Energy & Fuels 2000, 14, 839-844
839
Isolation and Characterization of the Saturate and Aromatic Fractions of a Maya Crude Oil Walter E. Rudzinski,* Tejraj M. Aminabhavi, Steve Sassman, and Linette M. Watkins Department of Chemistry and Waste Minimization and Management Research Center, Southwest Texas State University, San Marcos, Texas 78666 Received October 5, 1999. Revised Manuscript Received April 26, 2000
Group separation of Maya crude oil was achieved by the saturates-aromatics-resins-asphaltenes (SARA) method. The sulfur-containing compounds in the saturate and aromatic fractions were then separated using a ligand exchange chromatography method based on organosulfur affinity for Cu2+ and Pd2+, respectively. The separation into group types is effective as a prelude to the structural characterization of crude oil fractions using elemental analysis, Fourier transform infrared, and both 1H and 13C nuclear magnetic resonance spectroscopy. In addition, gel permeation chromatography was compared with atmospheric pressure chemical ionization/mass spectrometry (APCI/MS) in order to determine whether the APCI/MS method could provide a rapid means for the determination of the average molar mass.
Introduction The sulfur content in coal and petroleum products generally ranges from 0.025 to 11%, and thus creates a potential hazard upon combustion due to the production of sulfur dioxide, a major component of acid rain.1 As fossil fuel consumption increases, legislators have mandated stricter controls on sulfur dioxide emissions. These regulations have necessitated the development of methods that can be used to lower the sulfur content of fuels, both before and after processing. Inorganic sulfur can be successfully removed by a variety of physical separation methods, but organosulfur compounds are much more recalcitrant and more difficult to eliminate. Even though a variety of techniques2 have been used to isolate and remove the sulfur, further developments are necessary in order to meet the regulatory requirements. Since the properties of polyaromatic sulfur compounds are very similar to those of polyaromatic hydrocarbons, finding a suitable isolation procedure is a formidable task. In an effort to successfully isolate sulfur-containing compounds, Nishioka et al.3-6 have developed a unique separation method based on the saturates-aromaticsresins-asphaltenes (SARA) approach7-11 followed by * To whom correspondence should be addressed. Fax: 512-245-2374. E-mail:
[email protected]. (1) Speight, J. G. Fuel Science and Technology Handbook; Marcel Dekker: New York, 1990. (2) Whitehurst, D.; Isoda, T.; Mochida, I. Adv. Catal. 1998, 42, 345471. (3) Nishioka, M.; Tomich, R. S. Fuel 1993, 72, 1007-1010. (4) Nishioka, M.; Campbell, R. M.; Lee, M. L.; Castle, R. N. Fuel 1986, 65, 270-273. (5) Nishioka, M. Energy Fuels 1988, 2, 214-219. (6) Nishioka, M.; Lee, M. L.; Castle, R. N. Fuel 1986, 65, 390-395. (7) Later, D. W.; Lee, M. L.; Barlte, K. D.; Kong, R. C.; Vassilaros, D. L. Anal. Chem. 1981, 53, 1612-1620. (8) Andersson, J. T. Anal. Chem. 1987, 59, 2207-2209. (9) Leontaritis, K. J. Proc. Int. Symp. Oilfield Chem. 1997, 421440.
ligand exchange chromatography (LEC) using PdCl2 and CuCl2 on silica gel columns. These results have shown that the ligand exchange of sulfur compounds with PdCl2 depends on their Lewis basicity. Alkyl sulfides and cyclic sulfides with aromatic rings are the most basic, followed by phenyl sulfides and phenyl disulfides. The Lewis base interaction of sulfur compounds with PdCl2 to form complexes has been documented in the literature.5,7,12-14 This approach has been used to isolate the polycyclic aromatic hydrocarbon (PAH) and polycyclic aromatic sulfur heterocycle (PASH) as well as polar sulfur compound (PSC) fractions in crude oil and petroleum products. These were further analyzed using gas chromatography and gas chromatography/mass spectrometry. Andersson and Schmid15 also used this approach for the separation of alkylated benzothiophenes, dibenzothiophenes, naphthothiophenes, and several other aromatics in shale oil. In an effort to isolate and analyze different fractions of a Maya crude oil and in view of the paucity of such published data, we have undertaken16,17 to isolate sulfur compounds in the aliphatic and aromatic fractions of a crude oil and to characterize them using a battery of (10) Mansfield, C. T.; Barman, B. N.; Thomas, J. V.; Mehrotra, A. K.; McCann, J. M. Anal. Chem. 1999, 71, 81R-107R. (11) Ali, M. F.; Bukhari, A.; Hassan, M. Fuel Sci. Technol. 1989, 7, 1179-1208. (12) Milenkovic, A.; Schultz, E.; Meille, V.; Loffreda, D.; Forissier, M.; Vrinat, M.; Sautet, P.; Lemaire, M. Energy Fuels 1999, 13, 881887. (13) Wright, B. W.; Peaden, P. A.; Lee, M. L.; Stark, T. J. Chromatogr. 1982, 248, 17-34. (14) Grang, B. Y. Anal. Lett. 1985, 18, 193-202. (15) Andersson, J. T.; Schmid, B. J. Chromatogr. 1995, 693, 325338. (16) Rudzinski, W. E.; Rodriguez, R.; Sassman, S.; Sheedy, M., Smith, T.; Watkins, L. M. Prepr. Symp.- Am. Chem. Soc., Div. Fuel Chem. 1999, 44 (1), 28-31. (17) Rudzinski, W. E.; Sassman, S. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, pp 3013-14 June 13th, 1999.
