546
Energy & Fuels 1996, 10, 546-551
Mo1 ssbauer Study of Iron Fischer-Tropsch Catalysts during Activation and Synthesis K. R. P. M. Rao,*,† F. E. Huggins,† G. P. Huffman,† R. J. Gormley,‡ Robert J. O’Brien,§ and Burtron H. Davis§ Consortium for Fossil Fuel Liquefaction Science, 533 South Limestone Street, Room 111, University of Kentucky, Lexington, Kentucky 40506, Pittsburgh Energy Technology Center, U.S. Department of Energy, Pittsburgh, Pennsylvania 15236, and Center for Applied Energy Research, 3472 Iron Works Pike, Lexington, Kentucky 40511 Received October 11, 1995X
Mo¨ssbauer data indicate that activating an iron Fischer-Tropsch catalyst containing Cu, K, and kaolin with syngas for 24 h at 270 °C and 1.14 MPa produces a material that has about 60% of the iron in the form of Fe3O4 and the remainder as a mixture of χ-Fe5C2 and ′-Fe2.2C. A similar activation with CO produces a material that contains more than 80% of the iron in the form of χ-Fe5C2 and the remainder as Fe3O4. The CO-activated catalyst exhibits a high activity, whereas the syngas-activated sample has a low activity. Moreover, the CO-activated sample converts during 50 h of synthesis to a mixture that contains more Fe3O4 (40%) and less χ-Fe5C2 (60), whereas the syngas activated sample contains about 80% χ-Fe5C2 after 50 h of exposure to synthesis gas. Thus, both catalysts underwent significant changes in the bulk composition without a parallel change in CO conversion. Another catalyst formulation, containing more Cu, much less K, and no binder, showed significantly higher syngas conversion after activation by CO than by syngas even though the bulk phase compositions were similar. A catalyst, initially activated with syngas and exhibiting low activity, showed evidence for only Fe3O4. This low-activity material after a 24 h treatment in CO exhibited a dramatic increase of the carbide phase and catalytic activity. The analysis of phase composition is informative about changes in the bulk composition of iron catalysts, but these changes do not necessarily track directly catalytic activity.
Introduction Iron catalysts have been used extensively for the Fischer-Tropsch synthesis (FTS) and are currently the catalysts utilized in the commercial operations at Sasol.1 The iron catalyst permits the production of olefinic products when operated in either the low or high alpha mode.2 The iron catalyst, unlike the cobalt and ruthenium catalysts, forms a carbide under the reaction conditions normally employed in its use. Thus, the composition of the iron catalyst is one aspect that must be elucidated to develop a complete understanding of the FTS mechanism (e.g., ref 3 ). In general, activation of an iron catalyst can be accomplished by (1) reduction by CO, (2) reduction by H2, or (3) reduction by synthesis gas (e.g., refs 4-6 ). Much of the early work was accomplished for fused or sintered catalysts, and these could only be made active by first reduction to elemental iron in a hydrogen atmosphere.7 Exposure to synthesis gas converted the †
University of Kentucky. Pittsburgh Energy Technology Center. Center for Applied Energy Research. X Abstract published in Advance ACS Abstracts, April 15, 1996. (1) Dry, M. E. CHEMTECH 1982, 22, 744. (2) Rao, V. U. S.; Stiegel, G. J.; Bose, A. C.; Cinquegrane, G. J. Prepr. Pap.sAm. Chem. Soc., Div. Fuel Chem. 1992, 37, 184. (3) Ponec, V. In Coal Science; Academic Press: New York, 1984; Vol. 3, pp 1-62. (4) Zarochak, M. F.; McDonald, M. A. Proceedings of the Indirect Liquefaction Contractor’s Meeting; PETC, U.S. DOE: Pittsburgh, PA, Dec 1986; pp 58-82. (5) Hofer, L. T. E. In Catalysis; Emmett, P. H., Ed.; Reinhold Publishing: New York, 1956; Vol. 4, pp 373-442. (6) Report on the Petroleum and Synthetic Oil Industry of Germany; Ministry of Fuel and Power, His Majesty’s Stationery Office: London, 1946; pp 96-100. ‡ §
0887-0624/96/2510-0546$12.00/0
