Effect of Water Vapor on the Reduction of Ru-Promoted Co/Al2O3

Journal of Catalysis 188, 281–290 (1999) A rticle ID jcat.1999.2666, available online at http://www.idealibrary.com on Effect of Water Vapor on the Reduction of Ru-Promoted Co/Al2O3 Yulong Z hang,1 D ongguang Wei,2 Sonia H ammache, and James G. G oodwin, Jr.3 D epartm ent of Chem ical and Petroleum E ngineering, University of Pittsburgh, Pittsburgh, Pennsylvania 15261 R eceived March 16, 1999; revised August 17, 1999; accepted August 18, 1999 The effect of water vapor on the reduction of a calcined Rupromoted Co/Al2O3 catalyst was investigated by introducing water vapor during the standard reduction procedure (H2, 350◦ C, 10 h) and during TPR. The presence of added water vapor (up to 3%) during standard reduction resulted in a decrease in the amount of Co able to be reduced. A peak located at ca. 500◦ C was observed in the TPR profiles of the catalyst after standard reduction in the presence of water. The amount of cobalt reducible in this temperature range increased with water vapor concentration. The reduction of this cobalt species, which was probably due to Co interacting strongly with the Al2O3, was inhibited during standard reduction in the presence of water vapor. Addition of water vapor during standard reduction also led to a decrease in the total amount of reducible cobalt (<900◦ C). The “nonreducible” cobalt existed mainly in the form of a cobalt aluminate with a reduction temperature higher than 900◦ C during TPR. It was observed that introduction of water vapor during TPR of the calcined catalyst also had a significant effect on the TPR profile. Two peaks could be observed in either the presence or the absence of added water vapor. The first reduction peak at ca. 230◦ C remained essentially unchanged; however, the second reduction peak temperature (400– 600◦ C) shifted up to 200◦ C higher and the peak area decreased to ca. one-third in the presence of increased water vapor pressures. The amount of cobalt able to be reduced during TPR to 900◦ C decreased with an increase in the amount of H2O added. The presence of water vapor during reduction appears to retard the reduction process by increasing the Co–alumina interaction and/or forming c 1999 Academic Press cobalt aluminates. ° Key Words: cobalt; Co/Al2O3; temperature-programmed reduction; reducibility; water vapor; Fischer–Tropsch. INTRODUCTION Supported cobalt catalysts are the preferred catalysts for the synthesis of heavy hydrocarbons from natural gas-based syngas because of their high Fischer–Tropsch (FT) activity, selectivity for linear hydrocarbons, and low activity for the 1 Permanent address: D epartment of Chemistry, Tongji U niversity, Shanghai, 200092, Peoples R epublic of China. 2 Permanent address: Institute of Coal Chemistry, Chinese Science A cademy, Taiyuan, 030001, Peoples R epublic of China. 3 To whom all correspondence should be addressed. water–gas shift reaction (1, 2). It is well known that the reduced Co metal, rather than its oxides or carbides, is the most active phase for CO hydrogenation (3). Investigations have been done on the nature of cobalt species on various supports: alumina (5–13), silica (11, 12, 14–26), titania (11– 12, 14, 27), magnesia (11, 12, 28), carbon (11, 12, 29), zeolites (30, 31). The reducibilities of various cobalt species in supported cobalt catalysts have been studied by TPR , XR D, E XA FS, E SR , FTIR , etc. (4–6, 16–17, 20–23). They can be affected by composition (17), support (11, 12, 14), preparation procedure (9, 19, 25), and pretreatment (5, 7, 32). Many authors (4, 5) have noted that compound formation between cobalt metal and the support can occur under reaction conditions. This is a possible deactivation route due to a decrease in the amount of metallic Co available in the catalyst. The impact of water concentration on catalyst performance has been investigated due to the high water concentrations that can exist at high reactant conversions. It has been discovered that the presence of water vapor during reaction can enhance the deactivation of cobalt catalysts due to surface oxidation or compound formation between the metal and the support (4, 5). The formation of cobalt silicates on Co/SiO 2 under hydrothermal conditions was extensively studied by Kogelbauer et al. (4). H ydrothermal treatment at 220◦ C led to a catalyst with lower reducibility due to the formation of both reducible and essentially nonreducible (at temperatures < 900◦ C) Co silicates. It was found that hydrothermal treatment of the reduced catalyst or hydrothermal treatment of the calcined catalyst in the presence of hydrogen produces Co silicates, while hydrothermal treatment of the calcined catalyst in air does not result in their formation. H ydrothermal treatment of the calcined catalyst in inert gas also has little effect. The deactivation of A l2O 3-supported cobalt Fischer– Tropsch catalysts as a result of the presence of water vapor during Fischer–Tropsch synthesis (FTS) has been studied by Schanke et al. (5). They found that oxidation of the supported cobalt is limited to the surface layers if H 2 is present during hydrothermal treatment. They also reported that a noble metal-promoted Co/A l2O 3 catalyst has a more rapid deactivation than an unpromoted catalyst. Several 281 0021-9517/99 $30.00 c 1999 by A cademic Press Copyright ° A ll rights of reproduction in any form reserved. 