J. Phys. Chem. 1981, 85, 496-498
496
estimated; it should be > l o 0 1 for attractive tritium recovery. For T/D separation using MPD of the v2 mode, the optical selectivity appears to reach a maximum of -30 near 9.3 pm, whereas for T / H separation the selectivity is much higher, -700 near 9.3 pm. Reference to Figure 2 suggests that burning out the 13Cisotope (1.1% natural abundance) may further increase the selectivity in CTF,/CHF,; however, this procedure will probably not improve the CTF3/CDF3selectivity. Based on the v2 band strength, a COPlaser fluence of -30 J/cm2 should lead to significant dissociation of CTF, at the Q branch peak (for T / H recovery). T/D separation is favorable with an NH3 laser at 12.08 pm to pump the v5 mode, where the optical selectivity is >250. However, since the v5 band strength in CTF, is much weaker than in CDF,, a much higher fluence of 100 J/cm2 will be required to dissociate CTF3, compared to the -20 J/cm2 needed to decompose CDFB near 10.3 pm.ll3 Since the v1 mode is fairly strong (absorption coefficient of 10-2/(cmtorr)) and the absorption spectrum-near 1930 cm-l is fairly c l g n in CHFB(a N 1 X lOT(cm torr)) and in CDF, _ (a_ N 6 X_ 10-51(cm torr)), _ ~ __ use of a CO laser may lead to successful tritium recovery. Since the effective molecular absorption coefficient decreases with increasing fluence in multiple-photon absorption, the optical selectivity at the fluences of interest will be lower than those quoted here. Still, there appears to be some promise for photochemical T / H and T / D isotope separation using trifluoromethane. It should be mentioned that, at the conclusion of this study, we learned of a parallel effort in tritium isotope separation involving MPD of halogenated methanes at The
-
-
Institute of Physical and Chemical Research in Japan.lSm In ref 19 the fundamental vibrational frequencies of a number of tritiated halogenated methanes were calculated by using a Urey-Bradley-type force field. Reference 20 presents the results of preliminary experiments on CTF,; no infrared spectrum of CTF, is reported in that work. The calculation of the CTF3spectrum reported here is in much better agreement with the observed spectrum (Figure 1and Table 11) than is the calculation of ref 19. In addition, there are quite significant differences in the prediction of the vibrational frequencies of the other monotritiated halogenated methanes when Urey-Bradley fields are used (ref 19) rather than general harmonic force fields (ref 7). These points emphasize the importance of using the more accurate general harmonic force field, preferably based on harmonic frequencies, in determining the halogenated methane fundamental frequencies. Work is continuing on the spectroscopy and infrared laser photochemistry of the tritiated halogenated methanes. Acknowledgment. We thank Dr.Roy Tsugawa, Larry Walkley, and Evelyn Fearon for their experimental and technical assistance in this investigation. This work was performed under the auspices of the US. Department of Energy by the Lawrence Livermore Laboratory under contract number W-7405-Eng-48. (19) Y.Ishikawa, S.Arai, and R. Nakane, J. Nucl. Sci. Technol., 17,