10.1021/ef990207h CCC: $19.00 © 2000 American Chemical Society Published on Web 06/17/2000
840 Energy & Fuels, Vol. 14, No. 4, 2000
analytical techniques. In our approach, the SARA method is used to separate the crude oil into saturate, aromatic, resin, and asphaltene fractions. CuCl2 silica is then used to isolate the sulfur-containing aliphatics (S-ALP)4,5 while PdCl2 silica is used to separate the PAH, PASH, and PSC fractions in the aromatic fraction. The fractions collected are then analyzed by a combination of elemental analysis, Fourier transform infrared spectroscopy, 1H and 13C nuclear magnetic resonance (NMR) spectroscopy, gel permeation chromatography (GPC), and atmospheric pressure chemical ionization/ mass spectrometry (APCI/MS) techniques. Experimental Section Reagents. Methanol and tetrahydrofuran (THF) were HPLC grade solvents (EM Science, Gibbstown, NJ), filtered three times through 0.45 µm nylon filters (Sigma-Aldrich, St. Louis, MO). Hexane, chloroform, dichloromethane (CH2Cl2), benzene, diethyl ether, and toluene were ACS grade solvents (EM Science, Gibbstown, NJ), distilled prior to use. Diethylamine, PdCl2, CuCl2, and CDCl3 (99.8% D) were purchased from Aldrich (Milwaukee, WI). Alumina (Brockman Activity I, 80/200 mesh) was obtained from Aldrich (Milwaukee, WI), while silica (100/200 mesh) was obtained from Fisher Scientific (Pittsburgh, PA). The alumina and silica were dried at 200 °C overnight prior to use. Chromium and acetylacetone were obtained from Fisher Scientific. Cr(acac)3 was synthesized according to well-established procedures.18 Maya Crude Oil Sample. Maya crude oil was provided by Mobil Oil Corp., Beaumont, TX. The crude oil was characterized by a series of distillations and was found to contain 38% light distillate (the fraction that distills below 200 °C under 1 atm pressure), 22% middle distillate (the remaining fraction that distills below 180 °C under 20 Torr pressure), and 40% residue (the fraction that remains). The corresponding densities were 0.804 g/mL for the light distillate, 0.917 g/mL for the middle distillate, and 1.013 g/mL for the residue. The API gravity values of the crude oil, light distillate, middle distillate, and residue were, respectively, 25.1, 46.6, 23.4, and 9.6. Isolation of Crude Oil Fractions. Volatiles were removed by dissolving the crude oil in toluene and then removing the solvent using a rotary evaporator. The last traces of toluene and volatile components were removed by passing nitrogen gas over the sample. The total mass decreased by 9.1 ( 0.4%. The SARA method9-11 was used to fractionate the Maya crude oil into saturate, aromatic, resin, and asphaltene fractions. Approximately, 0.2-0.5 g of the Maya crude oil was dissolved in 20 mL of hexane, filtered through Whatman #1 paper to remove the asphaltene fraction, concentrated to 5 mL of hexane, and then adsorbed onto 3 g of alumina. The solvent was removed from alumina by vigorously stirring the mixture under a gentle stream of dry nitrogen. The alumina with the adsorbed sample was then packed on top of 6-12 g of neutral alumina in a 11 × 300 mm column (sample/adsorbent ratio ) 1/30). The sample was eluted with 20-40 mL of hexane to remove the aliphatic hydrocarbons (saturates, 23.5% w/w) and 80 mL of toluene to remove the aromatic hydrocarbons (aromatics, 34.7% w/w). A 50 mL mixture of toluene:methanol (80:20) was then added and the remaining fraction collected (resin, 15% w/w). Sulfur Fractionation of the Saturate and Aromatic Fractions. To isolate organosulfur compounds, the method developed by Nishioka5 was employed. PdCl2-impregnated silica gel (PdCl2 silica) and CuCl2-impregnated silica gel (CuCl2 silica) were prepared by suspending approximately 20 g of (18) Szafran, Zvi; Pike, R. M.; Singh, M. M. Microscale Inorganic Chemistry; John Wiley & Sons: New York, 1991, pp 226-227.