282 Z H A NG E T A L. studies using steady-state isotopic transient kinetic analysis (SSITKA ) have shown that the deactivation caused by water vapor being present does not alter the turnover frequency, but decreases the number of active intermediates leading to products (33, 34). Cobalt catalysts are usually calcined and then reduced by H 2 before reaction to produce higher Co dispersions. The reduction process produces water as the oxygen in the cobalt oxides reacts with hydrogen. If reduction of the catalyst is carried out in a large, deep-bed reactor, a high concentration of water vapor can result at the outlet of the reactor. U ntil recently, studies of the effect of water vapor on reduction of cobalt catalysts were rare. H olmen and co-workers (5, 8) noted that the relative reaction activity on R e-promoted Co/A l2O 3 decreased to one-half with water vapor treatment before reaction. They also noted that the reduction of Co/A l2O 3 in TPR is promoted by wetting a mixture of loose mechanically mixed Co/A l2O 3 and R e/A l2O 3 before reduction. It was suggested that water may enhance the interaction between cobalt and R e. Z ielinski (43, 44) has studied the effect of water vapor on the reducibility of SiO 2- and A l2O 3-supported nickel oxide. H e found that nickel oxide is reduced in the same way as unsupported nickel oxide in the absence of water, and in a similar way to nickel hydrosilicates in the presence of water. TPR peaks were affected by the addition of water vapor. In this paper, we address the impact of water vapor on the reducibility of CoR u/A l2O 3 during reduction. O f particular interest is the interaction between cobalt and the A l2O 3 support resulting from the presence/absence of water vapor. It is well known that the active phase in cobalt Fischer– Tropsch catalysis is cobalt metal. H owever, spinel cobalt aluminate can be formed during pretreatment, reduction, and catalytic reaction. The spinel cobalt aluminate cannot be reduced to cobalt metal under normal reduction conditions. O bviously, spinel cobalt aluminate formation results in fewer surface cobalt metal atoms available for FTS. In this investigation, effects of water vapor during both standard reduction and TPR were studied. Significant changes in the reducibility of cobalt species and in the TPR profiles were observed as the concentration of water vapor was varied. The effect of water vapor in terms of thermodynamics, kinetics, and phase transformations is discussed. EXPERIMENTAL ca. 60 µm. Cobalt nitrate and ruthenium nitrosyl nitrate were dissolved in deionized water and coimpregnated into γ -A l2O 3 to form a catalyst with 20 wt% Co and 0.5 wt% R u on γ -A l2O 3. The catalyst was dried at 110◦ C for 12 h and calcined in air at 300◦ C for 2 h. Standard R eduction in the A bsence/Presence of Water Vapor Standard reduction of the catalyst was performed employing a H 2 flow (100% H 2, 30 ml/min) in an A ltamira A MI-1 system using a temperature ramp from ambient to 350◦ C at 1◦ C/min and holding at 350◦ C for 10 h. The amount of catalyst used was 50 mg for each run. To investigate the effect of water vapor on reduction, water vapor was introduced into the system by a syringe pump or a saturator. When a syringe pump was employed, the water was introduced into a line heated to ca. 250◦ C to guarantee its steady evaporation. D eionized water was used that had been deoxygenated by bubbling A r through it for 2 days. The system lines were heated to ca. 120◦ C to avoid condensation of water in the system. A ll the gases used were 99.999% purity and further purified by oxygen traps and silica gel/5A zeolite to exclude trace oxygen and water. Pulse oxygen titration was conducted at 400◦ C to measure the degree of reduction after the standard reduction. TPR was also performed after standard reduction using ramping from 350 to 900◦ C at 5◦ C/min to investigate the changes in the reducibility of the catalyst. These last two procedures were performed in separate experiments. T PR in the A bsence/Presence of Water Vapor TPR was also carried out in the A ltamira A MI-1 system. It was conducted using 50 mg of calcined catalyst and temperature ramping from 30 to 900◦ C at 5◦ C/min. The carrier gas was 5% H 2 in A r. A cold trap (−70◦ C) was placed before the detector to remove H 2O produced during TPR . The consumption of hydrogen in the TPR process was recorded by a thermal conductivity detector (TCD ). The hydrogen consumption measurement was calibrated by TPR of silver oxide (A g2O ) under the same conditions. For studying the effect of water concentration on TPR , a syringe pump or a water saturator was placed just before the reactor and a U -tube containing silica gel was placed before the cold trap to avoid ice blockage of the flow system. When the syringe pump was used, the water was injected into a line heated to 250◦ C to ensure evaporation. A ll system lines were heated to ca. 120◦ C to avoid condensation of water vapor. Catalyst Preparation The R u-promoted Co/A l2O 3 catalyst was prepared by the incipient wetness technique. The A l2O 3 support used was prepared from Vista B. The support precursor was calcined at 500◦ C for 10 h, producing γ -A l2O 3 having a specific surface area of ca. 200 m 2/g and an average particle size of X -R ay D iffraction A Phillips X’pert System X-ray diffractometer instrument with monochromatized CuK α radiation was used for the XR D measurements. The spectra were scanned at a rate of 2.4◦ /min. The average Co 3O 4 crystallite size of the 283 E FFE CT O F WATE R VA PO R O N R E D U CTIO N O F R u-PRO MOTE D Co/A l2O 3 calcined catalyst was calculated using the Sherrer equation to be 230 Å . A fter reduction in the absence or presence of water vapor, the samples were flushed with A r for 30 min and cooled down to room temperature in an A r stream. Then, they were placed into a container containing liquid paraffin to prevent their oxidation prior to XR D. Chem isorption Chemisorption was performed with a Micromeritics A SA P 2010C system. The standard reduction was carried out for the calcined catalyst in the system. A fter reduction, the catalyst was evacuated at 350◦ C for 90 min to desorb the chemisorbed H 2. The measurement of hydrogen chemisorption was conducted at 100◦ C with an equilibrium interval of 180 s. A fter the first set of hydrogen chemisorption meaurements, the catalyst was soaked in H 2 at 100◦ C for 30 min; then the system was cooled down to room temperature to measure the reversible uptake. The adsorption isotherms were obtained by the increasing pressure method. The pressure range used was 75 to 500 mm H g. E xtrapolation of the total and reversible adsorption isotherms to zero pressure gave the total uptake and reversible uptake for chemisorbed hydrogen. The “irreversible” chemisorption uptake was calculated from the difference between the total and reversible uptakes. Average particle size of the Co metal in the catalysts was calculated by assuming a stoichiometry of H irr /Co s = 1, an average Co atom surface area (in a plane) of 6.62 Å 2, and spherical Co particles with a density of 8.9 g/cm 3. A fter standard reduction, the average Co metal particle size determined by H 2 chemisorption was 218 Å . RESULTS E ffect of Water Vapor on Standard R eduction A fter standard reduction, the amount of Co able to be reduced during standard reduction in the absence or presence of added water vapor was determined by pulse oxidation at 400◦ C (see Table 1). In separate experiments, TPR was carried out to measure the amount of additional Co species able to be reduced up to 900◦ C. TPR profiles after standard reduction in the presence of various water partial pressures are shown in Fig. 1. A s can be seen in Table 1, in the absence of added water vapor, 92% of Co was able to be reduced during standard reduction, with only 3.9% additionally reduced during TPR to 900◦ C. It can be seen in Fig. 1 that the peak around 500◦ C increased with increasing water partial pressure during the standard reduction process. A s reported in the literature (8, 13, 35), the peak around 500◦ C in TPR of noble metal-promoted Co/A l2O 3 catalyst is due to the reduction of small particles of amorphous Co species interacting with the support. The presence of water vapor during standard reduction obviously increases the concentration of these cobalt species. It can be seen from Fig. 2 that the amount of this ca. 500◦ C reducible species increased more quickly at lower water concentrations. The temperature of this TPR peak remained essentially constant (Fig. 3) for all the water concentrations. Introduction of water vapor decreased the degree of reduction of cobalt during standard reduction as well as the total amount able to be reduced up to 900◦ C (Table 1). The degree of reduction decreased from 92 to 45% after introduction of 3% water vapor in the standard reduction process. A pproximately 25% of Co was able TABLE 1 Effect of Water Vapor on the Degree of Reduction during Standard Reductiona % R eduction b,c H 2O added R amping H olding P H 2O (atm × 100) % H 2O H 2O /H 2 ratio