
should be made to derive the “average molecular structure”
(7) Petrakis, L.; Allen, D. NMR for Liquid Fossil Fuels; Elsevier:
Amsterdam, Oxford, England, New York, Tokyo, 1987;pp 91-162 and
references therein.
(8) Aczel, T.; Williams, R. B.; Brown, R. A.; and Pancirov, R. J. In
Analytical Methods for Coal and Coal Products; Karr, C., Jr., Ed., Ac-
ademic: New York, San Fracisco, CA, London, 1978; Vol. 1, pp 499-539
and references therein.
Boduszy nski
from those data. A more detailed discussion of the limi-
tations of the “average molecular structure” determinations
can be found el~ewhere.~
The molecular weight has a very significant effect on the
“average molecular parameters”. Figure 2 shows molecular
weight distribution curves for two heavy petroleums: (top)
Kern River, San Joaquin Valley, California; (bottom)
Offshore California. The curves were obtained by plotting
number-average molecular weight values measured for
fractions having progressivelyhigher AEBP. Interestingly,
data in Figure 2 show higher molecular weights for the “less
heavy” (22.5 “API) Offshore California crude oil than for
the ”heavier” (13.6 OAPI) Kern River petroleum. (See ref
1 for more details.)
The average molecular weight values for distillate cuts
(up to about 1300 O F AEBP) were obtained by VPO in
toluene and were in excellent agreement with those ob-
tained earlier by using FIMS.’ Thus, these results dem-
onstrate that the VPO measurements for distillate frac-
tions are not affected by intermolecular associations.
The results for the SEF fractions, which were derived
from DISTACT “nondistillable”residues (about 1300 O F +
AEBP), were obtained by using VPO both in toluene and
in pyridine. Data in Figure 2 illustrate the dramatic effect
of intermolecular associations on the molecular weight
measurements by VPO in toluene. The latter produced
much higher average molecular weight values than those
obtained in solutions of pyridine. These results are con-
sistent with those reported earlier by Moschopedis et al.’O
The previously reported FIMS data’ for the SEF-1
fractions are in a fairly good agreement with the results
obtained by using VPO in pyridine. Unfortunately, in the
case of the SEF-2 and SEF-3 fractions (the SEF-4 fractions
were too small to analyze), such a comparison is impossible
because of the limited sample volatility (about 40-70%)
under the conditions of FIMS analysis, which did not allow
representative average molecular weight values to be ob-
tained. However, FIMS data obtained for those fractions
indicated relatively low molecular weights extending from
about 500 to 2000 amu.l The number-average molecular
weights for the SEF-2 and SEF-3 fractions obtained by
using VPO in pyridine, although significantly lower than
those obtained by VPO in toluene (see data in Figure 2),
are beyond the molecular weight range indicated by FIMS.
The average molecular weight values obtained for those
fractions using VPO in pyridine may still be affected by
intermolecular associations. This contention is supported
by the literature data on the effects of solvent polarity and
temperature on the VPO molecular weight measurements
for “asphaltenes”. For example, the measurements in
nitrobenzene gave significantly lower values than those
obtained in pyridine. A higher temperature further re-
duced the molecular weight values.1°
Data in Tables I1 and I11 illustrate the complexity of
heavy petroleums using “average molecular parameters”.
The results obtained for DISTACT-SEF fractions that
were derived from Kern River (Table 11) and Offshore
California (Table 111) AR’s are used as examples. The
results in Table I1 show that, for cut 1 (722 O F 50%
AEBP), on average every molecule is composed of 24.1
carbon atoms (6.4 aromatic and 17.7 aliphatic) and 39.3
hydrogen atoms. The data also indicate (if we assume one
heteroatom per molecule) that for every 1000 molecules,
there are 104 sulfur-, 75 nitrogen-, and 88 oxygen-con-
(9)Boduszynski, M.M.Liq. Fuels Technol. 1984,2,211-232.
(10)Moschopedis, S. E.; Fryer, J. F.; Speight, J. G. Fuel 1976,55,
227-232. Speight, J. G.In The Chemistry and Technology of Petroleum;
Dekker: New York, 1980;p 211, Table 7-6.
Composition of Heavy Petroleums Energy & Fuels, Vol. 2, No. 5, 1988 601
KERN RIVER CRUDE OIL (13.6 "API) Cut Point,

0 By VPO in Toluene
By VPO in Pyrldlne

Numbedverage Molecular Weight

NumberAverage Molecular Weight

Figure 2. Number-average molecular weight distributions for Kern River and Offshore California heavy petroleums.

Table 11. "AverageMolecular Parameters" for Kern River AR

50%
AEBP, cumwt% av no. of atoms/molecule
fraction O F from AR mol wt "2"' c, car €3 S N 0 V Ni Fe
cut 1 722 21.6 3375 -8.9 24.1 6.4 39.3 0.104 0.075 0.088
cut 2A 825 32.8 420" -12.0 30.0 8.0 48.0 0.134 0.156 0.118
cut 2B 887 42.8 471" -14.7 33.7 9.0 52.7 0.163 0.215 0.177
cut 3A 955 51.4 5275 -16.9 37.7 10.0 58.5 0.196 0.293 0.198
cut 3B 1023 58.3 594" -19.7 42.3 12.4 64.9 0.249 0.378 0.264
cut 4A 1091 64.3 682' -23.6 48.6 14.0 73.6 0.296 0.492 0.349 0.001 0.001
cut 4B 1158 68.9 755" -27.0 53.8 16.1 80.6 0.316 0.615 0.235 0.001 0.002
cut 5 1239 74.3 876" -31.5 62.4 17.7 93.3 0.356 0.745 0.515 0.008 0.001
SEF-1 (1365) 90.3 1464" -57.9 103.8 38.6 149.7 0.608 1.464 0.873 0.001 0.003 0.001
1112b -44.1 78.9 29.3 113.7 0.462 1.112 0.737 0.001 0.002 0.001
SEF-2 96.0 3064" -170.5 218.1 97.3 265.7 1.388 4.815 2.298 0.010 0.023 0.015
2518b -140.3 179.3 80.0 218.3 1.141 3.957 1.888 0.009 0.019 0.013
SEF-3 99.1 5625" -334.6 395.8 214.1 457.0 2.514 9.643 4.641 0.030 0.038 0.042
3619* -215.4 254.7 137.8 294.0 1.617 6.204 2.986 0.019 0.025 0.027
SEF-4 99.6
a Number-average molecular weight determined by VPO in toluene. * Number-average molecular weight determined by VPO in pyridine.
'"Z" in the general formula C,H,,,+zX.
taining molecules. This means that 26.7% of all molecules These are, of course, only average estimates. The actual
in cut 1 could be represented by non-hydrocarbons. distribution of heteroatoms in a molecule is not known.
602 Energy & Fuels, Vol. 2, No. 5, 1988 Boduszynski

Table 111. “Average Molecular Parameters” for Offshore California AR

50 %
AEBP, cumwt% av no. of atoms/molecule
fraction OF from AR mol w t “Z”‘ ct c, H S N 0 V Ni Fe
cut 1 697 18.9 329” -7.6 23.0 3.9 38.3 0.282 0.056 0.062
cut 2A 804 30.4 399” -10.1 28.1 5.5 46.1 0.353 0.108 0.090
cut 2B 887 38.4 469” -11.7 32.9 6.4 54.1 0.462 0.141 0.117
cut 3A 955 45.4 549” -13.8 38.3 7.8 62.7 0.624 0.188 0.137
cut 3B 1026 49.4 609” -16.8 42.7 8.5 68.6 0.807 0.239 0.164 0.001
cut 4A 1100 54.2 704” -21.3 48.8 10.7 76.3 0.990 0.332 0.189 0.004 0.001
cut 4B 1153 57.0 785” -25.6 55.1 14.3 84.6 1.185 0.387 0.235 0.007 0.001
cut 5 1243 62.7 955” -30.8 66.2 16.4 101.6 1.611 0.532 0.310 0.008 0.002
SEF-1 (1372) 78.4 2254“ -79.3 154.6 41.4 229.9 4.120 1.481 0.873 0.007 0.002 0.001
1911b -67.3 131.1 35.1 194.9 3.487 1.255 0.740 0.006 0.002 0.001
SEF-2 87.9 5697” -278.2 387.5 151.9 496.8 11.358 7.121 2.991 0.104 0.024 0.016
4146b -202.6 282.1 110.6 361.5 8.250 5.180 2.177 0.076 0.017 0.012
SEF-3 98.6 16348’ -908.9 1112.1 487.1 1315.3 34.739 23.938 12.261 0.452 0.103 0.088
6069* -337.5 412.9 180.8 488.3 12.871 8.882 4.552 0.168 0.038 0.033
SEF-4 99.7
a Number-average molecular weight determined by VPO in toluene. *
Number-average molecular weight determined by VPO in pyridine.
“2” in the general formula C,H2,+zX.

