Anal. Chem. 2005, 77, 2536-2543
Characterization of High-Molecular-Weight Sulfur-Containing Aromatics in Vacuum Residues Using Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Hendrik Mu 1 ller and Jan T. Andersson*
Institute of Inorganic and Analytical Chemistry, University of Mu¨nster, Corrensstrasse 30, 48149 Mu¨nster, Germany Wolfgang Schrader
Max-Planck-Institut fu¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mu¨lheim an der Ruhr, Germany
Millions of tons of vacuum residues are produced in refineries every year and could be a potentially valuable resource for generating electricity and has possible application as heating and marine fuel. In this work, the polycyclic aromatic sulfur compounds (PASHs) from the aromatic fraction of vacuum residue before and after partial hydrodesulfurization (HDS) were derivatized by methylation to the methylsulfonium salts. Fourier transform ion cyclotron resonance mass spectrometry provided high-resolution data on these high-molecular-weight sulfur compounds. Compounds containing one and two S atoms were found to dominate, with masses up to ca. 900 Da. Classification according to hydrogen deficiency and the number of heteroatoms showed extensive series of homologues for double bond equivalents from 5 to 20. The sulfur-containing aromatics were separated using a palladium(II) complex as a liquid chromatographic phase into two compound groups: one containing compounds with an unconjugated thiophene ring and another with a condensed thiophene ring. This, combined with the mass spectrometry (MS) data, allows for the identification of several parent structures. Partial HDS removed primarily compounds with one S atom, whereas those with two S atoms were largely unaffected. The distillation of crude oils in the world’s refineries not only produces heating and transportation fuels but also results in large amounts of vacuum residues that are used as marine fuel and in power plants for the generation of electricity. The reduction of emissions is essential for pollution control, and the European Commission has issued stringent specifications to limit sulfur emissions (600 mg SOx/Nm3 for new powerplants). The sulfur in marine fuel must not exceed 4.5 wt %. In the Baltic Sea (SOx control area), the sulfur content must be reduced to 1.5 wt %. Current proposals in the European Parliament are aimed at reducing sulfur levels in all European Union (EU) territorial seas * Corresponding author. Tel: ++49-251-8 33 31 02. Fax: +49-251-8 33 60 13. E-mail:
[email protected].
2536 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
to 1.5% by December 2010 and to 0.5% by December 2012. These legislative goals can only be achieved through desulfurization of the vacuum residues; however, this refinery process is often hampered by the large amounts of heteroatoms present in these heavy fractions. One of the major problems is the high level of sulfur in these residues, because the H2S generated is a poison for hydrocracking catalysts. In the lighter distillation fractions, polycyclic aromatic sulfur heterocycles (PASHs) are reported to be among the most recalcitrant compounds to the complex pathway that leads to hydrodesulfurization (HDS).1 Even in deeply desulfurized fractions, some of the sulfur species remain, mainly dibenzothiophenes alkylated in the positions adjacent to the S atom, namely in the 4and 6-positions, where they are thought to exert a steric hindrance to efficient approach to the catalyst surface. Such fractions can be readily analyzed using gas chromatography (GC) with sulfurselective detection.2-4 Desulfurization of heavier petroleum fractions is less efficient than that of lighter fractions. Although not much is known about the structure of the recalcitrant species in such materials, most of the persistent sulfur is found in the fraction of aromatic compounds.5 These compounds (∼50 wt %), together with the asphaltene and resin fractions, dominate after the hydrodemetallization (HDM).6 Better knowledge of the structure of such recalcitrant sulfur forms is highly desirable to assist in the development of new HDS catalysts and to identify high-yielding refinery process conditions. Defined analytical information on complex petroleum mixtures of molecular weights over ca. 400 is difficult to gain. Bulk parameters and some structural information have been obtained by chemical and pyrolytic degradation. A disadvantage of degradation is reagent selectivity and possible formation of secondary (1) Whitehurst, D. D.; Isoda, T.; Mochinda, I. Adv. Catal. 1998, 42, 345-471. (2) Schmid, B.; Andersson, J. T. Anal. Chem. 1997, 69, 3476-3481. (3) Macaud, M.; Milenkovic, A.; Schulz, E. J. Catal. 2000, 193, 255-263. (4) Ma, X. L.; Sakanishi, K. Y.; Mochida, I. Ind. Eng. Chem. Res. 1994, 33, 218-222. (5) Ancheyta, J.; Morales, P.; Betancourt, G.; Centeno, G.; Marroquin, G.; Munoz, J. A. D. Energy Fuels 2004, 18, 1001-1004. (6) Kressmann, S.; Morel, F.; Harle, V.; Kasztelan, S. Catal. Today 1998, 43, 203-215. 10.1021/ac0483522 CCC: $30.25
