Environ. Sci. Technol. 2008, 42, 1944–1947
Unraveling Heavy Oil Desulfurization Chemistry: Targeting Clean Fuels TUSHAR V. CHOUDHARY,* STEPHEN PARROTT, AND BYRON JOHNSON ConocoPhillips Company, Bartlesville Technology Center, Bartlesville, Oklahoma 74004
Received August 14, 2007. Revised manuscript received December 18, 2007. Accepted December 27, 2007.
The sulfur removal chemistry of heavy oils has been unraveled by systematically investigating several heavy oils with an extremely wide range of properties. The heavy oil feed and product properties have been characterized by advanced analytical methods, and these properties have been related to the sulfur conversion data observed in pilot hydrotreating units. These studies coupled with kinetic treatment of the data have revealed that the desulfurization chemistry of heavy oils is essentially controlled by the strongly inhibiting three and larger ring aromatic hydrocarbon content and surprisingly not by the content of the “hard-to-remove” sulfur compounds. Such enhanced understanding of the heavy oil sulfur removal is expected to open new avenues for catalyst/process optimization for heavy oil desulfurization and thereby assist the efficent production of clean transporation fuels.
1. Introduction Environmental concern has led to a continual push toward clean transportation fuels, with the emphasis on sulfur minimization (1, 2). Billions of dollars are anticipated to be spent for meeting the proposed stricter worldwide clean fuel regulations (3). Sulfur removal from heavy oils is an important process for production of clean ultralow sulfur fuels. Due to the limited supply of quality petroleum feedstocks, there is an increasing necessity to utilize low grade heavy oils that are unfortunately considerably more difficult to convert to clean transportation fuels (4). Further catalyst/process improvements related to heavy oil desulfurization are, therefore, critical for the continual efficient supply of clean fuels. The above, however, requires an excellent fundamental understanding of the heavy oil reaction chemistry and therein is the main problem. The complexity of petroleum feedstocks (in terms of type and number of molecules) increases exponentially with increasing boiling point, and therefore obtaining fundamental understanding about heavy oils, a high boiling petroleum fraction, is exceedingly challenging (2, 5); correspondingly realistic information pertaining to this topic is scarce. Development of an enhanced understanding about heavy oil desulfurization reaction chemistry is, therefore, very important from the viewpoint of continual efficent production of environmentally benign transportation fuels. Due to the enormous inherent complexity (appropriate feeds procurement, characterization, reactor testing, and product analysis) of this research area, a study of the required * Corresponding conocophillips.com. 1944
9
author
e-mail:
tushar.v.choudhary@
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008
magnitude has not been previously reported. Attempts to understand heavy oil sulfur removal in the past have been via studies that involved very limited feeds (model compounds or one/two feeds in most cases) and/or unrealistic experimental methods and/or extremely inadequate feed characterization (6, 7). However, we have found that obtaining a realistic understanding about the heavy oil desulfurization chemistry requires a systematic investigation of several heavy oil feeds that cover a very broad range of properties. Herein, an unprecedented number of heavy oil feeds (eight heavy oils with widely varying properties and different origins) were hydrotreated over a commercial nickel-molybdenum (NiMo) based catalyst in pilot reactors under an industrially relevant range of process conditions. The heavy oil feeds and selected hydrotreated products were characterized by our recently developed advanced sulfur analysis method (2) and several other heavy oil characterization methods. Insights into the heavy oil desulfurization chemistry were further obtained via kinetic analysis of the experimental data. The intention of this work is not to propose a comprehensive heavy oil desulfurization kinetic model but instead use kinetics as a tool to reveal the underlying heavy oil desulfurization chemistry.
