摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 253-261 GENERAL DISCUSSION Prof. J. B.Nagy (FUNDP, Numur, Belgium) opened the discussion on Dr Vedrine's paper: Could you show us the 31P-n.m.r. spectra with and without m.a.s.? The analysis of the spectra could help in studying the symmetry of the phosphorous surroundings. The relative intensity of the unshifted and the shifted n.m.r. lines could yield interesting information about the short-range organization of the two neighbouring phases VOPO, and (VO)2P207. Dr Vedrine replied: We agree that the 31P m.a.s. n.m.r. peak may greatly help us to study the symmetry of P surroundings. However, in our case the main constituent is (VO),P207 with vanadium ions in the +4 oxidation state i.e. d' ion. The paramagnetism results in a very broad 31P m.a.s.n.m.r. peak. In contrast, VOPO, with V5+ (do) ions gives a narrow 31P peak with a chemical shift depending on the structure of the different phases:' 6 = -19.5, -11.8, -18.8 and 1.6 ppm for p, y and 6 VOPO, phases with respect to H3P04. When VOPO, and (V0)2P207 phases are present in proximity, 31P peaks corresponding to the two phases are observed. If the VOPO, entities are interacting with (VO)2P207 the corresponding 31P is shifted with respect to the pure phases, 6 = -30 and -40 ppm for 6 and a l l phases, respectively. The latter case is observed for the best catalysts (>60% yield in maleic anhydride), which supports our conceptual view of active sites corresponding to topotactic changes as: P207 2 PO,. 1 M. David, F. Lefebvre and J. C. Volta, Actas def XI Simposio Iberoamericano de Catalysis, Guanajuato, ed.F. Cossio, G. del Angel, 0. Bermudez and R. Gomez (IMP, Mexico, 1988) p. 365. Prof. 0. V. Krylov (Institute of Chemical Physics, Moscow, U.S.S.R.) then said: It is my opinion that there are some contradictions between the data of the papers of Dr Vedrine and of Prof. Centi. Dr Vedrine showed formation of two V-P-0 phases during oxidation of butane to maleic anhydride. Prof. Centi claimed formation of only one active phase (V0)2P207. We have studied many processes of partial oxidation by physical methods in situ. In many cases we observed formation of new phases during catalysis. I think that multiphasity is very often principally important for selective catalysis. It allows us to use different phases for the acceleration of different steps of the complex catalytic process.I would like both speakers to comment on this statement. Dr Vedrine, in reply, said: As noted by professor Krylov our paper may appear contradictory to that of Centi et al. However, our view is that such catalysts behave as living materials with topotactic changes during catalytic reaction. The main constituent is (VO),P2O7 but for us the active surface sites correspond to a topotactic change under catalytic conditions (VO)*P*O, * 2VOP0,. This means that during the catalytic reaction the catalyst is in a slightly oxidized state with respect to (VO),P,O,, which is obviously not in agreement with the conclusions of Centi et al. However, the reaction mechanism was shown to be of Mars and Van 253254 General Discussion Krevelen type, i.e.it involves lattice oxygen and therefore a redox mechanism at the surface of the catalyst; I think that Dr Trifirb's group agrees with such a statement. Prof. Krylov also mentions that multiphase catalysts are important for selective oxidation, which seems to be in favour of our conclusion. This holds particularly true for multicomponent-type catalysts used on an industrial scale and which are much more complex than our system. Dr G. Centi (University of Bologna, Italy) added: It is important to remember that in the V-P-0 system the catalyst evolves as a function of time-on-stream from a biphasic system [ (VO)2P207 plus a Vv-phosphate in the starting catalyst] to a monophasic system [only (VO)2P207 after about 1000 h on stream in contact with butane/air flow].A parallel small improvement in the catalytic behaviour is noted. It is thus difficult to believe that the active catalyst is a biphasic system, as may be indicated when analyses are made only on VPO catalysts which have not been stabilized. We cannot exclude, however, that few Vv sites are present on the (VO)2P207 surface, although these are extremely limited in number, as also indicated by the very low rate of oxidation of V'" to Vv in the stabilized (VO)2P207 (see paper). Doping experiments with SO2' agree with this indication. Regarding the more general question that multiphasic systems are necessary for selective catalysts, this may be true in some cases (for example, acrylonitrile syntheses from propylene), but I do not believe that it can be taken as a general statement.In particular, regarding the selective oxidation of n-butane, the absence of effective pro- moters to the VPO system is clearly indicated in the patent literature. In general, I don't believe that a selective mechanism of synthesis requires a multiphasic system, because this would imply that the adsorbed intermediates must migrate on the surface with probable lowering of the selectivity due to the enhanced probability for parallel waste reactions. More realistic in a multiphasic system, the reaction occurs at the interface between two phases, but in this case the real active phase is not the multiphasic system but the new phase generated at the interface. 1 G. Centi, G. Golinelli and F. Trifiro, Appl.Caral., 1989, 48, 13. Prof. J. B. Moffat (University of Waterloo, Ontario, Canada) had a comment on the paper by Centi et al.: Dr Centi has provided valuable information on the effect of sample preparation of V-P-0 catalysts in butane oxidation. In our laboratory we are also working with M-P-0 catalysts, both stoichiometric and non-stoichiometric.' We have found that, for B- P-0 catalysts where P/ B > 1 , Brgnsted-acid sites predominate, while for P/B < 1 the acidic sites are primarily Lewis. Further, with Al-P-0 catalysts we have found that where Al/P> 1 a single-site mechanism predominates in the dehydration of 1 -methylcyclohexanol while where Al/P< 1 a two-site mechanism is more evident. Since in our laboratory we have considerable interest in the P-0-H grouping it would be useful if Dr Centi would care to provide further information on the part that surface Bronsted sites (P-OH) play in the mechanism of selective oxidation.1 J. B. Moffat, Rev. Chem. Intermed., 1987, 8, 1 and references contained therein. Dr Centi responded: Our preliminary evidence based on F.t.i.r. results and reactivity measurements in n-butane and n-pentane oxidation' suggests that Bronsted P-OH sites in the presence of O2 and water may be effective in the oxidation of aromatic rings. In particular, in the mechanism of synthesis of maleic anhydride from butane their role is related to the stage of oxygen insertion on the furan-like adsorbed intermediate, accordingGeneral Discussion 255 to the following tentative mechanism (only first stage up to lactone): H 0 / Y A similar mechanism may be written for the consecutive stage up to maleic anhydride.A more detailed and careful investigation is, however, necessary to demonstrate the mechanism. 1 G. Busca, G. Centi and F. Trifiro, manuscript in preparation. Dr J. Evans (Southampton University) said regarding Prof. Pinnavaia's paper: The i.r. data presented for the materials formulated as supported [ Ru(CO),(OAl-),I, are similar to observations for oxidised samples of [Ru,(CO),,]/silica. In that case, Ru K-edge EXAFS studies show that there are indeed no Ru-Ru bonds and also the carbonyl-containing centre has predominantly the coordination sphere of cis- Ru(CO)~(O-)~' when related to previous i.r. evidence.2 I wish to ask Prof. Pinnavaia for his comments on the differences in the i.r.frequencies between the oxidised centres on the two different clays. Are these likely to be related to the nature of the clay binding sites or may there be acid-base equilibria at the coordination centre? Related to that question, may I ask him about the evidence for the ruthenium being bound to the pillars, rather than other structural elements in the clays? 1 N. Binsted, J. Evans, G. N. Greaves and R. J. Price, Organometallics, 1989, 8, 613. 2 J. Evans and G. S. McNulty, J. Chem. SOC., Dalton Trans, 1984, 1123. Prof. Pinnavaia replied: I am pleased to learn that your recent EXAFS results establish a ~is-[Rh"(C0)~] configuration for the product formed by the reaction of Ru,CO,, on silica. On the basis of our FTIR results, an analogous functionality is most probably formed on APM and LDH.The CO stretching frequencies shifts for the APM and LDH supports are probably due to differences in the cis-[ Ru"(CO)~] coordination sphere. The number, symmetry, and type of oxygen binding sites are surely different for the two supports. With regard to the second part of your question, the ~ i s - [ R u ( C 0 ) ~ ] centres in APM are almost certainly grafted to the alumina pillars. Sodium montmorillonite reacts with256 General Discussion R u ~ ( C O ) , ~ to bind only trace amounts of grafted ruthenium, most probably at layer edges. In addition, we have found that 'stuffed' alumina pillared fluorohectorites with internal pores too small to accommodate R U ~ ( C O ) ' ~ react to form only trace amounts of grafted ruthenium.Thus, both the alumina pillars and the open-pore structure of APM are needed to bind ruthenium at the levels observed here. Dr R. A. van Santen (University of Technology, Eindhoven, The Netherlands) said: I would like to comment on the synthesis gas conversion results of the Ru-activated A1 pillared-clay catalyst. The paper reports the production of isomerized hydrocarbons from synthesis gas. It is proposed that this is due to the combined action of two functions. One is the production of linear alkanes or terminal straight-chain alkenes and the consecutive isomerization of the alkenes. To support this proposition the results presen- ted in fig. 3 (p. 233) of the paper are used. It may well be that the two reactions, chain growth and isomerization, are independent.Linear alkanes are formed over Ru. In a consecutive reaction the linear alkanes are converted via a bifunctional reaction. The results presented in fig. 3 indicate decreased isomerization with increasing H2/C0 ratio. It is likely that the ratio of the chain growth versus chain termination is decreased by the increased H2/C0 ratio. As a consequence the production rate of the longer hydrocarbons is decreased, which would result in a decreased rate of isomerization. Prof. Pinnavaia responded: We are indeed attributing chain-growth and chain- branching (isomerization) over Ru/APM to independent pathways, and yes, the ratio of chain-growth to chain-termination should decrease with increasing H2/C0 ratio. However, this should not contribute to the isomerization results described in fig.3 of the paper unless, as your question implies, the isomer distribution in the C4-C6 range arises from the hydrogenolysis of alkanes with higher carbon numbers. Alkanes do not react over Ru/APM to yield isomerized products. On the other hand, terminal alkenes are isomerized over Ru/APM at the CO hydrogenation temperatures employed in this study. Thus, acid catalysed rearrangements of the terminal alkenes formed in the chain propagation step are responsible for the observed branching and isomerization. Prof. A. Zecchina (Turin University, Italy) then asked: The i.r. spectra of [Ru(CO),(OMr)] complexes in APM and HT are substantially different (the two frequencies are ca. 30-35 c m - ' lower on HT).Can you add something more on this point? In particular: ( I ) What is the intensity ratio of the two peaks [this is relevant for understanding (2) Did you perform 12CO-'3C0 isotopic substitution experiments in order to eluci- (3) What is the explanation for the observed frequency difference? structure and stoichiometry of the Ru(CO), moiety]? date the stoichiometry of the complex? Prof. Pinnavaia answered: As noted earlier, differences in the coordination sphere of the ci~-[Ru"(C0)~] moiety are probably responsible for the CO frequency shifts on the two supports. We have not carried out l3CO substitution studies, but the relative intensities of the two bands are comparable and in agreement with the presence of a cis-[ Ru"(CO),] centre. I should like to emphasize that our F.t.i.r.results clearly demonstrate that ruthenium can be grafted to the surfaces of both supports. Whether the cis-[R~"(C0)~] moiety is linked to two surface OM groups as originally proposed or to four or more surface sites, as suggested by the recent EXAFS results for the silica-supported centres, cannot be determined. Furthermore, it is important to note that while the grafting reaction helps initially to disperse the ruthenium, the Ru'l sites are not the catalytic centres. The activeGeneral Discussion 257 hydrogenation catalyst on both supports is crystalline ruthenium, formed by hydrogen reduction of the grafted Ru” sites. Prof. M. Ichikawa (Hokkaido University, Sapporo, Japan) commented: I appreciate in this paper the remarkably higher selectivities towards alcohols in CO hydrogenation for catalysts derived from Ru clusters on pillared clays, in contrast to conventional Ru catalysts, which preferentially produce hydrocarbons. Concerning the chemical origin of alcohol promotion, we have previously proposed”’ a two-site CO activation model for Rh (Pt, Ir)-Fe’+ for promoting CO insertion into M-H and M-alkyl bonds, thus suppressing CO dissociation and resulting in the remarkable improvement of alcohol formation.I expect that some metal cations such as Mg’+, Na+ and Fe3’ contained in pillared-clay and basic-layered supports could be responsible for the alcohol promotion, due to the formation of local heteronuclear bimetal sites at the Ru particle interface. Is there any spectroscopic evidence for the surface coverage or local concentration of the promoter ions/oxides at the metal-support interface? I would suggest that i.r.data for CO chemisorption would offer insight into the surface decoration of Ru particles with these promoter ion~/oxides.~ 1 M. Ichikawa, A. J. Lang, D. F. Shriver and W. M. H. Sachtler, J. Am. Chem. SOC., 1985, 107, 7216; A. Fukuoka, M. Ichikawa, J. H. Hriljac and D. F. Shriver, Inorg. Chem., 1987, 26, 3643. 2 M. Ichikawa, Polyhedron, 1988, 7, (22/23), 2351. 3 M. Ichikawa and T. Fukushima, J. Phys. Chem., 1985, 89, 1564; W. M. H. Sachtler and M. Ichikawa, J. Phys. Chem., 1986, 90, 4752. Prof. Pinnavaia replied: Substantial oxygenate yields were observed only for Ru supported on the basic, non-pillared, LDH clay. Ru on acidic, alumina-pillared mont- morillonite exhibited high selectivity for branched hydrocarbons, but no oxygenates.It is possible that Mg’+ or A13+ from the LDH support participates in a two-site CO bridging mechanism similar to the one you have proposed for oxygenate formation over Rh supported on silica containing certain electropositive metal-ion promoters. We plan to investigate CO chemisorption on LDH-supported Ru under reaction conditions and to compare the oxygenate selectivity with Ru on conventional supports containing oxygenate promoters. Dr P. A. Sermon (Brunel University) regarding the paper by Prof. Baiker (communi- cated): You mention that one of the motivations for using metallic glasses in catalysis is that they are ‘ideally chemically homogeneous’. Analysis of Pd-Si samples shows that this was anything but the case; we assume the differences were predominantly caused by the surface of the ‘chill block’ on which the ribbons were quenched.Could you comment on the extent of surface impurity normally encountered in such samples. Prof. Baiker replied: I do not know which Pd-Si samples you are referring to. We have never used Pd-Si alloys. Generally, it is possible that the surface exposed to the rotating copper wheel becomes contaminated by traces of copper. If the alloys are not fabricated under vacuum or an inert gas atmosphere, the major surface impurity is oxygen leading to partial oxidation of the more electropositive component of the alloy. This surface oxidation is particularly severe with zirconium-containing alloys and can lead to complete coverage of the surface with a zirconium dioxide layer due to segregation of zirconium from the subsurface region to the surface.Dr M. E. Bridge ( Trinity College, Dublin) asked: Does Prof. Baiker have any comments on the surface dynamics of metallic glasses used as catalysts? In particular, does the surface topography or chemical composition change? Does one component of the glass aggregate? And does surface crystallization occur?258 General Discussion Prof. Baiker answered: Surface dynamics certainly play a very important role in catalysis on metallic glasses. Evidence for this emerges from investigations where the surface structure of the metallic glasses has been characterized by means of STM before and after exposure to reaction conditions.However, it should be stressed that direct observation of the surface dynamics under reaction conditions has not been accomplished so far. Thus what really can be said about surface dynamics stems from ex situ measurements which may not be absolutely representative, since part of the observed surface reconstruction may have been caused during sample transfer and exposure to air. However, there is evidence that the surface of these metastable solids undergoes dynamic changes more easily, since the driving force (change in free-energy) is generally larger than with equilibrated crystalline alloys. Hence, under reaction conditions, changes in the topography and chemical composi- tion of the surface are likely to occur already at temperatures considerably below the bulk crystallisation temperature.Prof. M. S. Spencer (University of Wales College of Cardig) said: The papers by Baiker and Baydal et al. lay emphasis on the catalytic importance of the interfacial regions in partially oxidised alloys and glasses. Reactions at interfaces in catalysts have been proposed before but they could well be more significant in these materials. In contrast to more conventional catalysts the metal particles are very small and embedded in oxidic matrices. How can these interfacial regions be more effective catalysts, sometimes by orders of magnitude? To take the Nd/Cu system as an example, two types of cause can be proposed:' (a) The properties of copper atoms bonded to oxide ions in the matrix would be expected to be sufficiently different from those of metallic copper.(b) The reaction of adsorbed CO and hydrogen with oxide ions at the interface could give a formate intermediate straddling the periphery. Hydrogenolysis of adsorbed formats appears to be the rate-limiting step in methanol synthesis under a wide range of conditiom2 Hydrogenolysis of a peripheral formate to give methanol should be more facile than the hydrogenolysis of formate adsorbed on either copper metal or an Nd oxide. 1 M. E. Fakley, J. R. Jennings and M. S. Spencer, J. Caral., 1989, 118, 483. 2 G. C. Chinchen, M. S. Spencer, K. C. Waugh and D. A. Whan, J. Chem. SOC., Faraday Trans. I , 1987, 83, 2193. Prof. Baiker responded: I appreciate your valuable comment and I agree completely with your statement regarding the importance of the metal-metal oxide interface in partially oxidised alloys and glasses. The Nd/Cu system you refer to is an interesting example illustrating the crucial role of the interfacial region; another example is shown in ref.(38) of this paper. Dr J. Dewing (Chester) remarked: In fig. 1 you compare activity data for a conven- tionally prepared catalyst with those of a catalyst derived from the metallic glass as Arrhenius plots. It appears that the slope of the plot for the metallic-glass-derived catalyst is ca. half that for the conventional catalyst. You also show in your paper that the texture of the metallic-glass-derived catalyst is much finer than that of the conventional catalyst although you provide no quantitative measurement of pore size distributions.If there were significant diffusional constraint within the pore structure of the metallic-glass-derived catalyst and the conventional catalyst were showing a true chemical kinetically controlled rate of reaction such a halving of the apparent activation energy is to be expected. Is it possible that the difference in apparent activation energy shown in fig. 1 is due to diffusional limitation and the true difference in activity for the two catalysts is even greater than shown in your paper?General Discussion 259 Prof. Baiker replied: I agree, it is tempting to conclude from the results presented in fig. 1 that the true difference in activity for the two catalysts is even greater than shown by the Arrhenius plots. In fact, this possibility cannot be ruled out, and therefore, tests with regard to the influence of intraparticle mass-transfer limitation (intraparticle diffusion) are presently being undertaken in our laboratory. In any case, the comparison of the activities shown in fig.1 is conservative, and the real difference of the activities of the two samples may be larger. Prof. B. E. Conway ( University of Ottawa, Ontario, Canada) commented: Prof. Baiker has referred in his interesting paper to aspects of the behaviour of glassy metals in various catalytic processes, including electrocatalysis. Certainly, in electrochemistry, these materials have attracted much interest in recent years, mainly on account of their resistance to corrosion (due to the absence of grain-boundaries), but also because of the possibilities of specific electrocatalysis originating from the unusual elemental composition tht can be generated in a single bulk phase ( cJ: table 1 in the author's paper).In examining glassy metal electrode materials, based on Ni-Mo-B-Fe and Ni-V-Ti compositions, we have found' little advantage over pure Ni or single-phase bulk Ni-Mo alloys (Mo at.% < 19) with regard to cathodic H2-evolution kinetics. Perhaps this is because this process is not what I would like to call a 'site-demanding' reaction since its rate-controlling desorption step (MH,,,+ H20 + e + M + H2 + OH-) involves only one metal (M) site per adsorbed H atom and hence, probably, per transition state. Glassy metals may be expected to be of more interest in electrocatalysis for 'site-demanding' reactions requiring, in the Balandin sense, ensembles of the surface- metal atoms, e.g.in some cathodic hydrogenations of cyclic unsaturated organic molecules [cf: ref. (2)] where the glassy metal surface would offer both geometrically unusual coordination arrangements as well as local surface compositions not attainable at bulk-phase alloy catalyst materials. In this regard, however, it is unfortunate that thermal crystallization of most glassy metal materials will not normally produce a corresponding well defined crystalline material of the same composition: glassy metals tend to crystallize into two- or multi-phase mixtures, the surfaces of which will not be expected to bear any relation in composition to that of the parent amorphous phase. Hence, direct comparisons between catalytic activities of glassy metals and corresponding single-phase crystalline alloys of the same composition, including single crystals, are only rarely feasible, with the exception perhaps of some limited-range compositions of binary glassy metals, e.g. as in the work of Smith et al.3 on Pd-Ge, quoted by Dr.Baiker. Another difficulty we have encountered in our own work on glassy metal cathode and anode materials, and which is probably quite general for gas-metal catalysis, is that the surface-region compositions, as determined by Auger electron spectroscopy, differ substantially from the corresponding bulk compositions, a factor that must be taken into account in characterization of relative catalytic activities of these materials, e.g. for a series of related bulk elemental compositions. Another point of current interest" is that amorphous metals of Ti-V compositions appear' to be good hosts for the sorption of H or D, providing alternative hosts to Pd for D concentration.1 Lu. VraCar and B. E. Conway, Electrochim. Acta, 1989 in press. 2 G. V. Smith, W. E. Brower, M. S. Matyjaszczyk and T. L. Pettit, in Proc. Int. Cong. Catal., ed T. Seiyama and K. Tanabe (Elsevier, New York, 1981), vol. A, pp. 355-363. 3 W. E. Brower, M. S. Matyjaszczyk, T. L. Pettit and G. V. Smith, Nature (London), 1983, 301, 497. 4 M. Fleischmann and S . Pons, J . EIecrroanaL Chem., 1989, 261, 301. Prof. Baiker, in response, said: I am grateful to Prof. Conway for his interesting comments with regard to the application of glassy metals in electrocatalysis.260 General Discussion Prof.G. C. Bond (Brunel University, Uxbridge) said: My comments are directed to the part of Prof. Baiker’s paper concerning Pd-Zr catalysts for CO oxidation. In interpreting the difference in activity between that derived from a Pd-Zr glass and that prepared by conventional impregnation, Prof. Baiker places emphasis on the greater metal-metal-oxide ‘interfacial area’ shown by the former, without actually showing that it is less in the latter. Assuming, however, that he is correct, one possible mechanism for explaining the effect might be the spillover of reactive intermediates from the Pd to the ZrO,; we demonstrated some years ago] that this occurred with Pd/SnO,, and our experience then would suggest that determination of the orders of reaction might provide a valuable clue to what is occurring.One of the limitations to the use of metallic glasses as precursors is the constraint imposed on varying the content of the active phase; this of course is readily accomplished when impregnation is used. This leads me to my specific question. I have always wondered whether metallic glasses represent an economically viable route to practical catalysts. I would expect production costs and the technical inconvenience involved in their use to render them unacceptable for large scale use, even as precursors to supported metals. What does Prof. Baiker think? 1 G. C. Bond, M. J. Fuller and L. R. Molloy, Proc. 6th Int. Congr. Catal., ed. G. C . Bond, P. B. Wells and F. C. Tompkins (The Chemical Society, London, 19771, vol.1, p. 356. Prof. Baiker replied: First I would like to stress that we have clear evidence from high-resolution electron microscopy [details will be reported in ref. (75) of our paper] that the Pd/Zr02 catalysts prepared from the amorphous Pd,Zr, precursor possess a much greater metal-metal-oxide interfacial area than conventionally prepared Pd/ZrO, . As to your suggestion that a spillover of reactive intermediates from Pd to ZrO, may be a possible mechanism for explaining the observed differences in activity, we have no experimental evidence for such a mechanism. The results from HREM, X.P.S. and TDS collected so far, strongly indicate that the different activities of the catalysts originate from different properties of the interfacial region.Concerning your statement that varying the content of the active phase is more difficult to accomplish with metallic glass precursors than when impregnation is used, I have a different view. Metallic glasses are metastable solids, far from being equilibrated. Thus the thermodynamical constraints with regard to composition are less stringent than with crystalline materials, i.e. the flexibility for the variation of the composition is much larger. Furthermore, the metallic glasses are ideally isotropic solids in which the constituents are molecularly mixed. This property should provide a better basis for the distribution of the active constituents than impregnation does. It is well known that the impregnation of a polycrystalline solid does not usually yield a homogeneous distribution of the immobilized species due to the surface heterogeneity of the carrier.Whether metallic glasses represent an economically viable route to practical catalysts cannot be conclusively answered yet. However, I agree that their large-scale use is rather improbable. In my view, their technical potential lies more in special applications, where the advantages these materials offer are predominant and where similar catalysts cannot be prepared by conventional techniques. Prof. A. K. Datye (University of New Mexico, Albuquerque, U.S.A.) asked: In view of the structure insensitivity of CO oxidation under your experimental conditions, it is surprising to find over an order-of-magnitude enhancement in the turnover frequency for the Pd/Zr02 catalyst derived from the amorphous precursor.Were the metal surface-areas measured before and after reaction, and could the differences in TOF arise due to potential errors in surface area measurement?General Discussion 26 1 The electron micrograph in plate 1 of your paper shows crystalline domains > 10 nm in size; however, it is not clear where the Pd is located. Could you indicate this or provide another micrograph that shows the Pd and possibly index the diffraction pattern showing the Pd and ZrO, reflections. How does this catalyst differ from one that has highly dispersed Pd on crystalline Zr02? Prof. Baiker replied: The palladium surface-areas were determined both before and after reaction using CO chemisorption as described in our ref. (56). No significant difference was found between surface-areas measured before and after reaction. There are several possible reasons that the activity of a structure insensitive reaction may change depending on the catalyst preparation ( e.g. metal-support interaction, spillover etc.). In our case, the difference in activity of the catalysts is attributed to a different metal-support interaction. We have recently shown by X.P.S. [ref. (75)] that in the catalyst exhibiting the large interfacial area between metal and oxide the chemical properties of the zirconia support are different, a significant quantity of zirconia exists as oxygen-deficient zirconia ( ZrOlPx). Whether the oxygen-deficient zirconia in the interfacial region is the reason for the enhanced activity is, however, not yet clear and needs further consideration. I agree that the electron micrograph in plate 1 does not clearly show where the Pd is located. A detailed study showing the differences in the structure of the catalyst derived from the metallic-glass precursor and the catalyst prepared by conventional impregnation will be shown elsewhere [ref. (75)].