© 2005 American Chemical Society Published on Web 03/05/2005
products. Therefore, chromatographic fractionation is a common tool to simplify petroleum samples; however, sample complexity prevents the isolation of individual sulfur heteroaromatic compounds from heavy fractions. Size exclusion chromatography (SEC) was used for molecular-weight analysis (bulk property), in combination with matrix-assisted laser desorption ionization-mass spectrometry (MALDI-MS).7 Differentiation between organic (sulfidic and thiophenic) and inorganic sulfur in petroleum samples, as well as quantification of both species, has been performed using X-ray absorption near-edge structure spectroscopy (XANES) and X-ray photoelectron spectroscopy (XPS).8,9 Structural parameters such as the number of fused rings in the sulfur-containing aromatic systems and the degree of substitution are only partly accessible, and, generally, only bulk parameters are provided. Numerous studies analyzing heavy petroleum fractions with mass spectrometry (MS) and chromatographic methods have been published. Thermospray ionization can be applied to interface high-performance liquid chromatography (HPLC) to MS for the analysis of high-boiling polyaromatic hydrocarbons (PAHs)10 but was not further exploited for PASHs. Particle beam ionization (PB)11 and atmospheric pressure chemical ionization (APCI)12,13 have been applied for ionization of heavy aromatics; however, PB suffers from sensitivity and linearity limitations and APCI-generated spectra are reported to result from various ionization mechanisms, yielding several possible ionic species, which complicates data interpretation.13 Low-energy electron ionization (EI),14 electrospray ionization (ESI),15 and field desorption16 have been combined with high-resolution Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) for petroleum analysis; however, the latter studies were focused on polar compounds and advancing the instrumentation used. Aromatic components in petroleum vacuum residues were ionized by inbeam EI and characterized by FTICR-MS.14 All aromatics, both hydrocarbons and heterocycles, were unselectively analyzed. It has been said that petroleomics, which is the characterization of all of the chemical constituents of petroleum, along with their interactions and reactivity,17 is the next grand challenge for chemical analysis, which can only be approached through an intimate combination of separation and spectroscopic techniques. In our previous studies of the PASH fraction of a vacuum residue, we have used a series of consecutive chromatographic dimensionss namely, adsorption, ligand exchange chromatography (LEC), and gel permeation chromatography18sbut, even so, the complexity of the resulting small subfractions is tremendous. Mass spectra (7) Li, W.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. J. Chromatogr. A 2004, 1024, 227-243. (8) Waldo, G. S.; Carlson, R. M. K.; Moldowan, J. M.; Peters, K. E.; Pennerhahn, J. E. Geochim. Cosmochim. Acta 1991, 55, 801-814. (9) Christensen, L. H.; Agerbo, A. Anal. Chem. 1981, 53, 1788-1792. (10) Hsu, C. S.; Qian, K. G. Anal. Chem. 1993, 65, 767-771. (11) Pace, C. M.; Betowski, L. D. J. Am. Soc. Mass Spectrosc. 1995, 6, 597-607. (12) Marvin, C. H.; McCarry, B. E.; Villella, J.; Bryant, D. W.; Smith, R. W. Polycyclic Aromat. Compd. 1996, 9, 193-200. (13) Roussis, S. G.; Proulx, R. Anal. Chem. 2002, 74, 1408-1414. (14) Miyabayashi, K.; Naito, Y.; Tsujimoto, K.; Miyake, M. Int. J. Mass Spectrom. 2002, 221, 93-105. (15) Qian, K.; Rodgers, R. P.; Hendrickson, C. L.; Emmett, M. R.; Marshall, A. G. Energy Fuels 2001, 15, 492-498. (16) Schaub, T. M.; Hendrickson, C. L.; Qian, K. N.; Quinn, J. P.; Marshall, A. G. Anal. Chem. 2003, 75, 2172-2176. (17) Marshall, A. G.; Rodgers, R. P. Acc. Chem. Res. 2004, 37, 53-59. (18) Mu ¨ ller, H.; Schrader, W.; Andersson, J. T. Prepr. Pap.-Am. Chem. Soc., Div. Fuel Chem. 2004, 49, 335-337.