Rudzinski et al. silica gel in an aqueous solution containing either 1 g of PdCl2 or 1 g of CuCl2 in 100 mL of distilled water. The modified silicas were then filtered, dried overnight at 95 °C, and activated by heating at 200 °C for 24 h. The saturate fraction was adsorbed onto 0.5 g of CuCl2 silica and then packed on top of 6 g of CuCl2 silica in a 11 × 300 mm column. The saturates were then separated into aliphatic (ALP) and sulfur aliphatic (S-ALP) fractions using the Nishioka and Tomich3 procedure. The ALP fraction was eluted with 50 mL of hexane while the S-ALP fraction was eluted with a 100 mL mixture of chloroform:diethyl ether (90:10). The aromatic fraction was adsorbed onto 0.5 g of the PdCl2 silica and then packed on top of 6 g of PdCl2 silica gel in a 11 × 300 mm column. A 30 mL mixture of hexane:chloroform (50:50) was used to elute the PAH fraction. A further 50 mL of the same eluent was used to elute the PASH fraction. Then, a 100 mL mixture of chloroform:diethyl ether (90:10) was used to elute the PSC fraction. The PASH and PSC fractions were reduced in volume to approximately 1 mL, and 50 µL of diethylamine was added to break up the palladium complexes formed. The PSC fraction was further cleaned by eluting it through neutral alumina with 50 mL of benzene.5 All the aromatic fractions (PAH, PASH, and PSC) were evaporated to dryness and reconstituted in 5 mL of CH2Cl2. Elemental Analysis. Elemental analyses (C, H, and S) were performed by using a combustion analyzer that was calibrated with organic analytical standards certified by NIST (Desert Analytics Inc., Tucson, AZ). Infrared Spectroscopy. All FTIR spectral measurements were taken on a Perkin-Elmer FTIR (Model 1600). The samples were dissolved in CH2Cl2, coated on KBr pellets, and then purged with dry nitrogen gas. The crude oil, aliphatic, aromatic, and resin fractions as well as the LEC fractions were scanned between 500 and 4000 cm-1. The transmittance data were converted into absorbance in order to calculate the length of the aliphatic side chains. The relative intensities of the 1460 and 1380 cm-1 bands (A1460/A1380) were used to compute the ratio of methylene to methyl groups (nCH2/nCH3) using the relationship proposed by Liang et al.:19
nCH2/nCH3 ) (2.93A1460/A1380) - 3.70
(1)
The average chain length, L, was then calculated from the nCH2/nCH3 ratio as reported by Liang et al.19 Nuclear Magnetic Resonance Spectroscopy. All NMR spectra were collected on a Varian UNITY INOVA instrument equipped with a Sun workstation. 1 H NMR signals were obtained at an absorption frequency of 400 MHz. For each fraction, a 10 mg portion was dissolved in 1 mL of CDCl3. Chemical shift ranges were determined in ppm relative to TMS, then the relative proton intensity was used to determine the fraction of aromatic as well as the fraction of several aliphatic hydrogens.20 The Brown-Ladner method was used to determine structural parameters based on 1H NMR.20 13C NMR spectra were obtained at an absorption frequency of 100 MHz. CDCl3 was used as the solvent for making 3035% w/v solutions and also provided the internal fieldfrequency lock signal. To obtain quantitative 13C NMR data, the relaxation agent [Cr(acac)3] was added in 0.075 M concentration. The nuclear Overhauser enhancement (NOE) was suppressed by operating the spectrometer in the inverse-gated, decoupling mode in which the protons are irradiated only during the processing of the free induction decay. A relaxation delay of 4 s was used for all measurements. The signal-tonoise ratio was maximized by taking a total of 3000 scans. 2,2,4-Trimethylpentane and ethyl benzene were used to calibrate the spectrometer pulse sequence settings. DEPT (dis(19) Liang, W.; Que, G.; Chen, Y. Energy Sources 1991, 13, 251265. (20) Brown, J. K.; Ladner, W. R. Fuel 1960, 37, 87-96.
