JOURNAL OF CATALYSIS 120, 487-492 (1989)
Preparation and Activity of Solid-State Hydrodesulfurization Catalysts
There has been much work done in the able HDS characteristics. This concept was field of Mo$-based hydrodesulfurization surprising since in previous studies the pro- catalysis trying to establish what phase is moter sulfides had shown little activity (7, responsible for catalytic activity. Much of 8). More recently Prins and co-workers this research has focused on the effect of have expressed doubt about the importance the Co promoter and whether its predom- of a Co sulfide phase as the active phase inant contribution is of the electronic or (II). Theoretical work by Harris and structural nature. Schuit et al. (1) proposed Chianelli (12) has indicated that the pro- that the promoter ions stabilize a molybde- moter atom plays a role in determining the num oxide monolayer which is responsible electronic state of the MO&. Consequently, for the observed catalytic activity. Far- the increased HDS activity has been attrib- ragher and Cossee (2) postulated that the uted to an electronic contribution from the promoter was “pseudo-intercalated” be- promoter atom. According to this model, tween successive layers of MO& and that only Ni and Co showed a positive activity this structural change was accompanied effect as they increased the electron density by a surface reconstruction triggering on MO. Fe displayed little effect and Cu HDS activity. Delmon et al. (3) argued in acted as a poison, essentially oxidizing MO. favor of Co&& and MO& coexisting as the Recently Ledoux et al. (13, 14) proposed active species. In this “remote-control the existence of two types of Co species at model” Hz is dissociatively adsorbed on the edges of the MO& structure, a distorted the cobalt sulfide and transferred to the tetrahedral phase responsible for HDS ac- MO& surface where it then reacts with an tivity and a rapid octahedral which they adsorbed sulfur-containing compound. As think acts as a glue, anchoring the tetrahe- another possible explanation for the pro- dral phase to the MO&. However, Prins er moter effect of Co or Ni, Pratt and Sanders al. presented EXAFS work which does not (4) and Vrinat and co-workers (5) proposed support this hypothesis (II). that the promoter stabilizes very small par- The present note reports the results of ticles of MO&. This notion was also sup- our attempt to grow unsupported bulk ported by more recent NMR experiments P-MO-S (where P is Fe, Co, or Ni) phases. performed by Ledoux et al. (6). Topsoe and The synthesis of these phases was based on co-workers (7, 8) were able to correlate solid-state reactions of elemental sulfur HDS activity with a unique but structurally with metallic promoter and MO powders. difficult to define “CO-MO-S” phase, in These bulk phases are easier to character- which it is hypothesized that the Co-pro- ize than those in conventional supported moter atoms lie at the edges of the MO& HDS catalysts. Therefore, they offer a bet- layers. More recent EXAFS work has pro- ter opportunity for establishing correlations vided evidence in support of this edge- between stoichiometry, structure, and ac- decorated MO& species (9). Other re- tivity. searchers (10) have put forward the notion A series of samples with a limited P/MO that cobalt sulfide alone might have favor- ratio and a general stoichiometry of P0.0s 487 0021-9517/X9 $3.00 CopyrIght ‘i; 1989 hy Academic Pre\\. Inc. All rights of reproduction m any form rewrved. 488 NOTES