iron in part or essentially completely to the carbide. Activation in CO resulted first in the reduction to Fe3O4 and then in slower conversion to carbides (e.g., refs 8-10 ). Activating in synthesis gas is more complicated because of the competition between synthesis and carbiding as well as the formation of water that retards the reduction/carbiding steps. Much of the early German work utilized atmospheric or low-pressure startup pressures when synthesis gas was used.6 It has been claimed that an activation in CO produces an active catalyst that does not require a complex start-up program and is essentially independent of the catalyst composition; however, the activation procedures used either with synthesis gas or with hydrogen reduction appear to depend upon the catalyst formulation.8 Much of the early characterization of the phases of iron present in FTS catalysts utilized techniques such as X-ray diffraction and magnetization; however, these were difficult to make quantitative.7 Mo¨ssbauer spectroscopy, because of its sensitivity for iron and its quantitative nature, provides a convenient technique for the characterization of iron FTS catalysts and the changes in composition that they undergo during synthesis. The utilization of a continuous stirred tank reactor (CSTR) permits all of the catalyst charge to be (7) Storch, H. H.; Golumbic, N.; Anderson, R. B. The FischerTropsch and Related Synthesis; Wiley: New York, 1951. (8) Davis, B. H. Proceedings of the Coal Liquefaction and Gas Conversion Contractor’s Review Meeting; PETC, U.S. DOE: Pittsburgh, PA, Sept 1994; pp 373-387. (9) Baltrus, J. P.; Diehl, J. R.; McDonald, M. A.; Zarochak, M. F. Appl. Catal. 1989, 48, 199. (10) Bukur, D.; Nowicki, L.; Lang, X. Energy Fuels 1995, 9, 620.
© 1996 American Chemical Society
Iron Fischer-Tropsch Catalysts
Energy & Fuels, Vol. 10, No. 3, 1996 547
Table 1. Description of Catalysts and the Activation and Synthesis Conditions Used in This Study activation conditions run
catalyst
oil
1 2
UCI 1185-57 UCI 1187-57
C30 C30
3
UCI 1185-149-2
4
UCI 1185-149-2
distilled Allied Signal C28
5 6
UCI 1185-149-3 UCI 1185-149-3
7 8 9
100 Fe/3.6 Si/0.71 K 100 Fe/3.6 Si/0.71 K 100 Fe/3.6 Si/2.6 Cu/0.71 K
C28 distilled Allied Signal C30 C30 C30
synthesis conditions
flow, normal L/h‚g(Fe)
T, °C
P, MPa
CO H2:CO ) 0.7 CO
2.0 2.0
270 280
1.31 1.14
1.4
270
H2:CO:He ) 0.74:1.06:1.0 CO CO
2.4
H2:CO ) 0.7 H2:CO ) 0.7 H2:CO ) 0.7
gas
exposed to the same gas composition and, with appropriate experimental technique, to withdraw catalyst samples for characterization at intervals during the activation and synthesis operations. These techniques have been employed in the work that is reported in this paper. Experimental Section Two of the catalysts utilized in this study were prepared for use in a screening study designed to provide data for the selection of a catalyst for use in the bubble column reactor at LaPorte, TX. These two catalysts were prepared by United Catalyst, Inc., according to compositions provided them by the consortium to operate the LaPorte facility. These catalysts were formed into spherical particles nominally of about 4060 µm diameter using a spray-drying technique. Two other catalysts were formulated at the Center for Applied Energy Research (CAER); one batch contained copper and the other did not. These latter two catalysts were prepared for studies to define the role of copper in determining the phases present in the catalyst. The CAER catalysts were prepared by continuous precipitation at a pH of about 9 from aqueous solutions of Fe(NO3)3‚9H2O and concentrated NH4OH. The precipitate was thoroughly washed with distilled-deionized H2O and dried at 120 °C. Activity data were obtained so that an attempt could be made to define the role of catalyst phase in the determination of the catalytic activity. A listing of the catalysts and the activation and synthesis conditions are compiled in Table 1. For a typical test at the CAER, catalyst slurry consisting of 72.7 g of precipitated catalyst and 290 g of C28 (purified octacosane), C30 (Shell decene trimer), or C40+ (vacuum distillation residue of Allied Signal polyethylene) slurry oil was mixed inside a 1 L autoclave operated as a continuous stirred tank reactor (CSTR). The catalyst slurry was heated to 270 °C at 1.5-2.0 °C/min under a flow of CO [2.0 normal L/h‚g(Fe)]. CO activation continued at 270 °C and 1.31 MPa for 22-24 h. Following activation, hydrogen flow was started to give a H2/CO ratio of 0.7 with a flow rate of 