Studies on the Fractions of a Maya Crude Oil
Energy & Fuels, Vol. 14, No. 4, 2000 841
tortionless enhancement by polarization transfer) experiments were performed on each sample to determine the carbon type for each peak in the NMR spectrum.20 The fraction of aromatic carbon, fA, the total number of aromatic carbons, Ca, and the number of aromatic rings, Ra, were calculated as follows:
fA ) (integrated area from 115 to 150 ppm)/ total integrated area (2) Ca ) (M × %C × fA)/12.011
(3)
Ra ) (0.30Ca) - 1
(4)
where M is the molar mass obtained from APCI/MS (vide infra), the %C is obtained from elemental analysis data, and the Ra is obtained from the Brown-Ladner calculation of Ra based on a 60:40 ratio of peri-condensed to cata-condensed ring systems.19,20 The fraction of naphthenic carbon, fN, and the total number of naphthenic carbons, Cn, were calculated as follows:
fN ) (integrated area from 23 to 28.5 ppm)/ total integrated area (5) Cn ) (M × %C × fN)/12.011
(6)
Assuming three methylene groups per naphthenic ring, β or further removed from an aromatic ring, the number of naphthenic rings was calculated using the equation
Rn ) Cn/3
(7)
The fraction of parafinic carbon, fP, was calculated using the equation
fP ) 1.0 - fN - fA
(8)
The average chain length (L) was calculated by using the following equation:
L ) [(integrated area from 28.5 to 30.0 ppm) + 22.3 ppm]/ (integrated area at 13.7 ppm) (9) Gel Permeation Chromatography. The GPC experiments were performed at ambient temperature using a Tracor 951 LC pump with a Tracor 970A variable wavelength UV-vis absorbance detector set at 254 nm. The column used was a Progel TSK 61000 HXL (7.8 × 300 mm, dp ) 6 µm, Supelco, PA). THF was used as the mobile phase at a flow rate of 0.5 mL/min. A Rheodyne injector was used with a 20 µL sampling loop. The column was calibrated using naphthalene (M ) 128), benzo[k]fluoranthene (M ) 252), rubrene (M ) 533), 5,10,15,20-tetraphenyl-21H,23H-porphine (M ) 615), and polystyrene (M ) 1320) prepared in concentrations of 150 µg/L in THF. All the standards were obtained from Aldrich (Milwaukee, WI), except for the polystyrene, which was obtained from Shodex (Waters, Milford, MA). The GPC calibration curve was established and used to determine the log molecular mass corresponding to an elution volume of the unknown sample. The number-average molecular mass was determined by calculating the area under the peak within a specific mass interval and multiplying by the average molecular mass for that interval. Atmospheric Pressure Chemical Ionization/ Mass Spectrometry. To prepare solutions for the APCI/MS experiments, about 2 mg of the aromatic, PAH, PASH, PSC, ALP, or S-ALP fraction of the crude oil was dissolved in a mixture of hexane:CH2Cl2 (80:20). The mass spectra were obtained on a Finnigan LCQ iontrap mass spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source (Finnigan MAT, San
Table 1. Elemental Composition (in wt %) of Crude Oil and Its Fractions fraction
C
H
S
H/C
crude oil aromatic ALP S-ALP PAH PASH PSC
83.0 82.7 86.0 78.3 83.0 75.1 75.2
10.8 10.5 13.4 10.9 10.0 9.8 9.5
3.7 5.1 0.2 1.8 4.3 6.3 8.1
1.55 1.51 1.86 1.65 1.44 1.55 1.51
Jose, CA). A Gateway PC was used for control of the Finnigan LCQ as well as for data acquisition. Navigator version 1.2 software (Finnigan Corp., 1995-1997) was used for data acquisition and plotting. The liquid chromatography system used for sample introduction consisted of a Model AS3000 autosampler attached to a Model P4000 pump (Thermo Separation Products Inc., San Jose, CA). The analytical column was a LiChrosorb Amino column (4.6 × 150 mm, dp ) 5 µm, Phase Separations, Franklin, MA). The mobile phase consisted of a mixture of hexane:CH2Cl2 introduced as a linear solvent gradient from 100% hexane to a mixture of hexane:CH2Cl2 (60:40) over a period of 6 min. The flow rate was 1.0 mL/min, and the sample volume, 20 µL. The following APCI source parameters were used to determine the molecular mass distribution: discharge voltage, 5 kV; vaporizer temperature, 450 °C; nitrogen sheath gas pressure, 80 psi; nitrogen gas auxiliary pressure, 10 psi; heated capillary temperature, 150 °C; capillary voltage, 3 V, and tube lens voltage, 30 V. From the mass spectral data, the number-average molecular mass was calculated by taking the largest 1000 masses and multiplying them by their relative intensities.