TABLE 1 raised to 673 K under a flow of high-purity Surface Areas of Catalyst Samples helium at 20 ml per minute and held at 673 K for 30 min. Then, the temperature was Catalyst BET surface area (m*/g) lowered to the desired reaction tempera- ture, typically between 473-673 K. The He Feo.oM%.!& 14 flow was replaced by a gaseous feed stream 26 Coo.osMoo 975S2 containing 2.7% (by volume) of thiophene, Nio&oo.d2 15 the balance being hydrogen, at a flow rate MoS1.95 20 MO& 1 of 10 ml/min and atmospheric pressure. The high-purity He and HZ (<49 ppm impu- rities) were further purified by passage MO,& (with x being 0.975 or 0.95) was through a commercial oxygen trap (Mathe- prepared by combining stoichiometric son) and a bed of molecular sieves (5 A) to amounts of the elemental promoter and remove moisture. molybdenum powders with elemental sul- The effluent from the reactor was ana- fur, all three chemicals used being Fluka lyzed by gas chromatography. Product sep- products (purum). The powders were thor- aration and analysis were performed by oughly mixed by grinding them together to using an n-octanelPorasi1 C column in a insure intimate contact between the com- Varian 3700 gas chromatograph equipped ponents. The mixtures were placed into with a thermal conductivity detector. Peak quartz tubes which were then evacuated to areas were determined by a Hewlett- about 0.1 Pa and sealed. The tubes contain- Packard 3390A integrator. ing the mixtures were heated at a heating At 523 K the activity trends for the model rate of 10 K per minute from room tempera- catalysts were Ni(30.0%) > Co (6.3%) > Fe ture to 783 K, held at this temperature for (5.7%) > MoSg5 (5.5%) > MO& (~1%). 24 h and then rapidly quenched. Once the The specific thiophene conversions for treatment was finished, the quartz tubes these materials are given in parentheses. At were opened. The contents of the tubes, 573 K the activity trends for the model generally dark grey or black materials, catalysts remained, but the differences were removed and, if necessary, ground became more pronounced. After 3 h on into fine powders. Based on their nominal stream it was seen that the Nio.osMoo.g& composition the samples were named catalyst had a steady-state conversion just Feo.&oo.9d2, Coo.&oo.97&, and W.05 over 60%, the Coo.o~Moo.g& was in the MoO.g&. Unpromoted MO& and a nonstoi- mid-50% range, Feo.05Moo.9$$ was just chiometric sample, MoSg5 were also pre- over 40%, MoSg5 was near 30%, and MO& pared from a mixture of metallic molybde- was consistently below 1%. At higher tem- num and elemental sulfur following the peratures (673 K) conversions in excess of procedure mentioned above. The surface 90% were consistently observed for the areas of the freshly prepared samples were promoted samples and MoSi.g5. determined by using the BET method and a The temperature dependence of the HDS Monosorb Quantachrome Single Point In- activity for all the model catalysts is pre- strument with nitrogen as adsorbate (Ta- sented in Fig. la. The turnover frequen- ble 1). cies (TOF) were expressed in terms of The catalytic activity for the hydrode- moles of thiophene converted per gram- sulfurization of thiophene (Aldrich 99+%, atom MO per second. Again, it may be seen Gold Label) was measured in a i-in. diame- that the promoter trend of Ni > Co > Fe is ter stainless-steel continuous-flow reactor. observed in our solid-state system. Other The powdered catalyst (0.25-0.35 g) was researchers (15, 16) have reported similar loaded into the reactor and the temperature findings, although Fe-promoted catalysts NOTES 489
\ A ii x -10 0 X N1.05 MO 95 S2 + + A x + 0 Co 05 MO.975 S2 A Fe 05 MO 975 S2 x
-12 9- + MoS’ g5
0
t A I l MQS2 -16 1 1 4 1.6 1 8 20 2.2
l/T’1000
b) p&S, 95 ..- - -...... l...._.^l .I -. ^
q Fe 05 MO.975 S2
” CT a a c3; c4 rc4 lc4; c%ac4= Ic4 cc& Product