obtained by standard resolution instruments such as an ion trap mass spectrometer with EI and also time-of-flight MS showed signals at every nominal mass over a wide mass range (ca. 300950 Da). When the number of isomers for each exact mass is considered for compounds with a molecular mass of several hundred, in practical terms, a number approaching infinity will result. This complexity is beyond the resolution capability of even the best high-resolution chromatographic system, and the large molecular mass precludes the use of vapor-phase chromatography. FTICR-MS has been shown to be able to resolve thousands of masses in complex samples.19,20 In a crude oil, nitrogen and polar compounds were investigated and more than 11 000 exact masses were assigned to unique elemental compositions. The nitrogen compounds were protonated with acetic acid without any other sample preparation and analyzed in the mass spectrometer. Spectra in the combined positive and negative mode revealed more than 18 000 components.19 In this work, we use the very high resolving power of FTICRMS to study PASHs in mixtures with PAHs. The ionization of aromatic compounds with electrospray techniques is not very efficient;21,22 thus, derivatized compounds are desirable. For PASHs, palladium(II) has been used as a sensitivity-enhancing reagent in standard resolution experiments with ESI ion trap MS.22 However, this technique may show problems with samples of unknown sulfur content, because the concentration ratios of palladium(II) and sulfur seem crucial. In addition, instrumental parameters such as flow rate for injection of the sample and voltages must be finely tuned to suppress hydrocarbons. For such reasons, here, we investigate the derivatization of organic sulfur to methylsulfonium salts to achieve selectivity toward sulfur aromatics in the presence of PAHs.23-25 A larger number of parent systems can be expected among PASHs than among PAHs, because the S atom introduces an element of asymmetry into the molecule.26 We have previously found that some sulfur aromatics are retained on a stationary phase that contains Pd ions in normal-phase liquid chromatography.27,28 This LEC method is used here primarily to separate such compounds with an isolated thiophenic ring from PASHs, thus considerably increasing the informational content of the FTICRMS data. Coupling the selectivity of the palladium(II) stationary phase, the information obtained from FTICR-MS, and previous work on sulfur-containing compounds in fossil material allows us to suggest rather detailed information on the parent structures of highmolecular-weight PASHs in vacuum residues.28 First, comparative (19) Guan, S. H.; Marshall, A. G.; Scheppele, S. E. Anal. Chem. 1996, 68, 4671. (20) Schrader, W.; Klein, H.-W. Anal. Bioanal. Chem. 2004, 379, 1013-1024. (21) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Mass Spectrom. Rev. 1990, 9, 37-70. (22) Rudzinski, W. E.; Zhou, K.; Luo, X. M. Energy Fuels 2004, 18, 16-21. (23) Acheson, R. M.; Harrison, D. R. J. Chem. Soc. C 1970, 13, 1764-1784. (24) Green, T. K.; Whitley, P.; Wu, K. N.; Lloyd, W. G.; Gan, L. Z. Energy Fuels 1994, 8, 244-248. (25) Mu ¨ ller, H.; Andersson, J. T. Polycyclic Aromat. Compd. 2004, 24, 299308. (26) Andersson, J. T. Int. J. Environ. Anal. Chem. 1992, 48, 1-15. (27) Schade, T.; Roberz, B.; Andersson, J. T. Polycyclic Aromat. Compd. 2002, 22, 311-320. (28) Sripada, K. S.; Andersson, J. T. Anal. Bioanal. Chem. doi: 10.1007/s00216004-3026-y.