Results and Discussion The fractionation and characterization of a crude oil is a formidable experimental task, as the crude consists of hundreds of different classes of aliphatic and aromatic compounds. The aliphatics contain mostly linear nalkanes, branched-chain alkanes, and cycloalkanes, while the aromatics consist mostly of fused ring structures with numerous alkyl side chains. In addition, heterocycles containing sulfur, nitrogen, and oxygen are also found in the aromatic fraction as well as in the resin and asphaltene fractions.21-23 However, for a better understanding of how different molecular species react under processing conditions, it is necessary to determine the molecular composition. This problem can be partially resolved by using the well-known SARA method, which separates the Maya crude into saturate, aromatic, resin, and asphaltene fractions. Then using the LEC approach, the sulfur containing compounds can be fractionated though not completely removed from both the aliphatic and aromatic fractions. Elemental Analysis. The elemental analysis data for the crude oil, aromatic, and fractions obtained from the LEC experiments are given in Table 1. For the Maya crude oil and the aromatic fraction separated by the SARA method, the sulfur content is 3.7% and 5.1%, respectively. For the ALP and S-ALP fractions, derived from the saturate fraction of the crude oil, the sulfur (21) Miller, J. T.; Fisher, R. B.; Thiygarajan, P.; Winans, R. E.; Hunt, J. E. Energy Fuels 1998, 12, 1290-1298. (22) Miller, J. T.; Fisher, R. B.; van der Eerden, A. M. J.; Koningsberger, D. C. Energy Fuels 1999, 13, 719-727. (23) Shaw, J. E. Fuel 1989, 68, 1218-1220.
842
Energy & Fuels, Vol. 14, No. 4, 2000
content is very low, i.e., 0.2 and 1.8%, respectively. For the aromatic fraction, the percent sulfur increases from the PAH fraction, which contains 4.3% sulfur, to the PASH fraction, which contains 6.3% sulfur, to the PSC fraction, which contains 8.1% sulfur. Although there is greater variation in the elemental composition of sulfur in the saturate fraction, the sulfur level in both aliphatic fractions is much lower than in the aromatic fractions. These data are consistent with the fact that sulfur is generally found within polyaromatic heterocyclic ring systems (i.e., thiophenic) rather than within aliphatic chains (sulfides, disulfides, and thiols).15 In recent studies by Miller et al.21,22 and Shaw23 on the fractionation of Maya residuum, the sulfur content was 7.1%, while the C and H contents were 82.9% and 7.1%, respectively. Although a direct comparison cannot be made, in general, our data for C and S are close to these values. The H/C atomic ratio of the aliphatic fraction, the aromatic fraction, and the resin fraction vary according to the sequence aliphatic > aromatic > resin, suggesting an increasing hydrogen deficiency in this progression. This can be attributed to a larger degree of ring condensation. The H/C ratio of the resin fraction is 1.21 and this almost agrees with the H/C ratio of the asphaltene fraction from many geological origins, which show variations between 1.1 and 1.2, indicating a low hydrogen content.24,25 The H/C ratios for the ALP and S-ALP fractions are significantly higher and these are 1.86 and 1.65, respectively. This is expected because these fractions have a significant number of methylene moieties with a H/C ratio of 2.0. The H/C ratios of PAH, PASH, and PSC fractions vary between 1.44 and 1.55. These values circumscribe the H/C value (1.51) for the aromatic fraction and approximate the value of 1.56 obtained by Ali et al.11 for the aromatic fraction of an Arabian heavy crude oil. FTIR Structural Characterization. FTIR was used to obtain structural information for crude oil fractions. The FTIR spectra of the ALP and S-ALP fractions are representative of the aliphatic fraction with a nondetectable amount of aromatic moieties, as indicated by the complete absence of bands around 1600 cm-1 that are characteristic of ring vibrations. Also, the ALP fraction exhibits strong absorption bands at 2853, 2923, 1462, and 1377 cm-1, which are attributed to stretching and angle deformation vibrations of the CH2 and CH3 groups of alkanes.26 A peak at 3600 cm-1 with the S-ALP fraction may be due to the presence of atmospheric moisture. The S-H stretching vibration is observed at 2500 cm-1 with the S-ALP fraction, but no peak is observed at this wavelength in the spectrum of the ALP fraction, indicating the absence of the thiol groups. The total aromatic fraction separated by the SARA method was dark in color and its FTIR spectrum was quite complex, showing intense and rather broad bands at several frequencies. The bands observed at 1601 and 812 cm-1 are characteristic of aromatic moieties, (24) Weihe, I. A. Ind. Eng. Chem. Res. 1992, 31, 530-536. (25) Storm, D. A.; DeCanio, S. J.; DeTar, M. M.; Nero, V. P. Fuel 1990, 69, 735-738. (26) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. Spectrometric Identification of Organic Compounds, 4th Ed.; John Wiley & Sons: NY, 1981.
