Molecular Weight Distributions of Heavy Aromatic Petroleum Fractions

Research Department, Products and Chemicals Division, Imperial Oil, Sarnia, Ontario, Canada, N7T ... (ESI MS) to analyze heavy aromatic petroleum frac...
0 downloads 0 Views 162KB Size
Download PDF
Anal. Chem. 2002, 74, 1408-1414

Molecular Weight Distributions of Heavy Aromatic Petroleum Fractions by Ag+ Electrospray Ionization Mass Spectrometry Stilianos G. Roussis* and Richard Proulx

Research Department, Products and Chemicals Division, Imperial Oil, Sarnia, Ontario, Canada, N7T 8C8

The ability of electrospray ionization mass spectrometry (ESI MS) to analyze heavy aromatic petroleum fractions using silver nitrate as a reagent compound to form characteristic adduct ions has been examined. The complexation of aromatic compounds containing long alkyl substituents with the silver ion leads to the formation of abundant adduct ions such as [M + Ag]+ and [2M + Ag]+. The concentration of the [2M + Ag]+ ions can be reduced by increasing the sampling cone voltage. Molecular ions and other adduct ions may also be formed depending on the structure of the aromatic molecule. Results obtained from the analysis of representative heavy petroleum fractions and vacuum residues by the Ag+ ESI MS method and conventional ionization methods were in good agreement. The current method extends the applicability of electrospray ionization to the analysis of neutral hydrocarbons in heavy aromatic petroleum fractions. It is simple and compatible with widely available LC/MS instrumentation. The extreme complexity of the Ag+ ESI spectra will require the development of ultrahigh-resolution MS and MS/MS methods for the separation and elucidation of the structures in the vacuum residues. The development and improvement of processes for the upgrading of heavy petroleum fractions and bitumens into distillable components require detailed compositional information to determine the mechanisms and kinetics of the conversion reactions occurring in a given process and help evaluate its overall performance.1 Acquisition of detailed compositional information for heavy petroleum fractions is difficult due to the extreme complexity of the fractions and their nonvolatile chemical nature.2 Mass spectrometry (MS) is a method that can provide detailed molecular information for complex mixtures, but its ability to analyze nonvolatile petroleum fractions is limited by the sample systems available to introduce the heavy components into the mass spectrometer. The most widely used sample introduction method for heavy petroleum fractions is the heated direct insertion * Corresponding author: Research Department, Products and Chemicals Division, Imperial Oil, 453 Christina St. S., P.O. Box 3022, Sarnia, Ontario, Canada, N7T 8C8; (tel) 519-339-2441; (fax) 519-339-4436; (e-mail) [email protected]. (1) Gray, M. R. Upgrading Petroleum Residues and Heavy Oils; Marcel Dekker: New York, 1994. (2) Altgelt, K. H.; Boduszynski, M. M. Composition and Analysis of Heavy Petroleum Fractions; Marcel Dekker: New York, 1994.