ISSN:0301-7249    

DOI:10.1039/DC9898700253

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 263-273 Reactivity of CO on Stepped and Non-stepped Surfaces of Transit ion Met a1 s A. de Koster, A. P. J. Jansen and R. A. van Santen*T Laboratory for Inorganic Chemistry and Catalysis, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands J. J. C. Geerlings Koninklijke/Shell-Laboratorium Amsterdam (Shell Research b.v.), P.O. Box 3003, 1003 A A Amsterdam, The Netherlands Results of extended Huckel calculations show that coordination of CO to Group VIII transition-metal surfaces depends on a subtle balance of the interaction with the CO 5cr orbital, that tends to direct the CO molecule to the atop position and the CO 27~" orbital, that directs the molecule to higher coordination sites. In the atop position, the changes in bonding to different surfaces of the same metal can be mainly attributed to differences in the interaction with the CO SCT molecular orbital.The favoured dissociation path is such that carbon and oxygen atoms end in high coordination sites, sharing as few surface metal atoms as possible. The CO bond is activated by the metal atoms that are crossed upon dissociation. The main focus of this paper is the theoretical study of chemisorption and dissociation of CO on transition-metal surfaces. Chemisorption of CO has been extensively studied theoretically. Semiempirical'.' as well as ab initio quantum-chemical3 studies support the original proposal by Blyhol- der.4 The attractive interaction between CO and a transition-metal surface can be described as the sum of an electron-donating interaction between the doubly occupied 5 0 orbital of CO [the highest occupied molecular orbital (HOMO)] and empty metal orbitals, and the back-donating interaction between occupied metal orbitals and the empty 27" orbitals of CO [the lowest unoccupied molecular orbital (LUMO)]. More recent is the understanding of the repulsive interaction between CO and the transition metals.l(h),.5.6 Repulsive interactions derive from the interaction of overlapping doubly occupied orbitals. This Pauli repulsion energy is approximately proportional to the number of neighb~urs.~ However, repulsion is relieved if electrons can be removed from doubly occupied antibonding orbitals. This is often the case on transition-metal surfaces. 1 1.3.5.7 If the z axis is chosen perpendicular to the metal surface, bonding and antibonding orbital combinations are formed between the often highly occupied dz2 orbital of the metal surface atom bonded to CO and the doubly occupied 5u orbital.If the strength of the interaction is such that the antibonding orbital combination is pushed above the Fermi level, this orbital depletes electrons. As a result repulsion is reduced. We shall see later that differences in the details of these a-type interactions affect the relative stability of chemisorbed CO adsorption complexes significantly. They are an important contributing factor to the often found favoured bonding of CO to the atop position of transition-metal surfaces. f Also a t Koninklijke/Shell Laboratorium Amsterdam (Shell Research b.v.), P.O.BOX 3003, 1003 AA Amsterdam, The Netherlands. 263264 Stepped and No n -stepped Trans it ion - meta 1 surfaces Very little is known of the details of the CO dissociation path. Several studies are known based on LEPS-type potentials.* Recently important studies appeared by Shus- torovich.' It is well known'" that the back-donating interaction between the unoccupied CO 27r* orbital and the metal surface orbitals mainly determines the activation energy for dissociation, as the 2 7 " orbital is antibonding for the C-0 bond. This explains, for instance, the absence of a correlation between experimentally observed bond strengths and dissociation activation energy. CO binds more strongly to Pt than to Ni metal single surfaces," but the dissociation energy of CO on an Ni surface is much lower than it is on pt.1° An interesting question is raised by the work of Nieuwenhuys" and Banholzer13 on NO dissociation.Whereas Nieuwenhuys reported that Rh surfaces have the expected sequence in reactivity of Rh( 110) > Rh( 100) > Rh( 11 l ) , Banholzer found a different sequence on Pt. Similar differences are found in our work enabling an analysis of the cause of this difference in behaviour. We will see that it is dominated by subtle changes in the interaction of the 5 0 orbitals and metal-surface dZz orbitals with electron occupa- tion. This confirms earlier suggestions' based on estimates of changes in the dz2 LDOS at the Fermi level. We shall present results of quantum-chemical calculations using the ASEDI4 version of the extended Hiickel method.15 Clusters simulating the (1 1 I), (100) and ( I 10) surfaces of an f.c.c.metal will be discussed. Also the stepped Rh( 11 1) surfaces will be studied. The main limitation of the ASED method is the dependence on parameter choice and the approximate way that the repulsive energy contributions are calculated. This will be discussed. Nevertheless, the large number of calculations this method allows and the large metal clusters that can be used are the attractive features of the method. Theoretical Method and Cluster Models Used In the atom superposition and electron delocalization (ASED) molecular orbital methodi4 the total energy, Etot, is calculated by the summation of an attractive and a repulsive energy: 6 0 , = E a t , + Erep * (1) Eatt is found by applying extended Hiickel molecular orbital technique^.'^ Parameters used are listed in the Appendix.The repulsive energy, Erep, is an approximate expression derived from an analysis of corrections due to the electron-electron interaction^.'^ In table 2 (later) E,,, and Eatt are listed separately. Electronic effects of adsorption and dissociation are analysed by calculating the local density of states (LDOS): where Qi is the fragment orbital and qA is the calculated molecular orbital. Bond-order overlap population densities are calculated using the equation TI,( E ) = 1 c y c ; S l , G ( E - E , ) (3) I, where c: is the coefficient of fragment orbital i in molecular orbital k and S,, is the overlap between the fragment orbitals.In all plots, the calculated p I ( E ) and n,,( E ) are presented after convolution with a Gaussian distribution (a = 0.25 eV). As a model for an Rh(ll1) surface we used a cluster of 29 Rh atoms in two layers of 18 and 11 Rh atoms, respectively, which can be denoted as an (18, 11) cluster. Atoms in the centre of this cluster are fully coordinated by six Rh atoms in the same layer and three Rh atoms in the second layer. All adsorption and dissociation studies involveA. de Koster et a]. 265 those central atoms. We used the bulk value of 2.687 A as closest intermolecular Rh distance. A stepped Rh(ll1) is modelled by adding a third layer of six Rh atoms on top of the first layer, yielding a (6, 18, 11) cluster. According to the same general principles, an Rh( 100) surface is modelled by a (25, 16) two-layer cluster and an Rh( 110) surface by a (19, 14,9, 8) four-layer cluster.Several reaction paths for the dissociation of CO were studied. In each reaction path, the carbon atom was fixed on a certain site while the oxygen atom was moved according to the assumed reaction path. On several intermediate steps (usually 10-15 intermediate steps) the bonding energy was optimized by variation of the height of the oxygen atom in steps of 0.1 A. Theoretical Results General Features Donation of electrons from the 5u HOMO of CO to the metal, and back-donation of metal electrons into the 277" LUMO of CO are the dominating attractive interactions between the valence electrons of the metal surface and the adsorbate. In fig. 1 are shown the local density of states (LDOS) of the 1 T , 5 0 and 2n" CO orbitals before and after adsorption.Whereas the I T level only broadens, the average position of the 5cr and 27r" levels shifts. The downward shift of the maximum of the LDOS of 5 0 agrees with its bonding character and the upward shift of the maximum of the 27r" LDOS agrees with its antibonding character. The ratio of the upper and lower resonances of the 27r* LDOS decreases comparing atop with two-fold and three-fold positions. As fig. 1 shows, it decreases further if CO is chemisorbed next to a step. This agrees with increasing covalent interaction. In the limit of strong interaction the ratio of the two peaks becomes unity and their energy difference increases. Clearly the bonding interaction with the 277' increases with coordination number and adsorption to steps has a very favourable interaction.This can also be seen from the 2n" electron occupancy [table l ( n ) ] . The hi her the 27~' orbital occupancy of CO the lower its infrared stretching frequency.'" ' Thus the infrared stretching frequencies at higher coordination sites are predicted to decrease, with very low values to be found at steps. This prediction agrees with experimental o b ~ e r v a t i o n s . ' ~ ~ ' ~ Fig. 2 ( a ) shows the LDOS of the metal atomic orbitals involved in bonding. The metal d,- orbitals are found to become upwardly displaced. They participate in antibond- ing orbitals mainly by interaction with the CO 5 0 orbital. d,, and d,.: become bonding orbitals by interaction with the CO 27r* orbital.The interaction between two doubly occupied orbitals is repulsive. The interaction of CO 5cr and metal d,. orbitals is also repulsive as long as the Fermi-level position is such that the d-valence electron orbitals have a high occupancy and antibonding orbital fragments are occupied. This appears to be the case for the three-fold and two-fold positions, but not for the atop position [table 1, fig. 2(6)]. The depletion of the antibonding dz2 orbital in the atop position implies a much more favourable interaction with the 5rr CO orbital in the atop position than in the higher coordination sites. Comparison of the bond-order overlap population densities of the u symmetry orbital fragments involving d-valence orbitals [fig. 2( h ) ] shows that in the atop configuration the bond character remains bonding at higher d-valence band occupation than in bridge-coordination sites.On the other hand a similar comparison of the bond-order population densities of 7r symmetry involving the 2n* orbitals of CO shows that the chemical bond with the 277" orbital of CO remains bonding at the higher d-valence266 A Stepped a n d No n -stepped Tra n s i t io n - m eta 1 s u rfa ces I EF B -30 ( d ) 1- -20 -10 energy/eV C I Et 30 -20 -10 0 energy/eV El- Fig. 1. LDOS of the l x (A), 5u ( B ) and 2x" (C) CO molecular orbitals ( a ) in the gas phase, ( b ) one-fold adsorbed on Rh( 11 l ) , ( c ) three-fold adsorbed on Rh( 11 1 ) and ( d ) three-fold adsorbed on stepped R h ( l l 1 ) (distance of CO to step: 1.551 A).The Fermi level is indicated by E , .A. de Koster et al. 267 Table 1. ( a ) Gross population Hiickel Rh( 11 1) atop orbital adsorbate 111 100 110 bridge step 17T ads. 0.998 0.997 0.998 0.996 0.93 free 1 .o 1 .o 1 .o 1 .o 1 .o free 1 .o 1 .o 1 .o 1 .o 1 .o free 0.0 0.0 0.0 0.0 0.0 5 0 ads. 0.884 0.882 0.895 0.873 0.84 2 2 ads. 0.14 0.12 0.13 0.26 0.37 ( h ) Surface 4d,-. atomic orbital gross populations -~ ~ ~~ ~~ bridge atop symmetry 111 100 110 ff T after adsorption 0.67 0.65 0.67 0.86 0.9 1 before adsorption 0.92 0.9 1 0.88 0.93 0.92 ( c ) Bond-order overlap population densities between CO and surface group orbitals bridge atop symmetry 111 100 110 (T T d , 2 - 5 ~ 1 . 2 2 ~ l o - ' 1 . 2 4 ~ lo-' 1 . 1 7 ~ lo-' - 0 . 2 9 ~ l o - ' d,? - 27r" 0.42 x 10- I d,; - 2 ~ " 9.1 1 x lo-' 8.5 x lo-' 9.3 x lo-' 0.26 x 10- I d,, - 5~ 0 .6 ~ l o - ' band occupation for bridge coordination [fig. 2( c)]. At low electron band occupation the bonding character in the atop position dominates. The two effects together imply that if the d-valence electron occupancy increases, the interaction with the 5a orbital tends to favour the atop position and that the interaction with the 27r* orbital becomes favoured on the higher coordination sites. These trends relate to changes in LDOS distributions around the Fermi level." ' ' L ' . ~ As shown in table l ( c ) and fig. 2(6) and (c) in bridge coordination a significant bonding interaction is present between d,, and 5u orbitals increasing at lower valence- electron occupation, as well as the d_Z and 2 7 ~ * orbitals, reaching a maximum at higher electron occupation are present.In fig. 3 the resulting changes in total bond strength are plotted as a function of the number of electrons per metal atom. All other parameters are equal to the parameters used in the Rh calculation. One observes a decrease in bond strength as the d-valence electron band becomes filled. This is due to electron filling of antibonding orbital fragments. Secondly, whereas with the parameters used the three-fold position is always found to be favoured, the difference between atop and three-fold energies is least for Tc. This behaviour is completely in line with the ideas described earlier. For elements left of Tc in the periodic table, the 5u orbital interaction favours high coordination.For elements right of Tc, the interaction with the 27~" orbitals268 Stepped and Non-stepped Transition-metal surfaces B I A n I Et IE' I I I I I I I 0 l ~ ~ ~ ' l ' " ' l ' ' ' ' -30 -20 -10 -30 -20 - 10 0 energy/eV energy/eV Fig. 2. (A) LDOS of dz2 before ( a ) and after ( h ) one-fold adsorption of CO on Rh( 11 1). (B) Bond-order overlap population densities of CO on Rh( 11 1): ( a ) CO one-fold, d_Z - 5a; ( 6 ) CO two-fold dz2 - 5a; (c) CO two-fold, d,. - 5u. (C) Bond-order overlap population densities of CO on Rh( 11 1 ) : ( a ) CO one-fold, d,, - 27r"; ( 6 ) CO two-fold, d,; - 2 ~ * ; ( c ) CO two-fold, dz2 - 27r*. In all plots the Fermi level is indicated by E , .A. de Koster et - m mT; \ Y 4 1.1-- 1.0- 0 . 9 al. --I 4.0 c >, ‘- -v; : - 4* = - 3.0 ___~_____~_ Nb T c Rh -6 - 4 -2 0 2 I ’ l l ) “ I I f “ ” I I I ” 2.5 269 \ 1 1 3.5 Fig. 3.Attractive adsorption energy, E , , , , of one-fold adsorbed CO as a function of the occupation of the metal valence electron band ( - - - ) and the ratio of E , , , of CO adsorbed three-fold to CO adsorbed one-fold as a function of the occupation of the metal valence electron band (-1. The elements correspond to the total number of valence electrons according to the periodic system. dominates and favours the three-fold coordination site. The trend observed agrees well with experimental observation.6 Note that no repulsion effects are included in the energy terms considered, other than those due to the overlap between antibonding valence electrons.Elsewhere the total energies for chemisorption of CO and its dissociation products calculated according to the ASED method on Rh were presented.”’” In general one finds the same trends as expected on the basis of attractive energy considerations only. Adsorption to steps is an exception to this rule. In agreement with earlier results by Anderson,lX we find that the total interaction at steps with adsorbates is less than on the non-stepped surface, not withstanding the larger attractive energies found. This indicates that the ASED method has to be applied with care because experimental evidence indicates a favourable interaction with steps.’6 Face Dependence In fig. 4 the ratios of the attractive contributions to the bond energy for CO adsorbed atop to the ( 1 1 l ) , (100) and ( 1 10) faces of Rh are presented as a function of d-valence electron occupation.Only at medium-to-low band filling does one find the expected result that bonding is strongest to the most open surface. The observed change in sequence at high d-valence electron occupation is very interesting. From a comparison of the attractive bond strengths one finds for a metal containing nine electrons in the valence electron band that the (100) surface becomes most reactive, and both the (1 11) and ( 1 10) surfaces have lower reactivities. Analysing the (T- and n-orbital overlap population densities [fig. 2 ( h ) and ( c ) and table l ( c ) ] one observes that the differences in behaviour arise from the changes in interaction of the 27r” and 5 0 CO molecular orbitals with the d-metal atomic orbitals.Whereas the relative changes in bonding with the 50 orbital of CO change less,270 Stepped and No n -stepped Transition - me ta 1 surfaces t 1 . 0 + c i 0 . 9 -f r 1 t Nb Tc Rh Fig. 4. The ratios of E,,, of CO adsorbed one-fold on Rh(100) (-) and Rh( 110) ( - - - ) to CO adsorbed one-fold on Rh( 11 1) as a function of the occupation of the metal valence electron band. comparison with the differences in bond strength shown in fig. 4 for Rh show that the absolute values of these changes dominate the bond energy. The differences in bonding with the d=2 surface orbital can be rationalized on the basis of the changes in relative LDOS at the EF level.2'"'.5.h As can be seen from fig. 5 for Rh the LDOS at the Fermi level is highest at the (100) face and close for the (111) and (1 10) faces.At lower valence band occupations the LDOS at the (1 10) face increases and dominates. At higher valence-band occupation the LDOS at the (111) face domi- nates. These results are in line with the strong dependence of the orbital interaction energies on the d-valence electron occupation. This occupation not only depends on the total number of valence electrons, but also on the distribution of the electrons over the s, p and d valence-electron sub-bands (0.2,O.O and 8.8, respectively, for the particular Rh clusters studied). Therefore for a specific metal the parameters defining this relative distribution become important. In the present case the s and p band occupation is low, so that the stronger interaction at the (110) face may occur at a higher total valence- electron occupation if the s, p valence electron band contains more electrons. This is the situation that prevails on transition-metal surfaces.I t is of interest to note that changes in surface reactivity as predicted by the EHMO calculations have been reported by Niewenhuys12 and Banholzer13 for chemisorption of NO. Banholzer et al. explained the higher reactivity of the (100) surface found for Pt on the basis of the Bond surface d-dangling bond model." Application of this model to Rh results in similar predictions. Here we find that changes in electron occupation of the proper symmetry orbitals may explain the differences in behaviour of the Rh and Pt surfaces, implying that the Bond model requires some modification.CO Dissociation Table 2 gives the lowest activation energies for dissociation. These data are the result of calculations of many different reaction paths. In table 2 the attractive contribution to the bond energies and total energy contributions are separately listed.A. de Koster et al. 27 1 J I ' J ( c ) I ' ' ' ' I ' " I ' \ ' ' ' ' 1 -30 -20 -10 0 energy/eV Fig. 5. LDOS of d_2 of the Rh( 11 1) ( a ) , Rh( 100) ( b ) and Rh( 110) ( c ) faces. E , indicates the Fermi level. Table 2. Dissociation of CO on Rh single-crystal planes surface E,,, Eta E,,, 2T* Eat, population" Rh( 11 1) 1.98 1.98 0.55 Rh( 100) 1.48 1.43 0.65 Rh( 110) 1.90 1.53 0.56 '' 2rr* orbital gross population at the transition state. Dissociation on a non-stepped and a stepped (111) surface has been studied, as well as dissociation reaction paths at the (100) and (1 10) surfaces.The minimum activation energy is lowest on the (100) face and highest on the (111) surface. No overall decrease in activation energy caused by the presence of a step is found.''h' However, if one does not consider the dissociation energy with respect to chemisorbed CO, but with respect to gas-phase CO according to the changes in Eatt, a lowering of the overall activation energy for dissociation is found at the steps. From the many reaction paths considered, we have found a clear pattern for the lowest activation energy. Dissociation is favoured with the dissociated atoms ending in high coordination sites and sharing the least number of surface metal atoms.Secondly, for the dense faces crossing of the CO bond over a surface metal atom, when it stretches to dissociate, is favoured considerably.272 Stepped and Non-stepped Transition-metal surfaces The higher the CO antibonding 27r" molecular-orbital electron occupation becomes (table 2), the more the activation energy for dissociation is lowered. This observation explains why the activation energy for CO dissociation behaves parallel to the bond energy of CO in the atop position. As we discussed earlier the latter differences are also determined by differences in 27~" occupation. The correlation between 27~" occupancy and activation energy for dissociation agrees with similar observations based on UPS measurements. It also agrees with the promoting action of work-function- lowering coadsorbents such as alkali metals.2.1",30 Discussion The results of the extended Hiickel calculations presented here demonstrate that coordi- nation of CO to Group VIII transition-metal surfaces is determined by a subtle balance of the interaction with the CO 5v orbital, that tends to direct the CO molecule to the atop position, and the CO 27~" orbital, that directs the molecule to higher coordination sites.Interaction of the s-valence electrons with the CO 27~" is only possible in bridging coordination, and high d-valence electron band occupation also favours high coordina- tion sites. If the interaction with d-valence electrons is small, as with Ni".'' and Pd,33 experiments show that bridge coordination is favoured. Depletion of the d-valence electron band enhances the LDOS around the EF in the atop position, resulting in the favoured atop position of CO coordinated to CO,'~ Rh'5 and Ru." Since the work function of Pt is significantly larger than those of Pd and Ni, the relative contribution of 27~" back-donation is less and not only favours atop coordination on Ir, but also on Pt.It is found that in the atop position, the interaction with the 5a orbital determines differences in bond strength. Changes in CO 27~" electron density correlate with the minimum activation energy of dissociation. The favoured dissociation path is such that the carbon and oxygen atom end in high coordination sites, sharing as few surface metal atoms as possible and with the C-0 bond being activated by the metal atoms that are passed upon dissociation. Comparison of the activation energies for dissociation paths of lowest energy determined using the ASED method and that computed using Shustorovich BOC formulae' shows the predic- ted dissociation energies to be significantly lower, probably because of the extra stabiliz- ation of the 2i7" orbital of the dissociating molecule by the metal atom that is crossed.Appendix Atomic parameters: principal quantum number ( n ) , ionization potential (VSIP), orbital exponents (0 and respective coefficients (C,) (d only) used. S P atom n VSIP 6 n VSIP s C 2 20.00 1.658 2 11.26 1.618 0 2 28.48 2.246 2 13.62 2.227 Rh 5 8.09 2.135 5 4.5 7 2.100 n VSIP 5 Rh 4 12.50 4.290 0.5807 0.5685 1.97A. de Koster et al. 273 References 1 ( a ) S.Sung a n d R. Hoffmann, J. Am. Chem. Soc., 1984, 107, 2006; ( h ) C h . Zeng, Y. Apeloig and R. Hoffmann, J. Am. Chem. Soc., 1988, 110, 749; ( c ) N. K. Ray and A. B. Anderson, Surf Sci., 1982, 119, 35; ( d ) W. Andreoni a n d C. M. Varma, Phjx Rez:. B., 1981, 25, 437; ( e ) G. Doyen and G. Ertl, Surf Sci., 1977, 69, 157; ( f ) S. P. Mehandru and A. B. Anderson, Surf Sci., 1988, 201, 345. 2 ( a ) R. A. van Santen, Proc. 8th Int. Conf Cutal. (Verlag Chemie, Dechema, Frankfurt am Main, 19841, vol. IV, p. 97; R. A. van Santen, J. Phys. C, 1982, 15, L513; R. A. van Santen, J. Chem. Soc., Faraday Trans. 1 , 1987, 83, 1915; ( b ) A. d e Koster a n d R. A. van Santen, J. Vac. Sci. Technol., 1988, A6, 1128. 3 ( a ) D. Post a n d E. J. Baerends, J. Chem. PIij,.s., 1983, 78, 5663; ( h ) F.Raatz a n d D. R. Salahub, Surf: Sci., 1986, 176,219; ( c ) K. Hermann, P. S. Bagus and C . Nelin, Phys. Reo. B, 1987,35.9467; K. Hermann, P. S. Bagus and C. W. Bauschlicher, P h j ~ ReLi. B, 1985,31, 6371; ( d ) J. N. Allison and W. A. Coddard, Surf Sci., 1981, 110, L615. 4 G. Blyholder, J. Phj-s. Chem., 1964, 68, 2773. 5 E. J . Baerends and A. Roosendaal, in Quantum C'hemistrj-: The C'hallenge of' Transition Metals and Coordination Chemistrj., ed. A. Veillard, NATO AS1 Ser. 159. 6 R. A. van Santen, J. Mol. Strucf., 1988, 173, 157. 7 R. A. van Santen a n d E. J. Baerends, in Theoretical Models of' Chemical Bonding, ed. Z. B. Maksic (Springer-Verlag, Berlin, in press), part 4. 8 Ch-Y. Lee a n d A. E. d e Pristo, J . Chem. Phj,s., 1986, 85, 161. 9 E. Shustorovich, Suyf Sci. Rep., 1980, 6, 1; E. Shustorovich, Ace. Chem. Res., 1988, 21, 183. 10 ( a 1 G. Broden, T. N . Rhodin, C. F. Brucker, R. Benbow and Z. Hurych, Surf: Sci.. 1976, 59, 593; ( h 1 R. W. Joyner, Surf: Sci., 1977, 63, 291. 11 E. L. Muetterties, T. N . Rhodin, E. Band, C. F. Brucker a n d W. R. Pretzer, Chem. Rec., 1979, 79, 91. 12 H. A. C. M. Hendricks, A. P. J. M. Jongenelis and B. E. Niewenhuys, Sur$ Sci., 1985, 154, 503; H. A. C. M. Hendricks and B. E. Nieuwenhuys, Sut$ Sci.. 1986, 175, 185. 13 W. F. Banholzer, P. 0. Park, K. M. Mak and R. I. Masel, Surf: Sci., 1983, 128, 176; P. 0. Park, W. F. Banholzer and R. I . Masel, Surf: Sci., 1983, 119, 145; P. 0. Park, W. F. Banholzer and R. I. Masel, Surf.' Sci., 1985, 155, 341; 653. 14 A. B. Anderson, J. Chem. Plijx, 1975, 62, 1187. 15 R. Hoffmann, J. Chem. PIiJ-s., 1983, 39, 1397. 16 W. Erley. H. Ibach, S. Lehwald and H. Wagner, Surf.' Sci., 1979, 83, 585. 17 J. C. Bertolini and B. Tardy, Surf: Sci., 1981, 102, 131. 18 A. B. Anderson, R. W. Grimes and S. Y. Hong, J. Ph?is. Chem., 1987, 91, 4245. 19 G. C. Bond, Discuss. Faraduj, Soc., 1966, 41, 200. 20 J. K. Norskov, S. Holloway and N. 11. Wang, Sutf Sci., 1984, 137, 65; E. Wimmer, C. L. Fu and A. J . Freeman, Phys. Rev. Lett., 1985, 55, 2618. 21 S. Anderson, Solid State C'ommun., 1976, 20, 229. 22 W. Erley, bl. Wanger a n d H. Ibach, Surf: Sci., 1979, 80, 612. 23 A. Bradshaw and F. Hoffman, Surf Sci., 1978, 72, 513. 24 C. Backx, personal communication. 25 L. H. Dubois and G. A. Somorjai, Surf. Scr., 1980, 91, 514. Paper 8/04994G; Receiced 14th December, 1988