Rudzinski et al. Table 2. Hydrogen Distribution in the Crude Oil Fractionsa fraction
Har
HR
Hβ
Hγ
HT
aromatic ALP S-ALP PAH PASH PSC
0.06 0.02 0.05 0.10 0.03 0.02
0.17 0.04 0.05 0.14 0.14 0.21
0.59 0.64 0.69 0.49 0.68 0.64
0.18 0.31 0.19 0.28 0.15 0.12
1.00 1.01 0.98 1.01 1.00 0.99
a H , aromatic hydrogens; H , hydrogens in saturated groups ar R R to aromatic rings or double bonds; Hβ, hydrogens of methylene and methine groups β or further removed from aromatic rings or in methylene groups attached to terminal methyl groups; Hγ, hydrogens in methyl groups γ or further removed from aromatic rings or electron-withdrawing substituents; HT, total hydrogens in the molecule.
while the band at 749 cm-1 is attributed to an aromatic, out of plane C-H bend. For the PASH and PSC fractions, the bands appearing at 1313, 1230, and 1157 cm-1 are attributed to aromatic moieties as well as C-S vibrations. From the spectra of the resin fraction, it appears that the resin is mostly aromatic in nature. The broad absorption band near 3000 cm-1, extending to longer wavelengths than observed in the aromatic fraction, and an intense band at 1604 cm-1 suggest the presence of amides or amines.11 The average side chain length (L) for the aromatic fractions determined from the relative ratio of the CH3/CH2 absorption band intensities is as follows: 2.8 (PSC), 4.3 (PASH), 5.1 (PAH). The results indicate that L increases as the polarity of the fraction decreases where polarity is defined in terms of affinity toward PdCl2 silica. Hydrogen NMR. The proton intensities for Har, HR, Hβ, and Hγ were determined using the procedure of Liang et al.19 The 1H NMR spectral data for the various fractions are presented in Table 2. The ALP and S-ALP fractions contain a small amount of aromatic hydrogens. The values of HR, for the aromatic, PAH, and PASH fractions vary between 0.14 and 0.17, indicating that these fractions have about the same number of hydrogens in saturated groups R to aromatic rings or double bonds. The ALP and S-ALP fractions have no significant contribution from the HR term correlating with a lack of significant aromatic moieties in these fractions. The PSC fraction has a HR value of 0.21, which is high but to be expected because of the large number of polar moieties that cause a higher fraction of hydrogens to shift downfield and subsequently overlap with those R to aromatic rings. The Hβ contribution for the ALP, S-ALP, PAH, PASH, and PSC fractions varies between 0.49 and 0.69, indicating a high number of hydrogens in methylene and methine groups β or further removed from aromatic rings, or hydrogens in methylene groups attached to methyl groups. The Hγ term, which represents hydrogens in methyl groups or hydrogens at least three carbons removed from aromatic rings or electronwithdrawing substituents, is highest for both the ALP (0.31) and PAH (0.28) fractions, whereas it is lowest (0.12) for the PSC fraction. Carbon-13 NMR. Methyl carbons are found in the 10-20 ppm range. The carbon resonance at 13.7 ppm can be attributed to terminal methyl groups. Methylene carbons are found in the 20-31 ppm range. The signals
Studies on the Fractions of a Maya Crude Oil
Energy & Fuels, Vol. 14, No. 4, 2000 843
Table 3. Carbon Distribution and Molar Mass of Several Crude Oil Fractionsa fraction
Car
CN
CP
M
saturate ALP aromatic PAH PSC
0.12 0.11 0.35 0.40 0.32
0.09 0.11 0.06 0.06 0.07
0.79 0.78 0.59 0.54 0.61
nd 1020 1103 989 1105
a C , aromatic carbons; C , naphthenic carbons, β or further ar N removed from an aromatic ring; CP, paraffinic carbons; M ) molar mass from APCI/MS; nd ) not determined.