1408 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

probe.3-7 The low-pressure conditions of the ionization source (e.g., 10-6 Torr) and the ability to heat the probe to high temperatures (e.g., 650 °C) enable the vaporization of high-boiling crude oil components. Unfortunately, even under these conditions, considerable portions of the petroleum residue may remain unvolatilized. Additionally, large and thermally labile structures tend to decompose under the high probe temperature regimes and provide partial information, only about the molecular fragments and not the intact molecular species in the sample. An ideal sample introduction method would introduce heavy petroleum components into a mass spectrometer independently of their boiling properties. Field ionization mass spectrometry (FI MS) has been used to characterize heavy petroleum fractions8-12 based on its important property to produce intact molecular ions even for unstable molecules such as paraffins.13-15 However, the ability of FI MS to analyze nonvolatile petroleum fractions is limited since it relies on the use of thermal processes for sample introduction (e.g., direct insertion probe, all-glass heated inlet system, gas chromatographic interface, etc). The related technique of field desorption mass spectrometry16 (FD MS), where the sample is directly deposited on the FD emitter, has shown a greater potential for the analysis of nonvolatile petroleum components,17-20 but the difficulty associated with the manual loading of sample on a fragile (3) Schronk, L. R.; Grigsby, R. D.; Scheppele, S. Anal. Chem. 1982, 54, 748755. (4) Veloski, G. A.; Lynn, R. J.; Sprecher, R. F. Energy Fuels 1997, 11, 137143. (5) Grigsby, R. D.; Green, J. B. Energy Fuels 1997, 11, 602-609. (6) Fafet, A.; Bonnard, J.; Prigent, F. Oil Gas Sci. Technol. 1999, 54, 453-462. (7) Roussis, S. G. Rapid Commun. Mass Spectrom. 1999, 13, 1031-1051. (8) Boduszynski, M. M. Energy Fuels 1987, 1, 2-11. (9) Boduszynski, M. M. Energy Fuels 1988, 2, 597-613. (10) Wolf, P. H.; Severin, D. Fresenius Z. Anal. Chem. 1987, 329, 35-38. (11) Ueda, K.; Matsui, H.; Malhotora, R.; Nomura, M. Sekiyu, Gakkaishi 1991, 34, 62-70. (12) del Rio, J. C.; Philp, R. P. Org. Geochem. 1999, 30, 279-286. (13) Gallegos, E. J.; Fetzer, J. C.; Carlson, R. M.; Pena, M. M. Energy Fuels 1991, 5, 376-381. (14) Liang, Z.; Hsu, C. S. Energy Fuels 1998, 12, 637-643. (15) Briker, Y.; Ring, Z.; Iacchelli, A.; McLean, N.; Rahimi, P. M.; Fairbridge, C.; Malhotra, R.; Coggiola, M. A.; Young, S. E. Energy Fuels 2001, 15, 23-27. (16) Prokai, L. Field Desorption Mass Spectrometry; Marcel Dekker: New York, 1990. (17) Larsen, B. S.; Fenselau, C. C.; Whitehurst, D. D.; Angelini, M. M. Anal. Chem. 1986, 58, 1088-1091. (18) Aczel, T.; Laramee, J. A.; Hansen, G. J. Proceedings of the 30th ASMS Conference on Mass Spectrometry and Allied Topics, Honolulu, June 6-11, 1982; pp 808-809. 10.1021/ac011002k CCC: $22.00

© 2002 American Chemical Society Published on Web 02/14/2002

emitter tends to discourage its use for the routine analysis of samples. The use of thermospray ionization mass spectrometry (TSP MS) has been reported to produce results in good agreement with those obtained by FD MS (1120-1305 °F vacuum residue fraction).21 It was demonstrated that TSP can successfully interface mass spectrometry to normal-phase liquid chromatography for the selective ionization of high-boiling aromatic hydrocarbons. The approach has not been applied extensively to the characterization of heavy petroleum fractions, however, due perhaps to sensitivity limitations and problems associated with the analysis of polar components. Particle beam (PB)22 is another sample introduction method that interfaces liquid chromatography to mass spectrometry and has been used for the introduction of heavy hydrocarbons into the mass spectrometer.23 An important advantage of the PB interface is its ability to produce mass spectra with conventional ionization sources (e.g., electron and chemical ionization). Current PB interface designs, however, have been reported to suffer from sensitivity and linearity limitations.22-24 Atmospheric pressure chemical ionization (APCI)25,26 was developed for the interfacing of high-pressure chromatographic systems to mass spectrometers and has been used for the analysis of heavy polycyclic aromatic hydrocarbons.24,27,28 We have found, however, that the APCI spectra of hydrocarbon mixtures can be difficult to interpret due to the complexity of the competing mechanisms leading to the formation of several possible ions depending on the structures and sizes of the analyte molecules.29 In this work, we have examined the capabilities of electrospray ionization mass spectrometry (ESI MS)30 for the analysis of neutral (apolar) hydrocarbons in heavy petroleum fractions. Over the last 10 years, ESI MS has become an invaluable method for the characterization of polar and ionic compounds by successfully coupling high-pressure liquid chromatography to mass spectrometry. The original applications of ESI were mainly in the area of the health sciences.31 More recently, the method has been applied to the analysis of polar and ionic compounds in petroleum mixtures.32-34 However, little has been reported on its use in (19) Reynolds, S. D.; Aczel, T. Proceedings of the 33rd ASMS Conference on Mass Spectrometry and Allied Topics, San Diego, CA, May 26-31, 1985; pp 656657. (20) Aczel, T.; Dennis, L. W.; Reynolds, S. D.Proceedings of the 35th ASMS Conference on Mass Spectrometry and Allied Topics, Denver, CO, May 2429, 1987; pp 1066-1067. (21) Hsu, C. S.; Qian, K. Energy Fuels 1993, 7, 268-272. (22) Creaser, C. S.; Stygall, J. W. Analyst 1993, 118, 1467-1480. (23) Pace, C. M.; Betowski, L. D. J. Am. Soc. Mass Spectrom. 1995, 6, 597-607. (24) Anacleto, J. F.; Ramaley, L.; Benoit, F. M.; Boyd, R. K.; Quilliam, M. A. Anal. Chem. 1995, 67, 4145-4154. (25) Carroll, D. I.; Dzidic, I.; Stillwell, R. N.; Haegele, K. D.; Horning, E. C. Anal. Chem. 1975, 47, 2369-2373. (26) Duffin, K. L.; Wachs, T.; Henion, J. D. Anal. Chem. 1992, 64, 61-68. (27) Marvin, C. H.; McCarry, B. E.; Villella, J.; Bryant, D. W.; Smith, R. W. Polycyclic Aromat. Compd. 1996, 9, 193-200. (28) Lafleur, A. L.; Taghizadeh, K.; Howard, J. B.; Anacleto, J. F.; Quilliam, M. J. Am. Soc. Mass Spectrom. 1996, 7, 276-286. (29) Roussis, S. G.; Fedora, J. W. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, June 13-17, 1999; MPG 159. (30) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Mass Spectrom. Rev. 1990, 9, 37-90. (31) Cole, R. B. Electrospray Ionization Mass Spectrometry: Fundamentals, Instrumentation and Applications; John Wiley & Sons: New York, 1997. (32) Roussis, S. G.; Fedora, J. W. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, June 13-17, 1999; MPG 158.