ISSN:0301-7249    

DOI:10.1039/DC9898700263

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 275-289 A Comparative Study of the Behaviour of Single-crystal Surfaces and Supported Catalysts in NO Reduction and CO Oxidation over Pt-Rh Alloys Ronald M. Wolf, Jacobus Siera, Falco C. M. J. M. van Delft and Bernard E. Nieuwenhuys" Gorlaeus Laboratories, Leiden Uniuersity, P. 0. Box 9502, 2300 RA Leiden, The Netherlands The use of Pt-Rh-based three-way catalysts for automotive exhaust-gas control stimulated us to study various Pt-Rh alloy surfaces: well defined single-crystal surfaces and Si0,-supported catalysts. For the sake of com- parison data were also obtained for pure Pt and Rh single-crystal surfaces and catalysts. The surface composition and chemical properties have been studied using AES, FEM, TDS and XPS. The surface composition was studied under various experimental conditions, both in vacuum and in the presence of the relevant gases, NO, 02, CO and H?.We discuss the effects of the surface structure and the surface composition on the chemical proper- ties of the Pt-Rh alloys. The surface composition varies easily with changing experimental conditions (temperature and gas-phase composition). Clean Pt-Rh alloy surfaces are enriched with Pt after high-temperature annealing ( T S 1000 K). However, the surface composition of Pt-rich Pt-Rh alloys is almost bulk-like following low-temperature equilibration ( T -- 800 K). Adsorbates can easily induce segregation of Rh or Pt to the surface, and the chemical properties of the surfaces are changed accordingly. In par- ticular, the NO dissociation, which is often mentioned as the first step in the NO reduction with CO or hydrogen, is extremely sensitive to both the structure and the composition of the surface.For the pure metals the influence of the surface structure is different for Pt and Rh. The NO reduction and CO oxidation reactions were studied at low pressures on the single- crystal surfaces and at atmospheric pressure on the supported catalysts. We show that the chemical properties of the supported catalysts can be under- stood on the basis of the single-crystal results. 1. Introduction The first studies of the surface properties of Pt-Rh alloys were related to their industrial application: Pt-Rh gauzes have been used for many years as catalysts for the production of nitric acid via the catalytic oxidation of ammonia and the HCN synthesis.It has been known for more than half a century that the addition of rhodium to a platinum catalyst results in a drastic reduction of the Pt loss combined with an improvement of the reaction selectivity and activity.' Much fundamental research has been devoted to the characterization of these Pt-Rh catalysts in general.'-s By far the dominant and still growing usage of both Pt and Rh is as an automotive catalyst. In 1987 automotive catalysts already accounted for more than 35% of the world total Pt consumption and for 73% of the Rh consumption.6 The present generation of converters is based on bimetallic Pt-Rh catalyst^."^ The overall catalytic reactions that are important for control of the exhaust emission are given by the following equations: 2 c o + o , + 2 c 0 , ( 1 ) 275276 NO Reduction and CO Oxidation on Pt-Rh Alloys Table 1.Relevant data for Pt, Rh and Pt-Rh Pt Rh Pt- Rh ref. 32 1.39 1.34 - - 33 34 - s 0 35 AH:uhl 29x/kJ mol-’ 565 557 - r / 8, @ , W K 80“ 160“ - H,,,,/kJ mol-‘ - - W b ) / K 234 3 50 I’ (1 11) surface. 1.01 T T I K Fig. 1. The temperature dependence of the platinum surface concentration of ( a ) polycrystalline Pto h2-Rho 38 76 (A) and ( b ) Pto 25-Rho ,5( 100)2J (a). The full lines show the surface concentration calculated on the basis of the surface phonon softening model.-’‘ ‘HC’ + x 0 2 - yH,O + z C 0 , (2) 2 N 0 + 2CO - N-, + 2C02 ( 3 ) 2N0+2H, + N,+2H-,O. (4) Platinum is an effective catalyst for the oxidation of CO and ‘HC’. However, for the reduction of nitric oxide this metal is less effective.”’ One undesired reaction product is ammonia formed according to the equation Unfortunately, the selectivity of Pt to promote the NO reduction to N2 rather than to NH3 is poor.Rhodium is the essential ingredient in the three-way catalyst for the conversion of NO into N2. Despite the widespread use of bimetallic Pt-Rh catalysts, little is known about the performance of Pt-Rh alloy catalysts in the relevant reactions. This contrasts with the monometallic Pt and Rh catalysts, about which several investiga- tions have been reported in the literature.’-” The properties of both supported and well defined single-crystal surfaces have been studied. As a result the mechanisms of the relevant reactions and the specific differences between Pt and Rh in these reactions are reasonably well understood.However, there are strong indications that Pt and Rh may form alloy particles in the three-way catalyst.” Alloy catalysts often exhibit catalytic properties (e.g. selectivity, activity, stability or poison resistance) different from those of the constituent 2N0+5Hz - 2NH3+2H,0. ( 5 )R. M. Worf et al. 277 In view of the enormous importance of bimetallic Pt-Rh catalysts we have investigated in our laboratory the properties of several Pt-Rh alloy single-crystal and supported Pt-Rh alloys The major purpose of this paper is to review our recent studies concerning Pt-Rh alloy surfaces in relation to the processes taking place on the three-way catalyst surface. 2. Surface Composition Pt and Rh are completely miscible at all concentrations with a slightly negative (exo- thermic) enthalpy of mixing.Ordered phases have not been reported, although some studies may indicate the presence of some short-range According to many models the principle factors affecting surface segregation of an alloy AB are differences in bond strengths (A-A,A-B and B-B) and atomic Table 1 shows some relevant parameters for Pt-Rh. It shows that the differences in heats of sublimitation are very small (1-2'/0 difference), Pt and Rh are nearly of the same size (Rh is ca. 3% smaller than Pt) and that the enthalpy of mixing is only slightly negative. The surface composition was calculated using a Monte Carlo method.36 It appears that the computed surface composition is almost equal to the bulk composition due to the almost equal values of the sublimation enthalpies and the atomic sizes of Pt and Rh in combination with the almost zero enthalpy of mixing.Contrary to the predictions based on model calculations, most of the experimental data point to a large Pt surface enrichment. 17,22-28~37-41 The temperature dependence of the surface composition for the (100) face of a Pt0.2s-Rh0.7s single crystal and for a polycrystalline Pt0.62-Rh0,38 foil is shown in fig. 1. Similar results were obtained for the (410) surface of the same single Pt surface segregation is not confined to the close-packed surfaces, since FEM observations of Pt-Rh alloys suggest a Pt-like behaviour towards gases like nitrogen and carbon monoxide of the rough surfaces of a Pto.5-Rho.5 alloy and a strong Pt surface enrichment for Pto,12-Rh0.88 above 1500 K." The apparent work function of the clean Pto.12- Rho.,, tip increases with increasing annealing temperature. This observation suggests again that the amount of surface Pt increases with increasing annealing tem- perature, since the work function of Pt is ca. 0.5 eV higher than that of an Rh field emitter. This effect is even better illustrated by the adsorption of nitrogen. The work function change produced by molecularly adsorbed nitrogen at 80 K was measured as a function of the temperature at which the tip was annealed prior to exposure to N2. Following annealing above 1200 K the work function change is equal to that found on pure Pt. The work function change found after annealing at 600-700 K is between the values found for pure Rh and pure Pt.It suggests that the Pt surface concentration of atomically rough surfaces, as found on field emitters, increases with increasing annealing temperature. The largest change in surface composition occurs upon annealing around 1000 K. Summarizing, we can conclude that the experimental data suggest that clean Pt-Rh alloy surfaces, both atomically rough and smooth surfaces, show a pronounced Pt surface enrichment that increases with increasing annealing temperature. The Pt surface segrega- tion cannot be understood with the conventional surface segregation models. The small negative (exothermic) enthalpy of mixing cannot be responsible for the Pt surface enrichment, because Pt surface segregation has been observed for both Pt-rich and Rh-rich samples.A possible model that can explain the Pt surface segregation has been put forward by Van Langeveld and Niemant~verdriet.~' They noticed that the surface Debye tem- perature of Pt is extremely low compared to that of Rh and compared to the bulk Debye278 NO Reduction and CO Oxidation on Pt-Rh Alloys temperatures. The surface and bulk Debye temperature represent the vibrational proper- ties of the atoms on the surface and in the bulk, respectively. The values are included in table 1. In the surface segregation models discussed so far the effect of lattice vibrations on the surface composition was neglected. H o ~ h i n o ~ ~ showed that the surface phonon softening (i.e. the Debye temperature being lower for the surface than the bulk) may affect the surface composition of alloys.The reason is that surface segregation of the component with low surface Debye temperature leads to a larger entropy correspond- ing to the surface phonon softening. It should be noted that reliable values of the surface Debye temperature of Pt and Rh surfaces are not available and that the conventionally used method for determination of the surface Debye temperature is essentially not correct.34 However, qualitatively the Pt surface segregation can be correctly predicted by the surface phonon softening model as is shown in fig. 1.26,3’ It should be pointed out that the pronounced Pt surface segregation is not a definite proof that the surface phonon softening is the decisive factor.An interesting alternative model that may explain the Pt surface segregation has been put forward for the explanation of Pt surface segregation observed for Pt-Ni alloys. This alloy is another example of the few alloy systems whose surface segregation behaviour cannot be understood with the conventional surface segregation models. The models predict an Ni surface enrichment, whereas several studies confirmed a pronounced Pt surface enrichment for the ( 1 11) surface of Pt-Ni alloys.43344 Spencer4‘ suggested that the discrepancy between the calculated and the experimentally determined surface composi- tion may be due to neglect of changes of bond energies due to surface relaxation. According to De Temmerman et al.46 the Pt surface enrichment is driven by a strengthen- ing of the Pt-Pt bonds in the surface region.This model is consistent with the adsorption behaviour of the Pt-Ni( 1 1 1 ) surfaces. It has been found that the heat of adsorption of several gases is significantly lower on the Pt-Ni( 1 1 1) surface than on the (1 1 1 ) surfaces of pure Ni and Pt.43744 As we shall see, such a weakening of the adsorption bond does not occur on Pt-Rh surfaces. Moreover, the temperature dependence of the surface composition of Pt-Rh alloys is completely in line with the earlier mentioned model based on surface phonon softening. Evidently, the surface phonon softening model predicts that the Pt surface concentration becomes larger with rising temperature because of the increasing importance of the entropy contributions to the free energy at higher temperature.At lower temperatures the increasing importance of the enthalpy contribu- tions results in a lowering of the Pt surface concentration, as has indeed been observed. If the driving force for Pt surface segregation would be dominated by enthalpy contribu- tions the surface excess of Pt must decrease with increasing temperature, whereas the opposite effect is observed. The peculiar surface segregation behaviour of Pt-Rh alloys is due to the very small differences in the binding-energy parameters (Em-,, , ERh-Rh and EPt-Rh). As a consequence, the surface composition is extremely sensitive to the presence of adsorbate atoms. Any contaminant that has a slight preference for either Pt or Rh may induce surface segregation.It is known that the presence of a gaseous environment can modify the intrinsic surface segregation tendency of alloys.’”’’ In order to investigate the gas-induced surface segregation in more detail, several Pt- Rh alloy single-crystal surfaces were exposed to flows of several gases while the temperature was varied.’3 The surface analytical technique used was AES. First the temperature was increased in steps from 300 to 1200 K and then the temperature was decreased again stepwise. For the interpreta- tion of the AES results it is convenient to define the parameter 8: I ( Pt 64 eV) I ( Rh 302 eV) I(Pt 64 eV) I (Rh 302 eV) S = (in vacuum) - (in gaseous environment) with I the intensity of the relevant surface sensitive Auger signal. Several processes may contribute to changes in 8.For Pt-Rh S is positive in the case of a gas-induced RhR. M. Wolf et al. 0.2- 0.1 - 0 279 I 1 1 I I U 1 6 -0.2 -o*ll * T/K LOO 800 1200 T I K 6 -o.21 + I Fig. 2. AES results obtained for Pt-Rh(410) in a flow of 5 x lo-’ mbar O2 as a function of temperature for ( a ) an Rh-rich surface and ( b ) a Pt-rich surface. Auger 0 signal intensity as a function of temperature for ( c ) an Rh-rich surface and ( d ) a Pt-rich surface.’3 surface segregation or a selective adsorption on Pt and negative in the case of a gas-induced Pt surface segregation or a selective adsorption on Rh. Some of the results are shown in fig. 2 for Pt-Rh(410) in an oxygen atmosphere of 5 x lo-’ mbar. It can be concluded from the figure that at low temperatures oxygen is selectively adsorbed on the Rh sites, while heating in the oxygen atmosphere results in oxygen-induced Rh surface segregation. This process is observed in the temperature range ca.600-1000 K. The oxygen-induced Rh segregation is responsible for the fact that the maximum temperature at which oxygen is present on the surface does not depend on the equilibration temperature. The maximum in 6 is related to the decrease in oxygen coverage above 800 K. Oxygen is desorbed from the surface above 800 K, first from the Pt sites, and as a result S decreases above 800 K. Similar measurements have been performed in the presence of a flow of 5 x lo--’ mbar NO, CO and H2.23 For CO and H2 the change in S with increasing temperature was very small, hence these gases do not exert a significant influence on the surface composition. For NO a similar effect as for oxygen was observed but the induced surface segregation is much less pronounced than for 02.These results suggest that280 NO Reduction and CO Oxidation on Pt-Rh Alloys I 1 .ow 0.5- LOO 800 1200 T I K Fig. 3. Normalized 0 signal intensity as a function of temperature for 100 L O2 exposure on: ( a ) 17 pt and ~0.55-Rh0.45; ( b ) ~o.,z-Rho88. NO dissociates both on Pt-rich and Rh-rich alloy surfaces and the 0 formed can induce some surface segregation. 3. Adsorption Properties The adsorption properties of several polycrystalline Pt-Rh foils, pure Pt, pure Rh and various single-crystal surfaces cut from a Pto,zs- Rho,75 single crystal have been investi- gated in our laboratories.15*17,23,27347,48 The comparison of the behaviour of the various single-crystal surfaces allows one to examine the effect of the surface structure. The surface composition of the alloy surfaces was varied from Rh-rich to Pt-rich simply by variation of the equilibration temperature. In this review paper only some of our results are presented. The techniques used were TDS, AES, LEED, FEM and XPS. The adsorption of hydrogen and carbon monoxide is rather similar on all the Pt-Rh alloy surfaces, as well as on the pure Pt and Rh surfaces. Both the surface structure and the surface composition exert some influences on the TD spectra and hence on the heat of adsorption. These relatively small effects are, however, not relevant for the present discussion. No indication of CO dissociation was found.In addition to hydrogen adsorption hydrogen is also absorbed in large amounts in the bulk of Pt-Rh alloys. Striking differences have been observed in the behaviour of Pt-rich and Rh-rich alloys towards oxygen. As an example fig. 3 shows some AES results obtained for polycrystalline foils. After an exposure of 100 L t the oxygen was pumped out of the vacuum chamber and the 0 (510 eV) signal intensity was monitored during a stepwise temperature increase. In the figure the intensities normalized to their initial values at 300 K are shown. The initial amount of oxygen on Pt-rich alloys is smaller than on Rh-rich alloys. On Pt-rich alloys the small oxygen signal disappears at ca. 400 K, t l L = Torr s.28 1 500 ? 400 b 300 R. M. Worf et al.Tan,, = 1100 K (321) c (210) (410) o Tan, = 800 K 300 400 *500 Fig. 4. The temperature at which a dissociation percentage of 25% is obtained for Pt-rich Pt-Rh alloy surfaces versus the same parameter for Rh-rich Pt-Rh alloy s~rfaces.'~ probably by reaction with residual hydrogen and CO (base pressure in the ultra-high- vacuum system was 1 x lop9 Torri. consisting of 90% H2 and 10% CO). On Rh-rich alloys the amount of surface oxygen decreases upon heating in the temperature range up to 500 K, probably by reaction with the residual gas. However, above 500 K the oxygen surface concentration increases before it is desorbed in the temperature range 800-1200 K. Apparently, subsurface oxygen is present in large amounts and it diffuses to the surface above 500 K.Since the oxygen signal intensity becomes larger than the signal intensity at 300 K, the subsurface oxygen must already have been formed at the exposure temperature of 300 K. The Pt,.55-Rh0.45 sample shows Pt-like behaviour. For the R0.12-Rh0.88 alloy either Pt-like or the Rh-like behaviour was observed depending on the experimental conditions. Similar ambivalent behaviour has been found for the Pt0.25-Rh0.7s (100) surface. It appeared that subsurface oxygen is formed after low- temperature annealing (975 K, Rh-rich surface), whereas high-temperature equilibration (1425 K, Pt-rich surface) results in Pt-like behaviour. The effect of an exposure of 100 L NO on the various samples was studied in the same way as described for oxygen. The behaviour of the oxygen signal was identical to that found after exposure to oxygen for all samples.This indicates that NO dissociates on all surfaces, leaving oxygen adatoms. The NO dissociation is partly due to the intrinsic chemical activity of the surface and, most probably, partly induced by the primary electron beam. According to several studies, NO dissociation is very sensitive to the surface structure of Rh47 and, in particular, of Pt.48-50 For example, it has been reported that the Pt( 11 1) surface cannot break the NO bond, whereas the Pt(410) surface is very active in NO bond breaking.50 XPS has been used to investigate the NO dissociation on Pt-Rh single-crystal surfaces.27 The results are summarized in the fig. 4, which shows the following. ( a ) The dissociation activity is sensitive to the surface structure.The (410) and (210) surfaces are more active in NO bond breaking than the (321) surface, whose t l Torr= 101 325/760 Pa.282 NO Reduction and CO Oxidation on Pt-Rh Alloys activity is larger than that of the ( 1 1 1 ) surface. The effect of the surface structure is larger for Pt-rich than for Rh-rich surfaces. ( b ) The dissociation activity is higher for Rh-rich (equilibration at 875 K) than for Pt-rich surfaces (equilibration at 1100 K). 