for internal methylene carbons are found from 28.5 to 30.0 ppm, while methylene carbons once removed from the end of a chain resonate at 22.3 ppm. Only the carbon resonances between 23 and 28.5 ppm have been assigned to the methylene carbons of cycloalkanes, although some of the methylene carbons resonate in the interval between 28.5 and 30.0 ppm. A large number of very small signals found in the 31-55 ppm range have been assigned to methylene and methine carbons present in branched chain hydrocarbons. Aromatic carbons are found in the 115-150 ppm range. All assignments are based on the work of Michon et al.27 and confirmed by DEPT experiments. On the basis of these assignments, the Car, CN, and CP carbon distributions have been calculated. The data are presented in Table 3. Structural Parameters. Using the results of elemental analysis, FTIR, 1H NMR, and the average molar mass (from APCI data, vide infra), the average structural parameters were calculated using both the modified Brown-Ladner method18,19 and 13C NMR data. Average structural parameters were determined for the aromatic fraction, the resin fraction as well as all the fractions isolated using LEC. These data are presented in Table 4. The calculated 13C NMR values are reasonably close to the values obtained from the BrownLadner approach except for fA and fN. The 13C NMR approach consistently yields a fraction aromatic, fA, value that is higher than the value obtained from the Brown-Ladner approach. Since the fraction of aromatic carbons can be obtained directly from the 13C NMR experiment, fA should be more accurate than that obtained from the Brown-Ladner approach, which determines fA from assumptions based on the number of H nuclei. On the other hand, the fraction naphthenic, fN, obtained from 13C NMR data is lower than that obtained from the Brown-Ladner method, since the resonance frequency attributed to some naphthenic carbons overlaps with the resonance of methylene groups and the integrated area could not be included in the value of CN used for the calculation of the fraction naphthenic, fN. Despite the apparent limitations of both methods, the data are reasonably self-consistent. The ALP and S-ALP fractions have the lowest values for the fraction of aromatic carbons and the highest values for the fraction of paraffinic carbons. For the resin fraction, the highest value of fA (0.46) and lowest value of fP (0.32) suggest that the resin fraction has the most condensed aromatic structure with some alkyl side chains. Higher values of Ra than Rn are observed for the aromatic, resin, and PAH fractions, indicating the (27) Michon, L. C.; Netzel, D. A.; Turner, T. F.; Martin, D.; Pianche, J.-P. Energy Fuels 1999, 13, 602-610.
Table 4. Structural Parameters of Crude Oil Fractionsa methods
fA
fN
fP
Ra
Rn
RT
L
ALP (22.8 ( 1.2) NMR 0.11 0.11 0.78 1.4 2.7 4.1 6.0 1H NMR 0.09 0.07 0.84 1.1 1.8 2.9 13C
1H
S-ALP (0.7 ( 0.7) 0.21 0.10 0.69 3.0 2.2 5.2
σ
na 0.82
na
0.60
1.0
aromatic (34.7 ( 1.2) 13C NMR 0.35 0.06 0.59 7.0 1.5 8.5 7.5 1H NMR 0.29 0.14 0.57 5.0 3.6 8.6
na 0.75
na 0.59
1H
PAH (16.8 ( 1.4) NMR 0.40 0.06 0.54 7.1 1.4 8.5 6.2 NMR 0.35 0.11 0.53 5.5 2.7 8.2 5.1
na 0.69
na 0.41
1H
NMR
PASH (1.1( 0.2) 0.25 0.18 0.58 3.9 4.3 8.2 4.3
0.63
0.70
PSC (8.9 ( 0.8) 13C NMR 0.32 0.07 0.61 5.6 1.6 7.2 5.8 1H NMR 0.26 0.22 0.52 4.0 5.0 9.0 2.8
na 0.73
na 0.84
0.52
0.47
13C
1H
NMR
HAU/CA
NMR
resin (15) 0.46 0.22 0.32 7.1 5.0 12.1
fA ) fraction of aromatic carbon, fN ) fraction of naphthenic carbon, fP ) fraction of paraffinic carbon, Ra ) number of aromatic rings, Rn ) number of naphthenic rings, RT ) total number of rings, L ) average chain length from 13C NMR or from IR measurements (L values from IR data are italic and reported in the 1H NMR column), HAU/CA ) hydrogen to carbon atom ratio of the hypothetical unsubstituted aromatic ring system, σ ) degree of substitution of the aromatic system. na ) not applicable. a
presence of more aromatic ring structures in these fractions than naphthenic ring structures. For the ALP fraction, there are more naphthenic rings than aromatic rings. The results would be expected based on the relative polarity of this fraction. The components of all fractions average 7-9 rings except for the ALP (2.9), S-ALP (5.2), and resin (12.1) fractions. Further insight into the structure of different fractions was obtained by the calculation of the hydrogen to carbon atom ratio of the hypothetical unsubstituted aromatic ring system, i.e., HAU/CA. For the PAH, PASH, and PSC fractions, this ratio varies between 0.63 and 0.73, but for the ALP fraction, the ratio is 0.82, signifying the presence of less substitution in the aromatic ring structures. On the other hand, for the S-ALP fraction, the ratio is 0.60, indicating that the number of unsubstituted aromatic rings is less than that found in the PAH, PASH, and PSC fractions. The σ values, which represent the degree of substitution of the aromatic systems, vary in the following progression: PAH < resin < aromatic < PASH < PSC < S-ALP. The side chain length obtained from 13C NMR data varies from 5.8 to 7.5. These values are higher than those obtained from IR measurements, which vary from 2.8 to 5.1. The 13C NMR data may overestimate chain length because some methylene groups in naphthenic rings appear in the resonance frequency range associated with methylene groups. The scissoring of methylene groups in naphthenic rings is shifted to a lower frequency than that observed for methylene groups in paraffinic side chains, and so these methylenes would not yield a higher calculated value of side chain length from the IR data. Comparison of Average Molecular Mass Determination by GPC and APCI/MS. GPC and APCI/MS of the PAH fraction was carried out in order to see whether different molecular size fractions collected after
844
Energy & Fuels, Vol. 14, No. 4, 2000
Rudzinski et al.
