Cobalt

Role of the Alloy and Spinel in the Catalytic Behavior of Fe−Co/Cobalt Magnetite .... Christophe Lefevre , François Roulland , Alexandre Thomasson ...
0 downloads 0 Views 255KB Size
Energy & Fuels 2002, 16, 1271-1276

1271

Role of the Alloy and Spinel in the Catalytic Behavior of Fe-Co/Cobalt Magnetite Composites under CO and CO2 Hydrogenation F. Tihay,† A. C. Roger,† G. Pourroy,*,‡ and A. Kiennemann*,† LMSPC UMR CNRS 7515, 25, rue Becquerel, 67087 Strasbourg, France, and IPCMS UMR CNRS 046, 23, rue du Loess, 67037 Strasbourg, France Received March 7, 2002

The aim of this work is to compare the reactivity and the transformation of the Fe-Co alloy on the one side and the Co-containing magnetite on the other side in Fe-Co/Co-containing magnetite catalysts after CO/H2 and CO2/H2 reactions. Different Co-to-Fe ratios are studied. The catalysts undergo different modifications owing to their initial composition, but especially to the gas mixtures. X-ray diffraction (XRD) and high-resolution transmission electron microscopy (HRTEM) show that iron and cobalt carbides are formed in both cases: the iron carbides are formed first and are located on the particle surface while the cobalt carbides are formed in a second step and localized in the center. The spinel lattice parameter keeps constant values in CO/H2, while it decreases in CO2/H2. Furthermore the CO/H2 reaction occurs without a preliminary reduction unlike under CO2/H2 for which it is necessary. We propose that the carbides came preferentially from the Fe-Co alloy under CO/H2 and from the spinel under CO2/H2. CO2 formation during CO/H2 reactions could be related to the change of Fe-Co alloy and of the spinel.

Introduction Numerous works have been devoted to the FischerTropsch reaction (F-T) (e.g., refs 1-5 and references therein). Though, only the Co-based and Fe-based catalysts have been developed for industrial uses.6 These two metals have been successfully promoted by numerous additives, (alkalines, rare earth, metallic oxides which can be reducible or not) and used as bulk material or deposited on the usual supports (Al2O3, SiO2, TiO2...). They have been more rarely associated, and in this case, the properties are more likely due to the formation of an iron-cobalt alloy than the sum of their respective properties.7 This synergy effect depends on the initial Co-to-Fe ratio, but is difficult to understand on the basis of previous works on Co-Fe or Fe-Co catalysts. Some authors claim that the global activity of bimetallic catalysts decreases when the iron ratio increases7-9 while others claim the reverse.10-13 Various * Authors to whom correspondence should be addressed. (Kiennemann) Phone: 33 3 90242766. Fax: 33 3 90242768. E-mail: [email protected]. (Pourroy) Phone: 33 3 88 10 71 92. E-mail: [email protected]. † LMSPC UMR CNRS 7515. ‡ IPCMS UMR CNRS 046. (1) Anderson, R. B. The Fischer-Tropsch Synthesis; Academic Press: London, 1984. (2) Dry, M. E. Catal. Sci. Technol. 1981, 1, 1159-1255. (3) Vannice, M. A. Catal. Rev. Sci. Eng. 1976, 14, 153-191. (4) Pichler, H.; Schulz, H. Chem. Ing. Tech. 1970, 42, 1162-1174. (5) Riedel, T.; Claeys, M.; Schulz, H.; Schaub, G.; Nam, S. S.; Jun, K. W.; Choi, M.-J.; Kishan, G.; Lee, K. W. Appl. Catal. A 1999, 186, 201-213. (6) Jager, B. Stud. Surf. Sci. Catal. 1998, 119, 25-34. (7) Coville, N. J.; Duvenhage, D. J. Appl. Catal. 1997, A 153, 4367 and refs therein. (8) Villeger, P.; Barrault, J.; Barbier, J.; Leclercq, G.; Maurel, R. Bull. Soc. Chim. Fr. 1979, 9-10, 413-418.

