Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Energy & Fuels 2002, 16, 1571-1575 1571

Rapid and Accurate SARA Analysis of Medium Gravity


Crude Oils
Tianguang Fan and Jill S. Buckley*
New Mexico Petroleum Recovery Research Center, New Mexico Institute of Mining and
Technology, Socorro, New Mexico

Received May 31, 2002

Crude oils can be described compositionally by a number of methods. SARA analysis is widely
used to divide crude oil components according to their polarizability and polarity using a family
of related analytical techniques. Problems arise because the analytical techniques do not
necessarily produce identical results. Users of the data, however, rarely distinguish between the
different techniques, assuming that SARA fraction values generated by any of the commonly
used methods are essentially interchangeable. We examine this assumption for medium gravity
crude oils and three SARA analysis methods: gravity-driven chromatographic separation, thin-
layer chromatography (TLC), and high-pressure liquid chromatography (HPLC). Results for a
suite of six crude oil samples show that a significant volume of volatile material that contains
both saturates and aromatics is lost in the TLC analysis. An improved HPLC method is introduced
that gives analyses comparable to the ASTM-recommended chromatographic method in less time
than that required for TLC analysis. An internal consistency test is recommended for evaluating
SARA fraction data.

Introduction methods, first introduced by Suatoni and Swab.3 Early


HPLC techniques used silica or alumina columns to
Analysis of the composition of crude oils can be separate lighter petroleum fractions. The developments
endlessly complex; the amount of detail collected should in preparation of the bonded phase of HPLC columnss
be dictated by the application for which the data is especially NH2-bonded materialssmade it practical to
needed. One simple analysis scheme is to divide an oil separate heavier fractions of petroleum samples.4-8
into its saturate, aromatic, resin, and asphaltene (SARA)
HPLC techniques are faster, more reproducible, and
fractions. The saturate fraction consists of nonpolar
more readily automated than the ASTM column tech-
material including linear, branched, and cyclic satu-
nique. In both cases, however, it is necessary to remove
rated hydrocarbons. Aromatics, which contain one or
the asphaltene fraction before proceeding with the
more aromatic rings, are more polarizable. The remain-
ing two fractions, resins and asphaltenes, have polar chromatography. Asphaltenes are either irreversibly
substituents. The distinction between the two is that adsorbed or precipitated during the saturate elution
asphaltenes are insoluble in an excess of heptane (or step, and quantitative recovery cannot be achieved.9
pentane), whereas resins are miscible with heptane (or The fastest separation method uses thin-layer chro-
pentane). This classification system is useful because matography (TLC) on quartz rods that are coated with
it identifies the fractions of the oil that pertain to sintered silica particles. Unlike the column and HPLC
asphaltene stability; it thus should be useful in identify- techniques, asphaltenes need not be separated from
ing oils with the potential for asphaltene problems. other crude oil components before chromatographic
SARA analysis began with the work of Jewell et al.1 analysis. A popular technology known as the Iatroscan
Three main approaches have been used to separate that combines TLC with flame ionization detection
crude oils and other hydrocarbon materials into SARA (TLC-FID) was first applied by Suzuki10 to automate
fractions. A clay-gel adsorption chromatography method quantitative SARA separations, a method which has
is the basis of ASTM D2007-93.2 This method requires
a fairly large oil sample, is time-consuming and difficult (3) Suatoni, J. C.; Swab, R. E. J. Chromatogr. Sci. 1975, 13, 361-
to automate, and requires large quantities of solvents. 366.
(4) Miller, R. Anal. Chem. 1982, 54, 1742-1746.
Improved methods fall into two groups. In the first (5) Radke, M.; Willsch, H.; Welte, D. H. Anal. Chem. 1984, 56, 2538-
group are high-pressure liquid chromatographic (HPLC) 2546.
(6) Grizzle, P. L.; Sablotny, D. M. Anal. Chem. 1986, 58, 2389-2396.
(7) Félix, G.; Thoumazeau, E.; Colin, J. M.; Vion, G. J. Liq.
* Corresponding author. Fax: 1-505-835-6031. E-mail: jill@ Chromatogr. 1987, 10, 2115-2132.
prrc.nmt.edu. (8) Chaffin, J. M.; Lin, M. S.; Liu, M.; Davison, R. R.; Glover, C. J.;
(1) Jewell, D. M.; Weber, J. H.; Bunger, J. W.; Plancher, H.; Latham, Bullin, J. A. J. Liq. Chromatogr., Relat. Technol. 1996, 19, 1669-1682.
D. R. Anal. Chem. 1972, 44, 1391-1395. (9) McLean, J. D.; Kilpatrick, P. K. Energy Fuels 1997, 11, 570-
(2) ASTM D2007-93: “Standard Test Method for Characteristic 585.
Groups in Rubber Extender and Processing Oils by the Clay-Gel (10) Suzuki, Y. 21st Annual Meeting of the Japan Society for
Adsorption Chromatographic Method,” ASTM, 1993. Analytical Chemistry, 1972; 47 (in Japanese).