275 (1980). (20) Y.Makide, S. Hagiwara, 0. Kurihara, K. Takeuchi, Y. Ishikawa,
S. Arai, T. Tominaga, I. Inoue, and R. Nakane, J. Nucl. Sci. Technol., 17,645 (1980).
A Strong Metal-Support Interaction between Mononuclear and Polynuclear Transltion Metal Complexes and Oxide Supports Which Dramatically Affects Catalytic Actlvity Dennis A. Hucul and Alan BrenneP Department of Chemistry, Wayne State University, Detroit, Michigan 48202 (Received:October 27, 1980)
The interaction of carbonyl complexes with catalyst supports, primarily y-alumina, has been studied by temperature-programmeddecomposition. In all cases, including cluster complexes and complexes of noble metals, after heating to 600 "C in flowing He the catalysts are significantly oxidized due to a redox reaction between surface hydroxyl groups and the initially zero-valent metal. Contrary reports are probably incorrect and likely reflect the insensitivity of the experimental techniques used. For all but the most thermally unstable complexes, the oxidation occurs during the latter stages of decarbonylationindicating that there is no significant accumulation of bare zero-valent metal. Hence, decomposition does not in general provide a direct route to supported metals and, contrary to some claims, molecular cluster complexes cannot necessarily be used as precursors to supported metal clusters. Further, knowledge of this redox reaction is critical for understanding patterns of activity and for the development of improved catalysts. Introduction Transition metal complexes directly deposited on high surface area refractory supports (as alumina, silica, and molecular sieves) have been receiving wide attention in the last few years as a new class of catalysts. These materials physically lie at the frontier between traditional homogeneous and heterogeneous catalysts (note the metal is not insulated from the support by a chain of ligands and these are not immobilized homogeneous catalysts) and in fact have the potential of combining the better features of both types of catalysts: the wide gamut of catalyst precursors,
better characterization, and possible improved selectivity of homogeneous catalysts and the enhanced stability of heterogeneous catalysts. Particular attention has been focused on supported carbonyl complexes, both mononuclear and p~lynuclear.l-~It has also recently been demonstrated that in some cases carbonyl-derivedcatalysts can (1) Smith, A. K.; Basset, J. M. J. Mol. Catal. 1977,2, 229. (2) Ugo, R.; Psaro,R.; Zanderighi, G. M.; Basset, J. M.; Theolier, A,; Smith,A. K.Fund. Res. Homogeneous Catal., R o c . Int. Workshop 1979, 3, 579.
(3) Gates, B. C.; Lieto, J. CHEMTECH 1980, 10, 248.
0022-365418112085-0496$01.2510 0 1981 American Chemical Society
The Journal of Physical Chemistty, Vol. 85, No. 5, 198 1 487
Strong Metal-Support Interaction
be prepared with higher metal dispersions than for traditional heterogeneous catalysts (made by aqueous impregnation with metal salts followed by calcination and reduction) ,4i5 can have higher activity for hydrogenations and methanation,’ and can have improved selectivity for Fischer-Tropsch synthesk8 It has also been suggested that carbonyl cluster complexes might be used to prepare supported metals in which the metal is in well-defined and small ensembles which are not readily accessible by the traditional methods of catalyst s y n t h e ~ i s . ~ -However, ~J~ it has recently been shown that during thermal decomposition some supported carbonyl complexes of base metals can become oxidized by a reaction with surface hydroxyl groups of the support and thus zero-valent metal is not f ~ r m e d . ~ * ~Only J ~ - ~in’ a