2. Experimental Methods and Kinetics 2.1. Heavy Oil Feeds. Eight heavy oil feeds with widely varying properties and origins were used in this study. As seen from Table 1 the feed sulfur varied from 11518 to 35115 ppmw, feed nitrogen varied from 1321 to 6506 ppmw; total core aromatics varied from 21.9 to 30.8 wt %, hydrogen content varied from 10.5 to 12.06 wt.%, and specific gravity varied from 0.9226 to 0.9747. Also, a large range of heavy gas oils (some just pure components and some blends) have been used in this study. For example, HO3 is composed of purely coker gas oil, while HO1 is a blend of coker gas oil, vacuum gas oil, coker diesel, and heavy atmospheric gas oil. For the purpose of this study heavy oils are defined as petroleum fractions (asphaltene containing feeds are excluded) which have a 50% recovery temperature greater than 663 K. 2.2. Heavy Oil Characterization. 2.2.1. Analysis of Sulfur Distribution. In the first step, nonaromatic sulfide-like (NArS) molecules are selectively oxidized to sulfoxides and separated via adsorption over silica gel (2). The sample is heated in the oven at 363 K for 0.5 h. A 1 g amount of the sample is then reacted with 0.12 and 0.15 g of tetrabutylammonium periodate (TBAP) in 12 mL of toluene:methanol (5:1 (v:v)) at 358 K for 1 h. Following a water wash, the samples are then separated into two fractions on a silica gel column. The fraction containing sulfoxides is quantitatively analyzed for total sulfur. The remaining fraction (devoid of sulfoxides) is introduced onto a 5:95 diphenyl:dimethylpolysiloxane column (nonpolar) and then subsequently onto a 50:50 diphenyl: dimethylpolysiloxane (polar) column. The nonpolar first column separates according to boiling point, and a modulator reintroduces the sample onto the second column, which separates according to polarity; a sulfur chemiluminescence detector (SCD) and flame ionization detector (FID) are used for sulfur detection and hydrocarbon detection, respectively. The amounts of sulfur species determined by this analysis are combined with the amount of NArS and normalized to the total sulfur of the sample. This method is applicable to coker gas oils, vacuum gas oils, heavy atmospheric gas oils, heavy cycle oil, and other deashphalted heavy oils. Further details may be obtained from ref 2. 10.1021/es0720309 CCC: $40.75
2008 American Chemical Society
Published on Web 02/14/2008
TABLE 1. Heavy Oil Properties properties
HO1
HO2
HO3
HO4
HO5
HO6
HO7
HO8
specific gravity at 288.55 K H content (wt %) sulfur (ppmw) total N (ppmw) basic N (ppmw) total aromatics (wt %) 3+ ring aromatics (wt .%) IBP-FBP (K)
0.9279 11.60 32290 1691 661 26.2 8.6 440–815
0.9226 12.06 11518 1321 381 21.5 8.4 485–877
0.9747 10.50 11974 6506 2942 28.1 13.5 495–875
0.9670 10.98 35115 2582 945 30.8 12.0 586–895
0.9546 11.39 21238 1468 392 27.6 9.9 583–888
0.9313 11.95 12554 1444 441 22.2 8.9 587–892
0.9595 11.23 22850 3690 1346 28.5 14.4 593–879
0.9536 11.45 15859 2358 824 25 12.4 589–894
2.2.2.. Other Heavy Oil Analysis. Sulfur analysis was done using ICP, nitrogen by boat-inlet chemiluminiscence; boiling range distribution by gas chromatography; H content by NMR; specific gravity by hydrometer method, and aromatics analysis by nitric oxide chemical ionization gas chromatography/mass spectroscopy. Aromatics analysis was performed by Triton Analytics Corp. Most of the samples were analyzed in a single batch (single instrument calibration) for all the tests. The aromatic content reported in Table 1 is without the side chains. As a side-note, the 3+ ring aromatic content was found to strongly depend on the H content (or specific gravity) and boiling point of the heavier HO fraction (such as the average of T70, T80, T90, T95, FBP, or individual T80, T90, etc.). 