pubs.acs.org/Langmuir © 2010 American Chemical Society
Understanding the Effect of Cobalt Particle Size on Fischer-Tropsch Synthesis: Surface Species and Mechanistic Studies by SSITKA and Kinetic Isotope Effect† Jia Yang, Erik Z. Tveten, De Chen, and Anders Holmen* Department of Chemical Engineering, Norwegian University of Science & Technology, NO-7491 Trondheim, Norway Received April 19, 2010. Revised Manuscript Received June 23, 2010 Co/γ-Al2O3 catalysts with particle sizes in the range of 4-15 nm were investigated by isothermal hydrogenation (IH), temperature programmed hydrogenation (TPH), and steady-state isotopic transient kinetic analysis (SSITKA). Kinetic isotope effect experiments were used to probe possible mechanisms on Co/γ-Al2O3 with different particle size. It was found that CO dissociated on Co/γ-Al2O3 catalysts at 210 C. The total amount of CO2 formed following the dissociation depends on the cobalt crystal size. O-Co binding energy was found to be highly dependent on the Co metal particle size, whereas similar C-Co binding energy was found on catalysts with different Co particle size. Very strongly bonded carbon and oxygen surface species increased with decreasing particle size and acted as site blocking species in the methanation reaction. SSITKA experiments showed that the intrinsic activity (1/τCHx) remained constant as the particle size increased from 4 to 15 nm. The number of surface intermediates (NCHx) increased with increasing particle size. The apparent activation energies were found similar for these catalysts, about 85 kJ/mol. D2-H2 switches further confirmed that the particle size did not change the kinetically relevant steps in the reaction. The reactivity of the active sites on the 4 nm particles was the same as those on the 8, 11, and 15 nm particles, and only the number of total available surface active sites was less on the 4 nm particles than on the others.
Introduction The Fischer-Tropsch synthesis (FTS) is a heterogeneous catalytic process to produce high quality liquid fuels and chemicals from synthesis gas (CO þ H2) which can be derived from natural gas, coal, or biomass.1 Cobalt-based catalysts are preferred especially for converting natural gas derived synthesis gas owing to their low water-gas shift activity, high selectivity to long-chain hydrocarbons (i.e., C5þ hydrocarbons), and remarkable stability in the FTS processes.2 The performance of the catalyst depends on properties such as metal loading, particle size, and support.2-4 Since metallic Co0 are the active sites in FTS, much research has been devoted to improve the metal dispersion by changing the †
Part of the Molecular Surface Chemistry and Its Applications special issue. *To whom correspondence should be addressed. E-mail: holmen@ chemeng.ntnu.no. (1) Dry, M. E. The Fischer-Tropsch process - commercial aspects. Catal. Today 1990, 6 (3), 183-206. (2) Iglesia, E. Design, synthesis, and use of cobalt-based Fischer-Tropsch synthesis catalysts. Appl. Catal., A 1997, 161 (1-2), 59-78. (3) Khodakov, A. Y.; Chu, W.; Fongarland, P. Advances in the Development of Novel Cobalt Fischer-Tropsch Catalysts for Synthesis of Long-Chain Hydrocarbons and Clean Fuels. Chem. Rev. 2007, 107 (5), 1692-1744. (4) Storsæter, S.; Borg, Ø.; Blekkan, E. A.; Holmen, A. Study of the effect of water on Fischer-Tropsch synthesis catalysts. J. Catal. 2005, 231, 405-419. (5) Bezemer, G. L.; Radstake, P. B.; Koot, V.; van Dillen, A. J.; Geus, J. W.; de Jong, K. P. Preparation of Fischer-Tropsch cobalt catalysts supported on carbon nanofibers and silica using homogeneous deposition-precipitation. J. Catal. 2006, 237 (2), 291-302. (6) Okabe, K.; Li, X.; Wei, M.; Arakawa, H. Fischer-Tropsch synthesis over Co-SiO2 catalysts prepared by the sol-gel method. Catal. Today 2004, 89 (4), 431-438. (7) Martı´ nez, A.; Lopez, C.; Marquez, F.; Dı´ az, I. Fischer-Tropsch synthesis of hydrocarbons over mesoporous Co/SBA-15 catalysts: the influence of metal loading, cobalt precursor, and promoters. J. Catal. 2003, 220 (2), 486-499. (8) Concepcion, P.; Lopez, C.; Martı´ nez, A.; Puntes, V. F. Characterization and catalytic properties of cobalt supported on delaminated ITQ-6 and ITQ-2 zeolites for the Fischer-Tropsch synthesis reaction. J. Catal. 2004, 228 (2), 321-332.