analyzing neutral aromatic hydrocarbons in heavy petroleum fractions.35 ESI MS has the potential of two capabilities important to the analysis of heavy petroleum fractions: (1) accurate determination of molecular weight distribution profiles by eliminating the likelihood for thermal decomposition of the precursor molecules since vaporization and ionization are not based on thermal processes and (2) improved chemical speciation by direct coupling with liquid chromatographic methods of separation. Since neutral hydrocarbon molecules cannot be directly analyzed by ESI, we have examined the possibility of forming adduct ions by complexation reactions with selected reagent compounds. The reagent compounds are added in weak solutions of the hydrocarbon mixtures in suitable solvents or in the mobile phase used to transfer the analytes through the ESI interface. Several potential reagent compounds were examined: ammonium acetate, ammonium fluoride, silver nitrate, copper acetate, and others. The most promising results were obtained by using silver nitrate. In this work, the use of silver nitrate as a reagent system for the analysis of aromatic hydrocarbons in heavy petroleum fractions by ESI MS is presented. The methods used and the results obtained from the analysis of representative samples are given below. EXPERIMENTAL SECTION Mass Spectrometer. All experiments were carried out on a ZabSpec Ultima tandem double-focusing magnet sector/orthogonal acceleration time-of-flight (oa-TOF) mass spectrometer (Micromass Ltd., Manchester, U.K.).36 Mass spectra were obtained by scanning the fully laminated Ultima magnet (field strength ∼1.5 T) over the mass range of interest (e.g., m/z 800-100 or m/z 7000-100). Typical scanning rates were 1-20 s/mass decade. The maximum accelerating voltage used was 4000 V (i.e., upper mass limit m/z ∼8000). The potential difference between the electrospray needle and the counter electrode (pepper pot) was ∼3000 V. Nitrogen was used as both bath and nebulizer gas. The ESI interface temperature was maintained at 90 °C. The needle, sampling cone, skimmer lens, and ring electrode voltages were adjusted to optimize sensitivity. The degree of dimer adduct ion formation could be controlled by the sampling cone voltage. Sampling cone voltage settings ranged between 10 and 90 V. Poly(propylene glycol) (PPG) mixtures were used to calibrate the mass scale in the positive ion mode. The instrument was tuned to ∼1500 resolving power (10% valley criterion). Sample Preparation and Introduction. Solutions (∼1 mg/ mL) of silver nitrate, phenylnonane, phenyldecane, and chrysene were prepared by dissolving the samples in mixtures of methanol and cyclohexane (0.7:0.3). Solutions (∼1 mg/mL) of the petroleum fractions and coronene were prepared in toluene (100%). The hydrocarbon solutions were mixed in equal volumes with the silver nitrate solution and were injected directly or following further dilution with the methanol/cyclohexane solvent mixture. The same solvent mixture was used as the mobile phase for the (33) Roussis, S. G. Proceedings of the 48th ASMS Conference on Mass Spectrometry and Allied Topics, Long Beach, CA, June 11-15, 2000; WOB 3:40. (34) Qian, K.; Rodgers, R. P.; Hendrickson, C. L.; Emmett, M. R.; Marshall, A. G. Energy Fuels 2001, 15, 492-498. (35) Rudzinski, W. E.; Sassman, S. A.; Watkins, L. M. Prepr.-Am. Chem. Soc., Div. Pet. Chem. 2000, 45, 564-566. (36) Bateman, R. H.; Green, M. R.; Scott, G.; Clayton, E. Rapid Commun. Mass Spectrom. 1995, 9, 1227-1233.

Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1409

introduction of the samples into the instrument. Introduction of the model compounds was also possible by methanol (100%). A mixture of methanol and toluene (0.7:0.3) was equally able to introduce the vacuum residues into the instrument. Continuous infusion experiments were done using dilute solutions of the samples (1-10 µg/mL). Flow injection experiments were also done with dilute solutions of the samples (1-10 µg/mL) using the silver nitrate solution (1-10 µg/mL) as the mobile phase. Equivalent qualitative results were obtained by the different methods of sample introduction. Samples were introduced into the mass spectrometer via a Hewlett-Packard 1090 liquid chromatographic system or a Harvard model 22 syringe pump. Flow rates ranged between 2 and 50 µL/min. The injection volume was 20 µL. Samples. Model compounds were obtained from Aldrich. Heavy petroleum fractions and standards were obtained from Imperial Oil, Products and Chemicals (Sarnia, ON, Canada). RESULTS AND DISCUSSION The possibility of using ESI MS for the analysis of neutral polycyclic aromatic hydrocarbons (PAHs) was previously examined by Van Berkel and co-workers,37,38 who studied the formation of radical ions in solution by charge-transfer complexation reactions between the PAHs and selected reagent compounds such as trifluoroacetic acid (TFA) and antimony pentafluoride. They found that the degree of ionization and selectivity was dependent on the oxidizing strength of the electron-transfer reagent used. The approach, however, has not found extensive use due perhaps to the handling problems associated with strong oxidizing reagents such as antimony pentafluoride, which is highly toxic and corrosive. More recently, Airiau et al.39 reported the use of tropylium tetrafluoroborate for the postcolumn derivatization of PAHs and subsequent ESI-MS/MS analysis. The formation of radical cations by charge transfer was preferred over complex formation with the tropylium cation for the PAHs investigated (anthracene, pyrene, 1,2-benzanthracene). The capabilities of this reagent system for the ionization of petroleum fractions are not known. In the current work, we have used silver nitrate as reagent compound for the analysis of heavy aromatic petroleum fractions by ESI MS. The system is simple and has been used before in laser desorption (LD MS) experiments to analyze nonboiling, apolar hydrocarbon polymers (polystyrene, polybutadiene, polyethylene, etc.).40 Ag+ ESI Mass Spectra of Model Aromatic Compounds. Representative aromatic structures, commonly found in heavy petroleum fractions, have been employed to obtain general information about the features of ESI mass spectra when silver nitrate is used as reagent compound. The first set of model compounds examined (i.e., chrysene and coronene) are representative PAHs that provide information about the Ag+ ESI mass spectra of unsubstituted aromatic nuclei. The concentration of these structures in petroleum fractions is small, since the majority of the petroleum structures contain alkyl substituents or functional (37) Van Berkel, G. J.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1991, 63, 2064-2068. (38) Van Berkel, G. J.; Asano, K. G. Anal. Chem. 1994, 66, 2096-2102. (39) Airiau, C. Y.; Brereton, R. G.; Crosby, J. Rapid Commun. Mass Spectom. 2001, 15, 135-140. (40) Kahr, M. S.; Wilkins, C. L. J. Am. Soc. Mass Spectrom. 1993, 4, 453-460.

1410 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

Figure 1. Ag+ ESI mass spectra of chrysene (MW 228) acquired at different cone voltages: (A) 30 V; (B) 50 V.