10.1021/ef0201228 CCC: $22.00 © 2002 American Chemical Society


Published on Web 10/29/2002
1572 Energy & Fuels, Vol. 16, No. 6, 2002 Fan and Buckley

Table 1. Crude Oil Sample Properties


density at 20 °C MW n-C7 asph
oil °API (g/cm3) (g/mol) RI at 20 °C PRI (%)
A-95 25.2 0.8956 236 1.5128 1.4513 8.7
C-LH-99 22.6 0.9161 268 1.5137 1.4231 2.8
C-R-00 31.3 0.8673 235 1.4851 1.4444 1.9
S-Ven-39 28.8 0.8795 240 1.4976 1.4465 5.8
SQ-95 37.2 0.8409 213 1.4769 1.4223 1.3
Tensleep-99 31.1 0.8685 270 1.4906 ∼1.44 4.1

since been used extensively.11,12 Barman13 compared


SARA analyses of heavy hydrocarbon distillates by the
clay-gel and TLC-FID methods. TLC-FID uses very
small amounts of sample. SARA fractions in a crude oil
sample are often well-resolved using established devel-
opment procedures. Quantitative results are obtained
from peak areas, assuming that each SARA fraction has
an identical FID response factor.
Comparisons of SARA fraction measurements by
different techniques, usually from different laboratories,
can show large differences. To some extent, these Figure 1. Separation of paraffins from single- and double-
differences might be realscaused by the use of different ring aromatics.
samples. In this work, identical samples were examined
by several different techniques to illuminate the with a model 203 PeakSimple data system (SRI Instrument),
was used to scan silica-coated quartz rods (Chromarod-SIII,
strengths and weaknesses of each. An improved HPLC
Iatron Laboratories). The rods were 15.2 cm long and 1.0 mm
technique and a test for internal consistency in SARA in diameter, with a uniform coating of 5.0 µm silica particles
data are presented. (pore diameter 60 Å). The FID detector was operated with a
pure grade of hydrogen at a flow rate of 160 mL/min; air at a
Experimental Materials and Methods flow rate of 2.0 L/min was supplied by a pump; scan speed
was 60s/scan.
Crude Oils. Six medium-gravity, dead crude oils, varying Crude oil samples were dissolved in HPLC grade dichloro-
in API gravity from 22.6 to 37.2°, were used in this study. methane (DCM) at a concentration of 20 mg/mL. A 1 µL
Selected physical and chemical properties are shown in Table repeating syringe (Alltech Associate, Inc.) was used to spot
1 (reproducible to at least (5 in the least significant digit 10 to 20 µg of sample on freshly activated rods. Development
shown for each measurement). Densities were measured using steps included exposure to HPLC grade hexane for 30 min,
a Mettler/Paar DMA 40 with a circulating water bath for HPLC grade toluene for 10 min, and a 95:5 mixture of DCM
temperature control. API gravities were calculated from and methanol for 4 min. The rods were dried in air for 3 min
measured densities, corrected to 60 °F. Average molecular between solvent exposures. This treatment produced four well-
weight (MW) was determined by freezing point depression resolved peaks representing saturates, aromatics, resins, and
(Precision Systems Cryoscope 5009). Kinematic viscosities asphaltenes. Evaporative losses during the development steps
were measured in Cannon-Fenske viscometers and converted were evaluated by tests of similarly treated preparative thin-