16558 DOI: 10.1021/la101555u
preparation method5,6 and the support materials,7-9 and optimizing the activation processes.10 However, decreased catalytic activity accompanied by decreased selectivity toward higher hydrocarbons is found on highly dispersed Co catalysts. Some studies have attributed this phenomenon to the poor reducibility encountered with very small Co nanoparticles.11,12 Other studies claim that a lower intrinsic catalytic activity (reaction rate per surface Co0 site or TOF) could be responsible in highly dispersed catalysts.13-18 The absence of a clear correlation between cobalt particle size and intrinsic activity can also be found.19,20 Thus, the true influence of the cobalt particle size in FTS is still under debate. It has been widely accepted that the intrinsic (9) Martı´ nez, A.; Prieto, G.; Rollan, J. Nanofibrous γ-Al2O3 as support for Cobased Fischer-Tropsch catalysts: Pondering the relevance of diffusional and dispersion effects on catalytic performance. J. Catal. 2009, 263 (2), 292-305. (10) Sietsma, J. R. A.; Meeldijk, J. D.; Den Breejen, J. P.; Versluijs-Helder, M.; Van Dillen, A. J.; De Jongh, P. E.; De Jong, K. P. The preparation of supported NiO and Co3O4 nanoparticles by the nitric oxide controlled thermal decomposition of nitrates. Angew. Chem., Int. Ed. 2007, 46 (24), 4547-4549. (11) Khodakov, A. Y.; Lynch, J.; Bazin, D.; Rebours, B.; Zanier, N.; Moisson, B.; Chaumette, P. Reducibility of cobalt species in silica-supported FischerTropsch catalysts. J. Catal. 1997, 168 (1), 16-25. (12) Khodakov, A. Y.; Griboval-Constant, A.; Bechara, R.; Zholobenko, V. L. Pore size effects in Fischer-Tropsch synthesis over cobalt-supported mesoporous silicas. J. Catal. 2002, 206 (2), 230-241. (13) Reuel, R. C.; Bartholomew, C. H. Effects of support and dispersion on the CO hydrogenation activity/selectivity properties of cobalt. J. Catal. 1984, 85 (1), 78-88. (14) Barbier, A.; Tuel, A.; Arcon, I.; Kodre, A.; Martin, G. A. Characterization and Catalytic Behavior of Co/SiO2 Catalysts: Influence of Dispersion in the Fischer-Tropsch Reaction. J. Catal. 2001, 200 (1), 106-116. (15) Bezemer, G. L.; Bitter, J. H.; Kuipers, H. P. C. E.; Oosterbeek, H.; Holewijn, J. E.; Xu, X.; Kapteijn, F.; Van Diilen, A. J.; De Jong, K. P. Cobalt particle size effects in the Fischer-Tropsch reaction studied with carbon nanofiber supported catalysts. J. Am. Chem. Soc. 2006, 128 (12), 3956-3964. (16) Prieto, G.; Martı´ nez, A.; Concepcion, P.; Moreno-Tost, R. Cobalt particle size effects in Fischer-Tropsch synthesis: structural and in situ spectroscopic characterisation on reverse micelle-synthesised Co/ITQ-2 model catalysts. J. Catal. 2009, 266 (1), 129-144.
Published on Web 07/12/2010
Langmuir 2010, 26(21), 16558–16567
Yang et al.
Article Table 1. Preparation and Characterization of the 20 wt % Co/γ-Al2O3 Catalysts
catalysta
H2O/EG mass ratio
dispersion (%)
degree of reduction (%)
corrected particle size (nm)
BET surface area (m2/g)
pore diameter (nm)
pore volume (cm3/g)
MP_4 0.80 10.3 45 4.2 158 10.0 0.52 MP_8 0.93 7.0 59 8.1 136 11.5 0.48 MP_11 0.96 6.2 73 11.2 140 12.9 0.51 LP_15 0.96 5.1 81 15.1 99 24.9 0.6 a MP and LP refer to medium pore alumina (Sasol, Puralox SCCa series) and large pore alumina (Puralox TH high pore volume series), respectively. The numbers 4, 8, 11, and 15 are the cobalt metal crystal size.
reaction rate does not change with the particle size or the support, provided that the particle size is larger than 10 nm for supported Co catalysts.15,20,21 Recently, a study on completely reducible Co/ CNFs catalysts reported that the turnover frequency (TOF) increases with increasing cobalt particle size up to an optimum size of 6 nm (1 bar) or 8 nm (35 bar) and remains constant for larger particles.15 The same trend has been reported for monodispersed Co supported on zeolites with an optimum particle size of 10 nm.16 In addition, cobalt surface reconstruction occurred irrespective of the metal particle size, which is probably induced by adsorbed C adatoms (surface carbide species), derived from CO dissociation.16 Due to the dynamic nature of the cobalt surface in FTS environment, in situ methods are required for a better understanding of the particle size effect. Steady-state isotopic transient kinetic analysis (SSITKA) has been widely used to characterize surface parameters of heterogeneous catalysts.22 SSITKA can be used to determine the mean surface residence time and the abundance of surface intermediates. Most importantly, this method can decouple the contributions from the intrinsic rate constant (k) and the surface coverage (θ) to the TOF.22 Moreover, the relevance of SSITKA carried out at methanation conditions (1.85 bar; H2/ CO = 10) and normal FT conditions (1 bar and 35 bar; H2/CO = 2) has been proven.17,23 The study of Co/CNF catalysts with particle size ranging from 2.6 to 16 nm showed that a significant amount of irreversibly bonded CO molecules are present on small particles, blocking part of the Co surface.17 The irreversibly adsorbed CO is claimed to be associatively chemisorbed. Recently, theoretical calculation demonstrates that CO dissociation and CO*, C*, or O* hydrogenation have much lower activation energy on step sites than on terraces, and the reaction rate on the terraces is an order of magnitude less than that on the step sites and hence cannot compete with them.24,25 A supported (17) den Breejen, J. P.; Radstake, P. B.; Bezemer, G. L.; Bitter, J. H.; Frøseth, V.; Holmen, A.; Jong, K. P. d. On the Origin of the Cobalt Particle Size Effects in Fischer-Tropsch Catalysis. J. Am. Chem. Soc. 2009, 131 (20), 7197-7203. (18) Herranz, T.; Deng, X.; Cabot, A.; Guo, J.; Salmeron, M. Influence of the Cobalt Particle Size in the CO Hydrogenation Reaction Studied by In Situ X-ray Absorption Spectroscopy. J. Phys. Chem. B 2009, 113 (31), 10721-10727. (19) Borg, Ø.; Dietzel, P. D. C.; Spjelkavik, A. I.; Tveten, E. Z.; Walmsley, J. C.; Diplas, S.; Eri, S.; Holmen, A.; Rytter, E. Fischer-Tropsch synthesis: Cobalt particle size and support effects on intrinsic activity and product distribution. J. Catal. 2008, 259 (2), 161-164. (20) Iglesia, E.; Soled, S. L.; Fiato, R. A. Fischer-Tropsch synthesis on cobalt and ruthenium. Metal dispersion and support effects on reaction rate and selectivity. J. Catal. 1992, 137 (1), 212-224. (21) Schanke, D.; Vada, S.; Blekkan, E. A.; Hilmen, A. M.; Hoff, A.; Holmen, A. Study of Pt-promoted cobalt CO hydrogenation catalysts. J. Catal. 1995, 156 (1), 85-95. (22) Shannon, S. L.; Goodwin, J. G. Characterization of Catalytic Surfaces by Isotopic-Transient Kinetics during Steady-State Reaction. Chem. Rev. 1995, 95 (3), 677-695. (23) Bertole, C. J.; Kiss, G.; Mims, C. A. The effect of surface-active carbon on hydrocarbon selectivity in the cobalt-catalyzed Fischer-Tropsch synthesis. J. Catal. 2004, 223 (2), 309-318. (24) Ge, Q.; Neurock, M. Adsorption and Activation of CO over Flat and Stepped Co Surfaces: A First Principles Analysis. J. Phys. Chem. B 2006, 110 (31), 15368-15380. (25) Gong, X.-Q.; Raval, R.; Hu, P. CO dissociation and O removal on Co(0 0 0 1): a density functional theory study. Surf. Sci. 2004, 562 (1-3), 247-256.
