Compositional Study of Polar Species in Untreated and Hydrotreated

Oct 7, 2009 - Department of Chemistry, University of Joensuu, Post Office Box 111, 80101 ... Technology Centre, Neste Oil Oyj, Post Office Box 310, 06...
0 downloads 0 Views 3MB Size
Energy Fuels 2009, 23, 6055–6061 Published on Web 10/07/2009

: DOI:10.1021/ef9007592

Compositional Study of Polar Species in Untreated and Hydrotreated Gas Oil Samples by Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (ESI FTICR-MS) Timo Kek€ al€ ainen,† Jaana M. H. Pakarinen,† Kim Wickstr€ om,‡ and Pirjo Vainiotalo*,† †

Department of Chemistry, University of Joensuu, Post Office Box 111, 80101 Joensuu, Finland, and ‡Technology Centre, Neste Oil Oyj, Post Office Box 310, 06101 Porvoo, Finland Received July 21, 2009. Revised Manuscript Received September 22, 2009

Two gas oil samples, untreated feed and hydrotreated product oil, were analyzed. Both basic and acidic polar species were detected by electrospray ionization (ESI) Fourier transform ion cyclotron resonance mass spectrometry, and the detected species were characterized on the basis of their elemental compositions. Samples were real refinery samples, with one sample being a certain distillation fraction of crude oil (feed) and the other sample being a hydrotreated feed oil (product). Comparison of the compositions of untreated and hydrotreated oil provides insight into (1) compounds that are resistant to processing, (2) compounds that are removed/degraded by processing, and (3) new compounds that are produced during processing. N1 class compounds were found to be the most abundant basic species in both oil samples. In addition, the proportion of N1 class compounds was clearly greater in the product oil than in the feed oil, which indicates that these basic species must be resistant to removal by hydrotreatment. All basic NxOy-, NxSz-, and NxOySz-containing compounds that were detected in the feed oil were completely removed by hydrotreatment. However, some of the OySz compounds remained in the oil after hydrotreatment. Negative-ion ESI revealed that the majority of the acidic polar species in the product sample were N1 compounds, which was also the predominant class in the feed sample. The second most abundant species were the Oy-containing compounds. All of the O1 compounds that were detected in the feed oil were degraded during the hydrotreatment process, as were the O2 compounds with double-bond equivalent (DBE) values >4. Acidic NxOy-, NxSz-, OySz-, and NxOySz-containing compounds were not completely removed by the processing, but the relative abundances of these species were no longer significant in the product oil.

chemical ionization (CI),6 field desorption/field ionization (FD/FI),7 matrix-assisted laser desorption ionization (MALDI),8 laser-induced acoustic desorption (LIAD),9 atmospheric pressure laser ionization (APLI),10 atmospheric pressure chemical ionization (APCI),11 and atmospheric pressure photoionization (APPI),12 have been successfully used to analyze petroleum samples. The complexity of petroleum poses a considerable challenge to oil refining and processing (as well as for the analysis of petroleum samples). In particular, despite the fact that they constitute only a small proportion of petroleum, polar NxOySz (nitrogen/oxygen/sulfur-containing) compounds have significant implications for oil refining. Thus, polar NxOySz species cause problems in oil production, refining,

Introduction A lack of knowledge of the chemical reactivity and properties of petroleum species has limited the improvement of refining efficiency. A new field of petroleomics started from the premise that the development of better technology and catalysts requires the detailed identification of petroleum species. Petroleomics includes the idea that the specific chemical composition should correlate with the properties and behavior of petroleum and its products.1-3 Electrospray ionization (ESI) interfaced with Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) is a suitable tool for the analysis of petroleum samples; ESI selectively ionizes polar heteroatomic compounds within the hydrocarbon matrix of petroleum samples, and FTICR-MS provides a high mass accuracy and mass resolving power.4 Moreover, several other ionization techniques, such as electron ionization (EI),5