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2537
Figure 1. Preparation of aromatic Fractions 1 and 2 by ligand exchange chromatography (LEC) for (a) “feed” and (b) “effluent” after partial desulfurization (HDS).
data on such residue before and after partial HDS over an HDM catalyst are also reported. EXPERIMENTAL SECTION Vacuum Residue Samples. Aromatic vacuum residue fractions of a Safanyia crude oil before (“feed”) and after (“effluent”) partial HDS, using HMC841 (with NiMo on an alumina carrier as active phase) from Axens, were investigated. The catalyst used is tailored to remove most of the asphaltenes and resins and mainly produce aromatics (50 wt %) in high concentration; some resins (30 wt %) are still present in the effluent, as well as saturates (15 wt %). The aromatics were isolated by first removing unconverted asphaltenes by precipitation with n-heptane and then isolation by liquid chromatography;29 this sample was provided to us by the Institute Franc¸ ais du Pe´trole in Vernaison, France. Elemental analysis showed that the aromatic fraction contained 4.66% sulfur before the partial HDS process and 2.91% sulfur after the partial HDS process. Ligand Exchange Chromatography. Ligand exchange chromatography (LEC) was used for the group separation of PASHs from PAHs on a palladium(II)-bonded stationary phase with 10 µm (100 Å) silica in a column with dimensions of 125 mm × 4.6 mm ID, as described previously.27 The residue samples (before and after HDS) were separated into two distinct fractions each, using eluents of different polarity. Figure 1 shows the separation steps used in this study; 12.7 and 17.0 mg of the feed and the effluent, respectively, were separated in several runs (to avoid overloading of the stationary phase). For the feed, a total of 4.40 mg of Fraction 1 and 7.40 mg of Fraction 2 were collected. The effluent yielded 6.15 mg and 9.55 mg, respectively. A second set of separations prepared two fractions of the effluent for sulfur determination. (29) Later, D. W.; Lee, M. L.; Bartle, K. D.; Kong, R. C.; Vassilaros, D. L. Anal. Chem. 1981, 53, 1612-1620.
2538
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
Sulfur Determination by Inductively Coupled Plasma (ICP) Spectroscopy. Sulfur in the LEC fractions of the effluent was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) after combustion of a 18.15 mg sample of Fraction 1 and a 7.38 mg sample of Fraction 2 to sulfate in an oxygen atmosphere.30 External calibration with sodium sulfate was utilized. Sulfur was detected at S(I) 142.503, 143.328, 147.400, 166.669, 180.731, and 182.034 nm. Fraction 1 contained 2.0 ( 0.2 wt % sulfur, and fraction 2 contained 2.8 ( 0.2 wt %. Methylation of Sulfur Compounds. Sulfur compounds in all fractions were selectively methylated at the S atom.21 All four residue fractions, containing between 10-2 and 4 × 10-3 mmol sulfur and 1 mmol of iodomethane, were dissolved in 3 mL of dry 1,2-dichloroethane (DCE). A solution of 1 mmol silver tetrafluoroborate in 2 mL of DCE was added and yellow silver iodide precipitated immediately. After the mixture was allowed to react for 48 h, the precipitate was removed by centrifugation/filtration and washed with 3 mL of DCE. The DCE and excess iodomethane were distilled off from the combined reaction and washing solutions under reduced pressure. The resulting sulfonium salts were dried under vacuum. The influence of alkyl substitution near the S atom on the methylation reaction was tested with seven reference compounds: dibenzothiophene, 4-methyldibenzothiophene, 2,4,6,8-tetramethyldibenzothiophene, 2,7-dimethylbenzothiophene, cholestano[2,3-b]-5,6,7,8-tetrahydronaphtho[2,1-d]thiophene (C34H54S),25 and phenanthrene as control. Except for the commercial dibenzothiophene and phenanthrene, the compounds were synthesized in our laboratory.31 Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Mass spectra were recorded using an APEX III FTICRMS (Bruker Daltonics, Bremen, Germany) system that was equipped with a 7 T magnet and an Agilent ESI source. The methylated samples were introduced as a solution in dichloromethane/acetonitrile 1:1 (v/v) and injected in the infusion mode with a flow rate of 2 µL/min at an electrospray voltage of 4.5 kV. The ions were collected for 0.5 s in a hexapole before release into the cyclotron cell. At least 64 scans were accumulated for each spectrum, to improve the signal-to-noise ratio. Internal and external mass calibration was performed using the Agilent electrospray calibration solution, covering the mass range of the sample with the exact masses 322.04812, 622.02896, and 922.00980 Da. Calculation of Compound Classes from Accurate Mass Data. For each mass signal, the most probable elemental compositions was calculated, based on the accurate mass determined from the [M+CH3]+ ion. Based on the assigned elemental compositions, the heteroatom content (class) and hydrogen deficiency Z (type) were determined for each mass signal. Oxygenand nitrogen-containing compounds were not observed. In a second step, the mass signals were transformed to the Kendrick mass scale (see Discussion section).32 The Kendrick nominal mass (KNM) and the Kendrick mass defect (KMD) were calculated and series with the same KMD were grouped. A (30) Ehrenberger, F. Quantitative Organische Elementaranalyse, XXI Edition; VCH: Weinheim, Germany, 1991. (31) Andersson, J. T. www.pash-standards.de, accessed January 24, 2005. (32) Kendrick, E. Anal. Chem. 1963, 35, 333-340.