Langmuir 2010, 26(21), 16558–16567
catalyst has normally a heterogeneous surface, including terraces, steps, and kinks. The fraction of these surface sites is determined by the particle size and the shape. Normally, the smaller the particle is, the higher the concentration of steps and kinks. However, studies related to the nature of active sites for catalysts with different particle size and the dependence of rate determining steps (RDS) on particle size are limited. In the present work, isothermal hydrogenation (IH) and temperature programmed hydrogenation (TPH) are applied to distinguish different types of surface species and their distributions for cobalt catalysts at working conditions. SSITKA reveals the intrinsic activity and the number of active intermediates in situ. In order to study the dependence of RDS on Co particle size, D2-H2 exchange experiments are performed.
Experimental Section Catalyst Preparation and Characterization. Co catalysts (20 wt %) were prepared by incipient wetness impregnation of two types of γ-alumina supports (Sasol, Puralox SCCa series and Puralox TH high pore volume series, shown in Table 1) with solutions of Co (NO3)2 3 6H2O. Different amounts of ethylene glycol were added in order to get the desired cobalt particle size. After impregnation, 28 g of catalysts was dried in air at 100 C for 4 h and then calcined in air (100 mL/min) at 300 C for 16 h with a ramping rate of 2 C/min. Oxygen titration was performed on a Micromeritics AutoChem II 2920 unit. The detailed procedure has been described elsewhere.26 The degree of reduction (DOR) was calculated assuming that all cobalt metal was oxidized to Co3O4. Hydrogen chemisorption was performed in a Micromeritics 2020 unit at 40 C, although recent studies have shown that a few degrees higher temperature could have been used.27 Since we have previously carried out the H2 chemisorption measurements at 40 C, it was decided to use the same conditions also for these experiments.26 The cobalt dispersion was calculated by assuming dissociative adsorption of hydrogen on the cobalt metal surface. The cobalt particle size (d(Co)) was calculated based on the following equation: d(Co) (nm) = (96/D (%)) DOR, where D is the dispersion. Transient Experiments. A total of 200 mg of catalyst (53-90 μm) was diluted with 400 mg of SiC (75-150 μm) and loaded in a quartz microreactor (4 mm i.d.). Reduction was performed at 350 C (ramping rate of 1 C/min from ambient temperature) under H2 flow at 10 mL/min for 15 h. After reduction, the reactor was purged with Ar (33.5 mL/min) at 350 C for 30 min before cooling to 210 C under Ar flow. The setup has been described in detail elsewhere.28 The fresh reduced catalysts were exposed to C18O by a step switch from Ar (33.5 mL/min) to Kr/C18O (33.5/1.5 mL/min) mixture at 210 C, 1.85 bar. When steady state was reached, the (26) Borg, Ø.; Eri, S.; Blekkan, E. A.; Storsæter, S.; Wigum, H.; Rytter, E.; Holmen, A. Fischer-Tropsch synthesis over γ-alumina-supported cobalt catalysts: Effect of support variables. J. Catal. 2007, 248 (1), 89-100. (27) Xiong, J.; Borg, Ø.; Blekkan, E. A.; Holmen, A. Hydrogen chemisorption on rhenium-promoted γ-alumina supported cobalt catalysts. Catal. Commun. 2008, 9 (14), 2327-2330. (28) Frøseth, V.; Storsæter, S.; Borg, Ø.; Blekkan, E. A.; Rønning, M.; Holmen, A. Steady state isotopic transient kinetic analysis (SSITKA) of CO hydrogenation on different Co catalysts. Appl. Catal., A 2005, 289 (1), 10-15.
DOI: 10.1021/la101555u
16559
Article
Yang et al.
reactor was purged with Kr (33.5 mL/min) for 20 min. Isothermal hydrogenation was then performed at 210 C, 1.85 bar under a H2 flow of 15 mL/min. Subsequently, TPH was conducted by heating the catalyst from 210 to 700 C at a rate of 10 C/min under a pure hydrogen flow of 15 mL/min at 1.85 bar. A comparative experiment was performed in the same way except that the methanation reaction followed by Ar purge was introduced right after the Kr purge and before the IH and the TPH experiments. All the transient responses were monitored with a Balzer QMG 422 quadrupole mass spectrometer (MS).