For example, should oxygen-containing species be repre-
sented only by carboxylic acids (two oxygen atoms per
molecule) and nitrogen-containing species be represented
only by porphyrins (four nitrogen atoms per molecule),
then the number of oxygen- and nitrogen-containing
molecules in cut 1 would be reduced by a factor of 2 and
a factor of 4, respectively.
Data for the higher boiling cut 3B (1023 OF 50% AEBP)
derived from the same crude oil (Table 11) show that the
“average molecule” is composed of 42.3 carbon atoms (12.4
aromatic and 29.9 aliphatic) and 64.9 hydrogen atoms, and
for every lo00 molecules, there are already 249 sulfur-, 378
nitrogen-, and 264 oxygen-containing molecules. This
brings the concentration of non-hydrocarbons in cut 3B
to a total of 891 molecules/lOOO or 89.1%. For comparison,
“an average molecule” in cut 3B (1026 OF 50% AEBP)
derived from Offshore California AR (Table 111) is com-
posed of 42.7 carbon atoms (8.5 aromatic and 34.2 ali-
phatic) and 68.6 hydrogen atoms, and for every 1000
molecules there are 807 sulfur-, 239 nitrogen-, and 164
oxygen-containingmolecules and one vanadium-containing
molecule. This brings the total concentration of hetero-
atoms to 1.21 per molecule, implying that some molecules
must contain more than one heteroatom.
The number of heteroatoms per molecule increases
rapidly with increasing AEBP. Results in Table I11 show
the following trend for Offshore California AR:1.516 het-
eroatoms/molecule for cut 4A (1100 OF 50% AEBP), 1.815
heteroatoms/molecule for cut 4B (1153 OF 50% AEBP),
2.462 heteroatoms/molecule for cut 5 (1243 OF 50%
AEBP), and 5.490 heteroatoms/molecule for the SEF-1
fraction (1372 OF 50% AEBP).
Two sets of data are reported in Tables 11 and I11 for
the SEF fractions derived from “nondistillable” residues.
The first set of “average molecular parameters” was ob-
tained by using erroneously high average molecular weight
values determined by VPO in toluene. The second set of
data is based on average molecular weight values deter-
mined by VPO in pyridine, which, as discussed earlier, are
considerably lower. (See also Figure 2.)
The results illustrate the dramatic effect of intermole-
cular associations on VPO molecular weight measurements
and, consequently, on the calculated “average molecular
parameters”. For example, data in Table I11 show that an
“average molecule” in the SEF-3 fraction of Offshore
California crude oil is composed of the following number
of atoms: 412.9 carbon (of which 180.8 are in aromatic
rings), 488.3 hydrogen, 12.9 sulfur, 8.9 nitrogen, 4.5 oxygen,
0.17 vanadium, 0.04 nickel, and 0.03 iron when the num-
ber-average molecular weight value of 6069 determined by
VPO in pyridine is used. However, when the number-
average molecular weight value of 16 348 determined by
VPO in toluene is used, the average number of atoms per
molecule for the same SEF-3 fraction is as follows: 1112.1
carbon (of which 487.1 are in aromatic rings), 1315.3 hy-
drogen, 34.7 sulfur, 23.9 nitrogen, 12.3 oxygen, 0.45 vana-
dium, 0.10 nickel, and 0.09 iron.
In the past, erroneously high average molecular weights
have led to many misconceptions regarding the molecular
nature of heavy crudes and, particularly, of so-called
“asphaltenes” (a fraction insoluble in alkane solvents) for
which molecular weight values extending up to 100000 or
more have been reported.lG15
“Compound-Class Composition”. Many attempts
have been made in the past to separate “heavy ends” of
petroleum into groups or types of compounds. Numerous
separation methods and entire separation schemes have
been developed over the years.16 Characterization ap-
proaches based on chromatographic separations describe
heavy oils in terms of weight percent concentrations of
operationally defined fractions, providing so-called
“compound-class”or “compound-type” composition. The
SARA method, S = “saturates”, A = “aromatics”, R =
“resins”, and A = “asphaltenes”, is a classic example.”
This kind of compositional data, although useful for a
rough comparison of different feedstocks, does not reveal
the complex chemistry of heavy oils. It is obvious that both
the concentration of those fractions and the molecular
nature of their components are likely to vary with in-
creasing AEBP and the origin of the crude oil.
(11) Bunger, J. W. Prepr.-Am. Chem. SOC.,Diu.Pet. Chem. 1977,
22(2), 716.
(12) McKay, J. F.; Amend, P. J.: Cogswell, T. E.; Harnsberger, P. M.;
Erickson, R. B.; Latham, D. R. In Analytical Chemistry of Liquid Fuel
Sources, Tar Sands, Oil Shale, Coal, and Petroleum; Uden, P. C., Siggia,
S.,and Jensen, H. B., Eds.;Advances in Chemistry Series 170; American
Chemical Society: Washington, DC, 1978; pp 128-142.
(13) Bunger, J. W.; Thomas, K. P.; Dorrence, S.M. Fuel 1979,58,183.
(14) Boduszynski, M. M.; McKay, J. F.; Latham, D. R. Asphalt Pauing
Technol. 1980.49, 123-143.
(15) Chemistry of Asphaltenes; Bunger, J. W., and Li, N. C., Eds.;
Advances in Chemistry Series 195; American Chemical Society: Wash-
ington, DC, 1981.
(16)Altgelt, K. H.; Jewell, D. M.; Latham, D. R.; Selucky, M. L. In
Chromatography in Petroleum Analysis; Altgelt, K. H., Gouw, T. H.,
Eds.; Chromatographic Science Series: Dekker: New York and Basel,
1979; pp 185-214 &d references therein.
(17) Jewell, D. M.; Albaugh, E. W.; Davis, B. E.; Ruberto, R. G. Ind.
Eng. Chem. Fundam. 1974,13, 278.
Composition of Heavy Petroleums
I Sample I well-defined “compound-class” fractions for further mo-
+ HPLC-BA/AA
“Neutral“ “Basic” “Pyrrolic” ”Acidlc”
HPLC-NHs
Monoaromatics
HPLk-SiL
Figure 3. HPLC separation scheme.
One of the most recent systematic studies on the com-
position of “heavy ends” of petroleum was the American
Petroleum Institute Research Project 60.18-39This re-
search effort greatly advanced the knowledge of the com-
position of high-boiling petroleum fractions and demon-
strated the need for extensive separations followed by
analyses of the fractions in order to derive meaningful
compositional information.
In this work, the goal of chromatographic separations
was to prepare chemically meaningful and operationally
Hinds, G. P. Proc.-Am. Pet. Inst., Sec. 3, 1970,50, 279.
(18)
Hainea, W. E.; Ward, C. C.; Sugihara, J. M. Proc.-Am. Pet. Znst.,
(19)