groups; however, the results can be compared with available results from different studies in the literature. The second set of model compounds examined (phenylnonane, phenyldecane) is more pertinent to the characterization of petroleum fractions that contain mostly substituted aromatic structures. (1) Unsubstituted Aromatics. The Ag+ ESI mass spectra of chrysene obtained at 30- and 50-V cone voltages are shown in panels A and B of Figure 1, respectively. The lower cone voltage settings favor the formation of adduct ions, possibly involving the solvent molecules. For example, the spectrum in Figure 1A contains a peak at m/z 228, which corresponds to the molecular ion of chrysene. However, the most abundant peak in the spectrum of Figure 1A is at m/z 367/369, which corresponds to the [M + Ag + 32]+ adduct ion formed by the chrysene molecule, the silver ion, and a species of mass 32, which may be methanol. The attachment of water and methanol to Ag(I) complexes has been observed with amino acids.41 The presence of silver in the ion structure is readily recognized from its characteristic isotopic pattern (e.g., m/z 367 and 369). Increasing the cone voltage from 30 to 50 V considerably reduces the relative amount of the peak at m/z 367/369 (∼10% of base peak), producing the molecular ion peak at m/z 228 as the most abundant peak in the spectrum (Figure 1B). The facile dissociation of the peak at m/z 367/369 indicates that [M + Ag + 32]+ is a weakly bound complex. The [2M + Ag]+ dimer abundance appears to be unaffected by the changes in the cone voltage (Figure 1AB). The most abundant ion in the 30-V cone voltage spectrum of coronene corresponds to the [2M + Ag]+ dimer adduct ion at (41) Perera, B. A.; Ince, M. P.; Talaty, E. R.; Stipdonk, M. J. Rapid Commun. Mass Spectrom. 2001, 15, 615-622.

m/z 707/709 (base peak). Lower abundance peaks are observed for the molecular ion of coronene at m/z 300 (∼30% of base peak) and the [M + Ag + 32]+ adduct ion at m/z 439/441 (∼25% of base peak). The increase of the cone voltage to 50 V produces the molecular ion peak at m/z 300 as the most abundant peak in the spectrum, but there is little effect on the relative abundance of the [M + Ag + 32]+ adduct ion at m/z 439/441. The [2M + Ag]+ peak abundance at m/z 707/709 is reduced to ∼45% at the higher cone voltage setting. In the case of coronene, the [2M + Ag]+ dimer structure is less stable than the [M + Ag + 32]+ complex. In contrast to the abundant [2M + Ag]+ ion (base peak) observed in the spectrum of coronene (30 V cone voltage), the abundance of the [2M + Ag]+ ion in the spectrum of chrysene (30 V cone voltage) is less than 10% (Figure 1A). The big difference in the abundance of the [2M + Ag]+ ion in the two spectra can be attributed to the structural differences (size and electronic properties) of the two compounds. The formation of the [2M + Ag]+ ion is more favored in the case of the large coronene structure but not in the case of the smaller chrysene structure. The tendency of coronene to form dimer (sandwich) complexes with atomic ions is well known from studies employing laser desorption experiments.42,43 The present results indicate that it may be possible to derive structure-specific information from the spectra of unsubstituted PAHs. The relative peak abundances provide information about the stabilities of the different adduct ions and their probabilities of formation. Additional work with isomeric structures is needed to further investigate these observations. (2) Alkyl-Substituted Aromatics. The Ag+ ESI mass spectra obtained for phenylnonane at cone voltages 30 and 70 V are shown in Figure 2A and B, respectively. The most abundant peak in the 30 V spectrum at m/z 311/313 corresponds to the [M + Ag]+ ion (Figure 2A). Increasing of the cone voltage to 70 V reduced the [2M + Ag]+ peak from ∼20 to ∼10% of the base peak. The [M + Ag + 32]+ peak abundance was unaffected by the cone voltage changes (∼15% of base peak). Similar spectra were obtained for phenyldecane. The most abundant peak in the phenyldecane spectrum corresponds to the [M + Ag]+ ion at m/z 325/327. The [2M + Ag]+ dimer peak was reduced from ∼10 to ∼5% at the higher cone voltage. Work at 50-V cone voltage had a smaller effect on the reduction of the [2M + Ag]+ ion abundance. Simple mass spectra were obtained for the substituted structures examined, consisting primarily of the [M + Ag]+ ion (base peak) with small contributions from the [2M + Ag]+ (10-20%) and [M + Ag + 32]+ (5-15%) ions. The likelihood for the formation of the [M + Ag]+ structure is considerably higher than that for the formation of the [2M + Ag]+ and [M + Ag + 32]+ ion structures. The formation of stable adduct ions may be facilitated by the participation of solvent molecules. Detailed experimental and theoretical studies are needed to elucidate the mechanisms responsible for the observed spectra. Analysis of a Heavy Aromatic Fraction by Ag+ ESI MS. The spectra obtained for a heavy aromatic fraction (boiling range 343-565 °C) by Ag+ ESI MS at 30- and 70-V cone voltage settings are shown in Figure 3A and B, respectively. An abundant distribution of peaks is present in the mass range m/z ∼450 to (42) Pozniak, B. P.; Dunbar, R. C. J. Am. Chem. Soc. 1997, 119, 10439-10445. (43) Buchanan, J. W.; Grieves, G. A.; Flynn, N. D.; Duncan, M. A. Int. J. Mass Spectrom. 1999, 185-187, 617-624.