Steady-State Isotopic Transient Kinetic Analysis (SSITKA). SSITKA experiments were performed using a fixed-bed quartz reactor (4 mm i.d.). The catalyst (100 mg; 53-90 μm) was diluted with 200 mg of inert silicon carbide (75-150 μm) to improve the isothermal conditions along the catalyst bed. The catalyst was reduced in situ in 10 mL/min H2 with a ramping rate of 1 C/min to 350 C. After 16 h reduction, the catalyst was cooled down to 170 C. The temperature was then increased at a ramping rate of 1 C/min to 190 C. At 190 C, syngas H2/CO/Ar (15/1.5/ 33.5 mL/min) was introduced and pressure adjusted to 1.85 bar. The system was further heated to 210 C at a rate of 1 C/min. The concentrations of H2, CO, Ar, and C1-C7 hydrocarbons were analyzed with an online gas chromatograph (HP5890) equipped with TCD and FID. The isotopic switch (Ar/12CO/H2 to Kr/ 13 CO/H2) was performed after reaching steady state (5 h on stream) and monitored using the Balzers QMG 422 quadropole mass spectrometer. Kinetic Isotope Effect Measurements. H2/D2 isotope effects were measured using H2/CO/Ar (15/1.5/33.5 mL/min) and D2/ CO/Ar (15/1.5/33.4 mL/min) reactant mixtures. The following gases have been used: H2 (YARA, 99.999%), D2 (Cambridge Isotope Laboratories, 99.6%), CO (YARA, 99.97%), and Ar (YARA, 99.999%). The reactant gases were introduced separately and mixed before entering the reactor. The measurements were conducted after the SSITKA experiments using the same catalysts. After measurement with D2/CO/Ar as the reactant, the feed gas was changed back to H2/CO/Ar to check whether there were any irreversible changes to the catalyst.
Results and Discussion Co/γ-Al2O3 catalysts with particle size ranging from 4 to 15 nm were prepared by incipient wetness impregnation according to a previously described method.19 The advantage of this method is that catalysts with different metal particle size can be obtained by only adjusting the ratio between ethylene glycol (EG) and water while keeping the cobalt loading constant. The physical properties of the final catalysts are displayed in Table 1. The mean diameter of the Co particle size on the catalyst is calculated based on the dispersion, corrected for degree of reduction. It is observed that the degree of reduction is lower for the catalyst with smaller Co particles. Reactivity and Stability of Surface Carbon and Oxygen Species. The labeled C18O was used to study CO dissociation and distribution of surface species on the cobalt catalysts with different particle size. The reduced catalyst was first exposed to C18O, and the resulting deposited surface species were then characterized by IH and TPH. The labeled CO (C18O) was used to identify the source of the oxygen on the catalyst: either from the gas phase CO or from lattice oxygen of unreduced Co oxides. Significant amounts of CO2 were produced upon CO exposure at 210 C for all the catalysts tested as shown in Figure 1, and the total amount of CO2 increased with increasing cobalt particle size of the catalyst. All three isotopologues of CO2, C16O2, C16O18O, and C18O2 were indentified by the MS. The quantity of these corresponding CO2 isotopologues were in the order of C16O2 > C16O18O>C18O2. The source of 16O could possibly be from either 16560 DOI: 10.1021/la101555u
Figure 1. Type and amount of CO2 formed during C18O exposure to fresh reduced cobalt catalysts at 210 C, 1.85 bar. C16O2, 44; C16O18O, 46; C18O2, 48.
the unreduced cobalt oxides or the alumina support. A reference experiment was performed by passing C18O through the γ-Al2O3 support at the same operating condition, and no trace of C16O was observed, implying that no isotopic oxygen exchange occurred between C18O and the surface oxygen on the Al2O3 support. However, it has been reported that there is a fast oxygen exchange between CO2 and the surface oxygen on alumina support through the formation of carbonate,29,30 which may be responsible for the formation of C16O18O and C18O2. Due to the similar chemical properties of 16O and 18O, the oxygen exchange between CO2 and alumina will not change the total amount of CO2 formed on the Co surfaces. Figure 1 clearly shows that the catalyst with larger Co particle size facilitated the formation of more CO2, which is probably due to CO dissociation and the subsequent recombination of adsorbed CO and surface oxygen from CO dissociation. After C18O exposure at 210 C, the catalyst surface was flushed with Ar to remove weakly bonded CO* on the Co surface. Surface deposited species were characterized by IH followed by TPH to determine intermediately bonded and strongly bonded surface species, respectively. CH4 and H218O signals were detected with the MS and thereby monitoring the hydrogenation of surface carbon species (C*) and surface oxygen species (18O*). The IH profiles for the effluents CH4 and H218O are shown in Figure 2a and b, respectively. Methane was observed immediately in the effluent as soon as H2 was introduced into the system. The flow of CH4 was close to an end within 30 s, as shown in Figure 2a. Similar peak shape and position were found for catalysts with the larger particle size, that is, 8, 11, and 15 nm, whereas a slight lag was found for the Co catalyst with 4 nm nanoparticles. The similar peak position with respect to time for the first three catalysts indicates similar C-Co binding energies, whereas a comparative stronger C-Co bond is found for the 4 nm Co catalyst. In addition, the peak area of CH4 reflects the amount of intermediately bonded surface carbon species. The total amount of methane assigned to the surface carbon species determined by IH experiment (C*-IH) can be calculated based on the CH4 effluent over the 30 s period as shown in Table 2. It is clear that more intermediately (29) Morikawa, Y.; Amenomiya, Y. Exchange of oxygen between carbon dioxide and alumina. J. Catal. 1977, 48 (1-3), 120-128. (30) Krupay, B. W.; Amenomiya, Y. Alkali-promoted alumina catalysts: I. Chemisorption and oxygen exchange of carbon monoxide and carbon dioxide on potassium-promoted alumina catalysts. J. Catal. 1981, 67 (2), 362-370.
Langmuir 2010, 26(21), 16558–16567
Yang et al.