comparison with tabled KMDs made it possible to assign each series of signals to a certain class and type.19,33 Carbon distribution numbers were calculated and checked to be integer numbers to confirm the assigned class and type. Assignments were in good agreement with results from the first step. At this stage, unassigned mass signals were matched up to “gaps” in the series of homologues obtained. The assignment was based on the difference between the measured and calculated masses of the compound in question being smaller than 1.5 mDa and unrivaled assignment. To enable distinct formula assignment, for an 800 Da compound, a precision of ∼1.9 ppm is required.33 The assignment was considered unrivaled when no other possible class with more than half the number of initially assigned members in the competing homologue series was within 1.5 mDa. RESULTS AND DISCUSSION The first step in our strategy for analyzing PASHs in fossil material is the isolation of a fraction containing the nonpolar polycyclic aromatic compounds that is further fractionated on a stationary phase containing palladium(II) ions in a complex with silica-bonded 2-aminocyclopentene-1-dithiocarboxylic acid.34 It has been shown previously, using GC with a sulfur-selective detector, that the aromatic compounds in volatile materials such as diesel fuel are completely separated into a fraction that only contains hydrocarbons (Fraction 1) and with a more polar eluent that only contains PASHs (Fraction 2).27,28 The same fractionation step was used here for the much larger aromatics in the vacuum residue, and the two resulting fractions were then investigated separately. For these samples, such a clear-cut separation into sulfur-free and sulfur-containing compounds is not obtained, as evidenced by the elemental analysis of Fraction 1. Despite this, the palladium(II) phase can be used to aid in elucidating the parent structure of PASHs, as shown below. The presence of some sulfur-free aromatics in Fraction 2 is indicated by FTICR-MS; however, it is unclear which molecular feature retains them on the palladium(II) column.28 As discussed previously, polycyclic aromatic compounds are not efficiently ionized by ESI techniques, thus necessitating a derivatization of the analytes. PASHs are known to complex with palladium(II) ions, and this process recently was used for the selective enhancement of the ionization of PASHs with ESI.22 Although a certain selectivity was achieved, the published mass spectrum of a crude oil showed the dibenzothiophene ion to lie in a cluster of at least eight ions of similar magnitude. A definitive assignment of the ions seems to require MS/MS techniques, thus adding considerable complexity to the technique. Here, we circumvent such problems by methylating the S atom in the heterocycles, thereby selectively derivatizing neutral sulfur species through their conversion to sulfonium salts, as shown in Figure 2. We are particularly interested in PASHs that show high recalcitrance to HDS. Such compounds often possess alkyl groups in positions adjacent to the S atom,4,35,36 presumably acting as steric (33) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. Anal. Chem. 2001, 73, 4676-4681. (34) Pyell, U.; Schober, S.; Stork, G. Fresenius J. Anal. Chem. 1997, 359, 538541. (35) Andersson, J. T. J. Chromatogr. 1991, 585, 376-377. (36) Rodgers, R. P.; White, F. M.; Hendrickson, C. L.; Marshall, A. G.; Andersen, K. V. Anal. Chem. 1998, 70, 4743-4750.
Figure 2. Methylation of dibenzothiophene.