3.4 normal L/h‚g(Fe). Catalyst slurry samples were removed (10-15 g) from the reactor at various times of the activation and synthesis. Catalyst slurry samples were Soxhlet extracted using refluxing toluene or o-xylene to remove accumulated hydrocarbons. Activation in synthesis gas was accomplished by heating the slurry to 200 °C in He and then switching to synthesis gas (H2/CO ) 0.67) at 150 °C, after which the sample was heated to 280 °C at a rate of 7.0 °C/min. After the catalyst was held at 1.14 MPa and 280 °C in a flow of synthesis gas for 12 h, the pressure was increased to 1.48 MPa and the temperature was lowered to 265 °C. These conditions for the activation and synthesis were chosen to correspond to the conditions used for run 2 at the LaPorte, TX, slurry reactor demonstration
gas
flow, normal L/h‚g(Fe)
T, °C
P, MPa
H2:CO ) 0.7 H2:CO ) 0.7
3.4 2.5
270 280
1.31 1.48
1.31
H2:CO ) 0.7
2.4
270
1.31
280
1.14
H2:CO ) 0.7
2.4
270
1.31
1.4 1.4
270 270
1.31 1.31
H2:CO ) 0.7 H2:CO ) 0.7
2.4 2.4
270 270
1.31 1.31
3.1 3.1 3.1
270 270 270
1.31 0.10 1.31
H2:CO ) 0.7 H2:CO ) 0.7 H2:CO ) 0.7
3.1 3.1 3.1
270 270 270
1.31 1.31 1.31
plant and were similar to those utilized by Mobil Oil during earlier studies for the U.S. Department of Energy (U.S. DOE).11 CO and H2 conversions were determined by analyzing the exit gas stream with a Carle gas analyzer. This standard instrument for gas analysis utilizes a variety of gas chromatography (GC) columns and switching valves to sequence the transfer of various gases contained in the injected sample to the appropriate column during a prescribed time interval. For a typical test conducted at the Pittsburgh Energy Technology Center (PETC) with CO activation, conditions were slightly different than at CAER. Precipitated catalyst (36.5 g) and 370 g of wax were loaded (9 wt % slurry) into a 1 L autoclave. The slurry was heated under a flow of helium at 1.31 MPa to 150 °C and stirred at 1000 rpm. Then heating was continued at 1 °C/min under a flow of CO [1.4 normal L/h‚g(Fe)] to 270 °C. CO activation continued at 270 °C and 1.31 MPa for 22-24 h. Following activation, synthesis gas with a H2/CO ratio of 0.7 was introduced at a rate of 2.4 normal L/h‚g(Fe). The synthesis was carried out at 270 °C and 1.31 MPa. Catalyst slurry samples (∼1-3 g) were removed from the reactor. The solid wax kept the catalyst from oxidizing in air during handling. This has been verified by comparing the Mo¨ssbauer data of two samplessone withdrawn directly into the Mo¨ssbauer sample holder to prevent exposure to air and then analyzed within hours after collection and a second sample that was stored in a sealed screw-cap bottle for about a month without special precautions to prevent exposure to air. For the syngas activation conducted at PETC, the catalyst/ wax slurry was heated in flowing helium at 1.14 MPa to 200 °C. Then syngas was introduced at 1.8 normal L/h‚g(Fe), with an additional 0.6 normal L/h‚g(Fe) of helium. The temperature was raised to 280 °C and held for 14 h, again at 1.14 MPa. Following activation, synthesis gas with a H2/CO ratio of 0.7 was introduced at a flow rate of 2.4 normal L/h‚g(Fe). The pressure was raised to 1.31 MPa and the temperature lowered to 270 °C. The Mo¨ssbauer experiments were carried out using a Halder, GmbH, spectrometer equipped with dual 57Co sources at either end of the drive so that calibration spectra of metallic iron could be collected simultaneously to those of the unknown catalyst samples. The spectrometer was run in the symmetric constant acceleration mode with 100 µs of dwell time per channel. The spectra were collected over 1024 channels in mirror image format. The sources consisted of 57Co in a Pd matrix; the original strength of each source was 50 mCi. The samples were examined in powder form mounted in plexiglass compression holders that present a thin aspect to the γ-ray beam. All samples were investigated at room temperature, typically over a velocity range of (12 mm/s; a few selected samples were also run at cryogenic temperatures. Spectra (11) Kuo, J. C. W. Two-Stage Process for Conversion of Synthesis Gas to High Quality Transportation Fuels. Final Report, DOE/PC/ 60019-9, Oct 1985.