(6) Roussis, S. G.; Proulx, R. Energy Fuels 2004, 18, 685–697. (7) Schaub, T. M.; Rodgers, R. P.; Marshall, A. G.; Qian, K.; Green, L. A.; Olmstead, W. N. Energy Fuels 2005, 19, 1566–1573. (8) Trejo, F.; Ancheyta, J.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2007, 21, 2121–2128. (9) Crawford, K. E.; Cambell, J. L.; Fiddler, M. N.; Duan, P.; Qian, K.; Gorbaty, M. L.; Kentt€amaa, H. I. Anal. Chem. 2005, 77, 7916– 7923. (10) Schmitt-Kopplin, P.; Engelmann, M.; Rossello-Mora, R.; Schiewek, R.; Brockman, K. J.; Benter, T.; Schmitz, O. J. Anal. Bioanal. Chem. 2008, 391, 2803–2809. (11) Hsu, C. S.; Dechert, G. J.; Robbins, W. K.; Fukuda, E. K. Energy Fuels 2000, 14, 217–223. (12) Purcell, J. M.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. Anal. Chem. 2006, 78, 5906–5912.

*To whom correspondence should be addressed. E-mail: pirjo. [email protected]. (1) Marshall, A. G.; Rodgers, R. P. Acc. Chem. Res. 2004, 37, 53–59. (2) Rodgers, R. P.; Schaub, T. M.; Marshall, A. G. Anal. Chem. 2005, 77, 21A–27A. (3) Marshall, A. G.; Rodgers, R. P. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18090–18095. (4) Marshall, A. G.; Hendrickson, C. L.; Jackson, G. S. Mass Spectrom. Rev. 1998, 17, 1–35. (5) Fu, J.; Klein, G. C.; Smith, D. F.; Kim, S.; Rodgers, R. P.; Hendrickson, C. L.; Marshall, A. G. Energy Fuels 2006, 20, 1235– 1241. r 2009 American Chemical Society

6055

pubs.acs.org/EF

Energy Fuels 2009, 23, 6055–6061

: DOI:10.1021/ef9007592

Kek€al€ainen et al.

and storing because, for example, they contribute to catalyst deactivation,13-15 corrosion,16 and storage instability.17,18 In addition, nitrogen oxides (NxOy) and sulfur oxides (OySz), which are released during combustion, are two major atmospheric pollutants.19 Catalytic hydrotreatment is one of a few practical methods for the removal of polar NxOySz species from petroleum.20-23 The term hydrotreatment encompasses several different processes, such as hydrodenitrogenation, hydrodesulfurization, and hydrodemetalation, that occur simultaneously during its implementation. Molybdenum (Mo) with nickel (Ni) or cobalt (Co) has been mostly used as the catalyst, in which case the active form of the catalyst is molybdenum sulfide (MoS2). Aluminum oxide (Al2O3) or silica-alumina with zeolites have commonly been used as supports. The final results are dependent upon the catalyst used and the conditions of the hydrotreatment.24-27 Unfortunately, the complete removal of polar NxOySz compounds is very difficult. Hydrodenitrogenation occurs only at saturated rings, which means that nitrogen-containing unsaturated heterocyclic rings must first be saturated before carbon-nitrogen bond cleavage can take place.20,28,29 It is for this reason that some Nx-containing species are so difficult to remove from petroleum by catalytic hydrotreatment. In addition to being resistant to hydrotreatment,5,30,31 some Nx-containing compounds also promote catalyst deactivation through coke formation on the catalyst surface.14,15 In the case of sulfurcontaining species, the sulfur-carbon bond can be cleaved directly without saturation of the aromatic ring,28,29 and therefore, sulfur-containing species are much easier to remove from petroleum by catalytic processing. In this paper, we report the elemental compositions of polar species in gas oil samples, as studied by ESI FTICR-MS. The samples were real refinery samples: an untreated feed oil and a hydrotreated feed oil (product oil). Both acidic and basic polar species were detected, and the detected species in the

Table 1. Some Properties and the Compositions of the Feed and Product Samples

feed product

Mwa (g/mol)

densityb (kg/m3)

total Nc (mg/kg)

total Sd (wt %)

Br numbere (g of Br/ 100 g)