few cases have quantitative data been reported4J1-14and other groups have not noted this reaction.leZ1 It is currently unknown if this oxidation reaction also occurs for complexes of more noble metals. Further, in several cases large drops in catalytic activity have been ascribed to this reaction.6J1pn Thus, knowledge of this redox reaction is of the utmost importance in understanding the basic surface chemistry of these new catalysts, in explaining patterns of activity, for fundamental studies relating to the utility of cluster complexes as precursors to discrete metal clusters, and for the correct interpretation of the current literature in this area. In this paper we describe this redox chemistry for alumina-supported complexes of all elements which form stable carbonyls and further document the dramatic effect which this reaction can have on catalytic activity. Experimental Section The catalysts were prepared by physically dispersing a carbonyl on 7-A1203(Conoco Catapal SB, previously calcined at 500 “C) by impregnation from pentane solution or sublimation of the solid carbonyl. The catalysts were activated and characterized by temperature-programmed decomposition (TPDE). TPDE involves raising the temperature at a linear rate, p, of about 5 OC/min as He is swept through the reactor and the evolution of gases (primarily CO and Hz) is continuously monitored with a pair of thermal conductivity detectors. Other gases formed (4) Brenner, A.; Hucul, D. A. Znorg. Chem. 1979, 18, 2836. (5) Anderson, J. R.; Elmes, P. S.; Howe, R. F.; Mainwaring, D. E. J. Catal. 1977,50, 508. (6) Brenner, A. J. Mol.Catal. 1979, 5, 157. (7) Brenner, A.; Hucul, D. A. Proc. Znt. Conf. Chem. Uses Molybdenum, 3rd 1979,194. (8)Commereuc,D.; Chauvin, Y.; Hughes, F.; Basset, J. M.; Oliver, D. J. Chem. SOC.,Chem. Commun. 1980,154. (9) Anderson, J. R.; Mainwaring, D. E. J. Catal. 1974,35, 162. (10) Muetterties, E. L. Science 1977,196, 839. (11) Brenner, A,; Hucul, D. A.; Hardwick, S. J. Inorg. Chem. 1979,18, 1478. (12) Brenner, A.; Burwell, R. L., Jr. J. Catal. 1978,52, 353. (13) Brenner, A.; Hucul, D. A. J. Catal. 1980,61, 216. (14) Bjorklund, R. B.; Burwell, R. L;&. Colloid Interface Sci. 1979, 70, 383. (15) Smith, A. K.; Hugues, F.; Theolier, A.; Basset, J. M.; Ugo, R.; Znderighi, G. M.; Bilhou, J. L.; Bilhou-Bougnol,V.; Graydon, W . F. Inorg. Chem. 1979.18. 3104. (16) Tkatchenko, D. B.; Coudurier, G.; Mozzanega, H.; Tkatchenko, I. J. Mol. Catal. 1979, 6, 293. (17) Gallezot, P.; Coudurier, G.; Primet, M.; Imelik, B. ACS Symp. Ser. 1977, No.40, 144. (18)Bilhou, J. L.; Theolier, A,; Smith, A. K.; Basset, J. M. J. Mol. Catal. 1977, 3, 245. (19) Bilhou, J. L.; Bilhou-Bougnol,V.; Graydon, W. F.; Basset, J. M.; Smith, A. K.; Zanderighi, G. M.; Ugo, R. J. Organometal.Chem. 1978,
-15.7 - -, 72 .-.
(20) Howe, R. F. Inorg. Chem. 1976, 15, 486. (21) Howe, R. F. J. Catal. 1977,50, 196. (22) Brenner, A. “Relations between Homogeneous and Heterogeneous Catalysis”;Centre National Recherche Scientifique: Paris, 1978; p 195.
TABLE I: Oxidation of the Metal During TPDE in Flowing He t o 600 C of Carbonyl Complexes Supported o n 7-Alumina
% meti1
av gas evolution O.N. (per complex) (per metal ~ 0 . 5 H, CH, atom)
0.154 0.0614 0.051 0.228 0.113 0.370 0.274 0.297 0.0978 0.197 0.376 0.166 0.376 0.268 0.518 0.907 0.246
440 0.98 170 1.91 290 1.97 220 2.07 160 1.02 200 5.36 220 -0.88 230 1.65 230 3.4 260 5.1 270 4.8 240 1.91 380 4.2 400 3.65 400 2.11 410 3.37 320 0.52
loading complex
V( co 16
Cr(C0)6 Mo(CO), w(c0)6 Mn,(CO),, Re,(CO),, Fe(CO), Fe,(CO),a Fe,(CO),, Ru,(CO),,~ Os,(CO),, Co,(CO), CO,(CO),,~ Rh,(CO),,a Rh6(CO)16a Irf(CO),,a Ni(CO),O a
temr, ( O Cj where O.N.