4. CO Oxidation and NO Reduction on Single-crystal Surfaces In order to investigate the influence of the initial surface composition and of the gas-induced surface segregation on the reactivity, TPRMS and AES measurements have been performed in various gas mixtures of CO-0, and CO-NO using the Pt-Rh (410) surface.Fig. 5 shows the CO, production for the reaction of CO with O2 and for CO with NO starting with a Pt-rich and with a Rh-rich surface as observed during a stepwise increase of the temperature using stoichiometric gas compositions (CO/Oz = 2 and CO/NO = 1). For comparison the Auger N (380 eV) and 0 (510 eV) signal intensities as observed in a flow of 5 x lo-’ mbar NO are also shown in the figure. In the relevant temperature range (500-800 K) NO dissociation is very fast. The temperature of maximum C02 production (T,) is the same for both reactions. Hence, it can be concluded that NO dissociation is not the rate-determining step in the reaction CO + NO under these experimental conditions. The Rh-rich surface shows a maximum rate of C 0 2 production around 665 K. However, on the Pt-rich surfaces the maximum occurs at the significantly lower tem- perature of 615 K.Apparently, the surface composition has a significant influence on T,. This can be explained on the basis of the metal-oxygen bond strength,’ which is weaker for Pt than for Rh, as is also confirmed by fig. 5 ( e ) and (f). Fig. 6 shows the results of similar measurements for increasing, decreasing and again increasing temperature (first, second and third branch, respectively). In the first branch T, is higher for the Rh-rich surface, as has been discussed. The second branch shows a higher CO, production, with T, almost the same for the two surfaces. The third branch shows a higher C 0 2 production than the first branch, but lower than the second.T, is now equal for the initially Rh-rich and Pt-rich surfaces, and it is slightly higher than in the second branch and in between the values of the first branch. Apparently, the surface composition alters during the reaction and is not dependent anymore on the initial equilibration temperature. AES measurements, not shown in this paper, support this conclusion. The largest changes occur on the initially Rh-rich surface. In fig. 6(c) and ( d ) the C (272 eV) and 0 (510 eV) signal intensities are shown as a function of the temperature programme. The product of the carbon and oxygen signal intensities is at a maximum in the same temperature region where the CO, production is at a maximum, confirming the essential reaction step between CO(ads) and O(ads).’ The higher COz production in the second branch might be expected if, starting at high temperatures, O2 is the first gas that is adsorbed, whereas at low temperatures the surface is inhibited by excess CO (high sticking probability).However, the AES results in fig. 6 ( c ) and ( d ) do not support this model; here the carbon signal intensity is higher in the second branch, especially for the initially Rh-rich surface, whereas the oxygen signal intensity is smaller. Apparently, the enlarged CO, production for the second branch is due to a more favourable surface composition. The results presented suggest that the initial surface composition is altered by chemisorption-induced segregation and/or thermal segregation effects. 5. C0-Oxygen and CO-NO Reactions over Silica-supported Pt-Rh Alloy catalyst^*^'^^ Fig. 7 shows the temperature required to achieve a certain, fixed turnover frequency. The turnover frequency was calculated on the basis of CO adsorption experiments. FiveIAES (arb.units) A r u 0 1 IAES (arb. units) pCOz (arb. units) 0 pC0, (arb. units) 0 d p C 0 , (arb. units) J pC0, (arb. units) --. -;i Q, 0 0 X284 NO Reduction and CO Oxidation on Pt-Rh Alloys - Tequ, - 875 K 1100 K l l I l I 1 l LOO 600 800 LOO 600 800 T / K LOO 600. 800 T l K Fig. 6. TPRMS results for a stoichiometric flow of CO and 02: ( a ) Rh-rich Pt-Rh(410); ( b ) Pt-rich Pt-Rh(410) with increasing ( 1 ) decreasing (2) and again increasing (3) temperature. AES carbon and oxygen signal intensities in the same flow for (c) Rh-rich Pt-Rh(410) and ( d ) Pt-rich Pt- Rh( 41 0).'' different catalysts have been used, viz. pure Pt, 25 atom Oh Rh, 50 atom '/O Rh, 75 atom '/o Rh and pure Rh. The CO/O2 and NO/CO ratios are expressed in terms of the equivalence ratio ( A ) , this being the ratio between the O2/CO or NO/CO ratio used and the O2/CO or NO/CO ratio at stoichiometry (O2/CO = 1/2 and NO/CO = 1 ) . Three values of A have been used, varying from CO-rich ( A = 1/4) stoichiometry ( A = 1 ) to CO-lean ( A = 4). The relevant observations as illustrated in the fig. 7 ( a ) and (6) may be summarized as follows: ( 1 ) much higher temperatures are required for achieving a certain turnover frequency for the NO + CO reaction than for the CO + O2 reaction. (2) Synergetic effects like enhancement of the reaction rate for the Pt-Rh alloy catalysts compared with the pure component catalysts are absent.In all cases the activity of the alloy catalyst is between those of the pure components. (3) For both reactions the activities of Rh and the Rh-rich alloy vary only slightly with the flow composition whereas for Pt and the Pt-rich alloy the influence of the flow composition is large. The effect of changing flow composition is larger for the CO + O2 reaction than for the CO + NO reaction. (4) UnderR. M. Wolf et al. 285 / # c( I 100 1 0 25 50 75 Rh (bulk) (%) B Rh (bulk) (%) Fig. 7. Temperature for a turnover frequency of 0.05 s-' for Pt-Rh alloy catalysts as a function of the bulk composition for three different OJCO or NO/CO ratios: ( a ) CO+oxygen, ( b ) CO + NO.oxidizing conditions Pt and Pt-rich alloys are better catalysts for the CO + 0, reaction than Rh and Rh-rich alloys. ( 5 ) Rh and Rh-rich alloys are much better catalysts for the reduction of NO than Pt and Pt-rich alloys, both under CO-rich and CO-lean conditions. (6) For the CO+ 0, reaction the behaviour of Pt0.7s-Rh0.25 and Pt0.5-Rh0.5 was almost like that of pure Pt. The Pt0.2s-Rh0.75 catalyst has an activity which is intermediate between those of Pt and Rh. (7) For the NO+CO reaction the activity of the Pto,zs-Rho,7s catalyst is almost equal to that of the pure Rh catalyst, independent of A. For stoichiometric and CO-lean mixtures the PtO,s-RhO.s alloy has an activity for the NO + CO reaction almost equal to that of Rh. Under reducing conditions, however, its activity is between those of Pt and Rh.The Pto,7s-Rho,25 catalyst shows an activity equal to that of pure Pt under net-reducing conditions. Under stoichiometric and net-oxidizing conditions its activity is between those of Pt and Rh. In order to explain the results obtained for SO,-supported Pt-Rh alloys it is useful to discuss first briefly the behaviour of the individual Pt/SiO, and Rh/SiO, catalysts. It is well established now that the main reaction pathway for CO oxidation on noble metals is a surface reaction between adsorbed 0 atoms and adsorbed CO For Pt the temperature for constant turnover frequency increases with increasing CO concentration in the feed as a result of the well known CO inhibition of the reaction rate. The Pt surface is predominantly covered with CO, leaving very little room for oxygen adsorption.Our results suggest that for Rh the effect of CO inhibition is smaller. Consequently, the Rh catalyst is the better catalyst under net-reducing conditions. However, under net-oxidizing conditions Pt is the most efficient catalyst, probably because of the lower metal-oxygen bond strength.'286 NO Reduction and CO Oxidation on Pt-Rh Alloys The present results illustrate again that Rh is a much better catalyst for NO reduction by CO than Pt, an observation that is consistent with literature The most likely mechanism of the NO + CO reaction is NO dissociation followed by reaction of adsorbed CO molecules and 0 adatoms and by combination of N adatoms to molecular n i t r ~ g e n . ~ - ~ ” ’ The dissociation step of NO is usually considered as one of the slowest steps in the reaction sequence.NO dissociation is more difficult on Pt than on Rh.9 Hence, the relatively low activity of Pt for the NO + CO reaction is consistent with this mechanism. Under our experimental conditions the effect of change of the feed composi- tion was very small for Rh. For Pt our results point to a significant CO inhibition, although much smaller than for the CO+O, reaction. This result may indicate that adsorbed CO does not only inhibit the dissociative adsorption of oxygen but also the dissociation of NO on Pt. NO dissociation may require an ensemble of several vacant platinum atoms. In sections 1-4 we have shown that the surface composition of Pt-Rh alloys varies strongly with the experimental conditions such as the equilibration temperature used and the gas-phase composition. Oxygen in the gas phase induces a Rh surface segregation because of the high Rh-0 bond strength relative to that of Pt-0.15 Hence, an initial surface composition as achieved by equilibration at a chosen temperature can easily be altered.The almost equal activity of the Pt, Pt0.7s-Rh0.2s and PtO.s-Rho.s catalysts for the CO+O, reaction suggests that all the surfaces of these catalysts are Pt-rich, after reduction with hydrogen and subsequent exposure to the reaction mixture up to 550 K. Similar behaviour has been found for the well defined Pt-rich alloy surfaces. The Pt0.2s-Rh0,7s catalyst shows for that reaction an activity between those of Pt and Rh. Hence, it is reasonable to visualize the surface of the Pto.zs-Rho.7s catalyst as being composed of both Pt and Rh.Again, this agrees with the results obtained for the Pt0.25-Rh0.7s crystal surfaces. For the CO + O2 reaction the thermodynamically more favourable Rh surface segregation under net-oxidizing conditions does not occur under our experimental conditions. This may be caused by the relatively low reaction tem- perature: complete CO conversion is already reached at temperatures far below 575 K. For Pt0.2s-Rh0,7s alloy single crystals we found 0-induced Rh surface segregation in the temperature range 600- 1000 K. At lower temperature 0-induced Rh surface segregation is kinetically hampered and at higher temperatures 0-induced Rh surface segregation is not thermodynamically feasible.A comparison of fig. 7 ( a ) and (6) reveals some interesting differences in the behaviour of the Pt-Rh alloy catalysts for the CO + O2 and for the CO + NO reactions. For the NO+CO reaction the PtO.s-RhO.s alloy catalyst behaves like pure Rh under stoichiometric and net-oxidizing conditions, whereas under net-reducing conditions its activity is between those of Pt and Rh. This suggests that the surface composition varies with the experimental conditions, from almost pure Rh under stoichiometric and net-oxidizing conditions to a perhaps bulk-like surface composition under net-reducing conditions. This effect of 0 (from O2 or NO)-induced Rh surface segrega- tion was indeed observed on well defined Pt-Rh alloy surfaces at temperatures above 600 K. The Pt0.75-Rh0,2s alloy catalyst behaves almost like pure Pt under net-reducing conditions, whereas its activity is between those of Pt and Rh under stoichiometric and oxidizing conditions.Again, it suggests that the surface composition changes with the feed composition using the relatively high temperatures required for this reaction. The composition changes from Pt-rich at A = 1/4 to Rh-rich at A = 4. The properties of the Pto,zs-Rho7s alloy catalyst are noticeable. For the CO+O, reaction its behaviour is intermediate between those of Pt and Rh, suggesting a surface with both Pt and Rh present in large concentrations. For the NO+CO reaction its behaviour is like that of Rh under stoichiometric and oxidizing conditions as expectedR. M. Wolf et al. 287 on the basis of our previous discussion.However, for that reaction this alloy behaves also like pure Rh under reducing conditions ( A = 1/4). It might be that the surface of this catalyst contains under the conditions of the NO + CO reaction with A = 1/4 many Rh atoms, as should be expected. The relatively high activity of this catalyst (Rh-like) might then be caused by the beneficial effect of the presence of both Pt and Rh: the easy dissociation of NO on Rh sites, the low CO inhibition on Rh sites and a beneficial effect of Pt, for example, on the amount of N on the surface. On Rh(ll1) the surface is populated by a high concentration of adsorbed N under reaction condition^.^' These N adatoms may act as a reaction inhibitor since sites are occupied that cannot be used for the reaction.On Pt surfaces the N surface concentration is much smaller than on Rh 6. Conclusion The surface composition of Pt-Rh alloys is unique and also flexible: it varies easily with changing experimental conditions (temperature, gas-phase composition). This has important consequences for the chemical behaviour of Pt-Rh catalysts. A strong Pt surface segregation is observed for clean Pt-Rh alloys, especially after high-temperature annealing ( T a 1000 K). However, for Rh-rich Pt-Rh alloys the surface composition is almost bulk-like following low-temperature equilibration ( T 800 K). Adsorbates can easily induce segregation of Rh or Pt to the surface, and the chemical properties of the surface are changed accordingly. The catalytic properties of Si0,-supported Pt-Rh alloys can be understood on the basis of the results obtained for well defined Pt-Rh alloy surfaces: (1) the surface composition as determined by the bulk composition, the temperature and the gas-phase composition and (2) the specific properties of the pure component metals. By (2) are meant ( a ) the inhibitive effect of CO on the CO-0, and CO-NO reactions at low temperatures, in particular on Pt, ( b ) the excellent activity of Pt in excess oxygen due to the relatively low Pt-0 bond strength and (c) the relative activities of the Pt and Rh surfaces in the dissociation of NO. Pt and Rh are most probably present in the form of Pt-rich alloy particles in the three-way catalyst.The surface composition of the alloy particles will depend on the temperature, the gas-phase composition and the presence of impurities. Under condi- tions of lean mixtures (oxidizing mixture) a severe Rh surface enrichment must be expected, eventually leading to the formation of a less active Rh oxide on the surface of the catalyst particle.In the case of an Rh-enriched surface the particle will show catalytic behaviour comparable with that of pure Rh, although it will be more resistant to oxidation due to tile presence of Pt. However, under rich conditions (reducing mixture) a Pt-rich surface is expected because of the Pt-rich bulk. Hydrocarbon fragments and trace of impurities (P, S, Si, B) from the gasoline and from the lubricant may severely alter the surface composition and, consequently, modify the catalytic behaviour of the alloy particles.Excursions to high temperature (above 1000 K) may result in a pronounced Pt surface segregation provided that the temperature is sufficiently high to obtain an uncovered surface. These considerations show that the surface composition of the catalyst particles in the three-way catalyst and, hence, the catalytic behaviour of these particles may be very variable. We thank Dr A. D. van Langeveld, M. C. Angevaare-Gruter, A. G. van den Bosch- Driebergen, M. J. Dees, A. van Dreumel, M. N. H. Kieboom, M. J. Koster van Groos, R. J. Vreeburg and G. H. Vurens for their valuable contributions. Thanks are also due to the Johnson Mathey Technology Centre (Reading, U.K.) for the loan of the Pt and the Rh salts. This research was supported in part by NATO via Grant No.86-352.288 NO Reduction and CO Oxidation on Pt-Rh Alloys References 1 S. L. Handforth and J. N. Tilley, Ind. Eng. Chem., 1934, 26, 1287. 2 T. Wang and L. D. Schmidt, J. Catal., 1981, 71, 411; 70, 187 and references therein. 3 A. R. McCabe and G. D. Smith, Proc. 8th. Int. Congr. Catal. 1984 (Berlin, IV-73; Plat. Met. Rev., 1988, 4 J. Pielaszek, Plat. Met. Rev., 1984, 28, 109. 5 K. G . Gough and B. L. Wibberley, Plat. Met. Rev., 1986, 30, 168. 6 Platinum 1988 (Johnson Mathey plc, London), pp. 29 and 59. 7 K. C. Taylor, Catal. Sci. Technol., 1984, 5 , 119. 8 K. C. Taylor, in Catalysis and Automotive Pollution Control, vol. 30 of Stud. Surf Sci. Catal., ed. A. 9 B. E. Nieuwenhuys, Surf: Sci., 1983, 126, 307 and references therein. 32, 1 1 . Crucq and A. Frennet (Elsevier, Amsterdam, 1987), p.97. 1 1 W. F. Egelhoff, Chem. Phys. Solid Surfaces and Heterogeneous Cataljsis, ed. D. A. King and D. P. 12 D’Arcy Lorimer and A. T. Bell, J. Catal., 1979, 59, 223. 13 W. C. Hecker and A. T. Bell, J. Caral., 1985, 92, 247. 14 S. H. Oh, G. B. Fisher, J. E. Carpenter and D. W. Goodman, J. Catal., 1986, 100, 360. 15 H. A. C. M. Hendrickx, A. M. E. Winkelman and B. E. Nieuwenhuys, Appl. Surf: Sci., 1987, 27, 458. 16 S. B. Schwartz, G. B. Fisher and L. D. Schmidt, J. Phys. Chem., 1988, 92, 389. 17 F. C. M. J. M. van Delft, G. H. Vurens, M. C. Angevaare-Gruter and B. E. Nieuwenhuys, in Catalysis and Automotive Pollution Control, vol. 30 of Stud. Surf: Sci. Catal., ed. A. Crucq and A. Frennet (Elsevier, Amsterdam, 1987), p. 229. Woodruff (Elsevier, Amsterdam, 1982), vol.4. 18 General Motors Research Labs., Warren, Michigan, personal communication. 19 H. J. Kelley and V. Ponec, Progr. Surf: Sci., 1981, 11, 139. 20 W. M. H. Sachtler and R. A. van Santen, Adv. Catal., 1977, 26, 69. 21 A. D. van Langeveld, Thin Solid Films, 1985, 129, 161. 22 B. E. Nieuwenhuys, in Chemisorption of Gases on Metal Films, vol. 32 of Stud. Sur$ Sci. Catal., ed. P. 23 F. C. M. J. M. van Delft, J. Siera and B. E. Nieuwenhuys, Surf: Sci., 1989, 208, 365. 24 F. C. M. J. M. van Delft and B. E. Nieuwenhuys, Surf: Sci., 1985, 162, 538. 25 R. M. Wolf, M. J. Dees and B. E. Nieuwenhuys, J. Phys. Paris, 1986, C7, 47; 419. 26 F. C. M. J. M. van Delft, A. D. van Langeveld and B. E. Nieuwenhuys, Surf: Sci., 1987, 189/190, 1129.27 F. C. M. J. M. van Delft, J. Siera, R. J. Vreeburg, M. J. Koster van Groos, A. D. van Langeveld and B. E. Nieuwenhuys, Proc. 9th Int. Congr. Catal., Calgary, 1988, vol. 3, p. 1124. 28 F. C. M. J. M. van Delft, B. E. Nieuwenhuys, J. Siera and R. M. Wolf, J. Iron Steel Inst. Jpn Internat., 1989, 29, in press. 29 A. G. van den Bosch-Driebergen, M. N. H. Kieboom, A. van Dreumel, R. M. Wolf, F. C. M. J. M. van Delft and B. E. Nieuwenhuys, Catal. Lett., 1989, 2, 73. 30 A. G. van den Bosch-Dreibergen, M. N. H. Kieboom, A. van Dreumel, R. M. Wolf, F. C. M. J. M. van Delft, and B. E. Nieuwenhuys, Catal. Letr., 1989, 2, 235 31 H. J. Goldschmidt and T. Land, J. Iron Inst., 1947, 155, 221 and references therein. 32 M. Kh. Karapetyants and M. K. Karapetyants, Handbook of Thermodynamic Constants of Inorganic and Organic Compounds ( Ann-Harbour-Humphrey Science, Ann Harbor, 1970). 33 G. A. Somorjai, in Chernisorption in Two Dimensions: Sugaces (Cornell University Press, Ithaca, 1981). 34 F. C. M. J. M. van Delft, M. J. Koster van Groos, R. A. G. de Graaf, A. D. van Langeveld and B. E. 35 A. K. Niessen, F. R. de Boer, R. Boom, P. F. de Chatel, W. G. M. Matters and A. R. Miedarna, Calphad, 36 G. H. Vurens, F. C. M. J. M. van Delft and B. E. Nieuwenhuys, Surf: Sci., 1987, 192, 438. 37 F. L. Williams and G . C. Nelson, Appl. Sur- Sci., 1979, 3, 409. 38 P. H. Holloway and F. L. Williams, Appl. Surf: Sci., 1982, 10, 1 . 39 F. C. M. J. M. van Delft, Ph.D. Thesis (Leiden, 1988). 40 A. D. van Langeveld and J. W. Niemantsverdriet, Sur$ Sci., 1986, 178, 880. 41 T. T. Tsong, Surf: Sci., 1988, 8, 127. 42 K. Hoshino, J. Phys. Soc. Jpn, 1981, 50, 577. 43 J. Massardier, B. Tardy, P. Delichere, M. Abon and J. C. Bertolini, 8th Int. Congr. Catal., Berlin, 1988, IV- 185. 44 J. C. Bertolini, B. Tardy, M. Abon, J. Billy, P. Delichere and J. Massardier, Surf: Sci., 1983, 135, 117. 45 M. S. Spencer, Surf: Sci., 1984, 145, 153. 46 L. de Temmerman, C. Creemers, H. van Hove and A. Neyer, Surf: Sci., 1987, 183, 565. 47 H. A. C. M. Hendrickx and B. E. Nieuwenhuys, SurJ Sci., 1986, 175, 185. 48 P. Leerkamp, R. M. Wolf and B. E. Nieuwenhuys, J. Phys. Paris, in press. Wissmann (Elsevier, Amsterdam, 1987), p. 476. Nieuwenhuys, Sur- Sci., 1987, 189/190, 695. 1983, 7, 51.R. M. Wow et al. 289 49 R. J. Gorte, L. D. Schmidt and J. L. Gland, Sur$ Sci., 1981, 109, 367. 50 R. I . Masel, Catal. Rev. Sci. Eng., 1986, 28, 335. 51 S . B. Schwartz, G. B. Fisher and L. D. Schmidt, J. Phys. Chem., 1988,92, 389. Paper 9/00277D; Received 23rd December, 1988