Diu.Refin. 1971, 51, 375.
(20) Jewell, D. M.; Weber, J. H.; Bunger, J. W.; Plancher, H.; Latham,
D. R. Anal. Chem. 1972,44, 1391.
(21) Hirsch, D. E.; Hopkins, R. L.; Coleman, H. J.; Cotton, F. 0.;
Thompson, C. J. Anal. Chem. 1972,44,915.
(22) McKay, J. F.; Jewell, D. M.; Latham, D. R. Sep. Sci. 1972, 7,361.
(23) McKay, J. F.; Latham, D. R. Anal. Chem. 1972,44, 2132.
(24) Cogswell, T. E.; McKay, J. F.; Latham, D. R. Anal. Chem. 1971,
43, 645.
(25) Coleman, H. J.; Dooley, J. E.; Hirsch, D. E.; Thompson, C. J.
Anal. Chem. 1973,45, 1724.
(26) Dooley, J. E.; Hopkins, R. L.; Hirsch, D. E.; Coleman, H. J.;
Thompson, C. J. “Compound Type Separation and Characterization
Studies for a 370 to 535 OC Boiling Distillate of Gach Saran, Iran, Crude
Oil”; BuMines RI 7770; US. Bureau of Mines: Pittsburgh, PA, 1973.
(27) Hirsch, D. E.; Dooley, J. E.; Coleman, H. J.; Thompson, C. J.
“Compound Type Separation and Characterization Studies for a 370 to
535 OC Distillate of Wilmington, Califomia, Crude Oil”,BuMiies RI 7893,
US. Bureau of Mines: Pittsburgh, PA, 1974.
(28) Hirsch, D. E.; Dooley, J. E.; Vogh, J. W.; Thompson, C. J.
“Compound Type Separation and Characterization Studies for a 370 to
535 OC Distillate of Recluse, Wyoming, Crude Oil”, BuMines RI 7945,
U.S. Bureau of Mines. Pittsbureh. PA. 1974.
(29) Dooley, J. E.; Hirsch, DrE.; Thompson, C. J.; Ward, C. C. Hy-
drocarbon Process. 1974,53(11). 187
(30) Haines, W. E.; Hirsch,-D. E.; Thompson, C. J. “Separating and
Characterizing High-Boiling Petroleum Distillates: The USBM-API
Procedure”: RI LERC/RI-75 / 2, BERC/RI-75/ 2, July 1975.
(31) Thompson, C.’J.; Ward, C. C.( Ball, J:S. “Characteristics of
World‘s Crude Oils and Results of API RP 60”, RI BERC/RI-76/8, July
1976.
(32) McKay, J. F.; Cogswell, T. E.; Weber, J. H.; Latham, D. R. Fuel
1975,54, 50.
(33) McKay, J. F.; Weber, J. H.; Latham, D. R. Anal. Chem. 1976,48,
891.
(34) McKay, J. F.; Amend, P. J.; Harnsberger, P. M.; Cogswell,T. E.;
Latham, D. R. Fuel 1981,60, 14.
(35) McKay, 3. F.; Harnsberger, P. M.; Erickson, R. B.; Cogswell, T.
E.; Latham, D. R. Fuel 1981,60, 17.
(36) McKay, J. F.; Latham, D. R.; Haines, W. E. Fuel 1981, 60, 27.
(37) Boduszynski, M. M.; McKay, J. F.; Latham, D. R. Prepr.-Am.
Chem. SOC.,Diu. Pet. Chem. 1981,865-881.
(38) Grizzle, P. L.; Green, J. B.; Sanches, V.; Murgia, E.; Lubkowitz,
J. Prepr.-Am. Chem. Soc., Diu. Pet. Chem. 1981,839-850.
(39) Sturm, G. P.; Jr.; Grindstaff, Q.G.; Hirsch, D. E.; Scheppele, S.
E.; Hazos, M. Prepr.-Am. Chem. SOC.,Diu. Pet. Chem. 1981,851-856.
Energy & Fuels, Vol. 2, No. 5, 1988 603
lecular characterization. One must bear in mind, however,
that even the most efficient chromatographic separation
can only produce operationally well-defined fractions. The
mere separation of a complex mixture into fractions does
not reveal their composition. I t accomplishes, however,
an important goal of “grouping”and isolating components
of interest from a complex sample matrix. Thus, it makes
the fractions more amenable to further molecular char-
acterization. Of particular interest was the effect of in-
creasing AEBP, and thus increasing complexity, upon (a)
the performance of chromatographic separations and (b)
the ability of mass spectrometry to describe the chroma-
tographic fractions in terms of homologous series of com-
pounds.
Figure 3 shows the HPLC separation scheme used in this
study. A schematic diagram in Figure 4 illustrates the
DISTACT-SEF-HPLC-FIMS separation and character-
ization steps, using Kern River petroleum as the example.
The UV/vis profiles for the “solubility” fractions (SEF)
derived from DISTACT “nondistillable”residue are shown
in Figure 5.
The first HPLC step (Figure 3) involves separation on
a dual basic alumina/acidic alumina column (HPLC-
BA/AA). Conceptually, this separation step is similar to
the API RP 60 scheme because it first separates the sample
on the basis of acidity and basicity. The major difference,
however, is the use of basic and acidic alumina instead of
anion- and cation-exchange resins. Although the latter
were believed to be more selective than alumina (due to
simpler in nature interactions with solute molecules), they
require tedious preparations, were difficult to reproduce
from batch to batch, and were reported to introduce ar-
tifacts due to the deterioration of r e s i n ~ . ~ ~ ~ * ~
The HPLC-BA/AA method uses fresh, “as received”,
basic and acidic alumina for each separation. The sepa-
ration requires less than 2 h to complete. The fractions
are operationally very well-defined due to the use of au-
tomated HPLC equipment. The method has been de-
veloped primarily for separations of DISTACT distillable
cuts covering a boiling point range from about 650 to 1300
OF AEBP. However, in this work the method has been also
used for the SEF-1 fractions, which were derived from
DISTACT “nondistillable” residues (see Figure 4). These
solubility fractions had a 50% AEBP of approximately
1370 OF as determined by simulated distillation methods.l2
The results obtained for fractions boiling up to about
1300 O F AEBP showed material recoveries of 98.5 wt %
or better. Losses for the SEF-1 fractions did not exceed
about 8 wt %. The separation of nitrogen-containing
species into “acidic”, “basic”, and “pyrrolic” compounds
using the HPLC-BA/AA method was excellent, with ni-
trogen balance of 95% or better.
Further separation of “neutral” compounds involved two
additional steps as shown in Figure 3. The HPLC-NH2
method provides separation of aromatics according to the
number of double bonds in the aromatic ring system with
the simultaneous survey and identification of aromatic ring
cores by the photodiode-array detector. This is illustrated
in Figure 6, which shows an example of the HPLC-UV
map for “aromatics” in a straight-run VGO (-650-1000
OF AEBP). The data show well-resolved ring-type frac-
tions for which the predominant aromatic ring cores,
namely, benzene ( l ) , naphthalene (2), phenanthrene (3),
and pyrene followed by chrysene (4), can be recognized.
(40) Green, J. B.; Hoff, R. J.: Woodward, P. W.; Stevens, L. L. Fuel
1984,63, 1290-1301.
(41)Strachan, M. G.; Johns, R. B. Anal. Chem. 1987, 59, 636-639.
604 Energy & Fuels, Vol. 2, No. 5, 1988 Boduszynski