Figure 3. High-resolution mass spectra of the first fractions obtained by LEC of before the partial HDS process (panel a) and after the partial HDS process (panel c); the three internal standards used for mass calibration are indicated by their exact masses. Panel b shows a mass scale expanded section from m/z 721 to 724 of the upper spectrum in panel a in more detail.
hindrance to the reaction. To test this influence, six reference compounds were methylated using the procedure described. Phenanthrene was used as a control to study the behavior of a non-sulfur species; however, it was inert under the applied conditions. The reaction23,24 gave a quantitative yield, even for 2,4,6,8-tetramethyldibenzothiophene, after 16 h. To take into account the possibly stronger steric hindrance that may be present in the large compounds in a vacuum residue, the reaction time was extended to 48 h. The resulting mass spectra of the PAH fraction before and after partial HDS are shown in Figure 3. Other ionization techniques such as APCI and MALDI were tried for underivatized compounds but showed poor signal-to-noise ratios and no selectivity toward sulfur compounds. Some reference compounds showed clusters of several molecules with Ag or tetrafluoroborate ions. The data analysis of the vacuum residues showed that the amount of clusters can be estimated to lie below 10% of all detected signals. Signals identified as adducts with Ag or tetrafluoroborate ions were defined as Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2539
Table 1. Abundancy of the Mass Spectrometric Signals for Compound Classes in the Vacuum Residue Fractionsa Class PAH
S1
S2
S3
unassigned
cluster
Fraction 1 Fraction 2
92 (15.9%) 102 (22.8%)
289 (71.7%) 256 (65.7%)
Feed 44 (7.8%) 30 (7.9%)
27 (4.9%) 12 (3.6%)
42 53
18 87
Fraction 1 Fraction 2
95 (16.1%) 144 (24.1%)
267 (68.5%) 325 (63.7%)
Effluent 64 (11.4%) 44 (8.4%)
13 (3.9%) 16 (3.7%)
38 58
52 48
a The number of signals and the signal magnitude over a class are corrected for signals resulting from elemental compositions containing 13C isotopes. The values in parentheses refer to the standardized magnitude over all signals identified in the given class.
clusters and excluded from further consideration. Even a larger abundance of clusters would not provide a problem, because they can be identified easily using the high-accuracy mass data and treated separately. The assignment of signals was based on strict criteria, and no elemental composition was assigned in cases where several possible compositions were of nearly the same probability and where an unknown would be a member of a homologous group with a similar number of members as another possible composition would have. No evidence for multiply charged ions could be found; therefore, we refer to each ion by its mass in Da rather than to its mass-to-charge ratio (m/z). Numbers of clusters, nonassigned mass signals, and assigned classes for the four investigated fractions are given in Table 1. The relative intensities over all types sharing a class are given in parentheses. The signals of 13C isotopes were excluded from the numbers of identified elemental compositions presented in Table 1, although those signals were used as support for the assignment of elemental composition. The data in the table refer to the thiophenium ions produced from the PASHs and may not exactly reflect the abundance of these in the sample. Compounds with one S atom dominate; structures with three S atoms are identified in negligible numbers. There is a slight increase in the magnitude of the hydrocarbon signal and a corresponding decrease in the S1 class relative abundancy after the HDS. The slight increase of S2 compounds in Fraction 1 is remarkable, which might be explained by the partial removal of sulfur from molecules of higher sulfur content. This will be the subject of future studies. A very compact way of visualizing the large amount of data in high-resolution mass spectra is the stacking of 1 Da segments of the spectra into a pseudo two-dimensional Kendrick mass plot, where the mass defect (i.e., the difference between the exact and nominal mass) is plotted against the nominal mass.19 In addition, conversion from the IUPAC to the “Kendrick” mass scale (i.e., multiplying each mass by 14.00000/14.01565) is done, thus effectively making the methylene group the basis of the molecular weight system. As a result, all members of a series of homologues feature the same mass defect and, thus, appear as horizontal lines in the Kendrick mass plot, spaced at equal intervals of 14 Da. The Kendrick mass plots of the sulfur-containing compounds of all four vacuum residue fractions are depicted in Figure 4. Through the analysis of the exact masses, it becomes obvious that a large number of sulfur compounds are not retained by the palladium(II) phase