548
Energy & Fuels, Vol. 10, No. 3, 1996
Figure 1. Iron phase distribution in the UCI 1185-57 catalyst versus time on stream (a) during activation with CO and (b) during FT synthesis: superparamagnetic and Fe3O4 (s); iron carbides (- - -). See Table 1 for reaction details.
Rao et al.
Figure 2. Comparison of syngas conversion for UCI 1185-57 catalyst activated with CO (9) and syngas (4). See Table 1 for reaction details.
were usually acquired for a minimum of 48 h before they were stored on a DOS-based PC and then transferred to a MicroVAX computer for analysis. All spectra have been analyzed by means of a least-squares fitting procedure that models the spectra as combinations of quadrupole doublets and magnetic sextets based on a Lorenzian line-shape profile. The individual absorption features were identified on the basis of their isomer shift, quadrupole splitting, and magnetic hyperfine field values; the areas associated with each absorption were used to calculate the percentage of iron present in each distinct form. Errors in the determined percent Fe values are about (3% for well-resolved spectra; in those that contain several iron oxide and carbide phases the uncertainty increases with the complexity of the spectrum.12 However, these complex spectra are obtained during the course of change from a predominantly iron carbide or iron oxide phase and conform to a general trend.
Results 64% Fe2O3/5% CuO/1% K2O/30% Kaolin (UCI1185-57). During activation in CO, this catalyst was first reduced to Fe3O4 and then converted to iron carbide (Figure 1). At the end of the 24 h activation period, the catalyst contained about 90% χ-Fe5C2, with the remainder of the iron being present in a superparamagnetic form (Figure 1). Low-temperature (-261 °C) Mo¨ssbauer measurements on one of the samples of withdrawn catalyst indicated that the superparamagnetic (spm) phase was Fe3O4, and not the carbide, which can also be present in a spm phase. During synthesis, the amount of χ-Fe5C2 decreased to about 60% after 30 h and then remained at about this level (Figure 1). The CO-activated catalyst exhibited a conversion (H2 + CO) in the 70-80% range during 150 h of synthesis (Figure 2). During activation in synthesis gas, the catalyst was rapidly converted to Fe3O4 and the Fe3O4 then slowly converted to form iron carbides. At the end of the (12) Huggins, F. E.; Huffman, G. P. In Analytical Methods for Coal and Coal Products; Karr, C., Jr., Ed.; Academic Press: New York, 1979; Vol. 3, pp 371-423.
Figure 3. Iron phase distribution in the UCI 1185-149-2 catalyst versus time on stream (a) during activation with syngas and (b) during FT synthesis: superparamagnetic and Fe3O4 (s); iron carbides (- - -). See Table 1 for reaction details.
activation period, about 40% of the iron was present as a mixture of χ-Fe5C2 and ′-Fe2.2C (Figure 3). As can be seen in Figure 2, the conversion of syngas was increasing during the activation period and had attained about 60%. The amount of carbide was also increasing and had attained a level of about 40%. Almost immediately after the change of conditions to those used for the synthesis, the conversion dropped to the 20-30% level. During 24 h of synthesis, the catalyst composition changed so that it contained almost 80% χ-Fe5C2 and ′-Fe2.2C. At these low conversion levels, the formation of carbide continued so that it exceeded that of the CO-activated catalyst at an equivalent time on stream. It appears that carbide formation may be controlled primarily by the level of conversion of the syngas and not by the activating gas composition. A high level of (H2 + CO) conversion produces CO2 and H2O, which are oxidizing gases. On the other hand, for low (H2 + CO) conversions, the catalyst is exposed to a reducing atmosphere.