Table 5. Comparison of Number-Average Molecular Mass by GPC and APCI/MS Experiments for Three Different GPC Elution Volumes elution volume (mL) 4.25-4.75 4.75-5.25 5.25-5.75
molecular mass × 10-2 GPC APCI/MS 17.9 ( 0.4 13.2 ( 0.02 9.8 ( 0.002
15.5 ( 0.7 11.2 ( 0.08 8.2 ( 0.2
GPC could be correlated with the molecular mass determined by APCI/MS. The GPC eluent was collected at 0.5 mL intervals and the number-average molecular mass calculated for that fraction and then compared with the value determined from the APCI/MS experiment. An envelope consisting of high molecular mass components was observed only for the first three eluted fractions of the PAH. For all the three fractions, the number-average molecular mass was determined. The molecular mass data along with the GPC elution volumes of the first three PAH fractions are given in Table 5. This number was then compared to the molecular mass obtained from the GPC experiments. The molecular masses obtained by the APCI/MS method are generally lower than those obtained from the GPC experiments by about 160-240 Da. APCI is a hard ionization technique and thus produces an aerosol that may contain fragments and long-lived adducts from the nebulization and chemical ionization processes. If the GPC standards (except for the polystyrene at 1320) give a good approximation of the structure of crude oil, which is known to have a number of fused ring structures, then the APCI/MS method underestimates the molar mass by about 15%. Other authors have noted problems inherent in the GPC determination of molecular weight distribution,28 apparent association of molecules into aggregates which might lead to higher apparent molecular weights,21 as well as discrepancies between results obtained from vapor pressure osmometry and mass spectrometry.25 The APCI/MS spectra of PAH, PASH, PSC, ALP, and S-ALP fractions were also obtained. Each fraction gave a single broad peak under the chromatographic conditions employed. Any attempt to use a weaker mobile phase (e.g., 100% hexane) to better resolve the multicomponent mixture resulted in the precipitation of some of the components. The mass/charge (m/z) scan of the (28) Snyder, L. R. Anal. Chem. 1969, 41, 1223-1227.
chromatographic peak for each fraction resulted in a broad mass spectral envelope. The number-average molecular masses obtained by the APCI/MS experiments are presented in Table 3 and decrease in the sequence aromatic ∼ PSC > ALP> PAH. Conclusions The enrichment of aliphatic and aromatic sulfur compounds from a Maya crude has been described. The results of this study indicate that the use of a multidimensional approach can be a valuable tool for the characterization of crude fractions. In this paper, the SARA method as well as a ligand exchange approach have been used to separate the sulfur-containing compounds in the saturate and aromatic fractions of the crude oil. The different fractions were analyzed to determine their structures using elemental analysis, FTIR, NMR, GPC, and APCI/MS techniques. NMR allowed an accurate determination of several structural parameters, such as the aromaticity and nature of the structure of an average polyaromatic molecule. GPC was compared with APCI in order to assess its value as a rapid means of molecular mass determination. Further research efforts are currently underway to analyze the attributes of the vacuum residue of the Maya crude oil.29 Acknowledgment. W. E. Rudzinski (W.E.R.) would like to thank the Environmental Protection Agency (Grant No. R825503) for the acquisition of the Finnegan LCQ mass spectrometer as well as for support of this project. W.E.R. would also like to thank the Institute for Industrial and Environmental Science of Southwest Texas State University for assistance in the management of this project. L. M. Watkins (L.M.W.) acknowledges the donors of the Petroleum Research Fund, administered by the American Chemical Society, for partial support of this research. T. M. Aminabhavi (T.M.A.) would like to thank the vice chancellor (Dr. A. M. Pathan) of Karnatak University, Dharwad, India, for granting him a sabbatical leave to participate in this research. W.E.R. would also like to thank Lorraine Spencer for her efforts in isolating some of the crude oil fractions. EF990207H (29) Rudzinski, W. E.; Aminabhavi, T. M.; Sassman S. Fuel Submitted for publication.