P H 2O (atm × 100) H 2O /H 2 ratio D uring standard reduction d 0 0.6 1.2 1.68 3 3 0 0 0.6 1.2 1.68 3 3 0 0 0.006 0.012 0.017 0.031 0.031 0 0 0.6 1.2 1.68 3 0 3 0 0.006 0.012 0.017 0.031 0 0.031 92 67 50 NA 45 77 69 a D uring TPR e (350–900◦ C) after standard reduction Total f (30–900◦ C) % Nonreducible (<900◦ C) 3.9 7.2 18.1 19.7 25.3 NA NA 95.9 84.2 68.1 NA 70.3 NA NA 4.1 15.8 31.9 NA 29.7 NA NA Standard reduction: ramping from ambient temperature to 350◦ C at 1◦ C/min, holding at 350◦ C for 10 h, pure H 2 with added water at 30 cm 3/min flow rate, total pressure of 1 atm. b Based on complete reduction of 20 wt% Co on the catalyst with Co 3O 4 as the calcined cobalt species. c E rror = ± 5% of measurement. d Measured by pulse oxidation at 400◦ C. e Measured by TPR after standard reduction. f Total reduction up to 900◦ C: measured by sum of the results of pulse oxidation and TPR . 284 Z H A NG E T A L. FIG. 1. TPR profiles after standard reduction in the presence of added water vapor. Concentration of water vapor in pure H 2 during standard reduction at 350◦ C: (a) 0% , (b) 0.6% , (c) 1.2% , (d) 1.6% , (e) 3.0% . to be additionally reduced during TPR to 900◦ C, while, ca. 30% remained as nonreducible cobalt species. Introduction of water vapor to the system during both the ramping and the holding periods had a significant influence on the degree of reduction of cobalt. H owever, introduction of water only to either the temperature ramp or the temperature-holding period of the standard reduction procedure produced just a partial decrease in the reducibility. first reported attempt to study the impact of water vapor during TPR of Co catalysts. The results are very interesting. There were two major peaks located at ca. 250 and ≥500◦ C in the TPR profiles (Fig. 4), designated as peaks 1 and 2. It can be clearly seen that the TPR profiles changed dramatically after introduction of a small amount of water vapor, especially for peak 2. Figure 5 shows the peak temperatures of TPR as a function of water concentration. The temperature of peak 1 remained essentially the same, E ffect of Water Vapor on T PR To study further the impact of water vapor on reduction of the cobalt catalyst, water vapor was added during TPR of the calcined catalyst. A s far as we can ascertain, this is the FIG. 2. E ffect of water vapor concentration on degree of reduction during TPR after standard reduction in the presence of water vapor. FIG. 3. E ffect of water vapor concentration during standard reduction on the peak temperature during subsequent TPR . E FFE CT O F WATE R VA PO R O N R E D U CTIO N O F R u-PRO MOTE D Co/A l2O 3 285 FIG. 4. TPR profiles for the calcined catalyst in the presence of various concentrations of water: (a) 0% , (b) 0.6% , (c) 1.2% , (d) 6.9% , (e) 13.8% , (f) 20.7% . while that of peak 2 increased with increasing water partial pressures. The shift was more dramatic at lower water partial pressures. The degree of reduction of cobalt during TPR in the presence of the various water vapor concentrations is shown in Fig. 6. It can be seen that water vapor had little impact on the amount of Co reduced at the lower temperature (peak 1). The total degree of reduction and the degree of reduction during peak 2 decreased with increasing water partial pressure. Figure 7 shows the cumulative degree of reduction at various temperatures as a function of water vapor concentration. Standard reduction leads to a 92% degree of reduction (Table 1). This is achieved during TPR in the absence of added water on reaching ca. 550◦ C. It can be thus supposed that the degree of reduction reached at 550◦ C during TPR is comparable to the degree of reduction during a standard reduction at 350◦ C for 10 h. From Fig. 7 it can be seen that, at 550◦ C, the degrees of reduction at 0.6 and 1.2% water concentrations are 70 and 50% , respectively. This is consistent with the results in Table 1. The results for TPR in the presence of water vapor are summarized in Table 2. FIG. 5. E ffect of water vapor concentration on peak temperatures during TPR of the calcined catalyst. FIG. 6. E ffect of water vapor concentration on degree of reduction during TPR of the calcined catalyst. 