Article
Figure 2. Formation of (a) CH4 and (b) H218O during isothermal hydrogenation of surface carbon and oxygen species deposited by the C18O
exposure experiment for catalysts with different particle size. The isothermal hydrogenation is carried out in pure H2 (15 mL/min) at 210 C, 1.85 bar.
bonded surface carbon is produced on the 15 nm Co catalyst after C18O exposure at 210 C, whereas similar values of C*-IH are found on the catalysts with a smaller Co particle size (4, 8, and 11 nm). The formation of water (H218O) was substantially slower than that of methane during isothermal hydrogenation as indicated in Figure 2b. The effluent profiles first reached a maximum for all these four catalysts and then decayed slowly. The results clearly showed that the water breakthrough was strongly dependent on the Co particle size, and the catalyst with the largest Co particle size (15 nm) needed the shortest time (10 min) to reach maximum and the signal decayed faster afterward. The broad distribution of water breakthrough indicates very different O-Co binding energy on different sized Co particles. Clearly, O-Co bond strength is more sensitive to the particle size change compared to C-Co bond strength. In addition, only very few of C18O or C18O2 in the effluent was detected in this isothermal hydrogenation experiment, suggesting few molecularly adsorbed CO on the surface in the isothermal hydrogenation. Due to the incomplete Langmuir 2010, 26(21), 16558–16567
release of 18O* as shown in Figure 2b, the total amount of 18O* species derived from isothermal hydrogenation was difficult to determine. TPH spectra revealed the amount and the distribution of strongly bonded surface species. Figure 3a presents the effluent profile of CH4 produced during the TPH. Two distinct methane peaks were found for each catalyst: One peak was in the range of 200-430 C, and the other one was between 430 and 700 C, implying that there were two types of strongly bonded carbon species on the Co surface after isothermal hydrogenation. The later peak can be ascribed to the hydrogenation of the carbonaceous deposition on the surface of the catalyst, while the former one may be assigned to cobalt carbide.14 The methane peak positions were similar for all the catalysts, suggesting a similar range of C-Co binding energy for all four catalysts with different Co particle size. However, the total amount of CH4 formed from the TPH experiment was strongly dependent on the Co particle size. The catalyst with smaller Co particle size released more CH4 compared with the catalyst with larger Co particles. The amount DOI: 10.1021/la101555u
16561
Article
Yang et al.
Figure 3. Formation of (a) CH4 and (b) H218O during TPH following IH for catalysts with different particle size. The TPH experiment was carried out in pure H2 (15 mL/min) with a ramping rate of 10 C/min from 210 to 700 C.
of the strongly bonded C* species is shown in Table 2, defined as the C*-TPH. It was found that the total amount of C* from both IH and TPH, defined as C*-Tot, was significantly higher than the total amount of CO2 formed during C18O exposure as shown in Figure 1. This means that the contribution of the CO disproportion reaction to the surface carbon species was relatively small and the surface carbon and oxygen species predominately resulted from the dissociation of CO. The CO dissociation probably happened mainly during isothermal hydrogenation with the presence of H2 at 210 C. The amount of CO dissociated (≈ 2C*-Tot) was in the same order of magnitude as the total number of active sites determined by H2 chemisorption. Therefore, CO could not have been dissociated only on steps, it dissociated also on terraces through the H-assisted route. Both experimental and
theoretical studies have shown that H-assisted CO dissociation is more favorable than direct CO dissociation on dense surfaces in FTS.31,32 On both flat Co (0001) and stepped Co(0001) surfaces, the activation energy for direct CO dissociation is higher than H-assisted CO dissociation and hence less favorable than the H-assisted route.33 Figure 3b shows the profiles of the released H218O on the Co catalysts during the TPH, and the main peak of H218O appeared in the range of 420-700 C, whereas another small peak appeared between 210 and 420 C. This observation is in contrast to the profiles of CH4 in the TPH experiments as shown in Figure 3a, where large amounts of surface carbon species were released at relatively lower temperature. The formation of H218O preferable at higher temperatures than methane suggests that the surface O
(31) Mitchell, W. J.; Xie, J.; Jachimowski, T. A.; Weinberg, W. H. CarbonMonoxide Hydrogenation on the Ru(001) Surface at Low-Temperature Using Gas-Phase Atomic-Hydrogen - Spectroscopic Evidence for the Carbonyl Insertion Mechanism on a Transition-Metal Surface. J. Am. Chem. Soc. 1995, 117 (9), 2606-2617.
(32) Shetty, S.; Jansen, A. P. J.; van Santen, R. A. Direct versus HydrogenAssisted CO Dissociation. J. Am. Chem. Soc. 2009, 131 (36), 12874-12875. (33) Inderwildi, O. R.; Jenkins, S. J.; King, D. A. Fischer-Tropsch Mechanism Revisited: Alternative Pathways for the Production of Higher Hydrocarbons from Synthesis Gas. J. Phys. Chem. C 2008, 112 (5), 1305-1307.
16562 DOI: 10.1021/la101555u
Langmuir 2010, 26(21), 16558–16567
Yang et al.