and, thus, appear in the first LEC fraction. Sulfur determination with ICP-OES confirmed the presence of 2540 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2.0% sulfur in the first fraction and 2.8% sulfur in the second fraction of the effluent. The fact that there are sulfur compounds in Fraction 1 is contrary to our extensive experience of the behavior of PASHs in lighter petroleum fractions where a gas chromatographic analysis of the fractions is possible, and sulfurselective detection shows that a complete separation between sulfur-containing and sulfur-free aromatics is possible. This is also true for sterically demanding compounds such as 4-ethyl-6methyldibenzothiophene27 or 4,6-diethyldibenzothiophene. In previous work, we have also shown that even such a crowded benzothiophene as cholestano[2,3-b]-5,6,7,8-tetrahydronaphtho[2,1d]thiophene was retained by palladium(II), as expected for a benzothiophene.25 Noncondensed thiophenes have been shown not to be retained by palladium(II), and we take this as a starting point for our discussion of the identity of the sulfur compounds that are present in this fraction. The Kendrick mass defect plots of the sulfur-containing aromatics in Fraction 1 in Figure 4a and 4c show a different pattern from that of the sulfur-containing aromatics in Fraction 2 (Figure 4b and 4d). The KMDs on the y-axis are related to the hydrogen deficiency index Z, which is defined as the number of H atoms less than a completely saturated compound with the same number of C atoms and without rings has: for a given elemental composition 12C(c-x)13CxHhSs, is Z ) h - 2c or CcH(2c+Z)Ss. Z gives an indication of the number of rings plus double bonds present in the molecule, because each ring and each double bond means a reduction in the number of hydrogens by 2.33 Benzothiophene has the composition C8H6S and, thus, Z ) -10. Similarly, all alkylated benzothiophenes will necessarily have two rings and four double bonds and, thus, the same Z number and therefore lie on the same horizontal line in the plots in Figure 4. However, with Z being negative and attaining a value of -2 for a compound with no rings or double bonds, this definition is inconvenient in the present discussions and we prefer to use the double bond equivalent (DBE), which is the sum of the number of rings (R) and the number of double bonds (DB): DBE ) R + DB. For PAHs and PASHs, the relationship between DBE and Z is simple: DBE ) - Z/2 + 1. The results obtained here allow us to draw some conclusions about the parent structures present. A KMD of ca. 130.6 translates to a DBE value of 6 (Z ) -10) for compounds containing one S atom, and, thus, the lowest row of points in Figure 4a and the corresponding rows in the other panels in the figure could represent benzothiophenes (but no other condensed polyaromatic systems) with an increasing number of C atoms in the side chains
Figure 4. Kendrick mass defects (KMDs) versus nominal Kendrick mass of all vacuum residue samples: (a) first LEC fraction, (b) second LEC fraction before HDS, (c) first LEC fraction, and (d) second LEC fraction after HDS. The size of the dots is proportional to the relative signal magnitude. Only signals that have been assigned with high probability to a heterocyclic compound with one S atom (“S1”) and classes are shown for clarity. The DBE scale given is valid for both panels.
toward higher Kendrick nominal masses. Benzothiophenes are retained by palladium(II), so they should appear in Fraction 2. The first thick point in that row in Figure 4b has a mass of 456 and, thus, an elemental composition of C31H52S. If it is a benzothiophene, the side chains in total must be composed of C23H47. However, the mass spectrum cannot determine if it is one C23 side chain or several shorter ones or whether the side chains are linear or branched. The largest molecular weight for a compound in this row in Figure 4b is 792, implying 47 C atoms in the side chains if the parent structure is benzothiophene. Compounds with a KMD of 144.0 have a DBE value of 7, i.e., they contain one ring or one double bond more than benzothiophene, and compounds with a KMD of 157.4 have a DBE value of 8. These groups of compounds lie on the second and third row of points in Figure 4a. Rather than to hypothesize what type of parent structures could fit the data, we point to previous investigations of ring systems in fossil materials. A thorough examination of fairly low-boiling PASHs in a shale oil from Austria showed the presence of a large number of different parent systems. A range of alkylated phenylthiophenes (DBE ) 7) was observed with up to 9 C atoms in the side chains in material boiling at 168-175 °C and 3.5 Torr.37 Separate experiments have shown that both 2- and 3-phenylthiophene elute in Fraction 1 from (37) Pailer, M.; Hlozek, V. Monatsh. Chem. 1975, 106, 1259-1284.