FIG. 1. (a) Temperature dependence of turnover frequencies (molecules of thiopheneimol of MO/S) for thiophene HDS over the various solid-state catalysts. (b) Product distributions obtained in thiophene HDS. The results shown were collected at 523 K on all samples, except for the Ni-containing catalysts where data are shown for a temperature of 473 K. tend to have low activities. However, in Of course, this does not rule out that the our solid-state catalysts the promoting ef- promoter may have an additional second- fect of Fe is far from negligible and is very order effect beyond facilitating nonstoi- close to that of Co. It should be noted that chiometry and sulfur deficiency in molyb- the substantial increase in HDS activity of denum sulfide. these solid-state catalysts cannot be attrib- Two possible scenarios for the formation uted solely to the presence of promoters. of nonstoichiometric molybdenum sulfide In fact, even in the absence of a promoter, could be envisioned. One is that the incor- significant increases in HDS activity could poration of lower charged promoter ions be achieved simply by introducing anionic into the MO& lattice requires the introduc- sulfur vacancies into MO&. Hence, one of tion of sulfur vacancies to maintain electro- the predominant roles of the promoter atom neutrality. The second possibility could be in these solid-state catalysts appears to be that the promoter acts as a scavenger dur- linked to the creation of sulfur vacancies. ing catalyst synthesis. The formation of NOTES
promoter bulk sulfides could thereby de- the MO& sample. The high n-butane frac- plete the sulfur pool required for the forma- tions seen for CO~.~~MO~.&S~ and MoS~.~~ tion of stoichiometric MO&. In support of are likely due to an increased H2 transport the second hypothesis, when presulfiding capability resulting from the presence of (HJHZS at 673 K for 24 h) a mechani- unreacted MO in the catalyst synthesis cal mixture of MO and promoter powder charge. Vrinat et al. (17) have shown that bulk promoter sulfides were preferentially Ni has greater hydrogenation activity than formed, leaving most of the MO unreacted. Co. However, in our case the Ni catalyst At this point the relative contribution of was run at a lower temperature than the each of these two scenarios is not yet clear. other samples and thus this characteristic The product distributions of the different hydrogenation trend was not as pro- hydrocarbons resulting from the hydrode- nounced. At higher temperatures (i.e., sulfurization of thiophene are given in Fig. > 523 K) all the promoted samples pro- lb. The data shown in this figure were duced predominantly n-butane. As interest- collected at conversions less than 8% of the ing as these subtle trends may be, they are total thiophene fed. Isobutene and bu- not the focus of this note and a more de- tadiene could not be separated under our tailed kinetic analysis will be required to experimental conditions and Fig. lb shows fully explain the observed differences. the sum of these two products. Due to the It is important to note that more active higher activity of the Ni-promoted sample samples consistently had larger surface the reaction temperature was 473 K rather areas than that of the lower activity MO& than the 523 K used for the other promoted (Table 1). However, a normalization of catalysts in order to keep the conversion activity based on BET surface area would and TOF values roughly equivalent. It must not be meaningful in view of the lack of be noted that product distributions tend to correlation between HDS activity and sur- be not only a function of conversion but face area previously reported (28-20). also of temperature. Therefore, the data Within the HDS research community, a shown in Fig. lb should only be used for consensus seems to be developing that a illustrating that our catalyst selectivities normalization based on edge area might be conform with those found typically in HDS more desirable. However, to accurately catalysts under comparable reaction condi- determine the edge area is not a trivial tions. task. In the case of our solid-state cata- Although each of the three promoters lysts, high-resolution electron microscopy used in this study provided similar in- showed that the exposed basal planes of creases in activity, there are subtle differ- MO& containing the MO ions remain undis- ences in the product distributions. Coo.0~ turbed when promoters are introduced. In Mo~.