Article
species are more difficult to remove than surface C species through the stepwise hydrogenation. This is consistent with the previous observation during isothermal hydrogenation where methane formation was much faster than H218O formation. Table 2. Amounts and Distribution of Surface Species on Cobalt Catalyst after C18O Deposition at 210 C, 1.85 bara catalyst
C*-IH (μmol/gcat)b
C*-TPH (μmol/gcat)c
18 O*-TPH (μmol/gcat)c
MP_4 58 115 273 MP_8 52 83 193 MP_11 59 71 176 LP_15 71 72 210 a The reaction is performed in the following sequence: C18O exposure-Ar flushing-isothermal hydrogenation (IH)-temperature programmed hydrogenation (TPH). b The amount of C* species that can be hydrogenated at 210 C. Quantitative data for 18O*-IH is not calculated because the water peak is not complete. c The amount of strongly bonded C* and 18O* species hydrogenated in TPH (210-700 C), defined as C*-TPH and 18O*-TPH respectively
The total amount of H218O released is also shown in Table 2. It is difficult to compare the relative amount of surface C* and O* species due to the incomplete H218O* profiles in the isothermal hydrogenation reaction and the difficulty in the calibration of water. However, we can still compare the relative amount of H218O formed on the different catalysts: the smallest Co catalyst (4 nm) gives the highest amount of water during the TPH. Moreover, the largest C* pool along with the lowest CO2 formation for the smaller nanoparticles (Figure 1) indicates that the CO dissociation pathway (CO f C* þ O*) predominates against the CO disproportionation (2CO f CO2 þ C*), specially for the catalyst displaying smaller Co particles. Carbon and Oxygen Species on the Co Catalysts after CO Exposure and Methanation. To investigate the effect of methanation on the number and distribution of surface carbon and oxygen species, methanation was carried out immediately before IH and TPH and after the CO exposure experiment at 210 C. Methanation was performed for 30 min, and the surface was purged with Ar for 20 min before subsequent IH and TPH. It was found that almost no surface carbon and
Figure 4. Formation of (a) CH4 and (b) H218O during TPH following methanation and IH for catalysts with different particle size. The TPH experiment was carried out in pure H2 (15 mL/min) with a ramping rate of 10 C/min from 210 to 700 C. Langmuir 2010, 26(21), 16558–16567
DOI: 10.1021/la101555u
16563
Article
Yang et al.
oxygen species were released during isothermal hydrogenation. This could be due to the removal of these intermediately bonded surface carbon and oxygen species by the CO/H2 mixture during methanation. Very similar profiles for CH4 and H218O in the TPH experiment, shown in Figure 4a and b, were found to be comparable with the results shown in Figure 3a and b. However, the amounts of surface carbon and oxygen species are lower, as summarized in Table 3. These surface species released in TPH after methanation are very strongly bonded species and can be regarded as the blocking species in the methanation reaction (210 C, 1.85 bar), defined as C*-M-TPH and O*-M-TPH. With the additional methanation reaction, the amount of C*-TPH was reduced by a factor of 2, whereas the amount of O*-TPH was reduced by a factor of 5-10. A significant decrease of the oxygen species implies that the CO/H2 can effectively remove the strongly bonded oxygen-Co species. Moreover, the surface sites blocked by C*-M-TPH on the 4 nm Co catalyst were slightly higher than those on the catalysts with larger Co particle size. A significantly higher amount of O*-M-TPH was observed on the 4 nm Co catalyst compared with the other catalysts. In all, comparatively larger amounts of site blocking species were found on the catalyst with the smallest Co particle size (4 nm). Intrinsic Reaction Rate from SSITKA Study. SSITKA was used to determine in situ number of surface species and the intrinsic reaction rate. A typical normalized transient curve after a Table 3. Number of Very Strongly Adsorbed Surface C* and O* Species Elucidated in TPH catalyst
C*-M-TPH (μmol/gcat)a
18
O*-M-TPH (μmol/gcat)a
MP_4 55 49 MP_8 41 17 MP_11 45 30 LP_15 35 18 a Number of very strongly bonded surface C* and O* species determined in TPH experiment after C18O exposure, methanation reaction, and isothermal hydrogenation, defined as C*-M-TPH and 18O*-MTPH.
switch from Ar/12CO/H2 to Kr/13CO/H2 is shown in Figure 5. The mean surface residence time (τ) of CO and surface intermediates leading to methane (denoted as CHx species) are calculated by the areas under the normalized transition curve Fi(t), corrected for gas phase holdup with the Ar as an inert tracer. Z
¥
τi ¼
Fi ðtÞ dt
ð1Þ
0
The mean surface residence time of CHx species is corrected for the chromatographic effect of CO as shown in eq 2 τCHx ¼ τCH4 ðmeasuredÞ - 0:5τCO
ð2Þ
The number of the reversibly adsorbed CO and CHx species is estimated form the mean surface residence time (τ) and the exit flow (Q) of CO and CH4. Ni ¼ τi Qi, exit
ð3Þ
The SSITKA results are summarized in Table 4. The reaction rates of CO conversion and CH4 formation are based on the GC analysis. A significantly lower rate of CO conversion was found on the 4 nm Co catalyst compared with other catalysts, where the rate of CO conversion was almost independent of the Co particles size when the particle size is larger than 8 nm. This trend was also reflected in the rate of methane formation. Similar amounts of the reversible adsorbed CO (NCO) were found for all these four catalysts, whereas the number of surface CHx species (NCHx) increased with increasing cobalt crystal size. The higher concentration of surface intermediates (CHx) would probably lead to higher chain termination to methane in the presence of large amounts of H2 and thus lower C5þ selectivity at methanation conditions. However, a detailed discussion of the dependence of CHx and C5þ selectivity is beyond the scope of this paper. The emphasis here is to investigate the dependence of the intrinsic activity on the cobalt particle size. The TOF and the intrinsic reaction rate constant (k) are plotted as a function of the Co particle size for these catalysts, as shown in
Figure 5. Typical steady-state isotopic transients of Ar, 12CO, and 12CH4 following a switch from Ar/12CO/H2 to Kr/13CO/H2 (210 C, 1.85 bar, H2/CO = 10).
16564 DOI: 10.1021/la101555u
Langmuir 2010, 26(21), 16558–16567
Yang et al.