Iron Fischer-Tropsch Catalysts
Energy & Fuels, Vol. 10, No. 3, 1996 549
Table 2. Catalyst Activation and Synthesis Activity [UCI 1185-149-2; 57.2 Fe/9.3 Cu/0.05 K; Activated in CO at 270 °C, 1.31 MPa, 22 h, 1.4 Normal L/h‚g(Fe); FT Synthesis Conditions: H2/CO ) 0.70, 270 °C, 1.31 MPa, 2.4 Normal L/h‚g(Fe); 9% Catalyst Loading in Distillation Residue of Allied Signal Polyethylene] (H2 + CO) conversion,a %
37
23
84.0
60
30
10
67.7
68
27
5
51.2
χ-Fe5C2
FT run TOS ) 84 h FT run TOS ) 235 h FT run TOS ) 305 h
40
a
% Fe in Fe3O4
spm
sample
FT synthesis runs were carried out at PETC.
Table 3. Catalyst Activation and Synthesis Activity [UCI 1185-149-2; 57.2 Fe/9.3 Cu/0.05 K; Activated in H2/CO/He at 270 °C, 1.3 MPa, 22 h, 1.8 Normal L of Syngas/h‚g(Fe); FT Synthesis Conditions: H2/CO ) 0.70, 270 °C, 1.31 MPa, 2.4 Normal L/h‚g(Fe); 9% Catalyst Loading in C28] sample activated for 14 h FT run TOS ) 26 h FT run TOS ) 49 h a
χ-Fe5C2, % Fe
Fe3O4, % Fe
spm phase
(H2 + CO) conversion,a %
36 84
48 14
16 2
16
91
7
2
6
Figure 4. Iron phase distribution in the UCI 1185-149-3 catalyst versus time on stream during activation with CO (022 h) and FT synthesis (>22 h) in C28 start-up wax: superparamagnetic (]), Fe3O4 (0), χ-Fe5C2 (O), and ′-Fe2.2C (4); CO + H2 conversion (b). See Table 1 for reaction details.
FT synthesis runs were carried out at PETC.
57.2% Fe/9.3% Cu/0.05% K. This catalyst differs from the one described above since it does not contain the kaolin binder (support); it also contains almost twice as much copper and only about 1/20 as much potassium. The catalyst was activated in CO under the conditions normally employed at the CAER (Table 1). During activation, the composition of the catalyst exhibited a decrease in the amount of spm Fe3O4 present as well as a decrease in the amount of magnetically ordered Fe3O4 and a corresponding increase in iron carbides. During synthesis, the amount of χ-Fe5C2 increased from the 30-40% level to about 70% (Table 2). At the same time there was a further decrease in the amount of spm Fe3O4 and Fe3O4; apparently the spm material is converted to the carbide more rapidly than the magnetically ordered Fe3O4. During synthesis, and while the carbide phase was increasing, the conversion dropped from about 90% to the 50% level. When this catalyst was activated in synthesis gas, it had a poor activity, with the conversion in the 5-20% range (Table 3). At the end of the 14 h activation period, the catalyst contained 36% χ-Fe5C2. After 26 h of synthesis, the amount of χ-Fe5C2 had increased to 84%, and this increased to 91% after 49 h. Activation of this catalyst in CO was conducted in two different slurry media: purified n-octacosane and the residue of a vacuum distillation of an Allied-Signal wax (comprised of predominantly C40+). Similar results were obtained in both solvents. In the C28 wax, the catalyst contained 86% χ-Fe5C2 following the 22 h activation and this phase declined to about 55% during 160 h of synthesis (Figure 4). During the synthesis period, χ-Fe5C2 appears to be converted predominantly to the spm material rather than to magnetically ordered Fe3O4. It appears that the fraction of iron present in the χ-Fe5C2 phase was slightly lower in the heavy solvent than in the lower molecular weight wax (64 versus 86%) (Figures 4 and 5). During the synthesis the carbide phase decreases with a corresponding
Figure 5. Iron phase distribution in the UCI 1185-149-3 catalyst versus time on stream during activation with CO (022 h) and FT synthesis (>22 h) in distilled Allied Signal startup wax: superparamagnetic (]), Fe3O4 (0), χ-Fe5C2 (O), and ′-Fe2.2C (4); CO + H2 conversion (b). See Table 1 for reaction details.