ISSN:0301-7249    

DOI:10.1039/DC9898700275

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 291-302 Decomposition (and Synthesis) of Ammonia on W{ loo} A Thermal Molecular Beam Study Patrick Alnot,? Albert CassutoS and David A. King*§ The Donnan Laboratories, University of Liverpool, Liverpool L69 3BX The adsorption and decomposition of NH, and ND, on W{ 100) have been studied in detail using a molecular beam technique. Sticking probabilities and desorption rates of gases resulting from the decomposition reaction have been followed at crystal temperatures between 300 and 910 K. Isother- mal desorption of D7 (H,) was followed as a function of ND3 (NH,) beam time on an initially clean surface. Variations in the shape of the resulting isothermal desorption curves as a function of surface temperature are explained via the changes in activation energies for desorption of D2 and decomposition of adsorbed ND, as a function of N coverage.A significant kinetic isotope effect is found in H2 and D2 isothermal desorption during the decomposition of mixed NH, and ND, beams only at high N(a) coverages. A kinetic model and energetic scheme are presented for the ammonia decomposition and synthesis reactions on W{ loo} which are con- sistent with all of the experimental data, and show that the synthesis reaction is blocked by a large activation barrier to the initial hydrogenation of adsorbed nitrogen atoms. The adsorption and decomposition of NH3 on tungsten metal has been widely studied in order to understand the catalytic activity of metal surfaces in the ammonia synthesis reaction.'32 Although tungsten has a poor synthesis activity, it is interesting to work with in view of the considerable amount of effort performed to characterise the different types of tungsten surfaces.NH,/W{001} is a particularly good system since W(OO1) exhibits high sticking probabilities for most of the common gases, e.g. NZ, H2, CO, NO and NH,, rendering the system amenable to study by molecular beam techniques, which can give very accurate sticking and reaction probabilities in the range 0.1 d s d 1. Previous studies have mainly been concerned with temperature-programmed desorption, LEED, work function and XPS measurements. Wide divergences exist regarding both experi- mental results and interpretation. The first major study was performed by Estrup and Anderson.3 Their results indicate that room-temperature adsorption of NH, on W{OOl} is non-dissociative and occurs with a sticking probability s 3 0.45.TPD spectra showed that, on raising the temperature, hydrogen is lost at ca. 800 K, leaving behind a c(2 x 2) structure which was ascribed to a f monolayer of NH2. This structure was destroyed at 1375 K, with the simultaneous loss of H2 and N2 from the surface. These authors deduced that the rate of hydrogen desorption is controlled by an N-H bond-breaking process rather than H-atom recombi- nation. Other studies strongly conflict with the above results. In particular Dawson and H a n ~ e n , ~ performing an FEM study, found that adsorption of ammonia on W yields hydrogen and nitrogen adatoms on the surface, the decomposition being complete at a surface temperature of 400 K.Wilf and Folman' arrived at the same conclusion. Their T Present address: Thomson-CSF LCR, Domaine de Corbeville, Orsay Cedex 91401, France. t Present address: Laboratoire Maurice Letort, Route de Vandoeuvre, F-54600 Villers Les Nancy, France. !4 Present address: Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 IEP. 29 1292 Decomposition of Ammonia on W{ 100) results indicate that when ammonia interacts with a clean W{lOO} crystal, surface decomposition begins below 300 K and is complete above 400 K, the only products being H and N adatoms. However, they also showed that when NH, is adsorbed onto the (100) plane precovered with N2, the decomposition process is initiated at higher temperatures than on the clean surface.They also concluded that intermediate species such as W2N3H and WNH exist and are stable up to 600 K. Matsushita and Hansen6 also differed from Estrup in their analysis of TPD data following NH, adsorption on polycrystalline W. After adsorption of a full monolayer of NH, and heating to 800 K they found a surface stoichiometry of NH,,,, . Results obtained by May et al.' seem to agree somewhat with those of Estrup. They studied NH3 on W{211}, and concluded that after saturation of the surface with ammonia at room temperature and heating to 500K, one third of the surface hydrogen is evolved, leaving a residue with the stoichiometry NH2. At 1200 K simultaneous desorption of H2 and N2 occurs. More recently Egawa et aL8 studied NH, adsorption and decomposition on W(100) with XPS and UPS.At 120 K non-dissociative adsorption was found. Upon heating to 220 K, the formation of NH2(a) and H(a) was deduced from XPS and UPS spectra. Upon heating to 400 K, further dissociation of NH,(a) to NH(a) and N(a) occurred. They also studied ammonia adsorption on a predosed c(2 x 2) - N structure, showing evidence for ammonia in molecular form at 300 K. Heating this structure to 430 K in vacuo produced NH( a). The rate-determining step of the overall ammonia decomposition reaction was thought to be the desorption of atomic nitrogen as NZ, as concluded much earlier by Tamaru.' The latter found that this assumption gave the explanation for zero-order kinetics with respect to P,,, for the decomposition of ammonia at high temperatures.The possibility of a different rate-determining step has been inferred from the observation of an isotope effect in the decomposition of NH, and -ND,. The original data by Jungers and Taylor" and Barrer," showing that NH, decomposes 1.6 times faster than ND,, was reproduced later by McAllister and Hansen." These authors interpreted their results by assuming that the rate-determining step involves the fission of an N-H bond. There are three possible sources for the above discrepancies: (i) experiments are run on different types of surfaces (e.g. polycrystalline, W{ loo}, W{211}); (ii) experiments were performed over differing pressure ranges, from 10-1 Torrt of NH, (when saturation of the surface with N could occur rapidly) to 10-'Torr; (iii) results have been misinterpreted.In the present work we report sticking probabilities as a function of coverage and surface temperature and a quantitative examination of ammonia decomposition, and report the conditions under which an isotope effect is observed in the decomposition of NH, and ND,. Experimental The molecular beam technique of King and Wells', was used in these studies. Modifications to the apparatus are described in an earlier paper.'" Well collimated beams of NH, and ND3 with measured intensities in the range 10'2-1013 molecule s-' cm-2 were produced using a series of five differentially pumped chambers. Beam 'temperatures', TB, referred to in this paper are calculated assuming a Maxwellian velocity distribution with TB = 2kB Tg . The sticking probability is given by the simple expression s = ( P2 - P ) / P , where P is the isotropic pressure in the cell during beaming when the beam is incident on the crystal and P2 is the pressure at saturation coverage (fig.l ) , when the sticking probability 'l 1 Torr = 101 325,' 760 Pa.P. Alnot, A, Cassuto and D. A. King 1.0 0.8 0.6 0 . 4 0.2 293 Fig. 1. Experimental sticking probability for ND, as a function of adatom coverage, at surface temperatures between 320 and 910 K. is zero. P2 was difficult to obtain by direct adsorption measurements since it has to be measured after a long beaming time to ensure surface saturation. Ammonia is also adsorbed on the glass walls of the adsorption cell, and P2 could not be obtained in the usual way by interrupting the beam with an inert (glass) surface.13 P2 was therefore measured by assuming that at high temperatures (>910 K) a steady state is reached in which the rate of adsorption of D as ND, and the rate of desorption as D2 are equal.Recombination of D atoms following ND, dissociation at such high temperatures leads to: d D 2 ) = A[ra(ND3)12 (1) where rd(D2) and ra(ND,) are the rates of desorption and adsorption of D2 and ND,. Taking into account the above expression for s and considering the adsorption curve for two different beam times (say t = 0 and 10 min) we can write: from which a value for P2 is obtained in units of the output signal of the mass spectrometer. (It is not necessary to convert this to pressure units since the sticking probability is obtained as a ratio of pressures.) The beam intensity was calibrated against a calibrated N2 beamI4 and found to be 2 .4 ~ 10l2 molecules s-l cm-' for a pressure behind the capillary of 40 Torr. The beam cross-section at the crystal is 1.85 x cm2. rd(D2, t=O)/rd(D2, t)={[P2-P(t =o)]/p2-P(t))Z Ammonia Adsorption and Desorption (i) Sticking Probability Dependence on Coverage The variation of the ammonia sticking probability, s, with surface cpverage, expressed as the total amount of N-containing species on the surface, N(a) = C,=, NNH,, is shown294 Decomposition of Ammonia on W{ loo} in fig. 1 for crystal temperatures between 320 and 910 K. The general trends displayed in these curves follow those expected for adsorption in a precursor at lower temperatures s is initially almost independent of coverage, while at high temperatures s decreases sharply with increasing coverage.Additionally, the initial sticking probability decreases with increasing substrate temperature above 500 K, which is again consistent with precursor models.13-14 The results are also consistent with the data of Kay and Raymond,” who found evidence for trapping-desorption of ammonia from an ammonia- saturated surface, with a mean residence time below 10 p s at 300 K. Our data are consistent with a trapping probability into the precursor state of unity. Reed and Lambert16 measured the initial sticking probability for NH, on W(100) as 0.4 at a crystal temperature of 300 K, which is in only fair agreement with our results. Their measurements were conducted using AES and an effusive beam; a well collimated beam, as in the present work, does have advantages in s measurements.(ii) Desorption Products: Adsorption at 300 K Adsorption of ND, was conducted on the clean W(100) surface at 300 K. Temperature- programmed desorption spectra were then obtained for both D2 and N2 desorption. The D2 desorption peak temperatures were lower than obtained after the adsorption of D2(H2) onto clean W{lOO}, as previously reported for coadsorption of H2 and N2 on W{ 100}.17718 The desorption energy is lowered by N coadsorption. In the present work, after a total ND3 uptake of 7 x 1014 molecule cm-I at 300 K, a single D2 desorption peak was observed at 400 K; no D2 desorption was observed above 600 K. N2 desorption spectra were identical to those obtained from pure N2 adsorption on W(100).(iii) Desorption of ND3 As shown by Egawa et aZ.,* the final state of chemisorbed NH, depends strongly on temperature. On adsorption at 120 K, chemisorbed ND3 stays mainly in its molecular form. We have checked for ammonia desorption after an ND3 beam exposure of 8 min on a clean crystal at 190 K, corresponding to an approximate coverage of 8 x lOI4 molecule cmP2. This experiment is particularly well defined, since the ND, beam is entirely focussed onto a portion of the front face of the crystal, thus eliminating any edge crystal-plane or support effects. A typical desorption spectrum after such adsorption is shown in fig. 2, which illustrates a desorption peak for ammonia at a peak temperature T = 285 K. Next, the surface temperature was increased to 800 K in order to remove all surface D.Then ammonia was readsorbed on the crystal surface for another 8 min beam time exposure. As shown in fig. 2, ammonia is again desorbed with a peak at T, = 285 K, but with a larger integrated intensity compared with the previous desorption experiment, indicating that twice as much ND, is desorbed after saturating a surface with preadsorbed N adatoms than from a clean surface. Since the amount of ND, desorbed from the clean surface is smaller than the amount desorbed after presaturation with N(a), partial ND3 decomposition of adsorbed molecules take place either during desorption or at a lower surface temperature. It is concluded that N adatoms suppress or poison the surface decomposition of ammonia molecules. Also, since ammonia desorption peaks at the same temperature on a clean surface as on an N-precovered surface, adsorbed ammonia can be assumed to occupy the same adsorption site on both types of surface.Since NH, adsorption is not blocked by preadsorbed N, we also conclude that the NH, adsorption site is not the same as that for N(a). Since N atoms are adsorbed into the four-fold hollow sites,” the favoured site is on top of a single W atom. The f.w.h.m. of the desorption peak is large at such temperatures, implying either a low pre-exponential factor for ND, desorption or a variation of Ed with ND, surface coverage. Assuming that Ed is independent of coverage,P. Alnot, A. Cassuto and D. A. King 295 0 200 400 Ts/ K 600 Fig. 2. Desorption spectra of ND3 after ND3 adsorption (8 rnin beam exposure) on a clean crystal ( * - -1 and after readsorption of ND3 on the same surface heated to 800 K to remove H (-), corresponding to a beam exposure of 8 min on the same crystal.The intensity of the molecular beam is 2.42 x 10l2 molecule cm-2 s-'. a simulation of both the peak temperature and the half-width is obtained with E d = 17 kJ mol-' and u = lo2 s-'. This anomalously low value for u, and the shape of the desorption peak, strongly imply that Ed varies with 8, giving rise to the broad peak observed . Isothermal Decomposition of ND3 and Residual Surface Stoichiometry The isothermal decomposition of ND3 was followed by monitoring the D2 pressure as a function of beam time at a range of crystal temperatures.The shape of these curves is very dependent upon surface temperature, as shown in fig. 3. The nitrogen coverage (as ND,) is shown as a function of beam time in fig. 4, as deduced from the measured ND3 sticking probability. The beam flux was always 2 . 4 ~ 10l2 molecule cm-' s-l in these experiments. At a surface temperature of 350 K D2 desorption begins only after 3 min of ammonia adsorption, when the surface ND3 coverage is 4~ lOI4 molecule cmp2. Two peaks are observed in the rate of D2 desorption, at t = 6 and 19.5 min. At 410 K D2 desorption is observed after only 30 s of beaming, when the ND3 coverage is 0.7 x loi4 molecule cm-', and only a single peak is observed in the rate of D2 desorption, at t = 3 min. As the temperature is increased further the delay time before D2 desorption is observed becomes immeasurably small, and the time taken to reach the first peak in the D2 rate decreases until at 910K the peak time coincides with the arrival of ND3 at the surface.At this temperature the decomposition to D2 proceeds to completion on impact and chemisorp- tion of an ND3 molecule on the surface; and the chemisorption is directly attenuated, or poisoned, by the surface residue, N(a). In addition, at 480 K a second D2 desorption rate peak is observed, at t = 10 min. The area under each of the curves in fig. 3 represents the amount of D2 desorbed in the decomposition process before it is completely attenuated; this amount can be quantified, since the pumping speed is known, and is shown in table 1. The amount of D2 formed is seen to pass through a maximum at T, = 480 K.In addition, since no N2 desorption occurs over the substrate temperature range investigated, the amount of N296 10 N I 5 Decomposition of Ammonia on W{ 100) - I I I I 350 beam time/min Fig. 3. Isothermal decomposition of ND, to D2 during ND, beaming at surface temperatures from 350 to 910 K. The intensity of the molecular beam is 2.42 x lo'? molecules cm-'s-'. 0 2 4 6 0 10 beam time/min Fig. 4. The increase in total N adatom coverage with ND3 beam time, as calculated from the experimental curves of the sticking probability versus beam time. TJK: ( a ) 320, ( b ) 400, ( c ) 540, ( d ) 910.P. Alnot, A. Cassuto and D. A. King 297 Table 1. Analysis of stoichiometry near termination of isothermal ammonia decomposition at substrate temperatures between 350 and 910 K amount of D2 desorbed amount of N adsorbed surface stoichiometry near saturation/molecule at termination/atom cm-' at termination of the T\/ K cm - decomposition, D : N - 7 350 0.65 x lo1 410 1.0 x lo1 480 3.3 x 10" 610 2.31 x l O I 5 700 1.80 910 1.54 10 x 1Ol4 l o x 1 O l 4 5 x 1ol4 10 x 8 x l O I 4 6.5 x 10" 1.7 1.2 0 0 0 0 adsorbed can be obtained during beaming at each temperature (fig.4) and, in particular, the amount of N adsorbed as the decomposition to D, nears completion is also given in table 1. At temperatures between 350 and 480 K this amount is constant, and it has been assumed, following Reed and Lambert,16 that the amount of ND, adsorbed at saturation at these temperatures is equal to the number of N adatom sites on the surface, i.e.l o x l O I 4 molecule ern-,. From the amount of N adsorbed and the amount of D2 desorbed at termination of the decomposition we are therefore able to calculate the surface stoichiometry at termination. At temperatures of 480 K and higher there is no D on the surface: the surface is saturated with N(a). At 410 K, the D : N stoichiometry is 1.2, and at 350 it is 1.7. These results provide quantitative support to the results of Egawa et ~ l . , ~ showing that NH(a) is the predominant species remaining on the surface after heating to 400 K. The amount of N adsorbed near termination of the ND3 decomposition decreases with increasing T, above 500 K (table 1). The reason is that the sticking probability of ND, decreases sharply with increasing T, above 500 K, as shown in fig.1; the saturation coverage of N adatoms is not achieved after 20min of beaming at these higher tem- peratures, and the ammonia sticking probability is so low that the D2 desorption rate becomes negligible. In fact, at these high temperatures the adsorption of ammonia is rate-determining in the production of D2. This also explains the decrease in the amount of D2 desorbed near termination of the decomposition process (table 1). However, over the temperature range 350-480 K the amount of D, formed increases with increasing temperature. Here the surface processes: ND,,(a) -+ ND,-,(a)+D(a); y = 1,2 or 3 (1) 2 D b ) - D2(g) (2) become the rate-determining factors, and the process is therefore very sensitive to surface temperature.No D, desorption is observed on ND3 adsorption at 300 K. Kinetic Isotope Effect: NH3 and ND3 Decomposition Under certain conditions we have observed a large isotope effect in the isothermal decomposition of NH3 and ND3. This effect was observed by monitoring the H,, HD and D, desorption rates at surface temperatures between 350 and 700 K, and samples of the data obtained at 350 and at 450 K are shown in fig. 5. At temperatures of 450 K and above, the rates of evolution of H, and D, were always within experimental error of each other. At these substrate temperatures there is no kinetic isotope effect in the ammonia decomposition. At temperatures above 450 K, therefore, it would appear unlikely that the rate-determining step in H2 or Dz evolution is eqn (1) above.We infer298 Decomposition of Ammonia on W{ loo} 0 2 4 6 8 10 beam time/min Fig. 5. The kinetic isotope effect in NH3/ND3 decomposition: products H2 ( a ) , D, ( b ) and HD( c ) as a function of beam time at 450 K (A) and 350 K (B). that at temperatures of 450-700 K, D, desorption, step (2), is rate-determining. As discussed above, at 910 K ND3 adsorption becomes rate-determining. However, at temperatures between 350 and 450 K the rate of H2 evolution was found to be higher than that for D2 evolution (fig. 5). At these temperatures the rate-determining step apparently involves the fission of N-H or N-D bonds, since no isotope effect has been observed in H2/D, desorption from clean W{100}.20 This statement can be quan- tified by writing the isotope effect as the ratio of the initial rates of H, to that of D2 production (or 2xrate of HD production to D, production), the initial rate only corresponding to the condition that the H: D surface stoichiometry is unity (as the decomposition proceeds the latter ratio decreases).We find, in fact, that the isotope effect is a strong function of N adatom coverage and of surface temperature and the curve extrapolates to unity, i.e. zero isotope effect, at zero N coverage. We believe the critical factor to be the N coverage: at lower temperatures, the onset of ND3 decomposi- tion is at higher N coverages, giving an apparent temperature dependence. We note that a simple decomposition process involving the direct formation of gaseous H, from adsorbed NH3 [NH,(a) + NH(a) + H,(g)] can be eliminated, since with an equimolar mixture of NH3 and ND3 we observe the gaseous HD product in statistical amounts.We conclude that formation of H2 in the gas phase proceeds through the formation of H(a). A simple simulation of the overall reaction has been performed considering the following mechanistic sequence: NH,(g) + NH,(a) (3) NH3(a)+NH2(a) + H(a) (4) NH,(a) - NH(a)+H(a) ( 5 ) W a ) -+H,(g) (6) where step (5) is rate-determining. This sequence would be consistent both with the observation by Egawa et dx that NH(a) remains on the surface at 400 K, and with our stoichiometric observations, and should apply to the evolution of H, (DJ at temperatures up to 400 K. The kinetic sequence was programmed with a set of fixed and variable15 10 5 O f P.Alnot, A. Cassuto and D. A. King 299 1 7 I I 0 5 10 15 20 2 5 beam time/min Fig. 6. A comparison between experimental (El) and theoretical (-) isothermal ND3 decomposi- tion spectra at 350 K. N ~ o v e r a g e / l O ' ~ atom cm-' Fig. 7. The best-fit functional dependences of the activation energies for desorption of D, ( a ) and decomposition of ND3 ( b ) on N adatom coverage. parameters: the desorption energy and pre-exponential factor for H2 (D,) at zero N(a) coverage were taken to be 136 kJ mol-' and lo-' crn-'sC', respectively; the activation energy and pre-exponential for decomposition of NH,(a) at zero N(a) were taken to be the same as that for desorption of NH3 (a) (see fig. 2). Pre-exponential factors were (arbitrarily) kept constant, but the desorption and decomposition energies were allowed to vary so as to produce a fit to the experimental data.The experimental dependence of NH3 (ND,) sticking probability on coverage (fig. 1) was inserted to describe step ( 3 ) . The best fit to the data for ND3 decomposition at 350 K is shown in fig. 6, and the fitted variations of the two energies with coverage are shown in fig. 7. Two peaks in the isothermal desorption curves are reproduced by the introduction of a step increase300 Decomposition of Ammonia on W{ loo} , N>5x101L t f Fig. 8. A schematic potential-energy diagram for ammonia decomposition (and synthesis) on W( 100); energies are in kJ mol-'. in the activation energy for ND3 decomposition and a step decrease in the desorption energy for D2( H2), at a total N coverage of 5 x 1014 atom cm '.These changes summarise the switchover in the slow steps. At low N coverages D2 desorption becomes rate-limiting, and there should be no isotope effect; at higher coverages ND2(a) decomposition is rate-limiting. The model reproduces a kinetic isotope effect at 350 K by insertion of a zero-point energy difference of 3.76 kJ mol ' for the breaking of each N-H (N-D) bond. However, on increasing the surface temperature to 410 K the model yields only a single isothermal D2 (H,) peak in agreement with experimental observation (fig. 3), and the isotope effect is reduced. At 480 K, two peaks are again observed in the experimental isothermal decomposition curves. The first peak is attributed to the above mechanism, corresponding to low N coverages where H2 ( D2) desorption is rate-limited; there is no isotope effect.However, at higher coverages the decomposition of NH(a) [ ND(a)] becomes rate-limiting, thus producing the second isothermal peak. At still higher temperatures the amount of ammonia adsorbed is reduced owing to the lowering of the sticking probability (fig. I ) , and this second peak is suppressed; at 910 K the rate-determining step is the adsorption of ammonia, since the maximum D' desorption rate is observed at zero coverage, as the beam is allowed to impinge on the clean surface. From the various activation energies deduced above, a schematic potential-energy diagram can be constructed for the decomposition and synthesis of ammonia on W{ loo}, which can be compared with that previously adduced for ammonia synthesis and decomposition on iron surfaces.' The diagram is shown in fig.8, for which the (approxi- mate) energies (in kJ mol-') are obtained as follows. ( i ) The heat of adsorption in the NH3 precursor state is taken from Kay and Raymond" as 45 kJ mol-I; crossover to the chemisorbed state occurs over a 27 kJ mol-' barrier, obtained from the temperature dependence of the sticking probability.'" ( i i ) The heat of adsorption in the chemisorbed state is obtained from the desorption spectra analysis in the present work as 55 kJ mol-'. Decomposition to NH(a) occurs at 220 K (Egawa et a / . ) at low N(a) coverage, and we therefore estimate the barrier to be ca. 50 kJ mo1-l. At high N(a) coverages this barrier height is increased.(iii) The stability of NH,(a) is not known, and is therefore indicated by a dashed line; however, decomposition of NH,(a) to NH(a) only occurs at ca. 400 K (Egawa et a/.) from which we estimate the barrier height as 80 kJ mol-'. (iv) Desorption steps involving H adatoms are taken from desorption spectra to be 140 kJ mol-I; thisP. Alnot, A. Cassuto and D. A. King 301 quantity is sensitive to the amount of adsorbed N(a). (v) Decomposition of NH(a) occurs over a relatively large barrier, since it occurs at 400-500 K; the barrier is estimated to be 90 kJ mol-'. (vi) The heat of adsorption of nitrogen is 340 kJ mol-1 at low N(a) coverage, decreasing to ca. 240 kJ mol-' at higher coverage. (vii) The enthalpy change for the reaction NH3(g) -+ $N2(g) +;H,(g) is 46 kJ mol-I.The potential energy diagram demonstrates clearly why tungsten is a poor ammonia- synthesis catalyst. The stability of adsorbed nitrogen on this surface leads to a very large activation energy barrier for the initial hydrogenation step, N(a) + H(a) -+ NH(a), of 240 kJ mol-' at low N(a) coverages and 190 kJ mol-' at higher coverages. Subsequent hyrogenation steps are relatively facile, but the reaction is virtually blocked by that initial step. Conclusion Many of the discrepancies in the literature can be resolved by recourse to the present analysis and to the data of Egawa et dx We find no trace of surface hydrogen, after ammonia is adsorbed at 300 K, at temperatures above 600 K. In our experiments with a collimated molecular beam NH3 is dosed only onto the front face of the crystal; supports are not exposed to NH3.We therefore attribute earlier observations of high- temperature H2 desorption to desorption from crystal supports (or possibly edges not { 100)-orientated). An isotope effect in ammonia decomposition is observed under a narrow range of conditions only, where N-H bond fission is rate-determining. Thus, the rate-controlling step of the overall reaction is the desorption of N2, since N(a) poisons the ammonia chemisorption step, and at high temperatures (>400 K) and low N coverages we find that the rate-determining step in the decomposition of adsorbed N H3 to gaseous H2 is H2 desorption. Only at high N coverages, where N H, ( a ) decompo- sition is inhibited, is the decomposition of the adsorbed NH, species determined by the rate of N-H bond fission.Previous work at high NH3 pressures would correspond to high N coverages, while at low pressures and higher temperatures the N coverage is low. From our data and the work of Egawa et al.' an approximate potential-energy diagram has been constructed for the ammonia decomposition and synthesis reactions on W{ loo}. Although decomposition to gaseous H2 can proceed readily on this surface, the synthesis reaction is blocked by the large barrier, 240-190 kJ mol-I, depending on N coverage, to the initial hydrogenation step N(a)+ H(a) -+ NH(a). We acknowledge financial assistance towards this work from ICI (Runcorn), and helpful discussions with Dr K. C. Waugh and Dr M. Bowker. References 1 R. M. Lambert and M. E. Bridge, in The Chemical fhj~sics of'Solid Surfaces and Heterogeneous Cafrrlni\ ed. D. A. King and D. P. Woodruff (Elsevier, Amsterdam, 1984), vol. 3B, p. 59. 2 M. Grunze, in The Chemical Phjisics of Solid Surfaces and Heterogeneous Catalj-sis, ed. I>. A. King and D. P. Woodruff (Elsevier, Amsterdam, 19821, vol. 4, p. 143. 3 P. J. Estrup and J. Anderson, J. Chem. Phjls., 1968, 49, 532. 4 P. T. Dawson and R. S. Hansen, J. Chem. Phj-s., 1968, 48, 623. 5 M. Wilf and F. Folman, J. Chem. Soc., Farday Trans. I , 1976, 72, 116s. 6 K. I . Matsushita and R. S. Hansen, J. Chem. Phys., 1969, 51, 472. 7 J. W. May, R. J. Szostak and L. H. Germer, Surf: Sci., 1969, 15, 3 7 . 8 C. Egawa, S. Naito and K. Tamaru, Surf: Sci., 1983, 131, 49. 9 K. Tamaru, Trans. Faruday Soc., 1961, 57, 1410. 10 J. C. Jungers and H. S. Taylor, J. Am. Chem. Soc., 1935, 57, 679. 11 R. M. Barrer, Trans. Faraday Soc., 1936, 32, 490. 12 J. McAllister and R. S. Hansen, J. Chem. f h j x , 1973 59, 414.302 Decomposition of Ammonia on W{ loo} 13 D. A. King and M. G. Wells, S u r - Sci., 1972, 29, 454. 14 P. Alnot and D. A. King, Surf. Sci., 1983, 126, 359. 15 B. D. Kay and T. D. Raymond, J. Chem. Phys., 1986, 85, 4140. 16 A. P. C. Reed and R. M. Larnbert, J. Phys. Chem., 1984, 88, 1954. 17 G. R. Thomas, Thesis (University of Liverpool, 1981). 18 W. Ho, R. F. Willis and E. W. Plummer, Surf: Sci., 1980, 95, 171. 19 K. Griffiths, D. A. King, G. C. Aers and J . B. Pendry, J. Phys. Chem., 1982, 15, 4921 20 P. Alnot, A. Cassuto and D. A. King, to be published. Paper 9/00278B; Received 23rd December, 1988

