Elucidation of Fischer-Tropsch reaction kinetics

Xiaodong Zhan, Heinz J. Robota, and Kym B. Arcuri Syntroleum Corporation 4233 S. 49th West Avenue Tulsa, OK 74107 Abstract Fischer-Tropsch reaction kinetics has been elucidated using a Syntroleum proprietary Co/Al2O3 catalyst. The study was conducted in CSTR reactors at 20-30 bar, 210-230 oC, H2/CO ratio = 2.1, and 10%-60% CO conversion. All data were collected after at least 100 hours of time-on-stream to eliminate the ambiguity between initial catalyst activity and stable activity. The kinetic expression in the simple power a law form of rco = kPco was adopted since the H2/CO ratio in the feed gas, and thus in the reactor tail gas, was always approximately constant. In this expression, the exponent, a, is, in fact, the sum of the reaction orders of hydrogen and carbon monoxide. This simplified method significantly reduces the amount of experimental work and is appropriate when process design considerations are the primary motivation for developing the kinetic expression. It is found that this empirical kinetic expression is adequate for both fresh and regenerated catalysts within a wide range of practical operating conditions. Introduction Fischer-Tropsch (FT) synthesis is a complex reaction that produces paraffins, olefins, and oxygenates. A large number of studies have been performed in an attempt to understand the mechanism and kinetics of FT reactions on cobalt-based catalysts in slurry reactors (1-3). However, there is little consensus among the different studies regarding the kinetic parameters. It appears that there is no one kinetic expression that can be applied universally due to differences in the type of catalysts and reaction conditions. A kinetic study is typically performed to discriminate between possible reaction mechanisms or to support process design and optimization. For industrial interest, the power law empirical expression, or detailed Langmuir-HinshelwoodHougen-Watson (LHHW) type expressions is often used to represent the kinetic correlation under certain processing conditions. It is believed that LHHW expressions predict rates over a wider range of conditions than the simple power law expressions due to their fundamental origins. However, for FT reactions, it is unclear if detailed LHHW expressions indeed do have any superiority since data from the literature are essentially not comparable with each other. For specific catalysts under proposed commercial operating conditions, a separate kinetic study is typically necessary, since the catalyst performance may change with a different dependence on reaction conditions, due to possible mechanistic variations.

As kinetic study is tedious and requires intensive experimental work, simplification is usually preferred when process design considerations are the primary motivation for developing the kinetic expression. Presented here is a simplified kinetic study method that provided appropriate data to support process design, yet the amount experimental work is significantly reduced. Experimental In this study, the experiments were performed in a continuous, mechanically stirred 1-liter tank reactor (CSTR) at 20-30 bar, 210-230 oC, H2/CO = 2.1, feed CO concentration 5%-16%, and CO conversion 15%-60%. The catalyst is a Syntroleum proprietary commercial-intent Co/Al2O3 catalyst. The spent catalyst was obtained from our Cherry Point test in a 42-inch slurry bubble column reactor with about 6000 run hours (4). The catalyst was reduced or regenerated prior to loading in a CSTR for evaluation. Typically, a catalyst volume of 120 cc was loaded into a 1-inch fluid-bed reactor at room temperature. The catalyst was subjected to N 2 drying prior to switching to hydrogen gas at the desired space velocity and pressure. During reduction, the reactor was heated at 0.5 o C/min to 350 oC and held for 8 hours. At the conclusion of reduction, the reactor was cooled to room temperature in the flowing H2 followed by N2 purging and pressure letdown. The catalyst was placed in molten Astor wax for CSTR evaluation; the operation was performed in a glove box to avoid catalyst exposure to air. The operational procedures in catalyst regeneration were similar to that of catalyst reduction described above. The spent catalyst was dewaxed so that it is in the form of free flowing powder before sequential oxidation and reduction. The detailed regeneration procedure has been described elsewhere (5). After loading the reduced catalyst-in-wax in a CSTR reactor, the reactor system was purged with N 2 and pressurized prior to introducing the syngas for the FT reactions. The reactor was ramped to 210 oC at 0.5 oC/min at a GHSV that is high enough to avoid early catalyst deactivation. The data reported here were collected after at least 100 hours of time-on-stream to eliminate the ambiguity between initial catalyst activity and stable activity. Results and Discussion For the FT process with a cobalt-based catalyst, the H2/CO ratio of about 2.1 in the feed gas is most efficient since this is approximately the stoichiometric gas consumption ratio, considering the catalyst productivity and selectivity. For the sake of industrial application, the simple power law kinetic expression in the form of Equation (1) continues to be of great interest, although the impact of the product water has

Petroleum Chemistry Division Preprints 2004, 49(2), 200-202

been suggested to be incorporated into the kinetic expression to reflect the catalyst deactivation by back-oxidation (6).
a' b' rco = k ' Pco PH

(1) In this work, the kinetic expression is further simplified to be in the form of Equation (2) since the H2/CO ratio in the feed gas, and thus in the reactor tail gas, was always controlled to be approximately constant.

rco = kP = k 0 e
a co

E RT

a Pco

(2)

that this is due to the fact that the absolute water partial pressure under our reaction conditions is not high enough to cause noticeable catalyst deactivation and thus departure of the reaction rate from the kinetic response. In a typical Syntroleum reactor, there is at least 50% inert gas in the feed syngas since our ATR is air blown. The H2O/H2 increases with increasing CO conversion but the absolute water partial pressure is substantially lower than that using oxygen blown systems. This is consistent with the literature reports that the deactivation of cobalt-based catalysts by water back oxidation depends on both of the H2O/H2 ratio and absolute water partial pressure at the same time (7).