Figure 2. Ag+ ESI mass spectra of phenylnonane (MW 204): (A) 30-V cone voltage; (B) 70-V cone voltage.

∼900 and a lower abundance (∼20%) distribution of peaks is present between m/z ∼900 and ∼1600 (Figure 3A). The relative peak abundances of the higher mass distribution decrease as the cone voltage is increased (Figure 3B). In-source dissociation of the adduct ions is possible by increasing the cone voltage. (Figure 3B). Given the lack of fragment ions in the spectra of the model alkylbenzenes at the higher cone voltage settings, the nature of the precursor ions leading to the observed fragment ions is not currently known. In-source fragmentation may be a function of the alkyl chain length or the degree of isomerization. The mass spectrum obtained for the aromatic fraction by Ag+ ESI (70 V cone voltage) (Figure 4A) is compared to the low-energy (∼10 eV) mass spectrum (Figure 4B) obtained using charge exchange (CE) chemical ionization with CS2 as reagent compound. The low-energy CS2 CE spectrum contains primarily molecular ions with negligible fragmentation7 and is used for the evaluation of the data obtained by Ag+ ESI. The direct insertion probe was used to introduce the sample for CS2 CE analysis. The details of the CS2 CE experiment can be found elsewhere.7 The probe temperature was raised from ambient to 380 °C at a rate of 10 °C/min. The ionization source temperature was 200 °C. A composite mass spectrum was obtained by the summation of all spectra acquired in the run. The CS2 CE and Ag+ ESI spectra are complex and visual comparison of the patterns is difficult (Figure 4). However, both spectra show similarities in that they contain clusters of peaks separated by 14 mass units, which correspond to homologues of the same compound types separated by one CH2 group. Both spectra show abundant peaks separated by two mass units as expected from structures differing by one degree Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1411

Figure 3. Ag+ ESI mass spectra of heavy aromatic fraction (boiling range 343-565 °C): (A) 30-V cone voltage; (B) 70-V cone voltage.

of unsaturation. An expanded view of the two spectra is shown in Figure 5. The mass at m/z 526 in the CS2 CE spectrum (Figure 5B) is known from previous high-resolution work to partly correspond to the C38H70+ alkylaromatic structure. Figure 5A contains an abundant peak at m/z 633, which can partly be attributed to the argentinated C38H70Ag+ ion structure. An equally abundant peak is observed at m/z 647, which corresponds to the heavier homologue C39H72Ag+. At a first approximation, the ion abundance in the polyisotopic spectrum (e.g., m/z 633 in Figure 5A) contains contributions from [M + 107Ag]+ (i.e., M ) 526 in Figure 5B) and [M - 2 + 109Ag]+ (i.e., M - 2 ) 524 in Figure 5B). By starting from the lowest mass peak [M + 107Ag]+ in the spectrum and assuming it contains no 109Ag+ contributions, it is possible to correct the polyisotopic spectrum by subtracting the [M + 109Ag]+ contribution of the lower mass from the abundance of the higher mass peak. Correction in that manner of the spectrum in Figure 5A reveals similar increasing ion abundance trends for the m/z 518-526 and 625-633 mass ranges in the Ag+ ESI and CS2 CE spectra, respectively (Figure 5A and B). The pattern repeats for the higher mass peaks (e.g., m/z 530-540 and 637-647 mass ranges). The results indicate that the masses of the peaks in Figure 5A correspond to the argentinated structures of the peaks shown in Figure 5B. This is consistent with the fact that the majority of the aromatic structures in petroleum fractions contain alkyl substituents. Based on the results with the model alkylaromatic compounds, the most abundant peaks are expected to be primarily due to the [M + Ag]+ monomers (lower mass distribution in Figure 3A), with small contributions from the [2M + Ag]+ dimers (higher mass distribution in Figure 3A) and the [M + Ag 1412 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

Figure 4. Molecular weight distributions of heavy aromatic fraction obtained by (A) Ag+ ESI MS (70-V cone voltage) and (B) CS2 charge exchange.