0.098 0.107 0.116 0.22 0.111 0.022 0.027 0.069 0.057 0.29 0.22 0.064 0.081 0.101 0.083 0.065 0.10
2.5 4.5 4.6 5.5 1.4 5.4 1.9 1.9 2.4 4.0 3.6 2.1 2.2 2.0 0.8 1.8 1.6
Catalyst prepared by dry mix method.
during TPDE (primarily CH4) are analyzed after the temperature programming by backflushing through a trap of SiOz (held at -196 OC during the run) which is between the detectors for CO and Hz. Details of the high-purity system and TPDE technology have been previously described.’l Results and Discussion Recent studies of supported carbonyls have shown that the evolution of H2during TPDE is due to a redox reaction between the initially zero-valent carbonyl and the surface hydroxyl groups of the support, which may be approximated as4J1-14
M(C0)j + n(u-OH)
A
(u-O-),Mn+
+ (n/2)Hz + $0
(1) Thus, the position of the H2 peak during the TPDE indicates the temperature at which a metal becomes oxidized and integration of the peak is a measure of the extent of oxidation. A small amount of methane is formedB (roughly 0.1 CHI/ complex) which represents additional reduction of H+ by the metal and hence must be considered in computing the final oxidation number (O.N.) of a catalyst after TPDE. However, the correction is small (about 0.6 units of oxidation per complex) compared to the oxidation implied by the Hz and separate experiments now show that the CH4 is always formed concurrently or at a higher temperature than the H2evolution. Therefore, the H2peak during TPDE still correctly portrays the temperature of the oxidation reaction. Excellent agreement (within &0.2 O.N.) has been reported for the oxidation number determined by gas evolutions and independent chemical titration for supported carbonyls of Fe4 and the groups 6B metals. 1-13 In contrast, catalysts of very active metals (Ru, Os, Rh, and Ir) prepared by impregnation gave unreasonably large evolutions of Hz. This is due to dehydrogenation of a tiny amount of pentane which is retained by the support to high temperatures. Where there was reason to suspect such a reaction (or if there were solubility or handling problems, as for Fez(C0)9,Ni(C0)4,and Co4(CO)lJ the catalysts were (23) Brenner, A.; Hucul, D. A. J. Am. Chem. SOC.1980, 102, 2484.
498
The Journal of Physical Chemistty, Vol. 85, No. 5, 1981
TABLE 11: Formal Turnover Frequencya for the Hydrogenation of Ethylene (2' = 0 C, P = 1 atm, H,/C,H, = 4 ) complex catalyst activation
200 "C, He 600 "C, He above t redoxb
W(CO), 0.02 9x 2 x 10-4
Re,(CO), 0 0.2
0.02 4
a s - l , per metal atom assuming 100%dispersion. d o x = 0, at 500 o C then H, at 600 o C.
Re-
prepared by a dry mix method to avoid such complications. Table I shows the final average oxidation number for each catalyst after TPDE to 600 "C in flowing He and the temperature at which the oxidation becomes significant (0.N 3: 0.5). The important finding is that in every case the intially zero-valent metal becomes oxidized during TPDE to high temperature. It should be stressed that the oxidation numbers rest on accurate stoichiometric measurements and do not require detailed characterization of the surface species. Futher, TPDE is highly sensitive, catalysts of 0.01% metal loading (corresponding to a fractional surface coverage of -3 X being amenable to study. Is is unlikely that a single species is being formed. Inspection of Table I shows that the amount of oxidation is independent of the nuclearity of the precursor complex. The apparent discrepancy in the case of Rh4(CO)12vs. Rh6(CO)16most likely reflects the very high thermal stability of the latter complex which makes it difficult to disperse (as a dry mix) on the alumina prior to TPDE. Hence, large crystallites may well be formed which would inhibit reaction with a-OH. This also probably explains why the temperature of oxidation is higher for C04(CO)12 than for Co2(CO)&Other experimentsNalso show that the extent of oxidation is nearly independent of /3 (although the expected increase in peak temperature with an increase in ,d is observed4) and metal loading. (After calcination of 500 "C the alumina is 28% hydroxylatedz6which corresponds to a ratio of a-OH to metal atoms of roughly 60 a t a loading of 0.2% metal. Oxidation of Mo to Me+ requires about 6 a-OH/M.) In almost every case the H2evolution coincides with the final evolution of CO. The exceptions are the highly thermally unstable Ni(C0)4 and V(CO)6 for which CO evolution is nearly complete before oxidation starts (at 150 and 250 OC, respectively). These results indicate that the decarbonylated zero-valent metal is usually oxidized as soon as it is formed, as has been confirmed in more detailed studies of Mo(C0)6/A1203.12However, surface mobility of a-OH may also be required for efficient oxidation, and this would explain why thermally unstable complexes such as Ni(C0)4might be able to yield some supported metal before the temperature becomes high enough to allow mobility of the a-OH and consequent oxidation. The converse of eq 1 is essentially the reduction of a supported metal oxide (assuming that the HzO normally observed during reduction is originally in the form of a-OH which then condense to yield H20). For this reason, from work with supported M O ( C O ) and ~ ~ ~Ni(C0)414it was suggested that the ability of these complexes to undergo (24) Hucul, D. A. Ph.D. Thesis, Wayne State University, 1980. (25) Peri, J. B. J. Phys. Chem. 1965, 69,211.