KERN RIVER PETROLEUM - 13.6”API

Fraction Wt %* I.D. HPLC
O C O F
Homologous Series CnHzn+z
204 4001
Dicyclic 0.182 Wt ‘10
260 500
3 5 6-500F- ---
G3 Monocyclic 0.112 Wt %
Tricyclic 0.252 Wt O/o

Tetracyclic 0.385 Wt VO
h
0
m
W
3 1 2 4 -- 500-65OF- -- Acyclic 0.105 Wt Oh* Pentacyclic
0.399 Wt O/o
e
Hexacyclic
-- ]177-CUT I - 0.217 Wt O h

-
2 427 800 “Compound-Class”
Saturates

e
0
92--CUT 2A --@
- Fractions FlMS
/ \
2 482 900 8.2 - CUT 2 8

7.0 -CUT 3A -@B--

5.7 -CUT 38
i
n CnHzn-6 Tetracycllc Alkanes
g 593 1100 3 4.9 -CUT 4A
E 1.6
f 3.8-CUT 48
-
5

w 649 1200 I c30

0
v)

704 1300
=t
4.4 -CUT 5 ---e3 .n
L

I-”
I
0
L

7 6 0 1400 t 2 13.1-SEF-1
3 4.7-SSEF-2
K
n
0)

- 2.5 - SEF-3
Carbon Number

- 0.4 - SEF-4

*Wt % from Crude Oil

Figure 4. Schematic diagram of the DISTACT-SEF-HPLC-FIMS separation and characterization steps for Kern River petroleum.
Key to HPLC fractions: (1) “saturates”;(2) ”monoaromatics”;(3) “diaromatics”;(4) ‘triaromatics”, (5) “tetraaromatics”;(6) “pentaaromatics
and greater”; (7) “basic”compounds; (8) ”pyrrolic”N compounds; (9) “acidic”compounds.
The effect of alkyl substituents on ring-type separation
is an obvious concern. Results obtained with model com-
pounds indicated the following trend. A paraffinic side
chain slightly decreases the retention time, while a na-
phthenic ring attached to the aromatic ring slightly in-
creases the retention time as compared with the unsub-
stituted aromatic hydrocarbon. When both types of alkyl
substituents are present, there is a tendency for those two
factors to cancel out. However, the effect of alkyl sub-
stituents upon the efficiency of this separation step be-
comes more significant with increasing AEBP. This is
illustrated in Figure 7A,B, which shows HPLC-UV maps
for fractions having progressively higher AEBP. The in-
creasing sample complexity with increasing AEBP and
consequently increasing degree of alkyl substitution on
aromatic rings cause some overlapping between adjacent
ring-type fractions. Analyses of the fractions derived from
distillates covering about 650-1000 O F AEBP range indi-
cate less than 20% overlapping between ring-type frac-
tions. The overlapping between the fractions boiling above
1000 O F AEBP is greater, and it is difficult to quantitate
because of the lack of adequate analytical methods.
The “neutral“ sulfur-, nitrogen-, and oxygen-containing
compounds generally show chromatographic behavior
similar to that of aromatic hydrocarbons having the same
aromatic ring number. The results show, however, that
only the sulfur-containingcompounds are major interfering
species during the separation step. The separation of
saturates from monoaromatics cannot be accomplished by
using the amino-bonded phase column (HPLC-NH2). The
third separation step, involving the HPLC-SIL method,
was required to provide good separation of saturates from
monoaromatics.
Figure 8 illustrates the “compound-class” composition
of Kern River petroleum. This is a composite picture
obtained by plotting the results of chromatographic sep-
arations for fractions having progressively higher AEBP
up to approximately 1400 O F . The data in Figure 8 show
a dramatic decrease in “saturates” (no. 1in Figure 8)with
increasing AEBP. The “saturates” account for 28.3 wt %
of Kern River crude oil. Approximately 41% of the total
“saturates” are represented by distillate cuts boiling below
650 OF AEBP. Almost 92% of the “saturates” do not
exceed lo00 O F AEBP, and 99% do not exceed about 1300
OF AEBP.
The “aromatics”having one through four aromatic rings
(mono, no. 2;di, no. 3; tri, No. 4; and tetra, No. 5 in Figure
8) account for 30.8 wt % of Kern River petroleum and
show a unique distribution pattern. Monoaromatics (no.
2) and diaromatics (no. 3) dominate. The distillation cuts
boiling below 650 OF AEBP contain only about 19% of the
total “aromatics”, while the 650-1000 O F AEBP portion
of this crude oil accounts for almost 62% of the total
“aromatics”. Approximately 98% of the “aromatics” do
not exceed 1300 O F AEBP.
The fraction designated “pentaaromatics and greater”
(no. 6 in Figure 8) was found to contain considerable
amounts of “neutral” nitrogen compounds. This fraction
Composition of Heavy Petroleums
SEF-1 SEF-2 SEF-3
81.o
648.0 Time, Min.
SEF-1 SEF-2 SEF-3
0.0‘
Time, Min. 81.0 648.0
Figure 5. UV/vis profiles for the “solubility” (SEF) fractions
derived from Kern River “nondistillable” residue.
20*ol
225.0
F45.0
Wavelength, nm
398.0 0.0 Time, Min.
Figure 6. HPLC-UV map of “aromatics” in a straight-run
vacuum gas oil: (1) monoaromatics; (2) diaromatics; (3) tri-
aromatics; (4) tetraaromatics.
accounts for 6.5 wt % on a crude oil basis. Approximately
25% of this fraction is represented by the VGO (-
650-1000 O F AEBP), and almost 50% is present in the
1000-1300 OF AEBP range.
The “basic” compounds (no. 7 in Figure 8) account for
7.5 wt % of Kern River petroleum. Data show that about
equal portions of this “compound-class” fraction, each
accounting for 28%, are represented by the 650-1000 and
1000-1300 O F AEBP ranges. The “pyrrolic” N compounds
(no. 8 in Figure 8) account for 10.1 wt % of this crude oil.
The VGO (-650-1OOO O F AEBP) accounts for about 30%
of the total “pyrrolic” N compounds, and the 1000-1300
O F AEBP range accounts for 37%. The “acidic” com-
pounds (no. 9 in Figure 8) account for 7.7 w t % on the
crude oil basis and show the following distribution pattern:
25% in distillates below 1000 O F AEBP, 35% in the
1-1300 O F AEBP range, and 40% in the fraction above
1300 O F AEBP.
The SEF-2, SEF-3, and SEF-4 fractions, derived from
“nondistillable” residue, together account for 7.6 wt % of
Kern River petroleum (see Figures 4 and 5) and by defi-
nition (all three being pentane insoluble) belong to the
Energy & Fuels, Vol. 2, No. 5, 1988 605
so-called pentane asphaltenes, which are conventionally
produced by precipitation. However, there is not a 1:l
correspondence between the above-noted SEF fractions
and the conventional pentane aspha1tenes.l These frac-
tions could not be separated by using the separation
scheme shown in Figure 3 without significant losses due
to irreversible adsorption.
Homologous Series of Compounds. The term ho-
mologous series is used to describe compounds belonging
to the same class and having the same general formula
CnH2rl+zX
where C and H are carbon and hydrogen, respectively, n
is the number of carbon atoms, 2 is the homologous series
value or hydrogen deficiency number, and X refers to
heteroatom(s) (S, N, 0, V, Ni, Fe).
Members of the homologous series differ by the ho-
mologous unit “CH2”(14 m u ) and are referred to as ho-
mologues. The carbon number homologues are defined by
their molar mass/charge value m/z. The isomers are not
distinguished. For example, the acyclic alkanes C5H12(m/z
72), C6H14 ( m / z 86), C7H16( m / z loo), etc., all belong to
the same homologous series, which is described by the
general formula CnH2n+2 and includes n- and isoalkanes
(paraffins).
The concept of “grouping” of compounds on the basis
of their general formulas has been utilized for years in
so-called “type The early method of Brown42
involved mass spectrometry capable of unit resolution up
to only a few hundred in molar mass. Soon, higher reso-
lution mass spectrometers allowed the analyzing of higher
molar mass mixtures.43 However, it became evident that
only a slight increase in the molar mass of the component
molecules in a complex mixture creates a formidable
problem. Namely, compounds with different structures
and belonging to different homologous series can have the
same general formula. This greatly complicates the
analysis. For example, alkylbenzenes and tetracyclic al-
kanes (naphthenes) both share the identical empirical
formula CnH2n-6.
In the case of relatively low-boiling hydrocarbon mix-
tures, the above problem was partly solved by the use of
chromatographic separations prior to the mass spectral
analysis. Separate “type analyses” were devised for aro-
and saturate& fractions. However, those methods
were still incapable of resolving such homologous series of
compounds as, for example, alkylnaphthalenes and tetra-
cyclic monoaromatics (e.g., alkyldecahydropyrenes),both
sharing the same formula CnH2n-12.Efforts have also been
made to develop the high-resolution electron-impact mass
spectrometry (HR EIMS) “type analysis”, which does not
require prior chromatographic separation.& However, all
“type analyses”, by their very nature, cannot provide ac-
curate compositional data. Those methods are also limited
to hydrocarbon mixtures boiling below about 950 O F
AEBP.
In this work, FIMS has been used as a routine method
to describe the HPLC-derived “compound-class” fractions
in terms of “apparent” 2 series and their carbon number
distribution profiles. This characterization approach,
(42)Brown, R. A. Anal. Chem. 1951,23,430.
(43)O’Neal, M. J., Jr.; Wier, T. P. Anal. Chem. 1951,23,830.
(44)Hastings, S. H.; Johnson, B. H.; Lumpkin, H. E. Anal. Chem.
1956,28,1243.
(45)Hood, A.; O’Neal, M. J., Jr. Proceedings of Joint Conference
Organized by Hydrocarbon Research Group, Institute of Petroleum, and
ASTM Committee E-14,University of London, September 24-26,1958.
(46) Gallegos, E. G.; Green, J. W.;Lindeman, L. P.; LeTourneau, R.
L.; Teeter, R. M. Anal. Chem. 1967,39,1833.
(47)Robinson, C. J.; Cook, G. L. Anal. Chem. 1969,37,792.
606 Energy & Fuels, Vol. 2, No. 5, 1988 Boduszynski

A B

1
100.0
592°F 50% AEBP
OF 50% AEBP
80.0 Avg. Carbon Number = 17
Carbon Number = 49

60.0

40.0

20.0

225.0 40.0 )\40.0

Wavelength, nm Wavelength, nm Y
398.0 0.0 Time, Min. 398.0 0.0 Time, Min.