Article Table 4. SSITKA Results for CO Hydrogenation (210 C, 1.85 bar, H2/CO = 10)
catalyst samples
rate of CO conversion [μmol/(gcat s)]
rate of CH4 formation [μmol/(gcat s)]
NCO [μmol/ (gcat s)]a
τ12CO (s)b
NCHx [μmol/ (gcat s)]c
τ12CHx (s)b
k (s-1)d
TOF (103 s-1)e
MP_4 1.2 0.7 56 5.7 6 8.2 0.12 3.3 MP_8 2.4 1.5 53 6.0 12 8.0 0.13 10.0 MP_11 2.5 1.7 63 7.2 13 7.6 0.13 12.1 LP_15 2.6 1.8 60 7.0 15 8.5 0.12 14.8 a Surface number of reversibly adsorbed CO. b Mean surface residence time. c Number of surface intermediates leading to methane. d Reaction rate constant for methanation assuming pseudo-first-order reaction (k = 1/τCHx) e TOF is calculated based on the CO reaction rate and H2 chemisorption results.
Figure 6. TOF and the pseudo-first-order reaction rate constant k (1/τCHx) for catalysts with different particle size. TOF is calculated based
on H2 chemisorption. Figure 6. The TOF increased dramatically when the Co particle size increased from 4 to 8 nm and increased less when the Co particle size increased from 8 to 15 nm. No clear leveling off of the TOF was observed for larger particle sizes as reported previously.2,16,17 The deviation is probably due to the larger pore size of alumina (LP) used for the 15 nm particle size catalyst. Studies on Co/γ-Al2O3 (MP) with a comparative particle size (15 nm) show that the TOF is almost the same as that for the 11 nm particle size catalyst in our study. It was reported that no clear relationship was found between particle size and TOF for Co/ Al2O3 catalysts supported on 26 different types of alumina, which differed in structure, phases, and impurities.19 We speculate that differences in the aluminas may shield the dependence of TOF on particle size. Since there is only a minor difference between the MP and LP alumina used in this study, we presume this would not influence the qualitative study of the particle size effect. The pseudo-first-order reaction rate constants (k = 1/τCHx) for the methane formation determined by SSITKA were essentially the same, as shown in Figure 6. The independency of the reaction rate constant (k) on the Co particle size is supported by the similar C-Co binding energy for the active sites as discussed in previously. It should be noticed that the amount of the inactive sites for these catalysts are strongly size dependent as shown in Figure 4. Small particles should have more surface active sites than larger particles as evidenced by H2 chemisorption. However, slightly lower reversibly adsorbed CO with the 4 nm catalyst was observed in the methanation, probably due to the fact that more sites were blocked by the very strongly bonded carbon and oxygen species. These very strongly bonded species should be responsible Langmuir 2010, 26(21), 16558–16567
for the lower amounts of CHx species for the catalysts with smaller particles as well. Therefore, the decrease of TOF in this cases could be related to the decrease of the active sites resulting from selective blocking by very strongly bonded C* and O* species. Interestingly, previous SSITKA studies on Co/CNF catalysts with particle sizes ranging from 2.6 to 16 nm attributed the decrease of TOF with small particles to two factors: a decrease in the number of active sites and a decrease of the reaction rate constant.17 Molecularly adsorbed CO on low coordination sites was proposed to be the site blocking species for small particles.17 The present work clearly evidenced that very strongly bonded C* and O* species blocked the sites with low coordination numbers. However, the unblocked active sites might be composed of the similar surface atomic ensemble in this study. The apparent activation energy has been determined to be around 85 kJ/mol and independent of the cobalt particle size. This is in agreement with an earlier study on Co nanoparticles prepared with colloidal method with particle sizes in the range from 3 to 15 nm.18 Kinetic Isotopic Effect (KIE). Deuterium kinetic isotope effects, defined as the ratio of rates with CO/H2 and CO/D2 as reactants (rH/rD), serve as versatile tools to unravel details about the reaction mechanism. A typical effect of the D2-H2 experiment on the CO conversion and the methane selectivity in CO hydrogenation is shown in Figure 7. The reaction was first operated under H2/CO/Ar flow. After about 200 min, the reaction reached pseudo-steady state and the reactants were switched from H2/CO/Ar to D2/CO/Ar. The conversion with D2/CO/Ar reactants was increased by a factor of 1.4, accompanied by a decrease in the methane selectivity. After switching back to the H2 feed, the DOI: 10.1021/la101555u
16565
Article
Yang et al.
Figure 7. Effect of D2-H2 experiment on the activity and selectivity of 20 wt % Co/γ-Al2O3 catalyst for Fischer-Tropsch synthesis at 210 C, 1.85 bar with feed gas composition of H2/CO/Ar (15/1.5/33.5 mL/min) or D2/CO/Ar (15/1.5/33.5 mL/min). Table 5. Kinetic Isotope Effect (rH/rD) Data for the Rates and Selectivities for CO Hydrogenation on the 20 wt % Co/γ-Al2O3 Catalystsa particle size 4 nm parameter
rH
rD
8 nm rH/rD
rH
rD
11 nm rH/rD
rH
rD
15 nm rH/rD
rH
rD
rH/rD
RCO [μmol/(gcats)] 1.16 1.52 0.77 2.38 3.26 0.73 2.54 3.48 0.73 2.56 3.66 0.70 0.66 0.66 1.01 1.49 1.56 0.95 1.73 1.81 0.95 1.77 1.91 0.93 RCH4 [μmol/(gcats)] 0.10 0.11 0.89 0.22 0.27 0.81 0.24 0.31 0.77 0.25 0.34 0.73 RC2 [μmol/(gcats)] 0.15 0.21 0.71 0.25 0.42 0.60 0.24 0.43 0.55 0.24 0.46 0.51 RC3 [μmol/(gcats)] 0.13 0.23 0.57 0.22 0.46 0.49 0.19 0.45 0.42 0.18 0.47 0.38 RC4 [μmol/(gcats)] 0.13 0.32 1.14 0.19 0.56 0.95 0.14 0.48 0.82 0.13 0.47 0.74 RC5þ [μmol/(gcats)] b 0.07 0.07 0.99 0.03 0.03 1.01 0.02 0.02 1.04 0.02 0.02 0.99 C2=/C2 0.99 0.98 1.01 0.35 0.34 1.02 0.24 0.23 1.02 0.20 0.20 1.00 C3=/C3b a Conditions: 210 C, 1.85 bar, H2/CO = D2/CO = 10. b KIE values are calculated by taking the corresponding experimental olefin/paraffin ratios for the reactant CO/H2 and dividing each one by the ratios for the CO/D2 reactant.