Figure 6. Iron phase distribution in the CAER 100Fe/3.6Si/ 0.71K catalyst versus time on stream during exposure to syngas (0-93 h), exposure to CO (93-115 h), and exposure to syngas (>115 h): superparamagnetic (]), Fe3O4 (0), and χ-Fe5C2 (O); CO + H2 conversion (b). See Table 1 for reaction details.
increase in the spm material. In both solvents a small fraction (0.06-0.12) of ′-Fe2.2C is formed during the initial period of the CO activation, but this phase is not observed following about 10 h of activation. The activities of the catalyst are similar in the two solvents during the first 100 h; however, the conversion is slightly lower in the higher boiling wax and declines in this wax more rapidly after 100 h on stream. 100% Fe/3.6% Si/0.71% K. This catalyst, following activation in synthesis gas, exhibited low activity.13 After 6 h of exposure to synthesis gas, the catalyst contained 25% magnetically ordered Fe3O4 and 75% spm phase (Figure 6). During the next 87 h, the amount of the spm phase decreased below the detectable limit, and (13) O’Brien, R. J.; Xu, L.; Spicer, R. L.; Davis, B. H. Energy Fuels 1996, submitted for publication.
550
Energy & Fuels, Vol. 10, No. 3, 1996
Figure 7. Iron phase distribution in the CAER 100Fe/3.6Si/ 0.71K catalyst versus time on stream during activation with syngas at 0.10 MPa (0-24 h) and during FT synthesis (>24 h): superparamagnetic (]), Fe3O4 (0), and χ-Fe5C2 (O); CO + H2 conversion (b). See Table 1 for reaction details.
the sample was essentially 100% magnetically ordered Fe3O4. During this period the synthesis gas conversion was in the range of 9-12%. Additional experiments have indicated that increasing the duration of the activation in syngas did not result in higher conversions even though the carbide content of the catalyst slowly increased with time. It is possible that some surface χ-Fe5C2 or spm χ-Fe5C2 species, undetected by the Mo¨ssbauer measurements, was responsible for the small syngas conversion that was observed. Following the 93 h exposure to synthesis gas, the hydrogen flow was stopped so that the sample was exposed to only CO at 270 °C and 1.31 MPa during a 22 h period.13 At the end of this 22 h period, the catalyst contained 58% magnetically ordered Fe3O4, 9% of the spm phase, and 33% χ-Fe5C2. Shortly after hydrogen was reintroduced so that the catalyst was again exposed to the synthesis gas, the conversion increased to 25% and within 4 h (119.5 h total) attained a level of 59%. Within 77 h following the CO treatment, the syngas conversion was 88% and remained at or above 80% during the next 500 h of the test. Thus, the CO treatment caused the conversion of about one-third of the iron to the carbide form, and this resulted in an increase in the conversion to greater than 80%. It is therefore apparent that the nearly 100 h exposure to synthesis gas did not inhibit the subsequent activation by CO. Furthermore, the subsequent CO activation produced a catalytic material that had an activity that was essentially the same as if the catalyst had been activated initially with CO. Another portion of the catalyst was activated in syngas under identical conditions as used above except that the total pressure was held at 0.1 MPa for a 24 h period before it was increased to the synthesis conditions. Results are given in Figure 7. After the 24 h period at atmospheric pressure, the catalyst contained 41% χ-Fe5C2, 27% spm phase, and 32% magnetically ordered Fe3O4. After 24 h at 1.31 MPa pressure, the conversion was 82%. The conversion declined slowly to 69% during the 384 h run.13 It is therefore apparent that the low-pressure treatment with synthesis gas led to the formation of a significant quantity of carbide, whereas this was not the case when the treatment was carried out at 1.31 MPa. In addition, the presence of iron carbide led to a catalyst with high activity, whereas the catalyst that did not contain a detectable amount of carbide had low activity. As a means of evaluating the role of copper upon the activation, the 100% Fe/3.6% Si/0.71% K catalyst was
Rao et al.