100.0- 100.0-

i
825°F 50% AEBP 1239°F 50% AEBP
80.0. 80.0.
Avg. Carbon Number = 30 Avg. Carbon Number = 62
60.0- 60.0.

40.0. 40.0-

20.0- 20.0-

,‘40.0

Wavelength, nm
398.0 0.0 l i m e , Min. 398.0 0-0 Time, Min.

100.0~
100.01I A 1

I
1‘G
955°F 50% AEBP 1365°F 50% AEBP
80.0-
Avg. Carbon Number = 38 Avg. Carbon Number = 85
60.0-

40.0-

20.0.

225.0% pi
e40.0
Wavelength, nm v
398.0 0.0 Time, Min.
Figure 7. HPLC-UV maps: (A) uaromatics”derived from DISTACT distillate cuts having 50% AEBP of 592 (top), 825 (center),
and 955 O F (bottom); (B) “aromatics”derived from DISTACT distillate cuts having 50% AEBP of 1091 (top) and 1239 OF (center),
and from the SEF-1 fraction having 50% AEBP of 1365 O F (bottom).
utilizing more efficient HPLC separation methods than
those available in the past, allows the unravelling of
changes in composition of heavy oils with increasing AEBP
up to about 1400 OF. Despite some ambiguity in the FIMS
data, the results provide improved understanding of the
immense complexity of these mixtures.
The great advantage of FIMS is that it produces es-
sentially fragmentation-free molecular ion spectra for most
organic compound^.^^^^ This technique has been well
~~
(48) Beckey, H. D. Field Ionization Mass Spectrometry; Pergamon:
Oxford, England, 1971.
(49) Scheppele, S. E.; Grizzle, P. L.; Greenwood, G. J.; Mariott, T. D.;
Pereira, N. B. Anal. Chem. 1976,48,2105.
(50) Scheppele, S. E.; Greenwood, G. J.; Benson, P. A. Anal. Chem.
1977,49,1847.
(51) Buttrill, S . E., Jr. Org. Coat. Plast. Chem. 1980, 43, 330.
(52) Yoshida, T.; Maekawa, Y.; Higuchi, T.; Kubota, E.; Itagaki, Y.;
Yokoyama, S. Bull. Chem. SOC.Jpn. 1981,54,1171.
(53) Yoshida, T.; Maekawa,Y.; Shimada, T. Anal. Chem. 1982,54,967.
(54) Boduszynski, M. M.; Hurtubise, R. J.; Allen, T. W.; Silver, H. F.
Anal. Chem. 1983,55, 232.
Time, Min.
398.0’
established a t SRI International, and it is available on a
routine bash5 However, the nominal-mass resolution of
FIMS ( M / U = -1400) is an obvious limitation.
“Saturates”.Figure 9 shows examples of FIMS molar
mass profiles for “saturates” separated from VGO’s (-
650-1000 OF AEBP) that were derived from three different
crude oils. All spectra have been corrected for the natural
abundance of the carbon-13 isotope. Each spectrum con-
sists of even-numbered molecular ion peaks that corre-
spond to the following “apparent”2 series: CnHk+,, acyclic
alkanes; CnHan,monocyclic alkanes; CnH2n-2,dicyclic al-
kanes; CnH2n-4,tricyclic alkanes; CnHan4,tetracyclic al-
kanes; CnH2n-8,pentacyclic alkanes; CnH2n-10, hexacyclic
alkanes.
The molar mass peaks belonging to each 2 series are 14
amu (CH,) apart. Intensities of the peaks have been
(55) Boduszynski, M. M.; Hurtubise, R. J.; Allen, T. W.; Silver, H. F.
Fuel 1985,64, 242.
(56) Boduszynski, M. M.; Hurtubise, R. J.; Allen, T. W.; Silver, H. F.
Fuel 1986, 65, 223.
Composition of Heavy Petroleums Energy & Fuels, Vol. 2, No. 5, 1988 607
Gravity, Fraction 50% AEBP,
“API “F

~~~oooF-” 451

“500-650°F” 592
-
g 20
722
cut 1
30 -
40 - Cut 2A 825

13.6 S 50 - Cut 28 887

Cut 3A 955
Cut 3 8 1023
Cut 4A 1091
cut 4 8 1158
cut 5 1239

SEF-1 (1365)
SEF-2
1 SEF-3
SEF-4
0 10 20 30 40 50 60 70 80 90 100
Cumulative Wt % from Fraction
Figure 8. “compound-class”distributions in Kern River petroleum: (1) “saturates”;(2) “monoaromatics”;(3) “diaromatics”;(4)
“triaromatics”;(5)”tetraaromatics”;(6)“pentaaromaticsand greater”;(7) “basic” compounds; (8) “pyrrolic” N compounds; (9) “acidic”
compounds; (10) losses.

converted to weight percent concentrations by using rel-
ative ionization sensitivities for each 2 series.
Data in Figure 9 demonstrate that a mere comparison
of weight percent yields of “saturates” (90.5 wt % in VGO
“A”, 36.3 wt % in VGO “B”, and 58.6 wt % in VGO “D”)
has limited analytical value because it does not reveal their
molecular nature. The FIMS results show dramatic dif-
ferences in compositions of the three “saturate” fractions.
The “saturates” from VGO “A”consist predominantly (72.6
wt %) of acyclic alkanes (paraffins, 2 = +2) covering the
carbon number range from C1, to about C4@Note that
isomers are not distinguished. Each molar mass peak may
be represented by a large number of isomers. (See Table
I.) The cycloalkanes (naphthenes) in this sample include
mainly monocyclics (2 = 0, 18.7 wt %) and dicyclics (2
= -2, 5.8 wt %). The higher ring-number naphthenes
account for a very small portion of this sample. (See Figure
9, top.)
Conversely, the “saturate” fraction from VGO “B” con-
tains mainly naphthenes and only 3.5 wt 90 of paraffins
(2 = +2). Tetracyclic naphthenes (2 = -6) are most
abundant in this sample (37.5 w t %) and show an unusual
carbon number distribution with peaks a t m l z 372,386,
400, and 414 that are characteristic of C27-C30steranes.
(See Figure 9, center.) Those were confirmed by the
fragmentation pattern in the EIMS spectra. The third
sample of “saturates”,derived from VGO “D” (Figure 9,
bottom), contains relatively high concentrations of both
paraffins and naphthenes.
Figure 10 illustrates the distribution of alkane homolo-
gous series in Kern River petroleum as a function of AEBP.
This immature and biodegraded crude oil has a very low
concentration of acyclic alkanes (paraffins), and it is rich
in cycloalkanes (naphthenes). The relatively high con-
centration of tetra- and pentacyclics (no. 5 and no. 6 in
Figure 10, respectively) is of particular interest because
those include such “biomarkers”as tetracyclic steranes and
pentacyclic triterpanes (e.g., hopane~).~’
(57) Tissot, B. P.; Welte, D. H. Petroleum Formation and Occurrence,
2nd ed.; Springer-Verlag: West Berlin, 1984.
The use of HPLC and FIMS for characterization of
“saturates” is particularly attractive for high-boiling pe-
troleum fractions. This is illustrated in Figure 11, which
shows an unusual carbon number distribution pattern for
tetracyclic naphthenes in high-boiling cuts derived from
Kern River petroleum. (See also Figure 10 for the dis-
tribution of the alkane homologous series.) Interestingly,
the FIMS data in Figure 11 reveal the presence of tetra-
cyclics covering the carbon number range from CI9 ( m l z
260) to CS (mlz 1198). These data are significant as they
greatly exceed the carbon number reported for
“biomarkers” and the carbon number that is amenable to
GC/MS (-<C40).
The nominal-mass resolution FIMS, however, does not
distinguish acyclic alkanes (CnH2n+2) from heptacyclics
(CnH2n-12), monocyclics (C,H,) from octacyclics (CnH%44),
dicyclics (CnH2,) from nonacyclics (CnH2n-16), and so on
(AM = 0.0938 amu). The resolution required to separate
those homologous series a t m l z 554 is about 6000. The
polycyclic naphthenes having seven or more rings, although
likely to be present in petroleum, would account for a very
small portion of the total “saturates”. (See Figure 10.)
It is also important to note that characterization of
“saturates” using FIMS requires a very good separation
of this “compound-class”fraction from monoaromatics. As
discussed earlier, different classes of compounds may be
described by the same general formula. For example,