selectivities in CH4 and C5+ versus Co/Fe ratio have been described owing to the authors. Probably, during the reduction step, the iron and cobalt oxide are changed into bcc alloys, but the reduction process does not allow a good control of its composition particularly on its surface which is iron-rich.7 To clarify this point, we have previously prepared Fe-Co/Co-containing magnetite catalysts14-16 for which the reduction step is avoided since the bcc Fe-Co alloy deposited on the spinel oxide is directly obtained during the synthesis and its composition can be determined. Although they are of submicronic sizes, these materials do not get oxidized in air. Such catalysts tested in the CO/H2 synthesis are active at 250 °C and mainly produce C2-C4 olefins.17,18 The CO2 selectivity can be low (e25 molar %) and depends on the initial Co/Fe ratio (from 0.23 to 0.53). A previous study has shown that at atmospheric pressure, the Fe-Co/Co-containing magnetite catalysts (9) Amelse, J. M.; Schwartz, L. H.; Butt, J. B. J. Catal. 1981, 72, 95-110. (10) Butt, J. B.; Schwartz, L. H.; Baerns, M.; Malessa, R. Ind. Eng. Chem. Res. Dev. 1984, 23, 51-56. (11) Arai, H.; Mitsuishi, K.; Seiyama, T. Chem. Lett. 1984, 12911294. (12) Chen, A. A.; Kaminsky, M.; Geoffroy, G. L.; Vannice, M. A. J. Phys. Chem. 1986, 90, 4810-4819. (13) Ishihara, T.; Eguchi, K.; Arai, H. Appl. Catal. 1987, 30, 225238. (14) La¨kamp, S.; Pourroy, G. Eur. J. Solid State Chem. 1997, 34, 295-308. (15) Yamegni-Noubeyo, J. C.; Pourroy, G.; Werckmann, J.; Malats i Riera, A.; Ehret, G.; Poix, P. J. Am. Ceram. Soc. 1996, 79, 20272032. (16) Cabet, C.; Roger, A. C.; Kiennemann, A.; La¨kamp, S.; Pourroy, G. J. Catal. 1998, 173, 64-73. (17) Tihay, F.; Pourroy, G.; Richard-Plouet, M.; Roger, A. C.; Kiennemann, A. Appl. Catal. 2001, 206, 29-42. (18) Tihay, F.; Roger, A. C.; Pourroy, G.; Kiennemann, A. Stud. Surf. Sci. Catal. 1998, 119, 143-148.

10.1021/ef020059m CCC: $22.00 © 2002 American Chemical Society Published on Web 07/31/2002

1272

Energy & Fuels, Vol. 16, No. 5, 2002

Tihay et al.

Table 1. Formulas of the Catalysts: Lattice Parameter and Composition of the Metal,a Weight Increase by Heating in Air and Co/Fe Global Ratios Determined by EDXS Analysis metallic phase catalysts

a (nm)

compositiona

∆m/m (%)

Co/Fe

formula

A B C D

0.2848(3) 0.2842(3) 0.2843(5) 0.2841(3)

Co0.63(3)Fe0.37(3) Co0.73(2)Fe0.27(2) Co0.70(2)Fe0.30(2) Co0.73(3)Fe0.27(3)

5.57(2) 7.17(2) 4.95(2) 6.23(2)

0.23(1) 0.33(1) 0.45(1) 0.53(1)

(Co0.63(3)Fe0.37(3))0.49(1)0[Co0.34(2)Fe2.66(2)O4] (Co0.73(2)Fe0.27(2))0.82(3)0[Co0.36(2)Fe2.64(2)O4] (Co0.70(2)Fe0.30(2))0.59(1)0[Co0.70(2)Fe2.30(2)O4] (Co0.73(3)Fe0.27(3))0.84(1)0[Co0.71(2)Fe2.29(2)O4]

a

Ref 25.