Figure 8. Iron phase distribution in the CAER 100Fe/3.6Si/ 2.6Cu/0.71K catalyst versus time on stream during activation with syngas: superparamagnetic (]), Fe3O4 (0), and χ-Fe5C2 (O); CO + H2 conversion (b). See Table 1 for reaction details.
impregnated with sufficient aqueous copper nitrate to give 2.6 atomic % Cu relative to Fe. This catalyst was then utilized for the synthesis [1.3 MPa, H2/CO ) 0.67, 270 °C, and 3.1 normal L/h‚g(Fe)] without prior activation.14 After 4.3 h, the sample consisted of 19% spm phase (carbide and/or oxide) and 81% magnetically ordered Fe3O4 (Figure 8). As the reaction continued, there was a decrease in the spm phase with an increase in the magnetically ordered Fe3O4. At 114 h of synthesis the spm phase corresponded to only 3% and there was 15% χ-Fe5C2; the carbide phase increased to 20% after 215 h on stream. The conversion of synthesis gas was higher when copper was present than when it was absent; this is true whether a carbide phase is present or not. Thus, after 19 h in syngas the Cu-containing material produced 26% conversion, whereas the same catalyst without Cu produced only about 10% conversion. The conversion continued to increase with time on stream when Cu was present, but this did not appear to be the case when Cu was not present. After 215 h on stream, the activity of the Cu-containing catalyst had increased to provide about 50% syngas conversion. Discussion For the kaolin-containing catalyst only the χ-Fe5C2 carbide phase was observed following activation in CO. When the same catalyst was activated with synthesis gas, both χ-Fe5C2 and ′-Fe2.2C were observed. Furthermore, the CO activation leads to a high fraction of the carbide phase and to high catalytic activity. With increasing time on stream the amount of the carbide phase decreases while the conversion remains essentially constant. Thus, the catalyst undergoes changes in chemical composition while the catalyst activity and selectivity13 both remain essentially constant. On the other hand, treating the catalyst in synthesis gas at 1.31 MPa leads to a material that produces low synthesis gas conversion. Under these low-conversion conditions, the fraction of iron present as a carbide phase gradually increases to exceed that of the CO activated sample. An iron catalyst containing much less potassium (0.05%) and no support produced a significant amount of carbide for both the CO and syngas activation. The activity of the catalyst activated with CO was significantly greater than the one activated in syngas. The activities for the material activated by the two gases were quite different when the catalysts contained the (14) O’Brien, R. J.; Xu, L.; Spicer, R. L.; Bao, S.; Milburn, D. R.; Davis, B. H. Catal. Today 1996, submitted for publication.
Iron Fischer-Tropsch Catalysts
Figure 9. Fraction of the initial carbide remaining after increasing time on stream: (2) UCI 1185-149 (syngas); (b) UCI 1185-149 (CO); (9) 100 Fe/3.6 Si/0.71 K (syngas, 1 atm). See Table 1 for reaction details.
same amount of carbide. Thus, when about 40% of the iron was present as a carbide, the CO-activated sample provided 84% syngas conversion (at 84 h on stream), whereas the syngas-activated sample produced a conversion that was very low (80%) conversion; however, the presence of carbide is not a sufficient condition for high activity. Furthermore, the current data are in agreement with the observation that the catalytic activity and selectivity will remain essentially constant while the catalyst undergoes significant chemical and/or phase changes. It is therefore not possible to predict the catalyst performance on the basis of the bulk analysis of the catalyst. However, when the bulk analysis of the catalyst, at the end of the activation step, shows no carbide present, the catalyst performance can be predicted to be poor. Acknowledgment. Portions of this work were supported by U.S. DOE Contracts DE-AC22-93PC93066, DE-AC22-91PC90056, and DE-AC22-94PC94055 and by the Commonwealth of Kentucky. EF9501962