tetracyclic alkanes and alkylbenzenes have the same gen-
eral formula, CnH2n-s.
“Aromatics”. In the past, the characterization of aro-
matic hydrocarbons in terms of the homologous series of
compounds using MS has been limited due to the fact that
the 2 value in the general formula CnH2n+Z is affected by
both the number of rings and the number of double bonds.
The 2 value can be calculated as follows:
2 = 2[1- ( R + DB)]
where R is the number of rings and DB is the number of
double bonds.
Thus, compounds having the same sum of rings and
double bonds cannot be distinguished. This is illustrated
by the following examples:
608 Energy & Fuels, Vol. 2, No. 5, 1988
CnH2n-12
The HPLC-NH, separation of “aromatics” according to
the number of double bonds in the aromatic ring system
with the simultaneous monitoring of the aromatic ring
cores by the photodiode-array detector greatly facilitates
the FIMS analysis of the ring-type fractions. However,
the limitations of the nominal-mass resolution FIMS for
analysis of “aromatics” are much more pronounced than
in the case of “saturates”. This is mainly due to the
presence of sulfur-containing species, which are rather
abundant in many heavy crudes and which cannot be
separated from aromatic hydrocarbons. This problem is
illustrated by the following examples:
C n H z n -20
m/z 21%
R=H
The homologous series of alkylbenzenes (CnHzn+),
benzothiophenes (C,H2,-loS), and monocycloalkyl-
phenanthrenes (C,H2,-20) would be represented in the
FIMS spectrum of the total aromatic fraction (unseparated
by ring number) by the same “apparent” 2 series with
peaks a t m / z 218, 232, 246, ..., etc. However, the results
of chromatographic separations using the HPLC-NH2
method showed that for samples covering the VGO boiling
range (-650-1000 O F AEBP), benzothiophenes (four
double bonds) are coeluted with diaromatics (five double
bonds) rather than monoaromatics (three double bonds)
and are readily separated from triaromatics (seven double
bonds). This somewhat facilitates interpretation of the
FIMS data.
LV/HR EIMS with a relatively moderate resolving
power of about 2500 at mlz 218 can easily distinguish both
C,H,,loS and C,H2n-20series from C,HZn+ (AM= 0.0905
amu and AM = 0.0938 amu, respectively) without necessity
for chromatographic separation. However, a very high
resolving power is needed (at m / z 218, a resolution of
about 64 000) to distinguish C,H2,-loS from the CnH2n-20
series (AM = 0.0034).8 The HPLC ring-type separation
can easily resolve these two series of compounds, and thus
it can help LV/HR EIMS analysis.
Figure 12 shows FIMS results for mono-, di-, and tri-
aromatics, respectively. The HPLC fractions were derived
from two different VGO’s. The compositions of the HPLC
fractions are given in terms of the “apparent” Z series.
Intensities of the peaks were converted into weight percent
concentrations assuming equal ionization sensitivities for
components of a given fraction. The overwhelmingwealth
of information provided by this analytical approach is
evident.
The results in Figure 12 reveal dramatic differences
between the two feedstocks. The monoaromatics (Figure
12, top) in VGO “A” account for only 4.3 wt % of the oil.
The most abundant homologous series (2 = -6) is repre-
Bodusz ynski
sented by 39.1% of the total monoaromatics (1.68 w t %
from VGO “A”). This homologous series covers the carbon
number range from C19to C&and may involve an immense
number of isomers. The monoaromatics in VGO “B” ac-
count for 16.1 wt % of the oil. The series 2 = -10 (24.2%)
and 2 = -12 (22.0%) are most abundant. These series
involve tri- and tetracyclics having one aromatic ring, re-
spectively.
The data in Figure 12 (center) show that the concen-
tration of diaromatics in VGO “B” is almost 7-fold higher
than in VGO “A”. The most abundant diaromatic ho-
mologous series in VGO “A” and VGO “B” are 2 = -12
(33.8% of the total diaromatics) and 2 = -16 (26.3% of
the total diaromatics), respectively. Data in Figure 12
(bottom) show that the VGO “B”also contains much more
triaromatics (8.4 wt %) than VGO “A” (1.0 wt %). The
high concentration of tetracyclic species in VGO “B”is also
evident in the FIMS data for triaromatics. The homolo-
gous series 2 = -20 accounts for 37.6% of the total tri-
aromatics in this oil and shows an unusual bimodal dis-
tribution pattern, which implies the presence of triaro-
matized steranes.
The use of HPLC-NH2and FIMS is very attractive for
characterization of high boiling fractions. Figure 13A,B
shows FIMS molecular weight profiles for triaromatic
fractions isolated from a series of DISTACT cuts having
progressively higher AEBP extending from about 800 OF
up to about 1150 O F 50% AEBP. In order to understand
better the immense complexity of high-boiling petroleum
fractions, it is worthwhile noting that the concentrations
of these triaromatic fractions on the crude oil basis range
from 0.253 wt % for the “804 O F 50% AEBP” cut (Figure
13A, top) to 0.063 wt % for the “1153 O F 50% AEBP” cut
(Figure 13B, bottom). The spectra in Figure 13A,B have
not been corrected for the 13Cisotope effect and show both
even- and odd-numbered peaks. Despite the complexity,
the data reveal the most prominent series to be 2 = -22,
which apparently involves pentacyclics having the phen-
anthrene aromatic ring core. The unique distribution
pattern covering the carbon number range from Czl to C,l
implies a “biomarker” origin. These results are significant
as they greatly exceed the carbon number reported for
“biomarkers”. The 2 = -22 series is most likely repre-
sented by partially aromatized pentacyclic triterpenes
having the following ring structure:
&R CnH2n - 2 2
R=H mJz 272
R = CbHll mJz 342
R = CloHzl mJz 412
R = c&I1ol mlz 972
The components of the 2 = -22 series in the “1153 O F
50% AEBP” cut (Figure 13B, bottom) together account
for only about 150 ppm by weight on the crude oil basis.
Concentrations of other “apparent”2 series (resolved a t
nominal mass resolution) in this high-boiling distillate are
also at a “ppm level” and may include compounds be-
longing to different “specific” Z series (requiring a high
resolution). The immense complexity of heavy oils is ev-
ident.
Nitrogen Compounds. It is obvious from the earlier
discussion on “average molecular parameters” that in the
Composition of Heavy Petroleums Energy & Fuels, Vol. 2, No. 5, 1988 609
3.6
Concentration, Wt %
Apparent

0 +2 72.6 65.0
s 0
-2
10.7
5.0
16.9
5.2
.-s;
L
u)
-4
-6
1.6
1.o
I .4
0.9
c -0 0.3 0.3

-
Q
C -10 0.0 0.0
c

--
E
0
n = 20 T n = 30
Total: 100.0 90.5

L
0
a
I-
ill.
a
i 300 400 500 600 700 800
Mass, mlz
4.8
tGO "B" t
Concentration, Wt %
Apparent
? " 2 Series Saturates
0 +2 3.5 1.3
s n = 30 0 7.6 2.0
t -2 13.3 4.0
.-s;
U
u) 0
-4
-6
18.1
37.5
6.6
13.6
C -8 16.4 5.9
-
0,
L
c -10 3.6 1.3

-
c
0
* I t
Total: 100.0

I-

0
200 300
i
400 500 600 700 800
Mass, m/z
VGO 'ID" n = 205 Concentration, Wt %
Apparent
? ? " 2 Series
i ,
0 +2
n = 30 0
0
-2
-4
0 0 -6 12.6
-0
-10

I " " " " ' I ' "' ' ' ' ' ' ' I " " " I ' ' I I
IO 300 400 500 600 700 800
Mass, m/z
Figure 9. lSC-correctedFIMS of "saturates"in three different vacuum gas oils.