307 301

889.3 876.7

999 704

1.280 0.046

3.6 2.5) of the positiveion ESI mass spectra were imported into Microsoft Excel 2003 software, with which Kendrick masses, Kendrick mass defects, and DBE values were calculated. According to these values, the various homologous series were sorted. Molecular formulas were predicted using the Generate Molecular Formula (GMF) program of Data Analysis software. The molecular formulas were limited to 100 12C atoms, 200 1H atoms, 6 14N atoms, 4 32S atoms, and 10 16O atoms. 13C and 34S atoms were also taken into account, and the presence of some metal atoms, such as sodium, potassium, nickel, and vanadium, was also checked. The modified script of Target Analysis software (Bruker) was used to identify the negative-ion ESI mass spectra. The program goes through the whole peak list (peaks of relative intensity less than 0.35% are filtered out) and, within given parameters, suggests elemental compositions for each peak. The results were imported into Microsoft Excel software, with which Kendrick masses, Kendrick mass defects, and DBE values were calculated. According to these values and the assigned elemental compositions, the homologous series were identified and sorted. The parameters of Target Analysis were set as follows: tolerance, (3 ppm; DBE, 0-50; heteroatoms limited to 6 14N, 4 32S, (32) Moini, M.; Jones, B. L.; Rogers, R. M.; Jiang, L. J. Am. Soc. Mass Spectrom. 1998, 9, 977–980. (33) Kendrick, E. Anal. Chem. 1963, 35, 2146–2153. (34) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G.; Qian, K. Anal. Chem. 2001, 73, 4676–4681. (35) McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed.; University Science Books: Mill Valley, CA, 1993.

(36) McKay, J. F.; Weber, J. H.; Latham, D. R. Anal. Chem. 1976, 48, 891–898.

6057

Energy Fuels 2009, 23, 6055–6061

: DOI:10.1021/ef9007592

Kek€al€ainen et al.

Figure 2. Most abundant heteroatom-containing compound classes identified in feed and product samples by positive-ion ESI. Only those classes with >1% relative abundance are shown.

Figure 4. Relative abundances versus carbon number for three DBE series of basic N1 species.

compounds of the same DBE values by hydrogenation of species of higher DBE and/or by reduction of compounds with multiple heteroatoms. The second most abundant compound class in both oil samples was the N2 class, which amounted to ∼7% in the feed oil and ∼3% in the product oil. In contrast to the N1 compounds, it is clear that a considerable part of the N2 compounds was removed by the hydrotreatment. The detected DBE series were 7-18 and 10-17 in the feed and product oils, respectively. Compounds with lower aromaticity were removed from the oil and were likely degraded to N1 species. All basic NxOy-, NxSz-, and NxOySz-containing compounds that were detected in the feed oil were completely removed/degraded by hydrotreatment. In the feed sample, N1S1 compounds accounted for about 6% of the total relative abundances. The DBE distribution for the N1S1 compounds was 6-16, among which the most abundant DBE series were 10 and 11. Signals because of N1S1 species were no longer seen in the spectrum of the product sample. Three major NxOy/NxOySz classes were found in the feed sample: a N1O1 class (∼3%) with DBE series 10-18, a N2O1 class (∼2%) with DBE series 6-20, and a N1O1S1 class (∼2%) with DBE series 4-15. All species of these classes were absent from the product oil, and some previous reports support these observations.30,31 It should be kept in mind that the hydrotreatment process may only remove the oxygen and/or sulfur heteroatoms from NxSz, NxOy, and NxOySz species, thereby producing Nx-containing species.37 Three different OySz classes were detected: O1S1, O2S1, and O3S1, and each of these classes was identified in both the feed and product samples. The results indicate that these OySz compounds show some resistance to the hydrotreatment. However, the relative abundances of these compounds were remarkably lower in the product oil than in the feed oil. The O3S1 class compounds were the most abundant. The DBE distributions for this class were 2-11 in the feed and 2-8 in the product oil. The relative proportions were ∼4 and ∼1%, respectively. The alteration of DBE distribution was quite minor, but the relative proportions changed substantially. O1S1 compounds of DBE series 7-9 and 11-12 were identified in the feed sample, whereas the O2S1 compounds