In the above expression, the exponent, a, is, in fact, the sum of the reaction orders of hydrogen and carbon monoxide, a+b in Equation (1). This simplified method significantly reduces the amount of experimental work since only three parameters are to be determined. Figure 1 shows the dependency of CO conversion on spacetime at 210 oC and 21 bar total pressure with the fresh catalyst. The CO concentration in the feed gas was from 5% to 16%. It is seen that the data fit the GHSV curve very well, albeit that the CO partial pressures vary. When the CO conversion is over 60%, the dependency of conversion on GHSV becomes very weak. This could be attributed to the catalyst deactivation by water back-oxidation when the conversion is too high leading to high absolute water pressure and high H2O/H2 ratio in the reactor (7).

Figure 2 Determination of total reaction order at 210 oC and 21 bar total pressure on the fresh catalyst. The temperature dependence of the reaction rate on the fresh catalyst is obtained from an Arrhenius plot at temperatures ranging from 210 oC to 230 oC. Figure 3 shows that the apparent activation energy is 28.6 0.6 kcal/mol, which is comparable to 26.6 kcal/mol that has been reported previously by R.B. Anderson on Co/Al 2O3 (8).

Figure 1 CO conversion as a function of space-time at 210 oC and 21 bar total pressure with the fresh catalyst. The total reaction order (a) and the rate constant ( k) at 210 oC can thus be determined by plotting CO rate versus CO partial pressure in logarithm scale, as shown in Figure 2. The total reaction order is found to be 0.64 0.03 with a high correlation coefficient of 0.98. It appears that, under our reaction conditions, the impact of water is not significant as can be seen from the tight kinetic fit of the data. It is possible

Figure 3 Arrhenius plot of CO conversion at 210 oC230 oC and 21 bar total pressure with the fresh catalyst

Petroleum Chemistry Division Preprints 2004, 49(2), 200-202

The final FT reaction kinetics with the fresh catalyst thus takes the form of Equation (3)

rco = k 0 e

14385 T

0.64 Pco

(3)

where the units of T, Pco, and r co are Kelvin, bar (absolute), and v/v/h, respectively. It should be noted that the FT reaction kinetics reported here considers only the total CO consumption. No impact of the temperature on the gas selectivities has been taken into account. Nevertheless, a similar data treatment method can be applied to gas selectivities so that more thorough kinetics for the liquid productivity can be developed. The above FT reaction kinetics was obtained for the fresh catalyst. It would be interesting to examine its applicability to regenerated catalysts, as catalyst regeneration is a component of Syntroleums FT technology. Figure 4 shows the experimentally measured and kinetically predicted CO rates of some of the catalysts tested. The data are somewhat scattered at high reaction rates; however, the standard error of the kinetic prediction of all of the data is only 6%. The good agreement of the data fit indicates that the single kinetic expression can be used to describe all Syntroleum catalysts, independent of the catalyst history and pretreatment method. This may indicate that all of the catalysts have similar chemical nature towards FT reactions. The only difference between the catalysts lies in the variation of the collision frequency, which is related to the number of active sites available for FT reaction.

When the H2/CO ratio in a Fischer-Tropsch reactor is maintained approximately constant, a simplified power law expression can be used to represent the reaction kinetics by lumping the reaction orders of CO and hydrogen together. This simplified method significantly reduces the amount of experimental work and has been demonstrated to be appropriate when process design considerations are the primary motivation for developing the kinetic expression. It is found that this empirical kinetic expression is adequate for both fresh and regenerated catalysts within a wide range of practical operating conditions.
Acknowledgement. The authors are grateful to the Technology Department of the Syntroleum Corporation, Tulsa, Oklahoma, for the opportunity to have carried out this investigation and for permission to publish the results . References (1) Espinoza, R.L.; Steynberg, A.P.; Jager, B.; Vosloo, A.C. Applied Catalysis A: General 1999, 186, 13-26. (2) Zennaro, R.; Tagliabue M.; Bartholomew, C.H. Catalysis Today 2000, 58, 309-319. (3) Keyser, M.J.; Everson, R.C.; Espinoza, R.L. Ind. Eng. Chem. Res. 2000, 39(1), 48-54. (4) Schubert, P.F. et. al. Prepr. Pap. - Am. Chem. Soc., Div. Pet. Chem. 2000, 45(2), 214-217. (5) Huang, R.; Agee, K.L.; Arcuri, B.A.; Schubert, P.F. US Patent, 2002, US2002/0183403 A1. (6) Li, J.; Zhan, X.; Zhang, Y.; Jacob, G.; Das, T.; Davis, B.H. Applied Catalysis A: General 2002, 228, 203-212. (7) Hilmen, A.M.; Schanke, D.; Hanssen, K.F.; Holmen, A. Applied Catalysis A: General 1999, 186, 169-188. (8) Anderson, R.B.; Kobel, H.; Ralek, M. In The Fischer-Tropsch Synthesis , Academic Press, Inc., Orlando FL, 1984.

Figure 4 Experimentally measured CO reaction rates vs. kinetically predicted rates with both fresh and regenerated catalysts

Conclusions

Petroleum Chemistry Division Preprints 2004, 49(2), 200-202