case of heavy oils having high concentrations of hetero- resolving power with a low-voltage mode of operation and
atom-containing species, there is a need for both high- the presence of almost exclusively molecular ions (little
efficiencv "comr>ound-class"semrations and hiah-resolu-
tion mas; spectial analysis. L\i/HR EIMS witG dynamic (58) Lumpkin, H. E.; Aczel, T. In High Performonce Mass Spectrom-
resolving power UP about 50000 h a been used etry: Chemical Applications; ACS Symposium Series 70, American
in various A combination of the high Chemical Society: Washington, DC, 1976; pp 261-273.
610 Energy & Fuels, Vol. 2, No. 5, 1988 Boduszynski
Gravity, Fraction 50% AEBP,
“API “F

“500°F-” 451

“500-650°F” 592

cut 1 722

Cut 2A 825

Cut 28 887
Cut 3A 955
Cut 38 1023
Cut 4A 1091
Cut 48 1158
cut 5 1239

(1365)

SEF-3
SEF-4
0 10 20 30 40 50 60 70 80 90 100
Cumulative Wt O/O from Fraction
Figure 10. Distributions of alkane homologous series in Kern River petroleum: (1) acyclic alkanes; (2) monocyclics; (3) dicyclics;
(4) tricyclics; ( 5 ) tetracyclics; (6) pentacyclics; (7) hexacyclics.

722°F 50% AEBP 1023°F 50% AEBP 1239°F 50% AEBP

1030

I
1002

200 300 400 500 300 400 500 600 700 800 900 500 600 700 800 900 100011001200
Mass, m/z Mass, mJz Mass, m/z
Figure 11. Carbon number distributions of tetracyclic alkanes in FIMS of ”saturates” derived from high-boiling distillates of Kern
River petroleum.

or no fragmentation) greatly facilitates the analysis. When
combined with chromatographic separations this technique
provides extremely valuable compositional i n f ~ r m a t i o n . ~ ~ NMR, HR EIMS, and GC/MS methods demonstrate that
However, the LV/HR EIMS method is not free of com-
plications. The presence of 13Cisotope peaks, which cause
considerable peak broadening, complicates the analysis.e2
For example, to resolve C22H27N( m / z 305.21434) from
13CC21H,6N( m / z 305.20987), a resolution of -68 200 is
required (AM = 0.00447).
Characterization of heteroatom-containing species using
nominal-mass resolution FIMS has obvious limitations.
However, results of this work revealed that chromato-
graphic separations of nitrogen-containing compounds can
greatly simplify this problem. FIMS spectra (Figure 14)
indicate that the majority of N compounds in the VGO
boiling range (-650-1000 OF AEBP) is represented by
(59) Aczel, T.; Bradley, P. K.; Colgrove, S. G.; Scheppele, S. E.;
Grindstaff, Q.G.; Sturm, G. P.Proceedings of ASMS, 31st Annual
Conference on Mass Spectrometry and Allied Topics;ASMS: East
Lansing, MI, 1983; pp 448-449.
(60) Schmidt, C. E.; Sprecher, R. F.; Batts, B. D. Anal. Chem. 1987,
59, 2027-2033.
(61) Sturm, G. P., Jr.; Green, J. B.; Tang, S. Y.;Reynolds, J. W.; and
Yu, S. K-T. Prepr.-Am. Chem. Soc., Diu. Pet. Chem. 1987, 369-378.
(62) Gallegos, E. J., Chevron Research Co., unpublished data.
mononitrogen compounds. Furthermore, the results of
extensive analyses of the HPLC fractions using FTIR, 14N
the HPLC-BA/AA method provides excellent separation
of the N compounds having the >NH group in a five-
membered heterocycle (“pyrrolic” N compounds) from
those in a six-membered nitrogen heterocycles (pyridines,
quinolines, etc.; “basic” compounds) and from the N com-
pounds having oxygen-containing functional groups such
as >C-0 and/or -OH (“acidic” compounds). The dis-
tribution of those “compound-class” fractions in the ni-
trogen-rich Kern River petroleum was discussed earlier.
(See Figure 8: “basic” compounds (no. 71, “pyrrolic” N
compounds (no. a), and “acidic” compounds (no. 9).)
Figure 14 shows FIMS spectra of the “pyrrolic” (top) and
“basic” (bottom) N compounds separated from a
straight-run VGO (-650-1000 O F AEBP). Both spectra
have been corrected for the 13Cisotope and consist almost
exclusively of odd-numbered peaks.
These spectra are consistent with the results of other
analyses of these fractions (elemental analysis, FTIR, LV
EIMS), which indicate that both fractions are essentially
100% concentrates of mononitrogen compounds.
Two homologous series have been marked in each
spectrum in Figure 14 to facilitate the discussion of the
Composition of Heavy Petroleums Energy & Fuels, Vol. 2, N o . 5, 1988 611

VGO “A” VGO “B”

2.0
n = 29
3?6
39.1 1.86

I
0 4
-8 25.0 1.07
tli T l t t -10

0
200 300 400 500 600 700 800 200 300 400 500 600 700 800
Mass, m/z Mass, m/z

380. ,n=28 Comontntlon, WI % 376t n=28

s
-
2;
I

m
* I I I
Q
C

-24
0.17
0.10
0.08
-
c
C
C -24
1.25
0.w
--m
e,

2
n=48 n=45

200 300 400 500 600 700 8 800

Mass, m/z Mass, m/z

“2”!
h h
0 -18 29.5 0.295
-26 30.1 0.301
-22 17.4 0.174
-24 6.7 0.M7
-28 5.2 0.052
-28 3.8 0.038

-
0
C

ell
L

c
0

n=47
0
260 300 400 500 600 700 200 300 400 500 600 700 800
Mass, m/z Mass, mlz -
Figure 12. 13C-correctedFIMS of “monoaromatics”(top),“diaromatics” (center), and “triaromatics” (bottom) derived from two different
vacuum gas oils.

data. These are C,H2n-15N and CnH2n-17N.The 2 = -15N
series in the top spectrum of “pyrrolic” N compounds
includes carbazoles covering the carbon number range from
C14 (C2-alkylcarbazole with the initial peak a t m / z 195)
to C30 (C17-alkylcarbazole; a peak at m / z 419). The
structural assignment for this homologous series is rather
straightforward and is supported by the results of HR
EIMS, GC/MS, FTIR, and 14N NMR ana lyse^.^^^^^
The 2 = -15N series in the spectrum of “basic” N com-
pounds involves tetracyclic compounds having nitrogen in
a six-membered pyridine ring as shown in Figure 14,
bottom. Obviously, many isomeric types are possible. The
2 = -15N series covers the carbon number range from c16
(the initial peak a t m/z 223) to about C40( m / z 559).
The 2 = -17N series in FIMS of “pyrrolic” N com-
pounds (Figure 14, top) extends from the initial peak a t
m / z 221 to a peak a t m / z 417 and can be assigned to
C1&30 cycloalkylcarbazoles (e.g., tetrahydrobenzo-
carbazoles). The 2 = -17N series in the FIMS analysis
of “basic” N compounds (Figure 14, bottom) covers the
carbon number range from C15(the initial peak at m/z 207)
to C40 ( m / z 557) and involves triaromatic N heterocycles
such as benzoquinolines.
The problem is much more complicated in the case of
molecules having several heteroatoms, e.g., NS, NO, N2,
etc. It has been our experience that a complete analysis
of the “acidic” compounds (see Figure 3 and Figure 8,
fraction no. 9) in the 650-1000 O F AEBP range in terms
of the specific homologous series is not possible using about
50000 resolution EIMS.62 The components of this
“compound-class” fraction (if derived from VGO) are
usually composed of nitrogen-, oxygen-, and sulfur-con-
taining compounds for which the distribution of hetero-
atoms in a molecule is not known.
As demonstrated in this discussion, the complexity of
high-boiling ( 1000 O F + AEBP) petroleum fractions is
N

so immense that characterization in terms of homologous

(63) Wilson, D.M.,Chevron Research Co., unpublished data.
(64) Thermodynamic Tables, Hydrocarbons; API Research Project
No. 44; TRC-Thermodynamic Research Center: Texas A&M University;
College Station, TX, 1974; Vol. V, Tables 23-2-(1.101)-k, pp k-10.10-k-
10.60.