ISSN:0301-7249    

DOI:10.1039/DC9898700291

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 303-309 GENERAL DISCUSSION Prof. D. A. King (University of Cambridge) began the discussion of the papers in this section: I wish to make two comments, the first relating to both the paper presented by Prof. van Santen and that by Dr Niewenhuys. ( 1 ) We have recently obtained, by F.t.-reflection-absorption i.r. spectroscopy, some new results throwing light on the interaction between CO and NO on Pd {llO}.' For each single-phase system the bridge site is preferred. When the two are coadsorbed, however, a mixed phase is formed in which the stable CO site is linear. The proximity of an adsorbed NO to a bridged CO destabilises the latter, while the linear state is stabilised by the interaction. We ascribe this dual effect to an electrostatic interaction between the adsorbed species.With NO, the molecule-surface bonding component of the 2 7 ~ * level is well below the Fermi level of the metal, and the molecule is therefore strongly negatively charged. With bridged CO this level is only partially occupied, while with linear CO the net charge transfer is dominated by u donation and the adsorbate is slightly positively charged. Thus, an adsorbed CO molecule placed in the dominant electrostatic field generated by the NO negative charge and its image will have its energy levels shifted upwards: in particular, the 27r* level is lifted above the Fermi level thus destabilising the bridge site. By the same token, the linear state is stabilised. This interaction model, based on the work of NGrskov et al.,' is considerably more relevant to extended metal surfaces, with a virtually infinite sink of valence electrons, than the notion of electron competition implied, for example, by the use of curly arrows.With charged species the interaction is predominantly electrostatic in nature. (2) Following on from this comment, I want to raise a question about the general relevance of cluster calculations of the kind described by Prof. van Santen to adsorption on extended metal surfaces. In calculations on a finite-sized cluster charge-transfer will occur between cluster and adsorbate. If, for example, an NO molecule is placed on the cluster a charge-transfer of ca. two electrons may be found to the NO molecule. This leaves the metal cluster with a charge of +2, which would strongly influence the bonding of a second molecule to the cluster.However, if the cluster were embedded in a semi-infinite sea of electrons, this charge on the metal atoms would average out to zero: the electron deficiency would be strongly localised at the site of the adsorbed molecule, and would not affect electron availability at neighbouring sites. In particular, one cannot therefore expect that cluster calculations with more than one adsorbed molecule would be relevant to extended metal surfaces. 1 R. Raval, S. Haq, G . Blyholder and D. A. King, J. Pbys. C, in press. 2 J . K. Nprrskov, S. Holloway and N. D. Lang, SurJ Sci., 1984, 137, 65. Prof. van Santen replied: The extended Huckel calculations presented by us have been performed on clusters with 30-45 Rh atoms.Since each atom contributes eight electrons, a change of two electrons per cluster amounts to ca. 1%. Our clusters have been chosen such that slab calculations or cluster calculations agree closely. Changes in electrostatic potential are not included in conventional extended Huckel calculations; so a change in dipole moment of a chemisorbed molecule will not result in a change in potential. If coadsorbate interactions are dominated by changes in electrostatic potential, such changes have to be explicitly incorporated in the potential parameters. We have demonstrated earlier,' in a study of the coadsorption of potassium and cobalt on the Pt"' surface, the electrostatic effects can be incorporated very well in extended- Huckel-type approaches. The change of CO by coadsorption of NO to the bridge site 3 03304 General Discussion as a result of the NO negative charge and its image, appears to be similar to the effect of coadsorbed positively charged potassium ions.1 R. A, Santen, f r o c . 8th Int. Cong. Catal. (Verlag Chernie, 1984), vol. IV, p. 97. Dr Nieuwenhuys also replied to Prof. King's first comment: The results described by Prof. King are very interesting. Again they show that co-adsorption of CO and NO can lead to the occupation of surface sites different from those used by the individual molecules CO and NO. Similar effects were reported for the first time for CO and NO on Ru ( O O l ) . ' Some preliminary results have been obtained in our laboratory2 by absorption infrared spectroscopy for mixtures of CO and NO on supported Pt-Rh alloys.On both metals CO prefers the atop sites. We found that at 300 K CO is adsorbed on Pt sites and NO on Rh sites. At higher temperatures the situation is more complex due to NO dissociation. 1 P. A. Thiel, W. H. Weinberg and J. T. Yates, J . Chem. f h j s , 1979, 71, 1643. 2 R. F. van Slooten and B. E. Nieuwenhuys, to be published. Dr Nieuwenhuys continued with a question for Prof. van Santen: Several years ago I collected literature values for the initial heat of adsorption of CO on the densely packed surfaces of the Group VIII metals.' It appears from these data that the differences in heat of adsorption on the various Group VIII metals are small. For surfaces with the same structure the differences are smaller than 20 kJ mol-' (or 20%) with the notable exception of Pd.Pd distinguishes itself from the other Group VIII metals by its preference to adsorb CO in bridged or multi-fold positions, as has been discussed in Prof. van Santen's paper. Its heat of adsorption is also significantly higher than on, for example, Ni, Rh or Pt. I wonder whether Prof. van Santen would care to speculate about the possible reasons for the relatively high heat of adsorption on Pd. 1 B. E. Nieuwenhuys, S u r - Sci., 1983, 126, 307 Mr P. R. Davies (University of Wales College of Cardig) said: Recently in Cardiff we have looked at the effects of the coadsorption of molecules at surfaces. An example of the kind of phenomenon that we have observed is the effect of submonolayer quantities of ammonia on the rate of oxidation of zinc by dioxygen at low temperatures (ca.100 K). An increase in rate by a factor of 10' has been observed in the presence of only 20% of a monolayer of ammonia.' Our studies suggest that the ammonia stabilises the transient surface species 0; (a), and so enhances the probability of its decomposition: O,(g) * O,(a) + O I ( ~ ) -+ O-(a) -+ O'-(a) We believe a 'through-metal' interaction is responsible for this effect: Would your model be capable of providing information on this interesting system? I A. F. Carley, M. W. Roberts and Song Yan, Catal. Lett. 1988, 1, 265. Prof. R. W. Joyner (University of Liverpool) then commented: Prof. van Santen has presented some very interesting calculations on the dissociation of carbon monoxide, which can be a very important step in catalysed reactions of synthesis gas.I would like to mention some calculations, using a quantum-mechanical adaptation of the effective- medium theory,' which relate to a nickel surface and to the reactivity of the carbidoGeneral Discussion 305 carbon species (chemisorbed carbon atom), which is the first product of CO dissociation. This can be readily hydrogenated, in which case methane or higher hydrocarbons are formed. It may also become part of a graphitic layer; this is much less reactive and indeed may be regarded as a catalyst poison. Calculations show that it is not possible for an ordered carbide layer to coalesce into a graphite monolayer; even at very high coverages the tendency of the carbido carbon is to dissolve into the bulk of a nickel catalyst.The only way in which a graphite layer may be formed is by a nucleation and growth process, and the calculations indicate that the nucleus must contain more than three carbon atoms. The calculations, which will be described in detail elsewhere,’ also demonstrate the power of the effective-medium approach for studying questions of catalytic relevance. 1 J. K. Norskov and N. D. Lang, Phys. Rev., 1980, B21, 2131. 2 G. R. Darling, J. B. Pendry and R. W. Joyner, Surf: Scz., in press. Prof. R. G. Copperthwaite ( University of the Witwatersrand, Johannesburg, South Africa ) regarding Prof. Joyner’s comments (communicated ): Prof. Joyner has reported some very interesting results on his effective-medium theory calculations for carbide interactions on nickel surfaces.Our group is particularly interested in the catalytic differences between cobalt and iron in the context of the Fischer-Tropsch reaction, and the undoubted role of carbon in this reaction. As is well known, iron forms several carbides in the presence of CO and, eventually, much graphite. Cobalt, on the other hand, has fewer bulk carbides, and these differences are reflected in, for example, hydrocarbon and oxygenate product selectivities over these two metals. It would there- fore be of some interest to apply effective-medium theory to this problem to see whether it can explain some of the important differences between the behaviour of iron and cobalt towards surface carbon species. Prof. Joyner responded: I accept Prof. Copperthwaite’s point and we may examine the contrast between iron and cobalt in the future. Previous theoretical work has been able to distinguish between iron and nickel, where carbido carbon is stable, and copper, where no stable surface carbide has been reported.’ It may be possible, and it would certainly be interesting to pursue the comparison between cobalt and iron.1 R. W. Joyner, G. R. Darling and J. B. Pendry, Surf Sci., 1988, 205, 513. Dr M. Bowker (University of Liverpool) then addressed Dr Nieuwenhuys: My com- ment relates to the apparent ease of oxidation of the surface of the Pt/Rh alloy surface as illustrated in fig. 3 of Dr Nieuwenhuys’ paper. The shape of curve ( b ) is rather surprising. First, the authors state that the surface oxidises very rapidly (only 1001 of gas dosed) implying that diffusion of oxygen in the lattice is facile even at room temperature.The driving force for adsorption and catalysis (and segregation) is the high surface free-energy at the interface, and so it is surprising that as the oxygen atoms are removed from the surface during the heating sequence of fig. 3 (presumably a very slow rate since the background pressure of these gases in the system is low) the subsurface atoms do not segregate to the surface to fill the vacated sites. Is it possible that the processes occurring during thermal treatment of the alloy surface are really more complex than has been considered? Dr Nieuwenhuys replied: The purpose of fig. 3 is to illustrate the difference in the behaviour of Rh-rich (or pure Rh) surfaces and Pt-rich (or pure Pt) surfaces towards oxygen.I agree that 0 diffusion from the bulk to the surface should be a fast process under our experimental conditions. The figure shows that the rate of oxygen removal from the surface is greater than the rate of oxygen supply from the bulk to the surface.306 General Discussion As we have discussed in our paper the most likely mechanism for the 0 removal is reaction with CO or hydrogen present as residual gases. Admission of oxygen always produces a higher CO background pressure. At tem- peratures above ca. 600 K the reaction rate of adsorbed 0 with CO probably decreases because of the lower surface concentration of CO (see fig. 6). As a result, the surface concentation of 0 increases above ca. 600 K. Dr P.A. Sermon (Brunel Uniuersity, Uxbridge) (communicated): The following results (fig. 1) obtained by Valerie Self in our laboratory suggest that F.t.i.r. microspectroscopy may in future enable differentiation and identification of various types of surface CO at steps and terraces of metals with spatial resolution during adsorption and catalysis. It may reveal the extent of formation of reactant islands on the surface, for example, and how the surface morphology affects the concentrations of CO in different surface configurations. Dr M. Bowker opened the discussion of Prof. King's paper: I would like to comment on Prof. King's proposal that it is likely to be the hydrogenation of nitrogen atoms on W{lOO} which would be rate-limiting for ammonia production. In work carried out with K.C. Waugh at I.C.I.' we calculated the rate of ammonia production using surface science data for the kinetic parameters involved, using a mechanistic scheme identical to that in fig. 3 of Prof. King's paper. We found that the rate-limiting step was the dissociation of adsorbed nitrogen molecules. where the last term is the number of free sites on the surface, 8 being the total surface coverage by adsorbates. The rate-limiting step was dominated by this term and the reaction was 'self-poisoning'; the surface is so greatly covered by adsorbate that there are very few sites left for adsorption. All other steps were at equilibrium and went very much faster than the rate-limiting step (see table 1). The depth of the summed adsorption energies for W are higher than on Fe even at half-monolayer coverage of nitrogen (see fig.3 of Prof. King's paper). Certainly, the potential barrier shown in that figure is very much bigger than we considered for Fe, but the exact height of the barrier is open to some speculation, as indicated by the dotted lines in the figure. Both the size of this barrier and the depth of the adsorptive well will probably lead to a similar 'self-poisoning' situation as on Fe, with almost complete blockage by Na although the kinetics proposed by Prof. King should be tested using the kinetics computer program to check if this is indeed so. 1 M. Bowker, 1. Parker and K. C. Waugh, Surf Sci., 1988 197, L223. Prof. King replied: There are two factors that lead us to conclude that the step N(a) + H(a) --* NH(a) is rate-limiting in ammonia synthesis over W{ loo}.(1) The zero coverage sticking probability of N2 on W(100) is in the range 0.2-0.6, depending on the substrate temperature;' for N, on Fe surfaces this sticking probability is about a factor of 10' smaller. (2) From our estimates of the potential barriers (fig. 3 of our paper) the step involving the hydrogenation of nitrogen adatoms involves a very large activation energy barrier, of 190-240 kJ mol-'. Thus, under reasonable conditions, the removal of nitrogen by adsorbed hydrogen would be so inefficient that the surface would rapidly saturate with nitrogen adatoms; this would presumably block sites for hydrogen adsorption, and we might even conclude that hydrogen adsorption was then rate-limiting ! However, by reducing the nitrogen pressure by a factor of about 10' the nitrogen adsorption rate could be made to mimic that observed for nitrogen on Fe surfaces.In that case the reaction would again be blocked by the N(a) hydrogenation step.General Discussion 307 CO chemisorption h I EO oxidation Fig. 1. Reflectance F.t.i.r.-microspectrometry during CO chemisorption and oxidation on Pt/HOPG; spectra were recorded at equally spaced points across the surface. HOPG is a crystal of highly orientated pyrolytic graphite. All measurements were after adsorption or catalysis at 423 K. Clearly the ratio of linear/bridge CO species @ / a ) varies across the surface in chemisorp- tion. In addition, during oxidation, it is the bridge species which appear to react, preferentially (possibly because of their ease of transition to CO-, or because these are at the edge of CO islands at whose perimeter the reaction occurs) but again surface CO? is not seen over the whole surface.308 Genera 1 Discussion Table 1.Rate-limiting steps for ammonia production on Fe( 11 1 ) coverage reaction rate major surface (% of reaction step log A E/kJ mol-' /mol cm-' s-' species total sites) H2 ---* H,(a) 13.2 t 13.2 H,(a) - 2H(a) 21.2 t 21.2 N3 ---* N,(a) 11.2 t 10.2 N,(a) ---* 2N(a) 15.2 t 21.1 N(a) + H(a) -NH(a) 20.8 t 21.0 NH(a) + H(a) +NH,(a) 20.7 t 21.0 NH2(a)+H(a) ---* NH,(a) 21.6 t 21.6 NH,(a) ---*NH3 13.0 t 12.0 0 0 0 92 0 46 31.3 198.7 64.9 18.8 64.9 18.8 64.9 18.8 52.7 0 1.050 x lo4 1 .o50 x lo4 2-638 x lo-' 2.637 x l o - ' H(a) 3.499 x 10' 3.499 x 10' N,(a) 2.326 x lo-' N(a) 6.928 x 10" NH(a) 1.017 x lo-' 6.928 x 10" 1.261 x lo-' 1.261 x l o - ' 1.954 x lo-' 2.694 x 10' 2.694 x 10' 1 *954 X lop2 12.1 1.5 84.4 1.9 This conclusion is supported by the computational work of Bowker et al.' who concluded, for the case of 'normal' values for the desorption pre-exponential (as would be the case for N2 from W{lOO}), that 'the energy well of the adsorbed nitrogen atoms is too deep, so that it is not the rate of dissociative nitrogen adsorption which is rate-determining in this model, rather it is that of the subsequent hydrogenation steps'. Our experimental work on W{lOO} is in very close agreement with this conclusion from computer modelling.1 P. AInot and D. A. King, Surface. Sci., 1983, 126, 1983, 126, 359.2. M. Bowker, I . B. Parker and K. C. Waugh, Appl. Catal. 1985, 14, 101. Dr R. A. Hadden (1.C.I. C&P Group, Billingham, Cleveland) commented: The sche- matic potential-energy diagram for ammonia synthesis (fig. S), which Prof. King has derived from a concise molecular beam study of the ammonia decomposition reaction, clearly shows that adsorbed nitrogen is strongly adsorbed on the tungsten surface. Curiously, however, it may be possible that this information is not always sufficient to make the claim that this is the reason for the poor synthesis activity of a metal surface. The ammonia synthesis reaction has also been studied on a variety of rhenium single-crystal surfaces.' The heat of adsorption of atomic nitrogen on these surfaces (263-210 kJ mol-I) is relatively high, yet the synthesis rate appears to be dominated by the slight temperature dependence ( EA = 6-14 kJ mol-I) and low sticking probability of the nitrogen adsorption process.* Indeed, more recent studies of the synthesis reaction on rhenium surfaces3 have shown that the steady-state surface nitrogen concentration during high-pressure ammonia synthesis is low. There is no evidence to suggest a build-up of nitrogen-containing species on the rhenium surfaces, as would be expected if hydrogenation of nitrogen was the rate-determining step in the synthesis reaction.1 M. Asscher, J. Carrazza, M. M. Khan, K. B. Lewis and G . A. Somorjai, J. Catal., 1986, 98, 277. 2 G . Haase and M. Asscher, Surj Sci., 1987, 191, 75. 3 R. A. Hadden, P. Ramirez de la Piscina and G. D. Somorjai, in preparation. Prof. King responded: Our molecular-beam results for the sticking probability of nitrogen on W{ loo}, published elsewhere,'*2 complement the ammonia decompositionGeneral Discussion 309 study over the same surface reported in the present paper. There, we demonstrated that the probability for dissociative adsorption is very high. In particular, at 300 K it is about 0.6 on a clean surface, and even at relatively high substrate temperatures the process is efficient. Thus, at a crystal temperature of 1000 K the dissociative sticking probability is 0.1 on a clean surface. Our conclusion regarding the bottleneck in the ammonia synthesis process on W{ 100) is therefore not only based on our derived potential-energy diagram; it is also derived from our knowledge of the efficiency of dissociative nitrogen chemisorption on the same surface. 1 D. A. King and M. G. Wells, Proc. R. Soc. London, Ser. A, 1974,339, 245. 2 P. Alnot and D. A. King, Sud Sci. 1983, 126, 359.