286 Z H A NG E T A L. TABLE 2 Degree of Reduction during TPR in the Presence of Watera TPR Catalyst Co 3O 4 CoR u/A l2O 3 H 2O partial pressure (atm × 100) H 2O /H 2 T ( ◦ C) % R eduction b,c T ( ◦ C) % R eduction b,c Total % reduction b,c (30–900◦ C) 0 3.0 0 3.0 0 0.6 300d 333 25 24 372 397 75 76 100 100 0 0.6 1.2 6.9 13.8 20.7 0 0.6 1.2 6.9 13.8 20.7 0 0.12 0.24 1.2 2.55 4.14 255 259 264 283 272 293 39 39 37 32 32 36 483 547 619 659 672 696 61 59 56 47 31 23 100 97 93 90 63 59 % H 2O Peak 1 Peak 2 a Carrier gas: 5% H 2 in A r, T = 30–900◦ C, ramp rate = 5◦ C/min, calcined catalyst. Based on complete reduction of the cobalt with Co 3O 4 as the original cobalt species. c E rror = ± 5% of measurement. d Shoulder. b For comparison, the results for TPR of unsupported Co 3O 4 are also included. The presence of water vapor had little impact on either peak temperature or degree of reduction for TPR of unsupported Co 3O 4. X R D M easurem ent XR D patterns for the CoR u/A l2O 3 catalyst after different treatments are shown in Fig. 8. It can be seen that XR D for calcined CoR u/A l2O 3 was similar to that for unsupported Co 3O 4 with some small contribution from the γ -A l2O 3. A fter completion of peak 1 in TPR (stopping TPR at 400◦ C), diffraction peaks at 31.4◦ and 59.4◦ for Co 3O 4 FIG. 7. Cumulative degree of reduction with temperature for various added water vapor concentrations during TPR of the calcined catalyst. FIG. 8. XR D patterns: (a) γ -A l2O 3 support, (b) unsupported Co 3O 4, (c) calcined catalyst, (d) catalyst after completion of peak 1 for TPR in the absence of added water vapor, (e) catalyst after standard reduction in the absence of added water vapor, (f) catalyst after standard reduction in the presence of 3% added water vapor. E FFE CT O F WATE R VA PO R O N R E D U CTIO N O F R u-PRO MOTE D Co/A l2O 3 287 disappeared. Instead, diffraction lines at 42.4◦ and 61.5◦ for CoO appeared. A fter standard reduction of the catalyst, diffraction peaks for Co 3O 4 disappeared (Figs. 8e and 8f). Compared with the XR D pattern for the partially reduced catalyst during TPR (Fig. 8d), the diffraction lines for CoO decreased greatly. It can be seen from Fig. 8 that the XR D patterns are similar for the catalyst after standard reduction in the absence and presence of added water vapor. H owever, perhaps a little more CoO phase may have been present for the sample reduced in the presence of added water vapor. D iffraction due to cobalt metal could not be resolved because of broadening and overlapping with the diffraction peaks of γ -A l2O 3 and CoO phases. A n XR D pattern for cobalt aluminate spinel was not observed, suggesting that any cobalt aluminate present was in a highly dispersed state. DISCUSSION The origin of the TPR peaks observed for a calcined alumina-supported Co catalyst has been discussed in several papers (8, 13, 35). The lower-temperature peak is usually assigned to reduction of crystalline Co 3O 4 particles. The higher-temperature peak is attributed to the reduction of highly dispersed amorphous cobalt oxide which is interacting strongly with the alumina support. These reduction peaks are shifted to lower temperatures by introducing a small amount of noble metal, such as R u, as a reduction promoter (4, 38–40). The peak assignment, however, is still somewhat controversial due to the complexity of the reduction of cobalt catalysts. To clarify the identity of the reduction peaks in TPR for Co/A l2O 3 catalysts, XR D measurements were conducted with the catalyst after completion of peak 1. It can be seen from Fig. 8d that diffraction peaks for Co 3O 4 disappeared, while diffraction peaks for CoO appeared after completion of peak 1 during TPR . This indicates that Co 3O 4 in the catalyst is reduced to CoO during peak 1 of TPR . Table 2 shows that H 2 consumption for peak 1 of the catalyst during TPR in the absence of added water vapor is much larger than that required for reduction of Co 3O 4 to CoO (equivalent to 25% reduction). It is suggested that reduction of some CoO to Co metal also takes place during peak 1, probably that existing in the large particles of cobalt. Thermodynamic calculations were carried out using A spen to explain the effect of water vapor on the reducibility of the cobalt catalyst. The calculated results are shown in Fig. 9. It can be seen that the degree of reduction decreases with increasing H 2O /H 2 ratio and temperature. H owever, thermodynamically, the effect of water vapor on the reduction of Co 3O 4 is very small. The near-total reduction of bulk Co 3O 4 in H 2 is essentially feasible even in the presence of large amounts of water vapor. The results of TPR FIG. 9. Thermodynamic limitations to reduction of Co 3O 4 as a function of H 2O /H 2 ratio at various temperatures: (a) 200◦ C, (b) 400◦ C, (c) 600◦ C, (d) 800◦ C. for unsupported Co 3O 4 are consistent, as expected, with the thermodynamic predictions. U sually, the particle size distribution of high-metalloading catalysts are bimodal rather than single modal. This 20 wt% Co catalyst was no exception, possessing both very large and small particles as evidenced by SE M (TE M is not useful for such a high loading of metal) and the results from H 2 chemisorption and XR D giving average Co 3O 4 (calcined catalyst) and Co (reduced catalyst) particle sizes. The reduction of a fraction of the cobalt present as large particles of Co 3O 4 on the CoR u/A l2O 3 catalyst during the reduction procedure used can be concluded, based