the palladium(II) column, so one might assign the second-lowest row of compounds in Figure 4a and 4c to alkylated phenylthiophenes. Of course, other structures for DBE ) 7 are also conceivable, such as tetrahydrodibenzothiophenes and cyclopentabenzothiophenes, the latter of which are also found in the Austrian shale oil. However, substituted representatives of both of these parent structures would probably appear in Figure 4b and 4d, because unsubstituted 1,2,3,4-tetrahydrodibenzothiophene is retained by the palladium(II) column and elutes in Fraction 2. Compound classes with DBE ) 8 include indenothiophenes and indanylthiophenes, which were reported to occur in the Austrian shale oil.38,39 Indenothiophenes can be regarded as bridged phenylthiophenes and, therefore, would presumably behave similar to those on the palladium(II) column (see previous discussion) and appear in Fraction 1 (see Figure 4a and 4c). Indanylthiophenes are substituted thiophenes and are also expected to elute in Fraction 1. A major group of compounds appears at DBE ) 9 (KMD of 170.8), as expected for dibenzothiophenes (DBT) and naphthothiophenes. Such compounds are among the most common sulfur species in fossil materials and are strongly retained by palladium(II). In Figure 4b, the heaviest signal in the DBT series (38) Pailer, M.; Grunhaus, H. Monatsh. Chem. 1973, 104, 312-337. (39) Pailer, M.; Bernerfe, L. Monatsh. Chem. 1973, 104, 339-351.
Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
2541
Table 2. Parent Structures for Homologous Seriesa
Figure 5. 13C-corrected relative intensities of the sulfur series in LEC Fraction 1 (a) before and (b) after HDS: (O) S1, DBE ) 6; (0) S1, DBE ) 7; (4) S1, DBE ) 8; (b) S1, DBE ) 9; and (9) S1, DBE ) 10.
a The last column shows the LEC fraction in which those series are expected.
has a molar mass of 828.7564 Da, which means that there are 46 C atoms in the side chains. The next higher benzologues are the benzonaphthothiophenes (DBE ) 12, KMD 211.0), which are often identified in crude oils; however, there are only relatively few points in Figure 4b that correspond to such ring systems. DBE ) 11 fits with phenanthro[4,5-bcd]thiophenes, which again is a class of compounds frequently found in fossil material. (See Table 2 for a description of the parent structures for the homologous series.) The overwhelming majority of the compounds show a DBE value between 4 (corresponding to one phenyl ring) and 21, which permits of up to eight condensed aromatic rings. At DBE values of >9, the number of possible parent systems becomes very large and a lack of published data on such high-boiling materials makes comparisons difficult. UV data on the entire PASH fraction show absorption bands up to 340 nm, so at least 3-4 condensed aromatic rings should be expected to be present in detectable concentrations. Naphthenic rings do not exhibit UV absorption but must contribute to the high DBE values. A large number of PASHs with several condensed naphthenic rings have been tabulated for crude oils.40 With the combination of analytical techniques presented here, much more detailed information can be gained than through chromatography alone. For instance, in a vacuum gas oil (boiling range 340-530 °C), GC coupled with mass-selective detection showed several classes of PASHs (with up to 16 side-chain C (40) Czogalla, C. D.; Boberg, F. Sulfur Rep. 1983, 3, 121-167.
2542 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005
atoms, in the case of benzothiophene),41 but those molecular weights only go up to ca. 400 Da. Furthermore, GC-MS data do not give information on parent systemssonly on the total molecular weight of the compounds. In all fractions (both of feed and effluent) before and after LEC, the large majority of the detected signals are derived from sulfurcontaining compounds with one sulfur atom (S1). Some species with two sulfur atoms (S2) and a few with three sulfur atoms (S3) can be identified through their masses, although they are considerably less abundant. An idea of their identity may be gleaned from their DBE values, which are in the range of 0-11 for S2 in Figure 4b and comparisons with previous studies. In the shale oil cited, parent structures with two S atoms include 1,2-bisthienylethane, dithienylmethane, dithienyl, thienothiophene, and dithienobenzene. For compounds with two S atoms, we found no MS evidence for the presence of doubly charged PASHs, which would mean that only one S atom reacts with iodomethane. It is unclear yet why PAHs are found in Fraction 2. This is not observed for lighter fractions, so other molecular features may be present in the high-molecular-weight compounds. The analytical procedure used here can also give information on changes in the PASH compound pattern, as a consequence of the partial HDS. The amount, as well as class and type, of sulfur compounds is changed during the partial HDS process. A visual inspection of Figure 5 shows that, in Fraction 1, the lighter S1 compounds with a DBE value of