2.3. Heavy Oil Hydrotreating Experiments. The hydrotreating experiments were undertaken in pilot reactors at the Bartlesville Technology Center, ConocoPhillips (COP). A commercial NiMo catalyst was used for hydrotreating experiments. The above catalyst has been extensively used in heavy gas oil hydrotreating units worldwide and therefore is an excellent representative catalyst for this study. Presulfiding was achieved using a distillate feed containing 0.88 wt % sulfur first at 561 K for 15 h followed by 589 K for 48 h (pressure ) 2.8 MPa; H2 rate ) 0.011 m3/h). Experiments were undertaken at varying reaction temperatures (620–670 K) and two different pressures (6.9 and 8.3 MPa); the liquid hourly space velocity was 1 h-1, and H2/feed ratio, 356 (m3/ m3). Four pilot units were used over a period of several months to obtain the data in this study. Analysis reported for each data point was obtained after at least 50 h time-on-stream at a given set of conditions. A fresh catalyst was used for each feed. Depending on the number of data points collected, sometimes two fresh catalysts were used for the same heavy oil feed to avoid any significant contribution from deactivation. An excellent reproducibility for different product properties was obtained between different pilot units for a given heavy oil under identical process conditions. 2.4. Kinetic Analysis. 2.4.1. Parallel First-Order Kinetics without 3+ Ring Aromatic Inhibition. The kinetic equation used for estimating the product sulfur using detailed feed sulfur distribution is shown below. The sulfur species were grouped into seven different groups on the basis of conversion data for the individual groups such that species with similar reactivity were grouped together. In the case of parallel firstorder kinetics methodology, each group/species is treated as if it obeys first-order kinetics (2, 8, 9). The kinetic parameters (Table 2) were obtained by fitting the experimental data set (sulfur distribution of products where available and total product sulfur) from all the heavy oils. total Sproduct )
∑ [S
f group(x)
exp((-kgroup(x) ⁄ (K1(N) + K2(S)))PH2Rgroup(x) ⁄ LHSV)]
where Sf ) feed sulfur, x ) 1-7 with group (1) ) thiophenes and benzothiophenes, group(2) ) C0/C1 dibenzothiophenes, group (3) ) C2+ dibenzothiophenes, group (4) ) phenath-
TABLE 2. Kinetic Parametersa (Preexponential Factor, A, and Activation Energy, Ea) for the Different Sulfur Molecular Groups properties
A (1/h)
Ea (kJ/mol)
group 1 group 2 group 3 group 4 group 5 group 6 group 7
1564185 1872381 19866549 2493877856682 661 26.2 8.6
-67.3 -71.2 -86.7 -150.2 -86.5 -86.5 -50.0
a See section 2.4.1, “Parallel First-Order Kinetics without 3+ Ring Aromatic Inhibition”.
rothiophenes, group (5) ) benzonapthothiophenes and five ring thiophenes compact, group (6) ) five ring thiophenes extended and six ring thiophenes, and group (7) ) nonaromatic sulfides; K1 ) inhibition constant for nitrogen compounds ) 0.70; K2 ) inhibition constant for sulfur compounds ) 0.0053; N ) nitrogen content; S ) sulfer content; R ) pressure dependence term; LHSV ) liquid hourly space velocity; kgroup(x) ) Agroup(x) exp(-Egroup(x)/RT); k ) rate constant; A ) preexponential factor; and E ) activation energy. 2.4.2. First-Order Kinetics without Sulfur Speciation but with 3+ Ring Aromatic Inhibition. The kinetic equation used for estimating the product sulfur without using detailed feed sulfur distribution is shown as follows (8). In this case only bulk sulfur information was used (total feed sulfur) along with inhibition from three+ ring aromatic compounds. The kinetic parameters shown below were obtained from the fitting of only three heavy oil feeds (HO1, HO7, and HO8). The selection of the three heavy oils for fitting was based on their 3+ ring aromatic content (covered a large range). Sproduct ) [Sf exp((-k ⁄ (K)(3 + R))PH2R ⁄ LHSV)] where Sf ) feed sulfur, K ) 3+ ring aromatic inhibition constant ) 0.066, 3+R ) 3+ ring core aromatic content, R ) pressure dependence term, LHSV ) liquid hourly space velocity, kgroup(x) ) Agroup(x) exp(-Egroup(x)/RT), k ) rate constant, A ) preexponential factor ) 84370 1/h, and E ) activation energy ) -56.5 kJ/mol..