to absolute viscosity units. The amounts of asphaltene pre- layer chromatography plates for which changes in weight could
cipitated by n-heptane (1 g oil:40 mL heptane) ranged from be measured and by comparison of peak sizes for undeveloped
1.3 to 8.7%. Refractive index (RI) was measured with an Index rods.
Instruments GPR11-37 automatic refractometer. PRI is the HPLC. The HPLC chromatographic separation system
RI of a mixture of oil and the least amount of heptane in which consisted of a Model 110A pump (Beckman), an R401 dif-
asphaltene aggregates can be observed microscopically (at a ferential refractometer (Waters), a U6K universal injector
magnification of about 320×). The difference between RI of (Waters), a Model 7040 high-pressure switching valve (Rheo-
the oil sample and PRI is a measure of asphaltene stability.14 dyne), and a Waters 486 UV detector (Millipore). Analogue
ASTM Column Separation. Each of the crude oils was signals from the RI detector were interfaced to the model 203
tested using the full ASTM-recommended procedure (ASTM PeakSimple data acquisition system (SRI Instrument). Two
D2007-932) using n-hexane to separate the asphaltenes. The 3.9 × 300 mm µBondapak NH2 columns with 10 µm packing
ASTM procedure is a chromatographic separation of the non- (Waters) were used in series for chromatographic separation
asphaltic oil components through two columns: an Attapulgite of crude oil. The UV detector was operated at a wavelength of
clay-packed column adsorbs the resins and a second column 254 nm to monitor elution of each fraction.
packed with activated silica gel separates aromatics from the Selectivity of the columns was tested using mixtures of
saturate fraction. A 50:50 mixture of toluene and acetone is known compounds. Mixtures of n-decane (C10), n-tridecane
used to recover the resin fraction from the clay packing. The (C13), n-pentadecane (C15), and n-octadecane (C18) in hexane
aromatics can be recovered by Soxhlet extraction of the silica eluted as a single peak, as did hexane solutions of toluene and
gel in hot toluene. Volatile components lost during the process 1,3-diisopropylbenzene (DIPB). Hexane solutions of a single-
are calculated by weight difference. ring aromatic (DIPB) with a two-ring compound (1-methyl-
TLC-FID. An MK-5 Iatroscan (Iatron Labs Inc., Tokyo), naphthalene or 1-MN) eluted as two well-resolved peaks.
equipped with a flame ionization detector (FID), interfaced Mixtures of the hexane solutions of DIPB, 1-MN, and C15
eluted as three distinct peaks (Figure 1).
(11) Karlsen, D. A.; Larter, S. R. Org. Geochem. 1991, 17, 603-617. Crude oil (1-mL) was weighed and dissolved in 40 mL of
(12) Vela, J.; Cebolla, V. L.; Membrado, L.; Andres, J. M. J. hexane in an open-top screw cap vial with a Teflon/silicon
Chromatogr. Sci. 1995, 33, 417-424.
(13) Barman, B. N. J. Chromatogr. Sci. 1996, 34, 219-225.
septum. After 48 h, a sample of maltenes dissolved in hexane
(14) Buckley, J. S.; Hirasaki, G. J.; Liu, Y.; Von Drasek, S.; Wang, was withdrawn through a 0.2 µm PTFE syringe filter. The
J. X.; Gill, B. S. Pet. Sci. Technol. 1998, 16, 251-285. withdrawn maltene solution was sealed in a 5-mL crimp-top
SARA Analysis of Medium Gravity Crude Oils Energy & Fuels, Vol. 16, No. 6, 2002 1573