are also active in CO2/H2 reaction, but only after H2 reduction unlike the CO/H2 reaction.19 Moreover, in the CO/H2 reaction, it has been demonstrated that cobalt magnetite could be preserved under test17 while Fe3O4 is destroyed and carbides formed.20-21 The aim of this work is to compare the reactivity of the Fe-Co/Co-containing magnetite composites under CO-H2 and CO2-H2 atmosphere, to investigate the respective transformation of the alloy and spinel phases after reactivity in particular in the CO-H2 and CO2H2 gas mixtures, to shed some light on the CO2 conversion into hydrocarbons. Experimental Section Catalyst Preparation. Iron(II) and cobalt(II) chloride solutions were prepared with various Co-to-Fe ratios of 0.25, 0.33, 0.50, and 0.66. The cation concentration (Fe + Co) in the chloride solutions is always 3 M. A 100 mL volume of solution of iron and cobalt chlorides was added to 350 mL of a 10 M boiling KOH solution.14,17 This preparation is very peculiar and is based on the disproportionation of Fe(II) which occurs either in a basic media or in the solid state in FeO and produces a metallic phase.22-24 The mixture was stirred for 1 h at constant temperature (T ) 115 °C). The solution was filtered, and the obtained precipitate was washed with water, alcohol, and then dried at 40 °C for 4 h. As chlorides are known to be a poison for Fischer-Tropsch catalysts, the precipitate was washed with water until no chlorides could be detected in the filtrate with silver nitrate and until the pH of the filtrate was neutral. The catalysts are named samples A to D for Co-to-Fe ratios of 0.25, 0.33, 0.50, and 0.66, respectively. Catalysts Characterization. The catalysts were studied by Scanning Electron Microscopy (SEM) on a JEOL scanning electron microscope. The Co-to-Fe ratio in the composite was determined by means of energy-dispersive X-ray spectroscopy (EDXS). X-ray diffraction data were collected at room temperature using a D500 Siemens diffractometer equipped with a quartz monochromator (Co KR1 ) 1.78897 Å). Thermogravimetric (TG) and differential thermal (DT) analyses were carried out in silica crucibles in air or in a vacuum by using a Setaram 92 apparatus. The temperature was increased by 5 °C/min from 25 °C up to 980 °C. The formulas have been determined as previously:18 the EDXS analysis gives Co-toFe ratio in the catalysts and X-ray diffraction measurements allow us to determine the metal lattice parameter, therefore its composition (Table 1).25 Following the method presented in refs 18 and 22, these two data in addition to the weight (19) Tihay, F.; Roger, A. C.; Kiennemann, A.; Pourroy, G. Catal. Today 2000, 58, 263-269. (20) Kato, H.; Sano, T.; Wada, Y.; Tamaura. Y.; Tsuji, M.; Tsuji, T.; Wiyazski, S. J. Mater. Sci. 1995, 30, 6350-6354. (21) Bukur, D. B.; Koranne, M.; Lang, X.; Rao, K. R. P. M.; Huffmann, G. P. Appl. Catal. A 1995, 126, 85-113. (22) Benard, J. Ann. Chim. 1939, 12, 5-92. (23) Shipko, F. J.; Douglas, D. L. J. Phys. Chem. 1956, 60, 15191523. (24) Schrauzer, G. N.; Guth, T. D. J. Am. Chem. Soc. 1976, 98, 3508-3513.

increase in air obtained by TG measurements allow us to give a general formula of the catalysts. Formulas of the catalysts, written (CoRFe1-R)γ (CoβFe3-βO4) in which (CoRFe1-R) is the metal, γ the metal-to-spinel ratio, and (CoβFe3-βO4) the spinel oxide, are given in Table 1. γ is increased from sample A to B, as well as from C to D, while the Co-to-Fe ratio in the spinel is increased in sample D with respect to sample B. Microstructural information was obtained by using bright and dark fields (TEM) and high-resolution transmission electron microscopy (HRTEM). Two methods were used to prepare the catalysts before analysis. In the first method, the powders were subjected to ultrasonic waves in alcohol, then a drop of the suspension obtained was deposited on a holey, carbon-coated, 1000 mesh copper grid. In the second method, the powders were ground and embedded in resin (Epon 812). Thin sheets of catalysts were obtained using an ultramicrotome equipped with a diamond knife. The 60-90 nm thick sections were collected on holey, carbon-coated, 200 mesh copper grids. Both types of preparations were examined under a TOPCON 002B microscope operating at 200 kV (point-topoint resolution of 0.18 nm) and equipped with an ultrathin window KEVEX EDX spectrometer. Energy-dispersive X-ray spectroscopy (EDXS) was used to determine the cobalt-to-iron ratio in the powder and the individual grain compositions. Reactivity Tests. Catalytic tests under CO/H2 were performed in a fixed bed reactor with a 1 MPa pressure. A 300 mg quantity of catalyst was heated to 220 °C (0.2 °C min-1) under a 2.3 L h-1 nitrogen flow. Then the N2 was replaced by the CO/H2 (1:1) mixture with a GSHV of 3000 h-1. The catalytic tests were carried out between 230 °C and 260 °C, for 60 h for each temperature. The steady state is rapidly reached since no reduction is needed. Reactions under CO2/H2 were carried out in a fixed bed reactor with a 1 MPa pressure and 300 mg of catalyst. The catalyst was heated under a 2.4 L h-1 nitrogen flow up to 300 °C (1 °C min-1). At this temperature, N2 was replaced by H2 (1.2 L h-1) for 45 min. The temperature was decreased to 200 °C. The CO2/H2 mixture (1:4) (GHSV 3000 h-1) was introduced and tests were carried out between 260 and 300 °C (60 h for each temperature). For both reactivities, the outlet gases were analyzed on-line by gas chromatography and the liquid products were collected in two traps, the first one heated at 60 °C and the second one cooled at 15 °C, and then analyzed by gas chromatography.