+ 32]+ structures. Contributions from the molecular and other ions are also expected. High-energy (400 eV) oa-TOF tandem mass spectrometry (MS/MS) experiments of selected precursor ions (e.g., m/z 605, 619, etc.) produced product ion spectra containing abundant alkyl fragment ions (e.g., m/z 43, 55, 57, 69, 71, etc.) and characteristic fragment ions of the alkylbenzene series (e.g., m/z 105, 119, 133, etc.). The MS/MS spectra also contained abundant peaks at m/z 107 and 109 (silver atom isotopes). The degree of fragmentation was controlled by the oa-TOF collision parameters. The MS/MS results are in agreement with the expected identity of the selected precursor ions in the heavy aromatic fraction. The molecular weight distribution maximum in the Ag+ ESI spectrum is obtained at m/z 647 (Figure 4A) whereas in the CS2 CE spectrum it is at m/z 484 (Figure 4B). Assuming that the predominant compound types in the sample are due to the CnH2n-6 alkylbenzene series, then the maximum of the distribution in the CS2 CE spectrum corresponds to C35H64 whereas the maximum in the Ag+ ESI spectrum corresponds to C39H72. The Ag+ ESI molecular weight distribution is shifted to the higher masses by four carbon numbers in comparison to the CS2 CE distribution. This may be due to differences in the relative sensitivities of the two methods for the various aromatic compound types or to discrimination effects during sample introduction. Uniform sensitivities are assumed for both methods. This is a good assumption for CS2 CE,44 but extensive studies with model compounds are needed to determine the relative sensitivities for Ag+ ESI. (44) Hsu, C. H.; Qian, K. Anal. Chem. 1993, 65, 767-771.

Figure 5. Expanded view of molecular weight distributions shown in Figure 4 for the heavy aromatic fraction: (A) Ag+ ESI MS (70-V cone voltage); (B) CS2 charge exchange.

Discrimination effects during sample introduction are less likely for Ag+ ESI and more likely for CS2 CE. Hsu and Qian44 found that the ionization source temperature affects the relative sensitivities of CS2 CE for the detection of model PAHs. A semiquantitative comparison of the results was obtained by using the low-resolution data obtained from the spectra in Figure 4 and correcting for the contributions of the silver isotopes using a computer program. Since low-resolution data were used for the calculations, only accumulated compound types of the series Z ) -6 to Z ) -18 in the general hydrocarbon CnH2n+Z formula were evaluated. More accurate comparisons will require high-resolution experiments to resolve the overlapping compound types. The contributions of the [2M + Ag]+, [M + Ag + 32]+, and [M]+ ions were neglected in the calculations. A good agreement was obtained between the two methods at the current level of approximations. The maximum difference in the percent amount determined by the two methods was obtained for the Z ) -6 series (∼3%) followed by the Z ) -18 series (∼2%). Accurate determination of the chemical compositions of the compounds in the Ag+ ESI mass spectrum will require advanced high resolving power experiments. Molecular Weight Distributions of Vacuum Residues by Ag+ ESI MS. The Ag+ ESI mass spectrum of a vacuum residue is shown in Figure 6A. The residue was collected after vacuum distillation of a crude oil at 565 °C atmospheric pressure equivalent temperature. The same vacuum residue was analyzed using FD MS. Approximately 1 µL of a dilute solution (∼0.5 mg/mL) of the residue in CS2 was deposited on the FD emitter, and the solvent was allowed to evaporate. The FD probe was then

Figure 6. Molecular weight distributions of petroleum vacuum residue obtained by (A) Ag+ ESI MS (70-V cone voltage) and (B) field desorption mass spectrometry (FD MS). The residue was obtained after vacuum distillation at ∼565 °C atmospheric pressure equivalent temperature.