ISSN:0301-7249    

DOI:10.1039/DC9898700303

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Furuduy Discuss. Chem. SOC., 1989, 87, 311-320 Adsorption and Reactions of Ethyne Effects of Modifiers and Formation of Bimetallics Nicolaas R. M. Sassen, Adrianus J. den Hartog, Fred Jongerius, Jacques F. M. Aarts and Vladimir Ponec" Gorlaeus Laboratories, Leiden University, P. 0. Box 9502, 2300 RA Leiden, The Netherlands Adsorption of C,H2 and C2H, has been studied with Pd(ll1) and Pd- Cu( 11 1) single-crystal planes using EELS. The formation of dissociatively adsorbed species is controlled by ensemble size effects. The catalytic experi- ments performed with powder catalysts reveal that formation of bimetallics from an active and an inactive metal or diminishing the particle size increase the selectivity to ethene. This supports the idea that the increase is due to suppression of dissociatively adsorbed, possibly multiply bound, species.Studies on selectivity of catalysts are becoming more frequent. Increasing prices of raw materials and energy on the one side and increasing environmental problems on the other force the industry to look for more selective catalysts. Knowledge concerning the selectivity in catalytic reactions is improving and its empirical character is slowly but steadily being converted into scientific theories.' However, we are still in the early stages of formulation of general principles governing selectivity. An interesting field to explore is the selective hydrogenation of ethyne to ethene. A reaction which has been first treated as a system of purely consecutive reactions (C2H2- C,H,-C,H,) appeared to be a reaction with two parallel pathways to ethane:2 one upon which intermediates do not leave the surface until they are fully hydrogenated and the other involving the already mentioned consecutive reaction.This paper attempts to identify the factors which influence the relative contributions of the two routes. Experimental Vibrational spectra have been obtained using an EELS apparatus (Leyboldt, Heraeus). A double-pass 127 O monochromator and analyser allowed measurements with a resolu- tion usually > 10 meV (80 cm-I). All spectra were recorded in a specular mode. Reactions of ethyne were studied in a static low-pressure apparatus (background pressure with powder catalysts in situ better than lo-, Torrt); C2H2/H2 = 0.3 and the total pressure was ca. 1 Torr.The products are analysed by means of a mass spectrometer through a continuous leak. The temperature range studied was 273-373 K. The products were analysed using the m / e peaks at 30, 27 and 26, respectively, for ethane, ethene and ethyne, whilst the whole fragmentation pattern was used to check the reliability of the analysis. The selectivity to ethene is defined as [C=C] [C=C]+[C-C] S = x 100%. The bimetallic powders were prepared by coimpregnation ( 5 wt '/o loading) of S O 2 with appropriate amounts of Ir(NHJ2CI, and CuCI, or Au compounds dissolved in aqua regia and drying at 400 K followed by reduction. Particle sizes were determined t l Torr = 101 325/760 Pa. 311312 Adsorption and Reactions of Ethyne by X-ray diffraction and were found to be ca. 7 nm for all catalysts. No extended alloy formation was found using this technique, by monitoring shifts in the peak position (and, of course, none are expected since these are systems with a limited solubility).Reduction at 670 K for 8 h ex situ was followed by reduction at 520 K for 8 h in situ. A series of Ir catalysts with varying particle sizes was prepared as follows. The catalyst with 0.6% Ir loading and 1 nm (TEM) Ir particles was prepared by ion exchange of Ir(NH,)2C16 with Na-zeolite (pH 12, NH40H) at 373 K. The catalyst with 0.7% Ir loading and 2 nm (TEM) particles was prepared by homogeneous precipitation,’ whereby Ir(NH4)$& is precipitated as Ir(OH), by slowly decomposing (373 K) urea on the surface of silica. The catalyst with 2.5% Ir loading and particles of 5.3 nm (XRD line broadening) was prepared by a classical impregnation technique.The largest-particle Ir catalyst, 13 nm (XRD line broadening) was prepared by reducing a mechanical mixture of the precursor and SiO, with 10 wt ‘/o Ir. All catalysts were calcined in oxygen (21 h) at 570-670 K, these temperatures being reached by slow heating (0.5-1.0 K min-I). After flushing the catalyst with N2 reduction took place, as with the bimetallics. Results The results presented below concern two topics: ( a ) chemisorption monitored by EELS when using single-crystal planes [ Pd( 11 1) and Pd/Cu( 11 l ) ] as adsorbents, and (6) catalytic reaction followed in a closed static low-pressure apparatus, using Ir and Ir-bimetallics on SiOz powder catalysts. A certain (albeit loose) relation exists between ( a ) and ( 6 ) which justifies the simultaneous presentation.Discussion of the conclusions on the catalytic reaction is separated, but the discussion concerning the identification of species by EELS is included in this section. EELS on C2H4 and C2H4 Adsorption Adsorption of ethene, the product of ethyne hydrogenation or disproportionation, was monitored first. Fig. 1A shows the most typical result obtained with Pd( 11 1). It should be mentioned that spectra of both ethene and ethyne show also a broad absorption at ca. 2900 cm-’, and some spectra also exhibit a band due to traces of (background) CO. Our results on ethene adsorption (also those not shown here) are in good agreement with the 1iteratu1-e.~” Bandy et al.’ collected convincing arguments that the 940 cm-’ band should be ascribed to .rr-complexed ethene.According to these authors T - complexed C2H4 is likely to be a precursor of other types of adsorption. On more reactive surfaces (e.g. Ni) the population of ?r-complexed C2H4 can be kept higher (at an observable level) by poisoning the surface; for example by deposition of a car- bonaceous layer on it. The pair of peaks at 1340 and 1090cm-’ has been a matter of a longer discussion. Originally the peaks were ascribed to ethylidene (=CH-CH3), but later stronger evidence was gathered for ethylidyne (=C-CH3) being responsible for these peaks. The analysis by Skinner et a1.6 of the i.r. data for the model compound CH,C-CO~(CO)~ was decisive. These authors identified the vibrational frequencies of the CCH3 group, which are in close correspondence with features observed in EELS for adsorbed C2H4.Moreover, the intensity ratio of the symmetrical CH3 deformation at 1340 cm-’ and the C-C stretch at 1090 cm-’, which both belong to the same symmetry species, was found to be near unity in the i.r. and EEL spectra, implying a similar rat0 for the relevant dynamic dipole moments. Fig. 1A also shows that when the system is heated the 7.r-complexed C2H4 disappears and the ethylidyne absorption grows. Fig. 1 shows the most relevant part of the results obtained upon adsorption of ethyne. One can observe the following important features: ( a ) at the lowest exposures onlyN. R. M. Sassen et al. 313 wavenumber/cm-' wavenumber/cm-' 9t0 1090 1340 I t I A I 100 200 energy loss/meV I 9401090 1340 I I I 1 I 100 200 energy loss/ meV Fig.1. (A) EEL spectra upon adsorption of CzH4 on Pd: ( a ) 3 L exposure at 253 K. Next to molecularly adsorbed (r-complex) C2H, (940 cm-'), ethylidyne is present (1340 and 1090 cm-'); ( b ) spectrum after slow heating to 293 K. (B) Adsorption of CzH2 on Pd. Changes in spectra with increasing exposure: ( a ) 0.2 L, ( b ) 0.6 L, ( c ) 3 L. .rr-complexed C2H4 is observed; (6) at higher exposures, formation of ethylidyne is observed, as characterized by the typical ethylidyne ratio P = 1.0. At the highest exposures, when the surface becomes more crowded, the two peaks at 1340 and 1090 cm-' no longer grow equally. P is now clearly greater than one. The most likely conclusion is that at least two species contribute to the pair of absorptions. Another interesting point is that the adsorption of the .rr-complexed C2H4 is diminished when ethyne is present at the highest exposure.The absorption at 940 cm-' clearly decreases, since molecular ethene is displaced by ethyne. It is not possible to ascribe at the moment the two other peaks (650 and 830cm-I) definitely. Let us mention only that molecular ethyne is expected to show bands in the 670-870 region,' and benzene, formed by trimerization, shows two strong losses in the region 700-900 cm-I. Formation of C2H4 and C,H,-* species upon adsorption of C2H2 shows that some molecules have gained an H atom. This can happen either by the known process of redistribution of hydrogen (disproportionation)' or by traces of endogeneous hydrogen not removed upon evacuation and present even at pressures of the order of lO-''Torr.Experiments with C2D2 showed that at least a part of the mentioned hydrogenerated species is formed by endogeneous hydrogen (C2H4 was formed from C2D2). Further adsorption of ethene and ethyne has been monitored on the (1 1 1 ) plane of a Pd/Cu alloy single crystal. According to an earlier study" the sample with 75% Pd in bulk should have ca. 70% Pd in the outermost layer. When ethene is adsorbed on the Pd-Cu (111) surface, the molecular, .rr-complexed C2H4 can be observed again (930cm-'). When the surface is covered by ethene at 253 K to a high extent (exposure314 Adsorption and Reactions of Ethyne wavenumberlcm-' 660 830 1090 1340 1420 n \ I 1 I 100 200 energy loss/meV Fig.2. EEL spectra upon adsorption of C2H2 on Pd/Cu (1 11): ( a ) surface saturated at 253 K (note the new loss at 1420 cm-'); ( b ) after exposure of 6 L at 293 K the gas phase evacuated (the layer dehydrogenated by that) and CzH2 admitted again (4 L) (note the unequal growth of the two losses at 1090 and 1340 cm-'); ( c ) exposure of 3 L at 293 K. 3 L)? and then slowly heated ( 1 K min-') molecular C2H4 was desorbed and no other species appeared. However, upon rapid heating (10 K min-') ethylidyne (1340, 1090 cm-I, P = 1 ) appears. This shows that the formation of ethylidyne, a species multiply bound to the metal surface is less easy on Pd-Cu than on Pd, and upon slow heating C2H4 is desorbed more rapidly than it is converted into ethylidyne.Results obtained upon adsorption of ethyne on Pd-Cu are shown in fig. 2. Upon adsorption at 293 K (3 L), ethylidyne is mainly formed. When a carbonacous layer is deposited first on the surface (6 L ethyne admitted and the gas phase evacuated for several hours) the exposure of additional ethyne (4 L) leads to a spectrum shown in fig. 2(b). We see that the ratio of intensities I ( 1340)/ I ( 1090) is considerably higher than unity, implying that another species is formed simultaneously with or instead of ethylidyne. We have seen that with pure Pd this situation could arise merely by admitting 3 L ethyne to the surface. With Pd-Cu [see fig. 2(c)] ethylidyne is the main product upon admission of 3 L ethyne, and a slightly more severe procedure is necessary to produce a surface (obviously a surface with a hydrogen-lean carbonaceous layer) on which the new species and/or ethylidyne are formed.The observation of a strong peak at 1340 cm-' and a weak one at 1090 cm-' points to an adsorbate containing a CCH3 group, but these peaks could also arise from a species which is different from ethylidyne. Although additional spectral evidence is lacking, we suggest that this species is ethylidene (=CH-CH3) which is bound only t l L= 10'Torr s.N. R. M. Sassen et al. 315 by a double bond (ethylidyne is triply bound) to a surface and which contains an additional H atom. A difference in the CC stretch and CH3 deformation dynamic dipole moments would be then responsible for the change in the I ( 1340)/ I ( 1090) ratio.The tilt of the molecular axis can also have the following effect on the spectral visibility of vibration modes. The unperturbed symmetric deformation vibration of CH3 would become less interacting and the perturbed asymmetric deformation vibration (a down- shift due to the interaction with the surface) would become more strongly interacting with the incoming field. When ethyne is admitted to saturation level ( i e . no further change is observed upon a further increase in exposure) at 253 K, a complex spectrum is observed. Molecular ethyne is most likely responsible for the absorptions at 660 and 830 cm-'; ethylidyne is present next to it and a new peak is observed at 1420cm-'. In compliance with the ideas expressed by other authors" we suggest that this is an absorption by a vinylidene group (C=CH2).Next to it, bands which we saw also with pure Pd (only they are more intense here) at 660 and 830 cm-' are observed. The industrial catalysts work under conditions at which the surface of the metal is most likely covered by a carbonaceous layer (a deposit with a very low H/C ratio). It is therefore interesting to see which changes occur in the spectrum after accelerated formation of such a layer. When a layer containing ethylidyne species is heated, ethylidyne decomposes and is possibly also desorbed. After heating to 473 K, the main absorptions which survive are at 750 and 2990 cm-' and they have been ascribed in the literature12 to M-C,-H,, vibrations (x > y ) . Such species have been observed with Pd, Pt, Ir and Ni, thus they.are formed quite commonly. Their effect on ethyne adsorption is shown in fig.3(b). When heating of the adsorbed layer to 473 K is repeated, the species produced by dissociative adsorption suppress entirely the formation of ethylidyne upon re-admission of ethyne. Heating an adsorbed layer to 723 K produces a surface showing no detectable absorption which could be ascribed to any C-C or C-H vibrations. However, the influence of such a layer on a subsequent readsorption is the same as that exerted by the layer formed at 473 K. The surface with carbonaceous layers again shows absorptions which correspond with 'C=CH2' like (vinyl and vinylidene group) vibrations. Obviously, these species appear upon ethyne adsorption only when the reactivity of the surface is first suppressed.Working industrial catalysts are not only modified by other elements (Pb, S) or carbonaceous layers but also by exposure to the air (02). It is interesting to observe the consequences such exposure may have. Fig. 4 shows that after adsorption of oxygen at 293 K (3 L) on Pd-Cu(lll), three peaks appear at 410, 830 and 1050 cm-'. Such losses were also observed with Pd(ll1) and Ir(ll1) and also polycrystalline C U . ' ~ According to the literature a band at 850 cm-' has been ascribed to peroxo (Os-) species and a band at 1035 cm-' to superoxo (0,) species; the metal-atomic oxygen bond on Pd( 11 1) absorbs at 500 cm-' and the metal-0, bond absorbs at 400 cm-'. However, we believe that the bands observed here (with Pd-Cu) should be preferably associated with those bands which have been observed with Ir( 11 1) after oxygen adsorption at 600-800 K.The latter paper ascribed the 800 cm-' loss to a surface oxide and the loss at 1005 cm-' has been related to the subsurface atomic ~ x y g e n . ' ~ The temperature (293 K) is rather high to expect molecular forms of 0, on Pd-Cu and, moreover, the observed bands remain in the presence of C2H2 and do not diminish at elevated temperatures. Oxygen vibrations disappear at ca. 800 K. Upon heating (293-800 K) the peak at 410 cm-' shifts to 390 cm-', that at 830 cm-' shifts to 700 cm-' and the highest-frequency loss shifts from 1000 to 1100 cm-'. When a surface precovered with oxygen (3 L exposure) is exposed to ethene only a small new peak appears at 1340 cm-'. This indicates that the dissociative adsorption or shifts of hydrogen atoms leading to ethylidyne are substantially suppressed by oxygen316 Adsorption and Reactions of Ethyne wavenumber/ crn- ' I energy loss/rneV Fig.3. EEL spectra upon adsorption of C2H2 on Pd/Cu( 11 1): ( a ) after exposure of 3 L at 293 K; ( b ) the layer from ( a ) heated to 473 K, and admitting C2H, after cooling down to 293 K (exposure 10 L); ( c ) obtained after exposure of 3 L at 253 K on a surface which was prepared in the following way to simulate the working surface of a catalyst: exposure of 3 L at 293 K, heating of a layer up to 723 K, sample cooled down in lo-' Torr Hz, sample evacuated; ( d ) the layer from ( c ) heated up to 293 K. preadsorption (3 L). If a surface covered with oxygen (3 L exposure) is exposed to 10 L ethyne at 293 K the spectrum shown in fig.4 is observed. We observe the oxygen losses again and next to them the dominating 1340 cm-' loss with a shoulder at 1420-1450 cm-'. Loss at 740 cm-' can most likely be attributed to a C,H bond vibration. The spectrum also has a very intense C-H loss at 2990 cm-' (not shown here). When this layer is annealed to 373 K, the 740 cm-' loss gains intensity at the expense of the 1340 cm-' loss. At 473 K the latter loss disappears completely. Above 473 K both C-H losses (740 and 2990 cm-') decrease and at 723 K the peak at 2990 cm- ' disappears completely. The peaks which remain (790 and 1090 cm-') are ascribed here to oxidic and sub-surface oxygen. Summarizing, we see again that the presence of blocking species (oxygen atoms) suppresses the formation of ethylidyne and promotes the detectability of ethylidene and vinyl-group vibrations. Catalytic Measurements We have seen in the foregoing section that alloying of Pd with a less active metal, or the blocking of the surface (by 0 or C) suppresses the formation of the species bound to the metal surface by a bond with the highest multiplicity (3) and probably requires the largest ensemble of active (Pd) atoms.We expected that in parallel with this otherN. R. M. Sassen et al. 317 wavenumber/cm-' I energy loss/meV Fig. 4. EEL spectra upon adsorption of C2H2 on Pd/Cu(lll) with preadsorbed oxygen: ( a ) preadsorbed oxygen, 293 K, 3 L. * = 1420-1450 cm-'; ( b ) adsorption of (10 L) C2H2 on the preadsorbed oxygen.100 Ir/Cu t 1 I 1 0 50 100 Ib (atom YO) Fig. 5. Effect of Cu and Au on the selectivity to ethene of the Ir catalysts. Selectivity as a function of the content (atom YO) of the Ib metal. (-) 373 K, (---) 323 K, (- - -) 295 K.318 Adsorption and Reactions of Ethyne 50- 25- 0 ; I I I 0 5 10 15 dlnm Fig. 6. Variation of the selectivity to ethene formation of reactions with C2H2-H2 mixtures, as a function of Ir particle size. intermediates (multiply bound or formed by dissociative adsorption of ethyne or ethene) on the surface which would lead to a non-selective hydrogenation up to ethane, would be suppressed by alloying or by diminishing the particle size (the latter is known to suppress the formation of multiply bound specie^).'^ However, Pd was expected to be a difficult system for which to find evidence by catalytic measurements since the selectivity with this metal is already high.With pure Pd without modifications the selectivity is nearing 100°/~. However, iridium, which intrinsically has a very low selectivity” offers more scope for observing selectivity changes caused by catalyst manipulation, therefore we have started with that metal. The effect of forming bimetallics (alloying) of Ir with Cu or Au is quite pronounced. A higher selectivity than with pure Ir is observed under comparable conditions with both bimetallic systems (fig. 5 ) . Selectivity to ethene also depends on the particle size of Ir. Smaller particles are more selective to ethene, as seen in fig. 6 . As expected, it was more difficult to arrive at clear-cut conclusions with Pd.I6 In the static system used by us the final selectivity with Pd is not reached immediately, but is a function of conversion. In the first stages of reaction more C, and C, hydrocarbons are formed than in the latter stages, and the selectivity to ethene (related to the total C2 hydrocarbon production) increases in the course of the reaction.Only at the end of the repeated run (with the same surface) does the selectivity to ethene level off. This final selectivity shows very little variation with the particle size. Two other metals were also studied as far as the variations of the selectivity with particle size are concerned. Rhodium behaves like Ir, but the variations are less pronounced than with IT,” and Pt resembles Pd.16 Discussion It is generally accepted’’ that a fraction of ethane is formed from ethyne by a consecutive reaction in which weakly adsorbed ethene plays the role as an intermediate.Ethene can be displaced from the surface by ethyne, which is more strongly adsorbed thanN. R. M. Sassen et al. 3 19 ethene. This is the so-called thermodynamic factor in controlling the selectivity to ethene. The step underlying this mechanism, the displacement of molecularly adsorbed ethene by ethyne, can be seen by vibrational spectra, as mentioned above. Alloying of Pd with Cu makes the formation of multiply bound ethylidyne from molecularly bound precursors (C2H2, C2H4) slightly more difficult (fig. 1 and 2). Other results presented above show that oxygen or carbonaceous (hydrogen-lean) layers make the formation of the triply bound ethylidyne more difficult too, whereas formation of the doubly bound ethylidene is less negatively affected, or is even promoted.This strongly suggests that multiple bonding or any dissociative adsorption (C-H dissoci- ation) can be influenced by blocking, i.e. it is regulated by the available ensemble size. Moreover, it is feasible to conclude that the higher the multiplicity of bonding between the adsorbate and the metal and the higher the degree of dehydrogenation of the species, the more important is the ensemble size effect in the formation of these species. Let us now turn our attention to the catalytic data. A modification of the surface by a metal which itself is not active (Au) or by a metal which has a low activity but a high selectivity to ethene (Cu), increases S (ethene) considerably.Both Cu and Au are known to be very poor at breaking C-H bonds.I8 It is known from our previous work that decreasing the particle size suppresses the formation of species multiply bound to the metal s~rface.’~ In compliance with this, S(ethene) for Ir catalysts is enhanced when the metal particle size of Ir is diminished. By combining the information on adsorption and hydrogenation reactions obtained here and in the literature2’ a conclusion can be formulated: the non-selective hydrogena- tion of ethyne to ethane proceeds via multiply bound (e.g. ethylidyne and ethylidene) and possibly dissociatively adsorbed (-Cr CH) species. The formation of these species can be controlled by the ensemble size effects. At the moment, a more detailed description cannot be suggested since it is not known whether the splitting in the pathways (ethene us.ethane) occurs at the stage of ethyne adsorption or ethene adsorption. References 1 W. M. H. Sachtler, Faraday Discuss. Chem. SOC., 1981, 72, 7~; M. W. Vogelzang, M. J. P. Botman and V. Ponec, Faraday Discuss. Chem. SOC., 1981, 72, 33. 2 A. S. Al-Ammer and G. Webb, J. Chem. SOC., Faraday Trans. 1, 1978, 74, 657; L. Guczi, R. B. la Pierre, A. H. Weiss and E. Biron, J. Cutal., 1979, 60, 83 3 J. W. Geus, Dutch Patent 6705 259 (1967). 4 H. Ibach and D. L. Mills, Electron Energy Loss Spectroscopy (Academic Press, New York, 1982); L. L. Kesmodel, L. H. Dubois and G. A. Somorjai, J. Chem. Phys., 1979, 70, 2180.5 B. J. Bandy, M. A. Chesters, D. I. James, G. S. McDougall, M. E. Pemble and N. Sheppard, Philos. Trans. R. SOC. London, Ser. A, 1986, 318, 141. 6 P. Skinner, M. W. Howard, I. A. Oxton, S. F. A. Kettle, D. B. Powell and N. Sheppard, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 1203; M. W. Howard, S. F. Kettle, I. A. Oxton, D. B. Powell, N. Sheppard and P. Skinner, J. Chem. Soc., Faraday Trans. 2, 77, 397. 7 J. A. Gates and L. L. Kesmodel, J. Chem. Phys., 1981,76,4281; E. S. Kline, Z. H. Kafafi, R. H. Hauge and J. L. Margrave, J. Am. Chem. Soc., 1987, 109, 2402. 8 G. D. Waddill and L. L. Kesmodel, Phys. Reu. B, 1985, 31, 4940. 9 S. Teratani and K. Hirota, 2. Phys. Chem. N.F., 1970, 69, 271. 10 A. D. van Langeveld, H. A. C. M. Hendrickx and B. E. Nieuwenhuys, Thin Solid Films, 1983,109, 179. 1 1 G. H. Hatzikos and R. I. Masel, Surf. Sci., 1987, 185, 479; J. A. Gates and L. L. Kesmodel, Surf. Sci., 1983, 124, 68; N. R. M. Sassen, Thesis (Leiden University, 1989). 12 L. L. Kesmodel, G. D. Waddill and J. A. Gates, Surf: Sci., 1984, 138, 464; T. S. Marinova and K. L. Kostov, S u d Sci., 1987, 181, 573 [see also ref. (4) and ( l l ) ] . 13 R. Imbihl and J. E. Demuth, Suyf Sci., 1986, 173, 395; T. S. Marinova and K. L. Kostov, Surf: Sci., 1987, 185, 203; J. L. Gland, Surf: Sci., 1980, 93, 487; K. Prabhakaran, P. Sen and C. N. R. Rao, Surf: Sci., 1986, 177, L971. 14 E. H. van Broekhoven, J. W. F. M. Schoonhoven and V. Ponec, Surf: Sci., 1985, 156, 894. 15 N. Yoshida and K. Hirota, Bull. Chem. Soc. Jpn, 1975, 48, 184. 16 M. Deng, unpublished results.320 Adsorption and Reactions of Ethyne 17 F. Jongerius, M.Sc. Thesis (Leiden University, 1987). 18 G . C . Bond, in Catalysis by Metals (Academic Press, London, 1962). 19 V. Ponec. Adv. Catal., 1983, 32, 149. 20 G . C . Bond, D. A. Dowden and N. MacKenzie, Trans. Faraday Soc., 1958, 54, 1537; H. Lindlar, Helv. Chim. Acta, 1952, 35, 446. Paper 8/04782K; Received 28th November, 1988