Results Texture of the Catalysts before Tests. All the samples have been examined by using TEM and HRTEM techniques. Catalyst A is given as an example. Oxygen analyses by EDXS technique make it possible to discern metal and oxides zones. Further nanodiffractions performed on these zones confirm these analyses. Round metallic zones of about 50-150 nm width are located on the spinel surface and both phases are well fitted into each other (Figure 1a and 1b). The edge of (25) Pearson, W. B. Handbook of Lattice Spacing and Structures of Metals; Pergamon Press: Oxford, U.K., 1964; 4, p 504.

Catalytic Behavior of Fe-Co/Cobalt Magnetite Composites

Energy & Fuels, Vol. 16, No. 5, 2002 1273 Table 2. Catalytic Results of Catalysts A, B, C, and D under CO/H2 (1/1) at 1 MPa Pressure, GHSV ) 3000 h-1 catalyst

A

T (°C) CO conversion molar selectivity

250 5.9 CO2 18.4 liquid hydrocarbons 57.9 gaseous hydrocarbons 23.7 chain growth probability (R) 0.69 weight distribution in C1 34.9 gaseous fraction C2) 10.5 C2 9.4 ) C3 24.0 C3 3.3 C4) 14.4 C4 3.5 ΣC2-C4 65.1 % olefins in the C2-C4 fraction 75.1 ratio O/P in the C2-C4 fraction 3.0

B

C

D

240 5.1 23.6 27.0 49.4 0.63 32.6 9.7 9.4 19.9 4.6 17.3 6.3 67.4 69.7 2.3

250 5.2 15.2 23.0 61.8 0.57 34.3 12.4 5.9 20.7 4.1 18.5 4.0 65.7 78.5 3.7

250 6.2 21.5 22.3 56.2 0.58 36.4 4.2 7.1 25.3 4.7 17.4 4.9 63.6 73.7 2.8

Table 3. Catalytic Results of Catalysts A, B, C, and D under CO2/H2 (1/4) at 1 MPa Pressure, GHSV ) 3000 h-1 catalyst T (°C) CO2 conversion molar selectivity

A

260 28.4 CO 5.2 liquid hydrocarbons 49.0 gaseous hydrocarbons 45.8 weight distribution C1 34.5 in gaseous fraction ΣC2-C4 65.5 O/P ratio in the C2-C4 fraction 1.2

Figure 1. Bright field image of the catalyst A. (a) General view, (b) nanodiffraction on a metal grain along the zone axis (1,-1,1), (c) HRTEM of the metal egde, (d) HRTEM of the spinel edge. S is assigned to spinel. M corresponds to domains where oxygen has not been detected by EDSX analysis. M is assigned to the alloy.

the metallic particles is covered by a layer of 2-5 nm width which can be either amorphous or crystallized. The lattice plane spacings and the angles of the crystallized part correspond either to cobalt hydroxide Co(OH)2 or to iron hydroxide R-FeOOH (Figure 1c). On the contrary, the spinel phase spread out till the edges of the crystal as shown on Figure 1d which displays the 311 and (1-1 3) planes.