introduced into the mass spectrometer, and the emitter current was manually raised to a maximum of 100 mA. The spectrum shown in Figure 6A was obtained by the summation of ∼70 mass spectra acquired in the continuum mode. To obtain the Ag+ ESI spectrum of the vacuum residue, 20 µL of a solution prepared by mixing equal volumes of the vacuum residue (∼1 mg/mL in toluene) and silver nitrate (∼1 mg/mL in methanol/toluene (0.7: 0.3)) was injected into the HP LC unit. A methanol/toluene (0.7: 0.3) mixture was used as mobile phase. Addition of toluene or cyclohexane to methanol was necessary to elute the heavy hydrocarbon components through the interface. The Ag+ ESI spectrum shown in Figure 6A was obtained by the summation of ∼15 mass spectra acquired in the continuum mode. The Ag+ ESI and FD spectra of the vacuum residue shown in Figure 6 are extremely complex. Determination of the chemical nature of the individual compounds in the spectra is not possible under the low resolving power conditions used in the current study. However, information can be obtained from the spectra about the molecular weight distributions of the vacuum residue. The molecular weight distribution obtained by FD MS (Figure 6B) ranges between m/z ∼400 and 2000 with a maximum at m/z ∼700. A similar molecular weight distribution was obtained by Ag+ ESI MS (Figure 6A) for the vacuum residue. However, fragment ions are observed in the low-mass range (i.e., below m/z 400) due to the relatively high cone voltage setting (70 V) used for the experiments (Figure 6A). The high cone voltage was used as in the analysis of the heavy aromatic fraction to reduce the Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

1413

indicating that there are compounds in the sample with molecular weights even higher than 7000. The nature of these compounds is not currently known. The high cone voltage setting was used to cause the fragmentation of the [2M + Ag]+ dimer complexes, which appears to be successful from the lack of a second peak distribution at m/z ∼1600. Due to the extreme complexity of the samples, it is highly difficult to detect differences in the Ag+ ESI spectra of the two residues shown in Figure 7. The only easily discernible differences in the spectra are the fragment ion patterns shown in the mass range below m/z ∼600. Several repeat analyses showed that it is more difficult to produce fragment ions from vacuum residue B than it is from vacuum residue A. Compositional characterization of the vacuum residues will require the use of ultrahigh resolving power instrumentation. The ability of Ag+ ESI MS to analyze heavy apolar aromatic hydrocarbons is significant because it extends the application range of the method. Most importantly, however, it provides an alternate method for the analysis of nonboiling petroleum fractions. The current experiments showed that vacuum residues have a molecular weight distribution that starts at m/z ∼400 and extends to higher than m/z 7000. The molecular weight distributions are in qualitative agreement with the FD MS results, which show a peak distribution maximum at m/z ∼700. The Ag+ ESI MS results support molecular weight findings using laser desorption methods of analysis.45,46 Figure 7. Molecular weight distributions of petroleum vacuum residues obtained by Ag+ ESI MS (70V cone voltage): (A) vacuum residue A; (B) vacuum residue B.

relative concentrations of the dimer complexes. As expected, the distribution maximum in the Ag+ ESI spectrum (Figure 6A) is shifted to higher masses, in comparison to the FD spectrum (Figure 6B), by ∼100 mass units (i.e., one silver atom). This mass shift corresponds to the formation of the [M + Ag]+ adduct ions. Although the signal in the FD spectrum returns to the baseline at m/z ∼2000 (Figure 6B), the signal in the Ag+ ESI spectrum remains at ∼15-20%, indicating the possible detection of higher molecular weight compounds by the ESI method. The detection of compounds having molecular weights up to ∼5000 is shown in Figure 7 for the analysis of two vacuum residues. A smaller portion of the vacuum residue A spectrum (mass range up to m/z 2000) was shown in Figure 6A. The signal for both residues does not return to the baseline (Figure 7). Extending the upper mass limit to 7000 produced similar results, (45) Seki, H.; Kumata, F. Energy Fuels 2000, 14, 980-985. (46) Suelves, I.; Islas, C. A.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2001, 15, 429-437.

1414 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002

CONCLUSIONS Heavy petroleum fractions can be analyzed by ESI MS using silver nitrate as reagent compound. Aromatic compounds react with the silver ion to form abundant adduct ions such as [M + Ag]+ and [2M + Ag]+. Characteristic spectra were obtained for unsubstituted polycyclic aromatic compounds. Formation of the adduct ions may involve the participation of the solvent molecules. Aromatic compounds containing long alkyl groups form primarily [M + Ag]+ ions and small amounts of [2M + Ag]+ ions. The abundance of the [2M + Ag]+ ions can be controlled by the ESI source cone voltage. Good agreement was obtained between conventional methods of analysis and Ag+ ESI for representative heavy and nonboiling petroleum fractions. A significant advantage of the Ag+ ESI method over FD MS and LD MS for the analysis of vacuum residues is its direct compatibility with LC methods of separation. Future work will deal with the development of LC/ MS and high-resolution MS methods for the detailed characterization of vacuum residues. Received for review September 17, 2001. Accepted December 21, 2001. AC011002K