on the reduction results and their low interaction with the A l2O 3 support, to not have been significantly affected by thermodynamic limitations even with the added water vapor. This fraction of the cobalt probably reduced during peak 1 of TPR of the calcined catalyst and during standard reduction with or without added water. The results obtained (Table 1), however, show that the presence of water vapor during standard reduction did have significant effects on the degree of reduction of the cobalt catalyst and on the subsequent reducibility of the standard reduced catalyst during TPR up to 900◦ C. The first effect relates to the formation of cobalt species able to be reduced during the following TPR (350–900◦ C). It is evident from Fig. 1 that the temperature of the reduction peak of this species (ca. 480–510◦ C) did not change significantly for different water vapor concentrations. Comparing this temperature with the TPR profile in the absence of water vapor (Fig. 4), it can be observed that the peak temperature in Fig. 1 (TPR after standard reduction) is similar to the temperature of peak 2 for TPR of the calcined catalyst in 288 Z H A NG E T A L. the absence of water vapor in Fig. 4. This suggests that the reducible cobalt species in the successive TPR after standard reduction in the presence of water vapor is similar to that reduced during peak 2 in TPR of the calcined catalyst in the absence of water vapor. The amount of this cobalt species remaining after standard reduction in the presence of water vapor increases with increasing water vapor concentrations (Fig. 2). It is suggested that the presence of water vapor has a strong inhibitive effect on the reduction of this species during standard reduction. A nother effect of the presence of water vapor during standard reduction is the formation of species that are nonreducible during the following TPR from 350 to 900◦ C. This nonreducible (<900◦ C) cobalt is concluded to be in the form of a cobalt aluminate. It can be seen from Table 1 that the amount of nonreducible cobalt species (cobalt aluminate) increases greatly on the addition of small concentrations of water vapor during standard reduction. The decrease in degree of reduction during standard reduction in the presence of water vapor could be due to kinetic and/or thermodynamic limitations on the reduction of the highly dispersed cobalt species. These Co species (Co 3O 4, CoO, etc.) can have a strong interaction with the γ -A l2O 3 support. From thermodynamic calculations discussed above, the thermodynamic limitations on the reduction of Co 3O 4 (in the absence of alumina) are essentially negligible. H owever, the reduction of well-dispersed cobalt species interacting with the support is a different case. Thermodynamic calculations for such a system are difficult because the Co species are not well defined and sufficient thermodynamic data are not available. Thus, a basis does not exist for concluding whether thermodynamic or kinetic limitations apply. H owever, we strongly hypothesize that the presence of water vapor during standard reduction enhances the equilibration of the interaction of Co with alumina. U sually, TPR is performed using a small H 2 concentration (ca. 5% ) mixed with argon or nitrogen (41). D uring TPR , water is formed and is purged by the carrier gas. The concentration of water in the reactor during TPR is, thus, very low, and the reduction proceeds under nonequilibrium conditions. Various parameters (H 2 concentration, amount of catalyst, ramping rate, flow rate, etc.) for TPR have been studied (41, 42). H owever, studies of the effect of H 2O, which is the only product of TPR , on TPR are few. Z ielinski (43, 44) has reported the effect of water vapor on the TPR of silica- and alumina-supported nickel catalysts. The effect of water vapor during TPR of supported cobalt catalysts has not been previously reported. In our study, it was found that water vapor has a significant influence on the TPR profiles (Fig. 4). The shoulder of peak 1 became a distinctly separate peak as the main part of peak 1 shifted slightly to higher temperatures with the addition of water. This phenomenon can also be observed in the TPR of unsupported Co 3O 4 as FIG. 10. E ffect of water vapor on TPR profiles for unsupported Co 3O 4: (a) 0% , (b) 3% . shown in Fig. 10. Thus, the effect of water vapor on peak 1 of TPR on CoR u/A l2O 3 is comparable to that on unsupported Co 3O 4. Since the ratio of the areas of these two peaks formed by introducing water vapor into the system in the TPR of unsupported Co 3O 4 approaches 1/3 (the theoretical ratio of H 2 consumption for reduction of Co 3 + to Co 2 + and Co 2 + to Co metal), this pair of peaks can be ascribed to the reduction of Co 3 + to Co 2 + and Co 2 + to Co metal. The similar effect during water addition on peak 