B

C

D

260 5.3 35.8 1.3 62.9 78.8 21.2 0.1

260 7.4 43.5 0.2 56.3 72.3 27.7 0.5

260 12.3 34.7 4.3 61.0 64.3 35.7 0.3

Catalytic Tests. CO/H2 Reactivity. The four catalysts have been tested for their ability to produce C2C4 alkenes from CO/H2. The catalysts were not subjected to any activation treatment. The results obtained are presented in Table 2 at iso-conversion of CO of almost 5% and at the steady state. It is reached at 250 °C for the catalysts A, C, and D and 240 °C for B. While the hydrocarbon distributions are similar for catalysts B, C, and D (50 to 62 molar % for the gaseous hydrocarbon fraction with 70 to 78% olefins in the C2C4 fraction), the behavior of A is very different: for this catalyst, the gaseous hydrocarbon fraction only represents 24% of the total hydrocarbons. The highest molar selectivity into hydrocarbons (85%) is obtained on catalyst C which is also the most selective catalyst for C2-C4 olefin production (O/P ) 3.7 in the C2-C4 fraction). Catalyst B is the less selective into hydrocar-

Figure 2. X-ray diffraction patterns of catalysts A, B, C, and D, before (a) and after the CO/H2 (b) and CO2/H2 (c) catalytic tests. b Spinel phase, O bcc Co-Fe alloy, 0 Co-Fe alloy isomorphous to R-Mn, 9 Co2C, and ( χ -Fe5C2.

1274

Energy & Fuels, Vol. 16, No. 5, 2002

Tihay et al.

Table 4. Lattice Parameters of the Spinel Phase of the Catalysts and Ratios of the (110) Diffraction Line of the Metal to the (311) One of the Spinel (R1, R2, R3) before and after Catalytic Testsa before tests

after CO/H2

after CO2/H2

catalysts

a (nm)

R1a

a (nm)

R2a

R2/R1

a (nm)

R3a

R3/R1

A B C D

0.8400(1) 0.8399(1) 0.8404(1) 0.8406(3)

0.23 0.50 0.30 0.45

0.8399(2) 0.8390(2) 0.8406(3) 0.8399(1)

0.22 0.09 0.36 0.39

0.95 0.18 1.20 0.88

0.8395(4) not determined 0.8388(4) 0.8389(2)

0.26 ∞ 0.36 0.90

1.13 ∞ 1.20 2.00

a

R1, R2, and R3 are given with a precision of 10%.

Figure 3. (a)TEM observation of the composite C after test in CO/H2. (b) The center of the particle (zone 1) is pure Fe-Co alloy. The Co/Fe ratio is close to 1 in the zone 2 and lower than 1 in zone 3. (c) χ-Fe5C2 is observed in the edge and Co2C in the aggregates center.

bons (76%) and the O/P ratio in the C2-C4 fraction is only 2.3. Concerning the hydrocarbon selectivity and the O/P ratio, catalysts A and D are intermediate between C and B. Let us note that the CO2 selectivity is low in all the cases, despite the high concentration of iron. CO2/H2 Reactivity. We have tested the catalysts under 1 MPa CO2/H2 pressure at 260 °C (Table 3). For catalyst A, CO2 conversion is high compared to CO in CO/H2 reaction and much liquid and gaseous hydrocarbons are produced. For B, C, and D, reactivity is lower, CO formation is important and mainly gaseous hydrocarbons are produced. Paraffins are favored with respect to olefins because the H2/CO2 ratio is higher than the H2/CO one. Comparison of the Compositions and the Textures after CO/H2 and CO2/H2 Catalytic Tests. After CO/H2. A spinel phase and a bcc Co-Fe alloy have been observed before testing as the X-ray diffraction pattern shows (Figure 2a). As the background is low, the occurrence of amorphous phases can be excluded. The spinel phase and the Co-Fe alloy are always present after the catalytic tests (Figure 2b). The lattice parameter of the spinel phase has not been modified under CO/H2 (Table 4). Additional diffraction lines corresponding to carbides, such as χ-Fe5C2 and Co2C are now observed.26,27 They are stronger for the B catalyst than for A, C, and D catalysts, showing that B interacts more strongly with the feedstock. These carbides result from the carburation of the metal, or the spinel, or both. Consequently, the diffraction lines of these latter phases