Table 3. Pairwise t-tests for Significance of Differences


between SARA Methodsa
p-valuesb
fraction ASTM vs TLC-FID ASTM vs HPLC
saturates <0.0001 0.0236
saturates + volatiles 0.0049 0.8192
aromatics 0.0064 0.1634
resins 0.0443 0.0914
asphaltenes 0.2214
a Statistically significant differences are highlighted in bold

italic type. b p is the probabilty that the samples come from


different populations; p < 0.05 indicates that differences between
the results of different measurement methods are statistically
Figure 2. HPLC separation of C-R-00 maltenes. significant.
Table 2. RI Detector Calibration would be consistently overestimated. Since the amount
class hydrocarbon signal (area/mg) of volatile material has been measured in this study, it
saturates n-heptadecane 4647 might be included in the saturate and aromatic frac-
1-ring aromatics DIPB 7088 tions, but additional information is needed on how to
2-ring aromatics 1-MN 12517 apportion it between these two fractions. Summation
3-ring aromatics phenanthrene 16022 of the volatile and saturate amounts from the TLC-
FID method would significantly overestimate the satu-
glass vial with an aluminum seal and PTFE/silicon septum. rate fraction and underestimate the amount of aromat-
The precipitated asphaltenes were recovered by filtration
ics. Only by separating all of the volatile material up to
through a 0.22 µm filter, dried, and weighed. Several 0.5 mL
aliquots of the maltene/hexane mixture were injected onto the some fairly high boiling point (Karlsen and Larter11
HPLC columns, using a gastight HPLC syringe (Hamilton). measured losses of approximately 50% of n-C14, which
Saturates and aromatics were eluted with hexane at a flow boils at 253.5 °C; whereas the loss of n-C16 , which boils
rate of 1.5 mL/min. The amounts of saturates and aromatics at 286.8 °C, was negligible) and analyzing that fraction
were calculated from peak areas using calibration factors separately, then combining the two sets of data would
measured for known compounds (Table 2). Standard deviations one be able to apply the TLC-FID technique to medium
for peak areas of known compounds were less than 1% of the gravity oil samples with any confidence, but the advan-
measured values. A typical chromatogram is shown in Figure tage of rapid analysissthe main reason for using this
2. techniqueswould be lost. Apportioning the high molec-
Resins were strongly adsorbed and did not elute with
ular weight polar materials between resins and asphalt-
hexane. A 30% (v/v) dichloromethane/hexane backflush was
used to elute the resins accumulated from at least three enes, which is always based, rather arbitrarily, on
replicate injections to ensure that sufficient resins could be solvency, inevitably will vary from the ASTM and HPLC
collected for accurate gravimetric determination of the amount methods as well since the solvents used are not identi-
of resin remaining after solvent evaporation. cal. On the basis of these comparisons, the TLC-FID
The principal improvements in the HPLC method described technique cannot be recommended for routine SARA
here over previously proposed HPLC techniques include the analysis of medium gravity oils.
following: (i) analysis of the saturate and aromatic fractions HPLC vs ASTM. Differences between the improved
without solvent evaporation, thus avoiding the uncertainties HPLC SARA fractions and those measured by the
due to evaporation of some of the more volatile material along ASTM-recommended method were mainly in the volatile
with the solvent; and (ii) improved gravimetric quantification
and saturate fractions. There were no losses to evapora-
of the amount of resins eluted by backflushing the columns
after repeated injections of maltenes. tion in the HPLC technique, unlike the ASTM method
in which there were some losses (during evaporation of
the solvent after removal of the asphaltenes and in the
Results and Discussion
various fraction recovery and solvent evaporation pro-
The SARA fractions measured by all three techniques cesses). However, if all of the volatile material lost in
are summarized in Figure 3. The results of the ASTM the ASTM measurements was assumed to belong to the
method provide a baseline against which other SARA saturate fraction, the amounts of saturates, aromatics,
separation methods can be compared. Table 3 shows the and resins determined by the ASTM and HPLC methods
results of t-tests applied to TLC-FID and HPLC results were statistically indistinguishable (see Table 3). The
paired with those obtained by the ASTM method for the measurement technique for the amount of asphaltenes
same oil to determine statistical significance of the is the same for both methods, hence asphaltenes were
differences in the data. determined only once. The major differences between
TLC-FID vs ASTM. Comparisons between ASTM the two techniques are the amounts of time, sample,
and TLC-FID results show consistent, statistically and solvent required. For both techniques, there is a
significant differences for all fractions except the as- two-day period required for separation of the asphalt-
phaltenes (Table 3). The amount of volatile material lost enes. Beyond that, however, the HPLC technique re-
before the detector response is recorded is not routinely quires only about 1 h per sample including three
recognized or measured. That amount is as high as 60% replicate separations of saturates and aromatics and
in this study, all of which comes from the saturate and backflushing of the column to elute resins, compared
aromatic fractions. If fractions were apportioned on the to at least 20 h for the ASTM separation. One gram of
basis of peak areas summed to 100%, ignoring the oil is adequate to give the same accuracy by the HPLC
volatile material, the amounts of resins and asphaltenes technique as the 10 g required by the ASTM test. The
1574 Energy & Fuels, Vol. 16, No. 6, 2002 Fan and Buckley