series and their carbon number distributions would require
the use of (1) highly selective and efficient physical sep-
arations, combined with (2) an extremely high-resolution
mass spectrometry capable of resolving “specific” 2 series
612 Energy & Fuels, Vol. 2,No. 5, 1988 Boduszynski
A 804°F 50% AEBP B 1026°F 50% AEBP
2.5 1.2
I 622
/ 692

6 1

100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000
Mass, mlz Mass, mlz
887O F 50% AEBP 1100°F 50% AEBP
0.8 692x
’ 762
-
i,, [552
Q
L
0
I-
C
622
, 0
6, L 832
I (

342, ,622 C
Q,

F
n I9O2

0
100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000
Mass, mlz Mass, mlz

1.o
955°F 50% AEBP
- 622
I 0.6 I
1153°F 50% AEBP
790-1 832

-
Q
c
I 02
I-
I iii I I C 692
0
972
L
C ik
F 622.
n

0, I ” ’ ” ’ I ’ ’ o i
100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1( 00
Mass, mlz Mass, mlz
Figure 13. FIMS of “triaromatics”: (A) derived from DISTACT distillate cuts having 50% AEBP of 804 (top), 887 (center), and
955 O F (bottom); (B) derived from DISTACT distillate cuts having 50% AEBP of 1026 (top), 1100 (center), and 1153 OF (bottom).

and, preferably, producing only molecular ions’(e.g., HR

FIMS or HR FDMS coupled with Fourier transform MS).
Unfortunately, such a technique is not currently available.
Conclusions
The results of this study help unravel the immense
complexity of heavy petroleums. The data show that the
molecular characterization of entire, unfractionated heavy
oils is beyond current analytical capabilities. This is
mainly due to the great diversity of molecular types in-
volving numerous homologous series of compounds, each
series having a very wide carbon number distribution and
very low relative concentrations of individual carbon
number homologues.
The data presented here demonstrate the need for and
the importance of “volatility” and “solubility” (DIS-
TACT-SEF) fractionations of heavy crudes to produce
operationally well-defined fractions having progressively
higher AEBP’s, which can be then analyzed and compared
on a consistent basis. The choice of a characterization
approach depends upon the AEBP range, the detail of
information required, and the available time and resources.
The results also illustrate the importance of HPLC
separations for preconcentrating and isolating compounds
of interest from a complex sample matrix. The data
presented in the previous paper in this series1reveal that
compounds comprising distillable fractions (below 1300
OF AEBP) account for the major portion of heavy petro-
-
leums. This portion of heavy crudes involves numerous
homologous series of compounds, each covering a wide
carbon number range extending up to about Cleo or so.
The concentrations oE carbon number homologues, each
represented by an immense number of isomers, decrease
rapidly with increasing AEBP from a few tenths of 1 w t
% to less than 1 ppm by weight on a crude oil basis.
The molecular characterization of the distillable portion
of heavy petroleums (- 1300 OF- AEBP) in terms of the
“apparent” 2 series can be accomplished by using the
HPLC and FIMS approach described in this paper.
However, as the concentration of heteroatom-containing
molecules increases with increasing AEBP, the ambiguity
of the data becomes greater. Characterization in terms of
“specific” 2 series requires the use of both HPLC and
high-resolution MS. The characterization is difficult and
tedious, and it is currently limited (at a resolution of
50000) to fractions not exceeding about 900 OF AEBP.
-
The results show that characterization of “nondistillable”
residues (-1300 OF+ AEBP) is greatly facilitated by
fractionation of this portion of heavy petroleum into a set
of well-defined “solubility”fractions (SEF), which can be
then analyzed and compared on a consistent basis. Data
show that the SEF-1 (pentane-soluble) fractions can be
Composition of Heavy Petroleums Energy & Fuels, Vol. 2, No. 5, 1988 613
5.2( un277
= 20 heavy crude components, comprising “nondistillable”
“Pyrrolic” N-Compounds
residues (-1300 O F + AEBP), may actually involve com-
s CnHzn-ld pounds having “on average” about 400 carbon atoms per
.-i
L
@ p R
molecule, of which about 180 carbon atoms are in aromatic
In
a
H R = C2H5,m/z = 195 rings and contain almost 30 heteroatoms per molecule,
-
L
C
C
,I 1111,lI CnHZn-lTN
depending on crude oil origin. However, these data should
be interpreted with caution because it is likely that mo-
-m lecular weight measurements by VPO in pyridine are still
L

c
0 H R = H, m/z = 221 erroneously high due to intermolecular associations.1°
Characterization of the SEF fractions in terms of the ho-
0
mologous series of compounds is beyond the capabilities
100 200 300 400 500 600 7 of currently available methods.
Mass, m/z Acknowledgment. Special thanks are due to R.
Malhotra and G. St. John of SRI International for their
1.60 assistance in providing FIMS data. I express my appre-
p p
333 n = 2 4
“Basic” N-Compounds ciation to C. E. Rechsteiner and T. C. Crabtree for their
assistance in processing FIMS data, to E. J. Gallegos for
s providing HR EIMS and GC/MS data, to D. M. Wilson
.->;
c
In for 14NNMR data, to D. C. Young and M. Kang for 13C
Q
NMR data, to X. Urrutia for FTIR data, to F. Su, A. F.
-
L
C
C Bentley, and G. C. Daugherty for VPO measurements, and
-m to T. H. Attoe, R. J. Clay, and J. R. Richter for their
L
0 technical assistance. I am grateful to Chevron Research
c
Co. for supporting this research and allowing publication,
of this paper.
0
100 200 300 400 500 600 7 Glossary
Mass, m/z
AEBP atmospheric equivalent boiling point
Figure 14. 13C-correctedFIMS of (top) “pyrrolic”N compounds AR atmospheric residue
and (bottom) “basic”N compounds derived from a vacuum gas DISTACT short-path distillation apparatus
oil. EIMS electron-impact mass spectrometry
FDMS field-desorption mass spectrometry
separated into “compound-class” fractions by using the FIMS field-ionization mass spectrometry
HPLC separation scheme described in this paper with FTIR Fourier transform infrared spectrometry
recoveries of 90 w t % or better. The HPLC separation GC/MS gas chromatography/massspectrometry
of the SEF-2, SEF-3, and SEF-4 fractions is not recom- HPLC high-performance liquid chromatography
mended because of large losses of material due to irre- HR EIMS high-resolution electron-impact mass spec-
versible adsorption on the columns currently available. trometry
Characterization of the SEF fractions in terms of the LV/HR EIMS low-voltage/high-resolution electron-impact
“average structural parameters”reveals that the complexity mass spectrometry
of those fractions increases rapidly with decreasing solu- m/z molar mass over charge
14NNMR nitrogen-14 nuclear magnetic resonance
bility. However, the “average molecular parameters” de- spectrometry
pend upon molecular weight measurementsthat are greatly SEF sequential elution fractionation
affected by intermolecular associations. Data based on VGO vacuum gas oil
average molecular weight values determined by VPO in z homologous series value (hydrogen deficiency
pyridine indicate that a relatively small percentage of number) in the general formula C,H2,+ZX