decrease. Thus, we have calculated the ratio of the intensity of the most intense diffraction line of the metal (110) to that of the most intense diffraction line of the spinel phase311 before testing (named R1) and after testing (named R2) (Table 4). Therefore, R2/R1 accounts for the relative intensity variations of the two phases. For instance, a R2/R1 ratio higher than 1 means that the spinel is more destroyed than the metal during the CO/H2 reaction. Conversely, a R2/R1 ratio lower than 1 means that the metal is more destroyed than the spinel. For catalyst B, the value of 0.2 means that the metal is very much destroyed compared to the spinel. For A and D, the R2/R1 ratios are close to 1, indicating that metal and spinel are destroyed at the same rate. For catalyst C, the value of 1.3 indicates that the spinel is slightly more destroyed than the metal. In addition, carbide formation is less in agreement with the higher stability of the two phases. The observation of the texture by TEM and HRTEM before and after the CO/H2 catalytic test shows that major changes have occurred. After testing, the round grains, which were observed before testing, near the spinel or partially included in it, have disappeared. Now, the spinel occurs either alone as before testing or near aggregates. Such an aggregate is displayed in Figure 3 for catalyst C. The Co/Fe ratio determined inside (zone 1 on Figure 3a) corresponds to that of the alloy. It decreases from the inside to the outside: Co/Fe is close to 1 in the zone 2 and iron is in majority in the zone marked 3. Nanodiffraction performed on the edge

Catalytic Behavior of Fe-Co/Cobalt Magnetite Composites

Energy & Fuels, Vol. 16, No. 5, 2002 1275

Figure 5. CO2 selectivity versus R2/R1

Figure 4. Catalyst C after testing in CO2/H2 atmosphere: the large particle named M corresponds to the iron-cobalt alloy. The SAD recorded on the zone delimited by the circle exhibits spots of the spinel phase and those of pattern of χ-Fe5C2.

exhibits the pattern of χ-Fe5C2. Thus, iron of the alloy reacts first with CO to form iron carbides. It is therefore drawn toward the edge, so that the inside becomes richer in cobalt. It is possible to conclude that alloy is more destroyed than spinel and that in the alloy iron tends to form carbides. After CO2/H2. The bcc metal and spinel oxide are always present for A, C, and D catalysts. Spinel of catalyst B completely disappears (Figure 2c), while metallic phase seems to be more preserved. Under CO2/ H2, as under CO/H2, B is the most transformed catalyst (spinel phase under CO2/H2, metal phase under CO/H2). This could be due to the poor composite character of B compared to A, C, and D, as observed by TEM.17 For A, C, and D, spinel phase has been modified as indicated by the decrease of the lattice parameter (Table 4). If R3 is defined as the ratio of the intensity, the (110) diffraction line of the metal to the 311 line of the spinel phase after test, the ratio R3/R1 higher than 1 indicates clearly that spinel has been more destroyed than metal, or totally destroyed as the infinite value shows for B (Table 4). TEM and HRTEM observations show that the catalysts have a completely different aspect after the CO2/H2 test. Catalyst C is shown as an example in Figure 4. The alloy occurs in broad zones of 100-500 nm wide (Figure 4a). HRTEM performed on the edge exhibits the same layer as it was observed before testing. The other particles that have undefined shapes contain mainly iron. A selected area diffraction pattern recorded on the particles with undefined shapes exhibits spots of the spinel phase and of χ-Fe5C2 (Figure 4b). (26) Rao, K. R. P. M.; Huggins, F. E.; Huffman, G. P.; Gormley, R. J.; O’Brien, R. J.; Davis, B. H. Energy Fuels 1996, 10, 546-541. (27) Shroff, M. D.; Kalakkad, D. S.; Coulter, K. E.; Ko¨hler, S. D.; Harrington, M. S.; Jackson, N. B.; Sault, A. G.; Datye, A. K. J. Catal. 1995, 156, 185-207.

Figure 6. CO2 selectivity versus olefins/paraffins

Therefore, these aggregates result from the reaction of the spinel with CO2/H2. It is therefore possible to conclude that spinel is more destroyed than alloy and that carbides are formed from the spinel. Due to the initial compositions of the spinel, iron carbides are favored compared to cobalt carbide. Discussion and Conclusions Comparing on the one side the structures and textures of the catalysts before and after the CO/H2 and CO2/H2 tests and on the other side the respective reactivities, allow some conclusions to be drawn: ‚While the metal is easily observed by HRTEM after CO2/H2 test, it is surrounded by carbides or difficult to observe after CO/H2 test. The reverse is observed for the spinel phase. Under CO2/H2, the spinel phase is surrounded by an amorphous phase while the metal surface does not change with respect to the raw catalyst.