3. Results and Discussion Depending on their structure, the reactivity of the sulfur molecules over hydrotreating catalysts can differ by more than an order of magnitude, and based on this, it is widely agreed that the structure and content of individual sulfur species predominantly define the desulfurization chemistry in petroleum feedstocks (10–14). Due to the enormous complexity of heavy oil, obtaining similar structure–reactivity information about heavy oils is very challenging (6, 7). Our previous studies have provided advanced capabilities, which have enabled for the first time structure-based quantitative discrimination between the different sulfur molecular groups VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1945
FIGURE 1. Comparison of experimental and estimated product sulfur for different HO under varying reaction conditions. Kinetic methodology: Parallel first-order kinetics using individual sulfur species data and inhibition from sulfur and nitrogen compounds. Error bar ) sulfur analysis repeatability. in heavy oils (2). On the basis of our earlier preliminary studies (two feeds), and influenced by the prevailing understanding about petroleum desulfurization, we were also led to believe that the nature and the content of the sulfur species was critical for determining the extent of heavy oil desulfurization (2). However, our present work involving a systematic investigation of eight heavy oils with widely varying properties has led to rather startling conclusions. If the nature and content of the different heavy oil sulfur species had been the main factor in defining heavy oil desulfurization chemistry, then heavy oils with a larger content of hard-to-remove sulfur compounds would be expected to be more difficult to desulfurize than heavy oils with a smaller content of hard-to-remove sulfur compounds. In other words by using the individual sulfur species information for the heavy oils, it would be possible to estimate their sulfur removal processability. If the above was true using parallel first-order kinetics involving the individual sulfur molecular groups it should be possible to numerically estimate the extent of desulfurization in heavy oils. However when we undertook such a kinetic study (described in Experimental Method and Kinetics), the nature/content information of individual sulfur species was surprisingly found to be inadequate for fitting the experimental desulfurization data (Figure 1). As mentioned previously the nature and content of the different sulfur compounds in the lower boiling petroleum fraction is known to predominantly define its desulfurization rate. If the above had also been valid for heavy oils, the distribution/content information about sulfur compounds should have provided an accurate extent of desulfurization for heavy oils. Since this is not the case, it is apparent that there is some other parameter that strongly influences heavy oil desulfurization. One striking feature of heavy oils is that they contain a considerably larger number of three+ ring aromatic compounds as compared to the lower boiling petroleum fractions. Compared to the one and two ring aromatic compounds, the three+ ring aromatic compounds have a considerably larger propensity to be adsorbed on the hydrotreating catalyst surface and thereby are expected to strongly compete with the sulfur compounds for the catalyst sites (15, 16). Efforts were therefore undertaken to study the effect of the inhibiting three+ ring aromatic compounds on heavy oil sulfur removal. Figure 2 illustrates the individual role of the three+ ring aromatic compounds in determining the extent of heavy oil desulfurization. Information pertaining to the structure and content of individual sulfur molecular groups was not used in this case; i.e. product sulfur estimation is based on kinetics involving bulk sulfur (no speciation) and inhibition from 1946
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008
FIGURE 2. Comparison of experimental and estimated product sulfur for different HO under different reaction conditions (only total sulfur content and three+ ring aromatic content was used for estimation). First-order kinetic parameters derived from the fitting of HO1, HO7, and HO8 were used to predict the desulfurization extent for the other five HO. Error bar ) sulfur analysis repeatability.