Figure 3. DBE distribution for basic N1 species in the feed and product oils and the relative abundances of various DBE series.

in the samples were quite similar, which supported the idea that N1 species are hard to remove from the oil (Figure 3). The relative proportions of N1 compounds of the different DBE series were greater in the product oil than in the feed oil up to a DBE value of 13. Thereafter, the inter-relationship was opposite, indicating that the most aromatic compounds were absent from the product oil. In considering the results, it should be kept in mind that they are based on relative abundances. They do not define the actual amounts of different heteroatom-containing compounds in the oil samples. This is because some compounds are less polar and, therefore, are not as easily ionized by ESI; thus, the different compounds have different ESI responses. However, when samples are measured under the same conditions, it is possible to work out the differences between them in terms of the polar compounds that they contain. It should also be noted that the relative abundance of a certain compound class increases if the species of this class are not removed from the oil during processing, but some other classes are removed. Furthermore, most basic mononitrogen compounds have a good ESI response, but apart from N1 compounds, there were essentially very few other polar heteroatom-containing species remaining in the product oil. Carbon number distributions of various DBE series (9, 12, and 15) of the N1 compounds are shown in Figure 4. The maxima of the carbon number distributions can be seen to have shifted from C25-C35 to C30-C40 during the hydrotreatment. This result is consistent with Figure 1, in which the mass distribution of the product oil was seen to shift ∼100 mass units to a higher mass range. However, the DBE distribution of the N1 compound class, which was by far the most abundant class, remained largely unchanged (Figure 3). An explanation that accounts for these results is that the catalytic process removes compounds of certain DBE values from the oil during processing, but it also generates new

(37) Angelici, R. J. Polyhedron 1997, 16, 3073–3088.

6058

Energy Fuels 2009, 23, 6055–6061

: DOI:10.1021/ef9007592

Kek€al€ainen et al.

Figure 6. Relative abundances of all heteroatom-containing compounds identified in oil samples by negative-ion ESI.

Figure 5. Negative-ion ESI FTICR mass spectra of the (a) feed oil and (b) product oil.

consisted of DBE series 7-8 and 10-12. The product sample was found to contain O1S1 and O2S1 species uniquely of DBE series 7. Thus, the species with a DBE value of 7 were more resistant to the process than the other O1S1 and O2S1 species. The structures of these species must differ in such a way that they are sufficiently stable to withstand the hydrotreatment. A possible explanation might be that steric factors prevent the species from fitting into catalytic sites. Whatever the explanation, the OySz species were partially resistant to the hydrotreatment, which is consistent with the conclusions of some previous papers,5,31 although opposite results have also been reported.30 Different catalysts may produce divergent results. Negative-Ion ESI. The acidic (and neutral) polar species were detected by negative-ion ESI FTICR-MS (Figure 5). Overall, there were no significant differences in the appearances of the mass spectra, with the tops of the mass distributions corresponding to a m/z range of ∼300-400. The spectrum of the feed oil consisted of about 3000 peaks, of which ∼2000 could be identified. The spectrum of the product sample consisted of ∼1800 peaks, of which ∼1300 were resolved. The sum of the relative abundances of the resolved peaks was about 90% of the total relative abundances for both mass spectra. When the peak numbers in the positive-ion and negative-ion spectra were compared, the feed sample displayed more peaks in its negative-ion spectrum than in its positive-ion spectrum, whereas the reverse was true for the product sample. Thus, the hydrotreatment removes relatively more acidic species than basic species. The predominant compound class comprised the Nx compounds, especially the N1 species, which accounted for about 55% of the total relative abundances in the feed sample and almost 80% of the total in the product sample (Figures 6 and 7). The result for the basic species was very similar (Figure 2). The samples also included N2, N3, and N4 compounds, albeit only in minor amounts. The DBE distributions of the N1 compounds in the samples were 6-20 for the feed oil and 6-18 for the product oil (Figure 8). A considerable difference may be noted between the relative abundances of DBE values 8 and 9, which is consistent with some previous reports.38,39 It might be supposed that a DBE