1 for TPR of CoR u/A l2O 3 and for the TPR of unsupported Co 3O 4 suggests that this Co species is similar to unsupported Co 3O 4. The reduction of ruthenium oxide cannot be seen in the TPR spectra because of its low relative loading (0.5 wt% R u) compared with that of Co (20 wt% ). XR D results in Fig. 8 also suggest that partial reduction of Co 3O 4 to CoO takes place in peak 1. The dramatic influence of water vapor presence on peak 2 of TPR for CoR u/A l2O 3 indicates that the reduction properties of the cobalt species in peak 2 are different from those of Co 3O 4. It, therefore, must be some type of cobalt species interacting with the alumina support. The effect of water vapor on the position of peak 2 of TPR for calcined R uCo/A l2O 3 (Fig. 4) probably has two mechanisms. A t lower water vapor concentrations, there is a dramatic shift in the peak temperature. This is possibly due to the change or modification of the cobalt species caused by presence of water vapor. O nly a slight additional shift was observed at higher water vapor concentrations. A computer simulation of TPR based on the model proposed by Malet and Caballero (42) suggests that a 10–15◦ C shift in peak temperature could have resulted due to the 20% variation in H 2 concentration and total flow rate with the addition of water vapor. Thus, the slight shifts observed for peaks 1 and 2 at higher water vapor pressures (Figs. 4, 5) as well as the shift for unsupported Co 3O 4 were probably E FFE CT O F WATE R VA PO R O N R E D U CTIO N O F R u-PRO MOTE D Co/A l2O 3 due only to the small changes in H 2 concentration and total flow rate. It can be noted that the peaks became narrower in the presence of water, and the onset of peak 2 shifted to a higher temperature. The onset temperature represents the initial formation of crystal nuclei of cobalt metal from cobalt species represented by the peak. Therefore, it can be postulated that the presence of water vapor during reduction may have affected the kinetics of crystal nucleation of small cobalt metal particles. It can be seen from Table 2 that the total degree of reduction up to 900◦ C decreased greatly with increasing water vapor concentration. It is suggested that water facilitates interactions between cobalt species and the γ -A l2O 3 support to form irreducible cobalt aluminate. It was observed that the catalyst turned dark blue after TPR in the presence of higher water vapor pressures, while it remained black in the absence of added water vapor. The former result is suggestive of cobalt aluminate formation. A similar effect of water vapor on SiO 2-supported nickel catalysts has been observed by Z eilinski (43). H e suggested that water vapor and the silica support jointly interact with nickel oxide, probably to form hydrosilicates, and thus retard the reduction of Ni. Interaction of cobalt and alumina has been observed by many authors (4–13) using various techniques including TPR , XR D, E XA FS, and XPS (E SCA ). The migration of cobalt ions into the alumina lattice sites of octahedral or tetrahedral symmetry is limited to the first few layers of the support under normal calcination conditions (6). The γ -A l2O 3 crystal structure is that of a spinel with a deficit of cations (36). D iffusion of cobalt ions into tetrahedral sites of γ -A l2O 3 can form a “surface spinel” in Co/A l2O 3 catalysts. The “surface spinel” structure cannot be observed by X-ray diffraction because it does not have long-range, three-dimensional order (6, 37). It has been suggested that cobalt ions occupying surface octahedral sites of γ -A l2O 3 are reducible, while cobalt ions occupying tetrahedral sites are nonreducible (6). A t lower calcination temperatures, filling of the octahedral sites is more favorable. Filling of the tetrahedral sites of γ -A l2O 3 may be enhanced by an increase in calcination temperature (6). It is suggested, based on this study, that this process can also be enhanced by the presence of water vapor. Water vapor may partially hydrate the cobalt oxides and γ -A l2O 3 support and, thus, facilitate the migration of cobalt ions into tetrahedral sites of γ -A l2O 3. CONCLUSIONS It can be concluded that water vapor has a significant effect on the reduction behavior of CoR u/A l2O 3 catalyst. A dded water vapor had little effect on unsupported Co 3O 4 and on peak 1 in TPR of calcined CoR u/A l2O 3, which is assigned to the reduction of large Co 3O 4 particles (similar to bulk Co 3O 4) to Co metal and partial reduction of highly 289 dispersed Co 3O 4 to CoO. H owever, it had a dramatic influence on peak 2 in TPR of CoR u/A l2O 3, which is assigned to the reduction of Co species that were well dispersed and strongly interacting with the support. A t lower water vapor pressures, the shift of peak 2 was probably due to a change or modification of highly dispersed