FIGURE 3. Relationship between three+ ring aromatic content of eight heavy oils and the corresponding sulfur conversion (reaction conditions: temperature ) 644 K; LHSV ) 1 h-1; pressure ) 6.9 MPa; H2/HO ratio, 356 (m3/m3)). three+ ring aromatic content only. It is noteworthy that the kinetic parameters derived from the fitting of just three heavy oil feeds (HO1, HO7, and HO8) was used to predict the extent of desulfurization of five other heavy oil feeds (HO2, HO3, HO4, HO5, and HO6). The excellent desulfurization predictability observed in Figure 2 cogently demonstrates that the heavy oil sulfur removal chemistry is predominantly controlled by the heavy oil three+ ring aromatic content. Since the extent of sulfur prediction is mostly within experimental error of sulfur analysis, the contribution from other known inhibitors clearly cannot be large. As expected, the product sulfur estimation was extremely poor when inhibition from the three+ ring aromatic compounds was not considered. A very straightforward approach to illustrate the importance of the proposed inhibition effect is to consider the relationship between experimental sulfur conversion of the different heavy oils under identical process conditions and their corresponding three+ ring aromatic content. An excellent correlation was observed between the sulfur conversion (644 K and 6.9 MPa) for the eight different heavy oil feeds and three+ ring aromatic content (Figure 3). The profoundly strong relationship between the three+ ring aromatic content and extent of desulfurization observed in Figure 3 unambiguously shows that the three+ ring aromatic content defines the desulfurization chemistry in heavy oils. Similar analysis confirmed that other commonly known desulfurization inhibitors (6, 10, 17–20) such as basic nitrogen,
(real feedstocks) under industrially relevant process conditions. The studies suggest that the factors that govern the desulfurization chemistry in the heavy oil range are significantly different than those in the diesel boiling range.
Acknowledgments We would like to thank J. Bares, R. Higbee, B. Dodd, J. Malandra, J. Green, D. Hsieh, M. Sardashti, B. Beever, J. Anderson, P. Meier, and D. Strope, for their assistance during this study.
Literature Cited
FIGURE 4. Relative distribution/content of sulfur compounds in different HO: T ) thiophenes; BT ) benzothiophenes; PhT ) phenanthrothiophenes; DBT ) dibenzothiophenes; 5RT-c ) five ring thiophenes compact; BNT ) benzonapthothiophenes; 5RT-e ) five ring thiophenes extended; 6RT ) six ring thiophenes. total nitrogen, total aromatic compounds, and sulfur compounds did not play a major role in defining heavy oil desulfurization. Although some previous studies have also suggested that aromatics can have a significant inhibiting effect on the desulfurization reaction (6, 10), it has never been suggested that inhibition from certain aromatic compounds can essentially define the extent of petroleum desulfurization. The most surprising conclusion from this study is that the structure/content of individual sulfur species (contrary to the prevailing understanding of petroleum desulfurization chemistry) does not effectively define heavy oil sulfur removal under industrially relevant conditions. This may be explained on the basis of the following: (a) There is a relatively uniform distribution of the different sulfur molecular groups in heavy oil (Figure 4), irrespective of the source. This is very interesting since the distribution of sulfur molecular groups in diesel feeds varies considerably depending on its source. (b) The difference in the relative reactivities between the different sulfur species in heavy oil is considerably smaller than that in lower boiling petroleum fractions. For example, at a reaction temperature of 633 K, benzothiophenes are ∼3 times more reactive than the dibenzothiophenes in heavy oil (2), while, in the case of diesel (lower boiling petroleum fraction), the benzothiophenes are ∼36 times more