value of 8 corresponds to two aromatic ring systems, whereas compounds with a DBE value of 9 comprise triaromatic systems. The order of the most intense DBE values was 12 > 9 > 10 in the feed sample but 9 > 10 > 12 in the product sample. Compounds with DBE values of 9 and 12 are likely to be mostly carbazoles and benzocarbazoles, respectively.39,40 Tetrahydrobenzocarbazoles and indole derivatives are possible structures for compounds with a DBE of 10.5 Tetrahydrocarbazoles have been reported to efficiently inhibit and deactivate a hydrodenitrogenation catalyst.41 The tops of the carbon number distributions for DBE series 9, 12, and 15 were seen to vary in the C22-C29 range (Figure 9). Overall, the carbon chains were slightly longer, and the most abundant species had slightly higher carbon numbers in the feed oil than in the product oil.

(38) Ter€ av€ ainen, M. J.; Pakarinen, J. M. H.; Wickstr€ om, K.; Vainiotalo, P. Energy Fuels 2007, 21, 266–273. (39) Hughey, C. A.; Rodgers, R. P.; Marshall, A. G.; Qian, K.; Robbins, W. K. Org. Geochem. 2002, 33, 743–759.

(40) Hughey, C. A.; Rodgers, R. P.; Marshall, A. G.; Walters, C. C.; Qian, K.; Mankiewicz, P. Org. Geochem. 2004, 35, 863–880. (41) Kanda, W.; Siu, I.; Adjaye, J.; Nelson, A. E.; Gray, M. R. Energy Fuels 2004, 18, 539–546.

Figure 7. Most abundant heteroatom-containing compound classes identified in the feed and product samples by negative-ion ESI. Only those classes with >1% relative abundance are shown.

Figure 8. DBE distributions for acidic (and neutral) N1-containing species in the feed and product oils and the relative abundances of various DBE series.

6059

Energy Fuels 2009, 23, 6055–6061

: DOI:10.1021/ef9007592

Kek€al€ainen et al.

Figure 9. Relative abundances versus carbon number for three DBE series of acidic (and neutral) N1-containing species.

It is generally well-known that petroleum includes carboxylic acids and phenols. Oy-containing compounds are detrimental to oil refining; naphthenic acids, for example, cause corrosion problems.16 In this study, Oy species were seen to be the second most abundant compound class after Nx species (Figure 6). It is very clear, if pKa values are considered, that nonbasic nitrogen species and most polar compounds are ionized less efficiently by deprotonation than, for example, carboxylic acids. Although Oy-containing species seemed to constitute a significant proportion of the total relative abundance in the feed sample, the actual amount of Oy-containing compounds could not be so significant to what the results indicated. In the feed sample, O1 class compounds were seen to constitute over 8% of the total sum of the relative abundances but all of these compounds were removed during the hydrotreatment process (Figure 7). The DBE distribution of the O1 species was 4-9, which relates to the fact that the detected O1 species were most likely phenols (Figure 10). The O1 class was problematic in that the resolution was insufficient to distinguish O1 and some unidentified classes and the peaks overlapped. The homologous series found in the feed oil were sorted between two classes on the grounds of intensities, carbon numbers, and DBE values. There were distinctive differences in the homologous series, and it was clear that both classes were present in the feed oil, but that the percentage values of these compound classes were not definitive. However, it is certain that none of these species were detected after hydrotreatment. O2 species were among the most abundant classes in the samples. The peaks of palmitic and stearic acids at m/z 255 and 283, respectively, have been removed from the results, because these acids are partially impurities.38 The DBE distribution was 1-11 (with DBE values 6 and 7 missing) in the feed sample, whereas the product sample included only species with DBE values of 1-4; thus, the most unsaturated compounds were removed from the petroleum during processing. Compounds with a DBE value of 1 were clearly the most abundant species (Figure 11). It could be supposed that some of the aromatic acids were degraded to non-aromatic acids, most likely naphthenic acids. However, the hydrotreatment had no significant effect on species of the O3 class (Figure 11). The DBE distributions (1-6 for both samples) were so similar that the O3 species could be deemed to remain

Figure 10. DBE versus carbon number for the O1 class compounds detected in the feed oil.