~& and MoSr.95 preferentially pro- striking contrast to that, other exposed duced n-butane, Feo.osMoo.975S2 favored crystallographic orientations containing the formation of trans-Zbutene, Nio.,,j sulfur ions, which could be considered MO&~ made primarily I-butene, while “edge areas,” show significant structural MO& produced significant quantities of disorder. From the lack of long-range order propylene, trans-2-butene, and cis-2- in these planes we can infer surface rough- butene. Only MO& and Nio.osMo0.9sSz ening on an atomic scale and the presence yielded any isobutylene or butadiene in the of sulfur vacancies. Figure 2 illustrates the product stream, and only MoSi.95 produced structural effect of the promoter (CO.~~ any isobutane. Lighter products (
491 NOTES broadening of reflections arising from sul- 12. Harris, S., and Chianelli, R., J. Catal. 98, 17, fur-containing planes. 1986. 13. Ledoux, M. J., Michaux, O., Agostini, G., and ACKNOWLEDGMENTS Panissod, P., J. Catal. 96, 189, 1985. 14. Ledoux, M. J., Michaux, O., Agostini, G., and M. A. Villa-Garcia wishes to express her gratitude Panissod, J. Catal. 102, 275, 1986. to the MEC/Fulbright Committee for financial sup- 15. Gobolos, S., Wu, Q., Andre, O., Delannay, F., port. The authors also wish to thank Dr. John Mans- and Delmon, B., J. Chem. Sot. Faraday Trans. 1 field of the Electron Microbeam Analysis Laboratory 82, 2423, 1986. at the University of Michigan for helpful discussions 16. Gobolos, S., Wu, Q., Delannay, F., and Delrnon, relating to electron microscopy. B., Polyhedron 5, 219, 1986. 17. Vrinat, M., Lacroix, M. Breysse, M., and Frety, REFERENCES R., Bull. Sot. Chim. Be/g. 93, 697, 1984. 1. Schuit, G. C. A., and Gates, B. C., AIChE J. 19, 18. Tauster, S. J., Percoraro, T. A., and Chianelli, 417, 1973. R. R., J. Catal. 63, 515, 1980. 2. Farragher, A. L., and Cossee, P., in “Proceed- 19. Furimsky, E., Catal. Rev. Sci. Eng. 22(3), 371, ings, 5th International Congress on Catalysis, 1980. Palm Beach, 1972” (J. W. Hightower, Ed.), p. 20. Villa Garcia, M., Lindner, J., Sachdev, A., and 1301. North-Holland, Amsterdam, 1973. Schwank, J., J. Cutal., in press. 3. Pirotte, D., Zabala, J. M., Grange, P., and Del- mon, B., Bull. Sot. Chirn. Belg. 90, 1239, 1981. J. LINDNER 4. Pratt, K. C., and Sanders, J. V., in “Proceedings, 7th International Congress on Catalysis, Tokyo, A. SACHDEV 1980” (T. Seyama, and K. Tanabe, Eds.), p. 1420. KodanshaiElsevier, Amsterdam/Tokyo, 1981. Department of Chemical Engineering 5. Vrinat, M. L., and De Mourgues, Appl. Catal. 5, The University of Michigan 43, 1983. Ann Arbor, Michigan 48109-2136 6. Ledoux, M. J., Maire, G., Hantzer, S., and Michaux, O., in “Proceedings, 9th International Congress on Catalysis, Calgary, 1988” (M. J. M. A. VILLA GARCIA Phillips and M. Ternan, Eds.), p. 74. Chem. Institute of Canada, Ottawa, 1988. Department Q&mica Organometdlica 7. Topsoe, H., Clausen, B. S., Candia, R., Wivel, Fact&ad de Quimica C., and Morup, S., J. Catal. 68, 433, 1981. Universidad de Oviedo 8. Wivel, C., Candia, R., Clausen, B. S., M&up, S., Oviedo-33071, Spain and Topsoe, H., J. Catal. 68,453, 1981. 9. Topsoe, H., Clausen, B. S., and Morup, S., Hyperjne Interact. 27, 231, (1986). J. SCHWANK 10. Duchet, J. C., van Oers, E. M., de Beer, V. H. J., and Prins, R., J. Catal. 80, 386, 1983. 11. Prins, R., Bouwens, S. M. A. M., Koningsberger, Department of Chemical Engineering D. C., and de Beer, V. H. J., “EXAFS and XPS The University of Michigan of Sulfided Co(Ni)-MO/C and Co(Ni)-Mo/Alz03 Ann Arbor, Michigan 48109-2136 Catalysts,” 1lth North American Meeting of the Catalysis Society, Dearborn, MI, May 7-l 1, 1989. Received May 22, 1989; revised August I, 1989