cobalt species in the presence of water vapor. The slight shifts of peak 1 and peak 2 (at higher water vapor pressures) for CoR u/A l2O 3 as well as the reduction peaks for unsupported Co 3O 4 were probably caused only by the small variations in H 2 concentration and total flow rate. Introduction of water vapor during standard reduction led to a decrease in the degree of reduction of cobalt probably in two ways: (1) inhibition of the reduction of welldispersed CoO interacting with the alumina support possibly by increasing the Co–alumina interaction; and (2) facilitation of migration of Co ions into probable tetrahedral sites of γ -A l2O 3 to form a nonreducible (≤900◦ C) spinel. Such an irreversible spinel formation results in a decrease in the amount of reduced cobalt metal atoms available to catalyze reactions. ACKNOWLEDGMENT The authors thank E nergy International (a division of Williams International) for financial support of this work. REFERENCES 1. Iglesia, E ., A ppl. Catal. A 161, 59 (1997). 2. Brady, R . C., and Pettit, R . J., J. A m . Chem . Soc. 103, 1287 (1981). 3. A nderson, R . B., “The Fischer-Tropsch Synthesis.” A cademic Press, O rlando, FL, 1984. 4. Kogelbauer, A ., G oodwin, J. G., Jr., and O ukaci, R ., J. Catal. 160, 125 (1996). 5. Schanke, D., H ilmen, A . M., Bergene, E ., Kinnari, K., R ytter, E ., A dnanes, E ., and H olmen, A ., Catal. L ett. 34, 269 (1995). 6. Chin, R . L., and H ercules, D. M., J. Phys. Chem . 86, 360 (1982). 7. Belambe, A . R ., O ukaci, R ., and G oodwin, J. G., Jr., J. Catal. 166, 8 (1997). 8. H ilmen, A . M., Schanke, D., and H olmen, A ., Catal. L ett. 38, 143 (1996). 9. van de Loosdrecht, J., van der H aar, M., van der Kraan, A . M., van D illen, A . J., and G eus, J. W., A ppl. Catal. A 150, 365 (1997). 10. Schanke, D., H ilmen, A . M., Bergene, E ., Kinnari, K., R ytter, E ., A dnanes, E ., and H olmen, A ., E nergy Fuels 10, 867 (1996). 11. R euel, R . C., and Bartholomew, C. H ., J. Catal. 85, 63 (1984). 12. R euel, R . C., and Bartholomew, C. H ., J. Catal. 85, 78 (1984). 13. A rnoldy, P., and Moulijn, J. A ., J. Catal. 93, 38 (1985). 14. Iglesia, E ., Soled, S. L., Fiato, R . A ., and Via, G. H ., J. Catal. 143, 345 (1993). 15. H o, S. W., H oualla, M., and H ercules, D. M., J. Phys. Chem . 94, 6396 (1990). 16. Kogelbauer, A ., Weber, J. C., and G oodwin, J. G., Jr., Catal. L ett. 34, 259 (1995). 17. A li, S., Chen, B., and G oodwin, J. G., Jr., J. Catal. 157, 35 (1995). 18. Ming, H ., and Baker, B. G., A ppl. Catal. A 123, 23 (1995). 19. Buchman, L. B., R autiainen, A ., Krause, A . O. I., and Lindblad, M., Catal. Today 43, 11 (1998). 290 Z H A NG E T A L. 20. van Steen, E ., Sewell, G. S., Makhothe, R . A ., Micklethwaite, C., Manstein, H ., de Lange, M., and O ’Connor, C. T., J. Catal. 162, 220 (1996). 21. van’t Blik, H . F. J., Koningsberger, D. C., and Prins, R ., J. Catal. 97, 210 (1986). 22. Sellmer, C., D echer, S., and Kruse, N., Catal. L ett. 52, 131 (1998). 23. E rnst, B., Bensaddik, A ., H ilaire, L., Chaumette, P., and Kiennemann, A ., Catal. Today 39, 329 (1998). 24. Jablonski, J. M., Wolcyrz, M., and Krajczyk, L., J. Catal. 173, 130 (1998). 25. H o, S. W., and Su, Y. S., J. Catal. 168, 51 (1997). 26. Khodakov, A . Y., Lynch, J., Bazin, D., R ebours, B., Z anier, N., Moisson, B., and Chanmette, P., J. Catal. 168, 16 (1997). 27. Martens, J. H . A ., van’t Blik, H . F. J., and Prins, R ., J. Catal. 97, 200 (1986). 28. Keyser, M. J., E verson, R . C., and E spinoza, R . L., A ppl. Catal. A 171, 99 (1998). 29. Noronha, F. B., Schmal, M., Nicot, C., Moraweck, B., and Frety, R ., J. Catal. 168, 42 (1997). 30. Bessel, S., A ppl. Catal. A 126, 235 (1995). 31. Bartholomew, C. H ., Stud. Surf. Sci. Catal. 64, 158 (1991). 32. H addad, G. J., and G oodwin, J. G., Jr., J. Catal. 157, 25 (1995). 33. R othaemel, M., H anssen, K. F., Blekkan, E . A ., Schanke, D., and H olmen, A ., Catal. Today 38, 79 (1997). 34. R othaemel, M., H anssen, K. F., Blekkan, E . A ., Schanke, D., and H olmen, A ., Catal. Today 40, 171 (1998). 35. van’t Blik, H . F. J., and Prins, R ., J. Catal. 97, 188 (1986). 36. R ichardson, J. T., and Vernon, L. W., J. Phys. Chem . 62, 153 (1958). 37. Cimino, A ., Lojcono, M., and Schiavello, M., J. Phys. Chem . 79, 243 (1975). 38. G oodwin, J. G., Jr., Prepr. ACS D iv. Petr. Chem . 36, No. 1, 156 (1991). 39. Schanke, D., Vada, S., Blekkan, E . A ., H ilmen, A . M., H off, A ., and H olmen, A ., J. Catal. 156, 85 (1995). 40. Vada, S., H off, A ., A dnanes, E ., Schanke, D., and H olmen, A ., Top. Catal. 2, 155 (1995). 41. H urst, N. W., G entry, S. J., Jones, A ., and McNicol, B. D., Catal. R ev. Sci. E ng. 24, 233 (1982). 42. Malet, P., and Caballero, A ., Chem . Soc. Faraday Trans. 1 84, 2369 (1988). 43. Z ielinski, J., Catal. L ett. 32, 47 (1995). 44. Z ielinski, J., Catal. L ett. 12, 389 (1992).