Figure 11. DBE versus carbon number for the O2 and O3 class compounds detected in the feed and product oils.

quite stable. The samples also included more highly oxygenated compounds (O4-O7), which behaved like the O3 species; DBE distributions were quite similar and the relative abundances in the product oil were slightly higher than in the feed oil; thus, the O4-O7 species seemed to be quite resistant to hydrotreatment. NxSz- and NxOySz-containing compounds in the oil were found to be almost completely removed/degraded during 6060

Energy Fuels 2009, 23, 6055–6061

: DOI:10.1021/ef9007592

Kek€al€ainen et al.

processing (Figure 6). One of the most abundant classes in the feed oil sample was N1S1-containing species (Figure 7). The compounds of the N1S1 class had DBE values of 11-15. No trace of N1S1 species was found in the product oil. NxOy-containing compounds were not completely removed from the sample oil by the catalytic process, but their proportion of the total relative abundance decreased from ∼3 to ∼2% (Figure 6). The feed oil included five different NxOy classes: N1O1, N2O4, N3O1, N3O2, and N4O1, whereas only the N1O1 and N4O1 classes were found in the product oil. The proportion of the N1O1 class compounds was clearly higher in the product oil than in the feed oil; therefore, it might be supposed that compounds with multiple heteroatoms yield N1O1 species by reduction (Figure 7). Five different OySz-containing classes were detected in the feed oil: O1S1, O2S1, O3S1, O4S1, and O1S2 classes. The two last mentioned classes were no longer found in the product sample. In addition, the O1S1 class was also largely removed by the hydrotreatment, with only the species having a DBE value of 8 persisting after the processing, whereas the DBE distribution of the O1S1 species in the feed sample was 6-13. On the other hand, the O2S1 and O3S1 classes seemed to largely withstand the processing. All O3S1 species detected in the samples had a DBE value of 4. Overall, the proportions of the different OySz classes were quite low.

processes, and thus, there is a need to identify petroleum species that are resistant to hydrotreatment. Two gas oil samples, untreated feed and hydrotreated product oil, have been analyzed by positive- and negative-ion 4.7 T ESI FTICR-MS, and the elemental compositions of polar species in the oil samples have been examined. It is sure that comprehensive results could not be provided by the used equipment, because the resolution has not been sufficient enough to separate peaks in all cases. However, the starting point for a study has been to compare two samples, and in most cases, the resolving power has been high enough to permit reliable results. The most abundant polar species were found to be N1-containing compounds. Both basic and acidic N1 class compounds have been found to be highly resistant to removal by hydrotreatment, and some compounds with multiple heteroatoms produced N1 species by reduction. The second most abundant acidic species were Oy-containing compounds, of which O1 species were the predominant class in the feed oil, but the hydrotreatment removed all of these species from the oil. The DBE distribution of O2 species was 1-11 in the feed sample, whereas the product sample contained only naphthenic acids (DBE values of 1-4). The other Oy species (y = 3-7) seemed to be quite resistant to hydrotreatment. Oy class compounds were not detected by positive-ion ESI. Basic NxOy-, NxSz-, and NxOySz-containing compounds were completely removed/degraded by hydrotreatment. In this respect, the catalytic hydrotreatment works effectively. On the other hand, acidic NxOy, NxSz, OySz, and NxOySz species and basic OySz species partially survived the processing.

Conclusions The purpose of this study has been to examine how efficiently catalytic hydrotreatment removes polar nitrogen-, oxygen-, and sulfur-containing compounds from gas oil. A knowledge of the compositional changes between feed and product oils is important for the development of catalytic hydrotreatment and better associated technology. Variations in petroleum composition directly impact all refinery

Acknowledgment. The authors are grateful for financial support from the Finnish Funding Agency for Technology and Innovation (Tekes) and Neste Oil Oyj.

6061