1276

Energy & Fuels, Vol. 16, No. 5, 2002

The decrease of the lattice parameter of the spinel is only pointed out under CO2/H2. ‚The ratio between the diffraction lines intensities of the metal and of the spinel phases has changed. The metal-to-spinel ratio has increased under CO2/H2 and decreased under CO/H2. At the same time, cobalt and iron carbides have been formed but it seems that carbides are preferentially formed from the alloy under CO/H2 and from the spinel under CO2/H2. ‚We have seen in CO/H2 reaction that the weight percent of C2-C4 does not vary with Fe/Co initial ratio. The catalyst B, where the metallic phase is the most destroyed (R2/R1 ) 0.2) gives the higher CO2 selectivity. Usually, at 250 °C and with CO/H2 ) 1, the water gas shift reaction is important and the water produced by the reaction is consumed if the conversion is low. Consequently, the CO2 selectivities are about 50% at the equilibrium. In our case, the CO2 formation is surprisingly low. Therefore, two assumptions can be made: either the water gas shift reaction is not fast enough to reach the equilibrium, or the CO2 formed is directly transformed into classical Fischer-Tropsch compounds on the Fe-Co alloy or on the cobalt magnetite. Furthermore, the variation of CO2 selectivity versus R2/R1 ratio in Figure 5 shows that the highest the R2/R1, the lowest the CO2 selectivity. In other words, the highest the metal amount after the CO/H2 test, the more the spinel has contributed to the CO2 decrease. When the CO2 selectivity decreases, the metal ratio on the surface increases and the possibilities for the olefin to hydrogenate into paraffin are increased. Therefore the O/P ratio decreases when the CO2 selectivity increases (Figure 6). ‚In CO2/H2 atmosphere, hydrocarbons and particularly liquid hydrocarbons are favored when the spinel phase is not as much attacked as the metal. Furthermore, the most active catalyst (catalyst A) is also the one for which the spinel phase is the less attacked. If it is destroyed and transformed into carbides, the activity will decrease, because neither the oxide nor carbides can transform CO2 into C surface. Let us note that the more active catalyst for CO2 conversion under CO2/H2 is that for which the obtained CO molar selectivity is the lowest (see Table 3).

Tihay et al.

‚Therefore, all these data show that the metal (FeCo alloy) and the spinel (Co-containing magnetite) do not play the same role under CO2/H2 and under CO/H2. R2/R1 and R3/R1, especially for catalyst B, exhibit this difference. A low R2/R1 after CO/H2 (catalyst B) indicates that it is the metal (Fe-Co alloy) which has been progressively transformed into carbides (Figure 2). Iron carbides are formed first, then cobalt carbides as deduced from Figure 3. The spinel is comparatively preserved and its role would be restricted to the transformation of a part of CO2 formed. ‚Under CO2/H2, high values of R3/R1 indicate that the spinel is totally destroyed and the metal is partly preserved (Figure 2). We can then suggest that, after a H2 prereduction in order to partly reduce the spinel and with a H2/CO2 ratio of 4/1 which may maintain this reduction, CO2 would react with the spinel and form C surface. The transformation of CO2 into carbon has been recently demonstrated on oxygen-deficient structures such as nickel ferrite, wu¨stite, or magnetite.28-30 The decomposition takes place in two steps: CO2 into CO, then CO into C. The first step is the limiting reaction and the whole is in agreement with what is generally known. The as-formed carbon can react with the spinel in order to form carbides or use hydrogen in order to participate to the chain growth of Fischer-Tropsch type. Because of the spinel composition, only the iron carbide is formed (Figure 2). The alloy role would be to dissociate hydrogen in order to allow the chain growth from C surface. It could also participate to the reaction by way of CO formation. It is important that nonstoichiometric spinel can be regenerated after reduction since previous works have shown that R-Fe and Fe1-xO are transformed into Fe3C and permanently deactivated.30 ‡ EF020059M (28) Zhang, C. L.; Liu, Z. Q.; Wu, T. H.; Jiang, Y. Z.; Peng, S. Y. Mater. Chem. Phys. 1996, 44, 194-198. (29) Zhang, C. L.; Li, S.; Wu, T. H.; Peng, S. Y. Mater. Chem. Phys. 1999, 58, 139-145. (30) Zhang, C. L.; Li, S.; Wang, L. J.; Wu, T. H.; Peng, S. Y. Mater. Chem. Phys. 2000, 62, 44-51.