reactive than the dibenzothiophenes (13). (c) Heavy oils have a considerably larger three+ ring aromatic content than the lower boiling petroleum fractions. In our previous study two heavy oils with almost identical three+ ring aromatic content (Table 1: HO1 and HO2) had been investigated, and hence the nature/content of individual heavy oil sulfur molecular groups had seemed important at that time (2). While previous studies have assisted in shedding light on the structure–reactivity and/or feed property relations of desulfurization in lower boiling petroleum fractions (17–20), herein we have provided an enhanced understanding of the desulfurization chemistry in the higher boiling heavy oil range. We have demonstrated that three +ring aromatic content (not nature and content of sulfur compounds) essentially defines the extent of sulfur removal in heavy oils
(1) Yang, R. T.; Hernandez-Maldonado, A. J.; Yang, F. H. Desulfurization of transpiration fuels with zeolites under ambient conditions. Science 2003, 301, 79–81. (2) Choudhary, T. V.; Malandra, J.; Green, J.; Parrott, S.; Johnson, B. Towards clean fuels: Molecular level sulfur reactivity in heavy oils. Angew. Chem., Int. Ed. 2006, 45, 3299–3303. (3) Venner, S. F. EU environmental laws impact fuels requirements. Hydrocarbon Process. 2000, 79, 51–60. (4) Gembicki, V. A.; Cowan, T. M.; Brierley, G. R. Update processing operations to handle heavy feedstocks. Hydrocarbon Process. 2007, 86, 41–53. (5) Ma, X. L.; Sakanashi, K.; Isoda, T.; Mochida, I. Determination of sulfur compounds in non-polar fraction of vacuum gas oil. Fuel 1997, 76, 329–339. (6) Whitehurst, D. D.; Isoda, T.; Mochida, I. Present state of art and future challenges in the hydrodesulfurization of polyaromatic sulfur compounds. Adv. Catal. 1998, 42, 345–471. (7) Choudhary, T. V. Structure-reactivity-mechanistic considerations in heavy oil desulfurization. Ind. Eng. Chem. Res. 2007, 46, 8363–8370. (8) Yui, S. M. Hydrotreating of bitumen-derived coker gas oil:kinetics of hydrodesulfurization, hydrodenitrogenation, and mild hydrocracking. AOSTRA J. Res. 1989, 5, 211–224. (9) Iannibello, A.; Marengo, S.; Burgio, G. Modelling the hydrotreating reactions of a heavy residual oil in a pilot tricklebed reactor. Ind. Eng. Chem. Process Des. Dev 1985, 24, 531– 537. (10) Girgis, M. J.; Gates, B. C. Reactivities, reaction networks, and kinetics in high pressure catalytic processing. Ind. Eng. Chem. Res. 1991, 30, 2021–2058. (11) Brunet, S.; Mey, D.; Perot, G.; Bouchy, C.; Diehl, F. On the desulfurization of FCC gasoline: a review. Appl. Catal., A 2005, 278, 143–172. (12) Song, C. An overview of new approaches to deep desulfurization for ultra-clean gasoline, diesel fuel and jet fuel. Catal. Today. 2003, 86, 211–263. (13) Ma, X. L.; Sakanashi, K.; Mochida, I. Hydrodesulfurization reactivities of various sulfur compounds in diesel fuel. Ind. Eng. Chem. Res. 1994, 33, 218–222. (14) Knudsen, K. G.; Cooper, B. H.; Topsoe, H. Catalyst and process technologies for ultra low sulfur diesel. Appl. Catal. 1999, 189, 205–215. (15) Korre, S. C.; Klein, M. T.; Quann, R. J. Polynuclear aromatichydrocarbons hydrogenation. 1. Experimental reaction pathways and kinetics. Ind. Eng. Chem. Res. 1995, 34, 101–117. (16) Choi, K. H.; Sano, Y.; Korai, Y.; Mochida, I. An approach to deep hydrodesulfurization of light cycle oil. Appl. Catal., B 2004, 53, 275–283. (17) Van Looij, F.; van der Laan, P.; Stork, W. J. H.; DiCamillo, D. J.; Swain, J. Key parameters in deep hydrodesulfurization of diesel fuel. Appl. Catal., A 1998, 170, 1–12. (18) Ho, T. C. Property-reactivity correlation for HDS of middle distillates. Appl. Catal., A 2003, 244, 115–128. (19) Ho, T. C. Inhibiting effects in hydrodesulfurization of 4,6diethyldibenzothiophene. J. Catal. 2003, 219, 442–451. (20) Shih, S. S.; Mizrahl, S.; Green, L. A.; Sarii, M. S. Deep desulfurization of distillates. Ind. Eng. Chem. Res. 1992, 31, 1232–1235.
ES0720309
VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1947