Figure 3. Comparison of SARA fractions determined by the ASTM method, by the improved HPLC method, and by TLC-FID.

solvent requirements are on the order of 300 mL and 2 data thus obtained were used to find regression coef-
L per sample for the HPLC and ASTM tests, respec- ficients that correlate the percentages of each fraction
tively. with the API gravities measured for the same samples.
Internal Consistency of SARA Fraction Data. The following empirical relationship was obtained:
There are many variations of the specific analyses, all
of which are referred to by the acronym of SARA, but ˚APIcalculated ) 74.5 - 0.306 × S - 0.385 ×
not all of which give equivalent results. Since the details A - 1.08 × R - 0.763 × As (r2 ) 0.68)
of the separation technique are not always provided, a
method is needed to evaluate SARA data from unspeci- where S is the wt % of saturates, A is the wt % of
fied tests so that ASTM and HPLC data can be distin- aromatics, R is the wt % of resins, and As is the wt %
guished from TLC-FID results. A simple evaluation of asphaltenes. A plot of measured vs calculated API
scheme is shown here, based on comparisons between gravity is shown in Figure 4.
the SARA fraction data and API gravity, which is almost To test the consistency of any set of SARA fraction
universally reported for stock-tank oil samples. data, similar plots can be prepared, as illustrated for
A group of 87 crude oils was evaluated using the the ASTM and the TLC-FID data sets. Despite some
HPLC technique described above. The SARA fraction scatter, the ASTM data clearly follow the predicted
SARA Analysis of Medium Gravity Crude Oils Energy & Fuels, Vol. 16, No. 6, 2002 1575

the gravity range of the samples that were used in its


development, i.e., from a low value of about 15 to a high
of about 40°API.

Conclusions
•Measurements of SARA fractions are highly depend-
ent on methodology. Results of different methods may
not be comparable. Caution is needed in comparing
results, even from similar methods.
•The TLC-FID technique cannot be used to test
medium gravity oils without considerable additional
analysis to account for components that boil at temper-
atures up to 250 °C. The apportioning of high molecular
weight, polar, aromatic material into asphaltenes and
resins does not correspond to the asphaltenes and resin
fractions defined by the ASTM method.
•An efficient HPLC methodology has been developed
and shown to correspond closely to the more time-
consuming ASTM-recommended method of SARA analy-
sis.
Figure 4. An estimate of API gravity obtained from SARA
fraction data can be compared to the measured API gravity •SARA fraction data can be tested for internal con-
as a test of the internal consistency of the SARA data. A sistency by comparing API gravity values calculated
correlation was developed with 87 different crude oil samples from the SARA fractions with measured values of API
spanning the range from about 15 to over 40°API, tested with gravity for the same oil sample.
the HPLC technique. The ASTM data set closely reproduces
the correlation, whereas there is no correlation between the Acknowledgment. The authors thank Dr. Jianxin
SARA fraction data from TLC-FID measurements and API
Wang for many useful discussions. This work was
gravity of the oil.
supported by the National Energy Technology Labora-
tory (NETL) of the US DOE through contract DE-AC26-
trend (r2 ) 0.76), whereas there is no correlation
99BC15204 and by support from industrial sponsors
between the TLC-FID data and API gravity measured
including BP, Chevron, Gaz de France, IFP, Norsk
for the same samples. Note that it is not sufficient to
Hydro, and TotalFinaElf. Crude oils were provided by
test a single oil sample. It is the trend for oils with
ARCO, Chevron, Shell, and the University of Wyoming.
varying properties over a range of API gravity that is
diagnostic. The correlation should be applied only over EF0201228

You might also like