conversion and selectivity returned to the pseudo-steady-state level, indicating that no deactivation occurred during the whole process. More details of the D2-H2 results are presented in Table 5 for the four catalysts with different Co particle size, where the KIE values (rH/rD) are calculated by taking the corresponding experimental rates (or olefin/paraffin ratios) for the reactant CO/H2 and dividing each one by the rates for the CO/D2 reactant. An inverse isotope kinetic effect (rH/rD < 1) was observed for the CO conversion and the C5þ formation for all the catalysts. However, the exchange of H2 with D2 did not change the rate of methane formation and the olefin/paraffin ratio (C2, C3). Therefore, it is speculated that the rate determining step (RDS) for methane formation and the overall reaction is different in FTS. The inverse kinetic isotope effect is essentially a consequence of an inverse equilibrium isotope effect being transferred to the rate determining step.34 The inverse isotopic kinetic effect normally indicates the involvement of a σ-complex [M(σ-XH)] or the product of oxidative addition [M(X)H], where X is hydrogen or carbon during the interaction of H-H and C-H bonds with the transition metal centers. The existence of a σ-complex, probably formed by Co and CHx surface species [Co(CHx)], seems to play an important role in chain propagation as evidenced by the inverse isotopic effect of CO conversion and C5þ formation. On
the other hand, it is well-known that the methanation reaction involves CO dissociation and stepwise hydrogenation of C* species. The KIE experiment excludes the stepwise hydrogenation being RDS, which would give a normal isotopic kinetic effect (rH/rD > 1). Thus, methane formation should be dependent on the CO dissociation ability, which is directly related to the C-Co binding strength. The similar C-Co binding strength found during isothermal hydrogenation is consistent with the constant intrinsic reaction rate constant (k) in SSITKA experiments. In addition, substituting H2 with D2 did not change the reaction preferences either to olefins or paraffins. A similar inverse isotopic effect has been reported on 21.9 wt % Co/SiO2 at 473 K, 2 MPa, H2/CO = 2 with a value of ∼0.8. The authors attributed this effect to the involvement of hydrogen adsorption-desorption equilibrium and to the C-H bond formation in the kinetically relevant FTS steps.35 Concerning the particle size effect, it would be expected that the KIE values would change significantly if the particle size changes the rate determining step. Instead, the kinetic isotope effect for CO consumption and CH4 formation was not much affected by the particle size of the cobalt metal, indicating that the kinetically relevant steps were the same for particles with different Co particle size.
(34) Parkin, G. Temperature-Dependent Transitions Between Normal and Inverse Isotope Effects Pertaining to the Interaction of H-H and C-H Bonds with Transition Metal Centers. Acc. Chem. Res. 2009, 42 (2), 315-325.
(35) Krishnamoorthy, S.; Tu, M.; Ojeda, M. P.; Pinna, D.; Iglesia, E. An Investigation of the Effects of Water on Rate and Selectivity for the FischerTropsch Synthesis on Cobalt-Based Catalysts. J. Catal. 2002, 211 (2), 422-433.
16566 DOI: 10.1021/la101555u
Langmuir 2010, 26(21), 16558–16567
Yang et al.
Article
The isotope effect for the rate of C5þ formation was more significant on large particles than on the smaller ones. Given that the isotope effect does not dependent on the conversion and the water content,35 it is speculated that the isotope effect for C5þ formation is related to the nature of the catalysts. The origin of the inverse isotope effect is that deuterium prefers to be located on the site corresponding to the highest frequency oscillator,34 and the lower C-D vibration frequencies of the agnostic C-D bonds in the ground state relative to the bonds in the transition state result in a smaller difference in the ground state relative to the transition state (ΔHD‡ < ΔHH‡).36 Probably, large particles can stabilize the transition state of chain propagation intermediate species to a larger extent than smaller particles, which indicates that larger particles have inherent higher chain propagation probability.
Conclusions CO dissociated on Co/γ-Al2O3 catalysts at 210 C. The total amount of CO2 formed following the dissociation depends on the cobalt crystal size. O-Co binding energy was found (36) Zeller, A.; Strassner, T. Inverse Isotope Effects of a Late Transition Metal Olefin Polymerization Catalyst: A DFT Study. Organometallics 2002, 21 (23), 4950-4954.
Langmuir 2010, 26(21), 16558–16567
to be highly dependent on the Co metal particle size, whereas similar C-Co binding energy was found on catalysts with different Co particle size. Very strongly bonded carbon and oxygen surface species increased with decreasing particle size and acted as site blocking species in the methanation reaction. SSITKA experiments showed that the intrinsic activity (1/τCHx) remained constant as the particle size increased from 4 to 15 nm. The number of surface intermediates (NCHx) increased with increasing particle size. The apparent activation energies were similar for these catalysts, about 85 kJ/mol. D2-H2 switches further confirmed that the particle size did not change the kinetically relevant steps in the reaction. The reactivity of the active sites on the 4 nm particles was the same as those on the 8, 11, and 15 nm particles, and only the number of total available surface active sites in situ was less on the 4 nm particles than on the others. Acknowledgment. The financial support form The Norwegian Research Council through the KOSK programme and the Norwegian University of Science and Technology is gratefully acknowledged. Statoil is acknowledged for providing catalyst samples.
DOI: 10.1021/la101555u
16567