Hucul and Brenner
reaction 1may correlate with the difficulty in reducing the corresponding metal oxides. However, it is now demonstrated that reaction 1 occurs for both metals yielding extremely difficult to reduce oxides (as W03) as well as for more noble metals yielding readily reducible oxides (as Ir02). Thus, even though in a sweep gas of He reaction 1 is both thermodynamically and kinetically allowed for all systems, this does not directly yield information on the ease of reversing the reaction in the presence of H2. A number of catalytically important elements (as V, Cr, Mo, W, and Mn; and Fe, Ni, and possibly other elements at low loadings) cannot be reduced back to the zero-valent state after they have become oxidized. However, many of the more noble metals can be fully reduced after TPDE (but under conditions unlikely to maintain the molecular nature of the precursor). Since the activity of the transition elements for hydrogenation (and other reactions involving soft reactants) is often maximized in the zerovalent state, a distinct dichotomy arises in the activity of catalysts derived from complexes of reducible vis-a-vis unreducible metals. This is illustrated in Table 11. As shown in Table I and by the TPDE chromatograms, activation at 200 "C causes substantial loss of CO (and possibly the concomitant development of coordinative unsaturation) with only slight oxidation. Hence, significant activity is displayed by both catalysts. To our knowledge, W(CO)6/A1203is the most active catalyst of supported W yet reported. After TPDE to 600 OC of fresh catalysts, the activity is extremely low due to oxidation. Only slight activity is regenerated by attempted reduction of the W catalyst due to its difficult reducibility.% (Note that this is equivalent to a "traditional" catalyst of reduced W03/A1203.) However, the more readily reducible Re becomes extremely active (note that all CO ligands have been removed). An analogous metal-support interaction between Fe3(C0)12and Y zeolites has also recently been reported to significantly affect the activity for FischerTropsch ~ynthesis.~' In conclusion, carbonyl complexes (and likely any organometallic unless it has a very low thermal stability) cannot in a straightforward way be used as a precursor for the preparation of zero-valent supported metals or discrete metal clusters. Literature claims to the contrary1621are probably in error and may reflect the insensitivity of the experimental techniques (usually infrared spectroscopy) to monitoring the oxidation reaction. In contrast, TPDE is seen to be a highly sensitive and quantitative probe of this strong and specific metal-support interaction. Further, interpretation rests only on stoichiometric measurements and thus is relatively free of the ambiguities often associated with spectroscopic techniques. Knowledge of this redox reaction is seen to be critical for understanding patterns of activity and for the designed synthesis of optimal catalysts. Futher studies demonstrating the means of controlling the oxidation number of the product catalyst (including the synthesis of zero-valent catalysts by appropriate support pretreatment) and detailing structure-activity relationships will be reported soon. Acknowledgment. Support of this research by the Department of Energy is gratefully acknowledged. (26) Biloen, P.; Pott, G. T. J. Catal. 1973, 30, 169. (27) Tkatchenko, D. B.; Coudurier,G.;Tkatchenko, 1. Prepr., Diu. Pet. Chem., Am. Chem. SOC.1980,25,755.