ISSN:0301-7249    

DOI:10.1039/DC9898700311

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989,87, 321-336 Ensemble and Ligand Effects in Selective Alkane Hydrogenolysis Catalysed on well characterised RhIr and RhFe Bimetallic Clusters inside NaY Zeolite Masaru Ichikawa," Lingfen Rao, Taro Ito and Atsushi Fukuoka Research Institute for Catalysis, Hokkaido University, Sapporo 060, Japan A series of metalfbimetal catalysts has been prepared from molecular car- bony1 clusters such as Rh6-yIr, (CO),,(x = 0-6) and Rh,Fe,(CO):, which were synthesized inside NaY zeolite supercages. Nay-entrapped Rh,, Ir, , Rh41r2, Rh,Ir, and Rh,Ir4 crystallites derived from the precursor carbonyl clusters have been well characterized by f.t.i.r., EXAFS, '29Xe n.m.r. and CO/ H chemisorption on their metallic structures and electronic states. The results suggested that the reduced crystallites inside NaY zeolite consist of cluster ensembles < 10 8, in size with controlled metal compositions.These ensembles are fairly stable through several cycles of oxidation, reduction and alkane hydrogenolysis. As the probing reactions, hydrogenolysis of n-butane and ethane, and benzene hydrogenation were conducted. There is a dramatic decrease of hydrogenolysis activity by four orders of magnitude across the series of clusters on increasing the number of Ir atoms in the active Rh ensembles, while a slight enhancement of activity is observed for benzene hydrogenation. The remarkable activity suppression of hydro- genolysis has been interpreted not simply in terms of the geometric Rh ensemble-size effect with a small amount of Ir atoms but of the different electronic states of the RhIr bimetallic clusters in decreasing the electron deficiency through the series clusters.This was demonstrated by the unusually large chemical shifts of n.m.r. signals of Iz9Xe adsorbed on Nay-Rh-rich clusters in reflecting the higher electron deficiency. For butane hydrogenolysis, the smaller Rh ensembles in bimetallic RhIr clusters gave the maximum selectivities towards the central C-C bond scission to give C2H,, higher than those on the Rh, and Ir, inside zeolites. In contrast, RhFe/ NaY derived from [ Rh,Fe,(CO),,]'-/ NaY exhibited conversely a high selectivity towards terminal C-C bond scission to give CH4+ C3H,. The Fe promotion for the terminal C-C bond scission is suggested to be associated with the electronic ligand effect on the heteronuclear sites consist- ing of Rh-Fe3+ located on the internal zeolite cages.Homo-/ hetero-metallic carbonyl cluster compounds have been used as the precursors for the preparation of size-controlled metal crystallites in the range <10 A, and with a controlled composition of bimetal particles highly dispersed on some suitable supporting materials such as S O z , A1203, MgO and carbon. 1 - 3 The structural characterization of the resulting cluster-derived metal catalysts has previously been conducted by means of EXAFS, coupled with Mo~sbauer,~ high-resolution electron microscopy,' XPS6 and TPD-i.r.' It has been also demonstrated that a remarkable promotion of Fe in CO hydrogenation8 and alkene hydroformylation' on Si0,-supported Rh4Fe2, Ir4Fe and Pd&, carbonyl cluster catalysts is associated with the cluster-derived heteronuclear adjacent Rh-Fe3+ (or Pd-Fe3+) sites which markedly enhance CO migratory insertion into metal-H and metal-alkyl groups and favour higher-alcohol formation.321322 Selective Alkane Hydrogenolysis The present work shows that the synthetic and structural methodology for well characterized metal/ bimetal cluster catalysts can be provided using suitable precursors such as zeolite-entrapped hexanuclear Rh carbonyl cluster compounds combining disper- ate elements, e.g. Ir and Fe. More significantly, it indicates that the catalytic properties derived are substantially linked to the specific cluster precursors having well defined metal compositions and discrete metal frameworks. This result offers the possibility of enhanced control and even the effective design of tailored bimetal catalysts having well characterized crystallite sizes and metal compositions, in probing the structure-sensitivity of some catalytic reactions such as alkane hydrogenolysis.There have been many reports in the last few years about the hydrogenolysis and isomerization of alkanes such as n-butane, hexane and ethane catalysed by unsupported and supported noble metals, e.g. Pt, Rh and IT,'' including metal films and single crystals' ' associated with mechanisms involving rearrangement or splitting of the hydro- carbon chains. A general conclusion is that in many cases the reactions are sensitive towards metallic structures relating the ensemble sizes, metal composition of the active sites and the different electronic states of bimetal 'alloy' ensembles compared with those of the pure metal components.Such activity changes have been observed in the catalytic behaviour of Group VIII-IB mixtures, e.g. Ni-Cu, Ru-Cu," Ni-Ag and Pd-Ag." The question arises as to whether similar behaviour would be expected if the incorporated foreign atom were actually similar to the host, although less active. Haller and co- w o r k e r ~ ' ~ previously studied Si0,-supported conventional RhIr bimetallic alloy catalysts (characterized by XPS and H-CO chemisorption) and suggested that the TOF of butane hydrogenolysis decreased in a non-linear manner on moving from Rh to Ir, which is less active, although the trend was less marked than that reported previously for Ni-Cu and Ru-Cu by Sinfelt and co-workers.'* This is due to the surface enrichment of active Rh on the conventional RhIr catalysts.They discussed the activity suppression of wbutane hydrogenolysis simply on the geometric basis of different numbers of atoms in the ensembles which constitute the sites. In a recent study, Sinfelt reported that alloying Ir with Pt and Rh results in catalysts for hydrocarbon conversion with markedly improved stability compared with the pure metals. Sinfelt and c o - ~ o r k e r s ' ~ later found in EXAFS studies that Al,O,-supported Pt-Ir and Rh-Ir mixtures form pt (or Rh) surface-enriched alloy particles even at low loadings. Ponec and Kuirjers,16 using Ir-Pt thin films and Auger spectroscopy, reported good agreement of the surface compositions with those predicted by the ideal solution model.In our studies, we prepared RhIr and RhFe bimetallic crystallites in the range < 10 8, with well characterized metal compositions by using the series of carbonyl cluster compounds presynthesized inside NaY zeolite cages. We have conducted the EXAFS, CO/ H chemisorption, '29X n.m.r., Mossbauer spectroscopies and F.t.i.r. measurements for probing the CO bonding to gain an insight into the structural and electronic characterization of the carbonyl precursors and the metallic Rh, Ir and RhIr clusters inside zeolite cages. The catalytic activity and selectivity for C-C bond scission were tested using the hydrogenolysis of n-butane and ethane, compared with the hydrogena- tion of benzene.The ensemble and electronic ligand roles of the second elements such as Ir and Fe in the well characterized Rh bimetallic cluster catalysts are discussed with respect to the site requirements in controlling the catalytic performance for both the probing reactions. Experimental Catalysts and Catalyst Preparation [ Rh,(CO) 16]/ Nay and Ir6( c o ) 1611 Nay : Rh3+/ NaY and Ir4+/ Nay (2-4 wt % metal loading in adjusting numbers of metal atoms g-' of NaY zeolite) were obtained by ion exchange of RhCI, - 3H20, Rh( NHJ5C12+ and Ir( NH3)&12+ with IrCl,H20M. Ichikawa, L-J: Rao, T. Ito and A. Fukuoka 323 ( a ) I96 0 >Rh-Rh< A 1 1758 I I I I I 2500 2000 1800 1600 2! I I I I I 0 2000 1800 1600 wavenumber/cm-' 21 12097 ( c > I I I I I 0 2000 1800 1600 Fig.1. 1.r. spectra of CO for NaY zeolite-entrapped Rh6(CO),6 ( a ) , IT,(CO)~~ ( b ) and Rh31r3(C0)3 ( c ) prepared from Rh3+/ Nay, Ir4+/ NaY and [ Rh3+ + Ir4+]/ NaY (4 nmol g-' Nay) under CO-H2 ( 1 : 2 molar ratio, 1 atm) at 473 K for 20 h. The medium intense bands at 1642 cm-' are assigned to the residual chemisorbed H 2 0 inside Nay. Table 1. 1.r. bands of CO of NaY zeolite-entrapped Rh6-.y~r,x(co)~6(X = 0-6) carbonyl clustersU Nay-entrapped sample carbonyl cluster linear CO face-bridged CO w.m.p.h.h [I1 Rh6(C0),6 2098 (s), 2062 (w) [VI Rh51r(Co) 16 2098 (s), 2060 (w) [VII Rh41r2(CO) 16 2098 (s), 2062 (w), 2036 (w) r VIII Rh3 Ir3(CO) 16 2098 (s), 2048 (w), 2021 (w) 1111 1r6(co)16 2098 (s), 2060 (w), 2038 (w) on the external NaY and in crystal Rh6(C0),6 2076 (s), 2025 (w) 1r6(co)16 2080 (s), 2050 (w), 2020 (w) 2098 (s), 2058 (w), 2000 (w) 1760 (m) 1756 (m) 1752 (m) 1744 (m) 1734 (m) 38 42 43 45 41 1805 (m) 1780 (m) (face-bridged isomer) 1830 (m), 1845 (sh) (edge-bridged isomer) - a s, strong; m, medium; w, weak; sh, shoulder.' Width at medium peak height of the face-briding co. (4 mmol dmP3) aqueous solutions at 363 K for 12 h," at pH 6-7, with NaY pow- der (LZ-Y52, surface area = 960 m2 g-', Si02/A1203 = 5.6, from Alpha Products Co). According to the literature method, Rh,(CO),, and Ir6(co)16 l 9 were stoichiometrically synthesized inside NaY by the carbonylation of Rh3+/NaY and Ir4+/NaY under an atmosphere of CO-H2 ( 1 :2 molar ratio, 1 atmt) at 473 K for 24 h and/or CO-H20 (23 mmHgS H20, total pressure 450 mmHg) at 303-393 K.The synthetic procedures and conditions were set in order to obtain the maximum yields of NaY-[Rh,(CO),,] and [Ir6(CO),,], following the formation of their characteristic carbonyl bands by in situ f.t.i.r. as shown in fig. 1 and table 1. t 1 atm = 101 325 Pa. $ 1 mmHg = 101 325/760 Pa.324 Selective Alkane Hydrogenolysis [ Rh6-xIrx( co) ,,I/ N a y (x = 1,2,3,4, and 5) were synthesized similarly as for Rh,(CO),, and Ir6(CO)16 by heating the double-ion-exchanged Nay, [ Rh3 + Ir4+]/NaY (4 mmol dm-3 of total metal atoms g-') having different metal atomic ratios under CO-H2 (1 : 2 molar ratio, 1 atm) at 473 K for 24 h. The corresponding hexanuclear RhIr bimetallic carbonyls which showed the characteristic carbonyl i.r. bands (table 1) were formed stoichiometrically in the prolonged reaction for 20 h via the formation of intermediate species such as Rh4-xIrx(CO)12 (x = 0-4), Rh(C0)2 and Ir(C0)3 inside Nay.The optimum preparation conditions were established by in situ i.r. observations. CO- H2/CO-H,0 Rh3+/ NaY [Rh6(Co)161/NaY CO-H2 CO-H, h4+/ NaY - [ h 6 ( c o ) ,611 N a y CO-Hz (6-x)Rh3++xIr4+/NaY - [Rh6-,Ir,(C0),,]/NaY; x = 1,2,3,4 Reduced Rh6, Ir, and Rh,-,Ir,/NaY:~ the precursor [Rh,-,Ir,(CO),,]/NaY (x = 0-6), 1.0 g were charged in Pyrex-glass reactor tubing and decarbonylated by mild oxidation with pure 0, (1 atm) at 373 K, followed by subsequent H2 reduction in flowing H2 (1 atm) with programmed heating from 300 to 473 K (2 h) to 673 K (2 h) after the residual H 2 0 had been carefully removed by evacuation at 373 K in each stage of the H2 reduction.[Rh4Fe2(CO)16]2-/NaY and [RhFe]/NaY [HFe3(CO),,]/NaY (3.5 wt% Fe) was pre- pared2' by the reaction of Fe2(C0)9 with 1.Og of hydrated NaY at 343 K in vacuo. [HFe3(CO),,]/NaY was allowed to react with Rh,(CO),, at 353 K until the evolution of CO and CO, stopped, resulting in the formation2' of [Rh4Fe2(CO)16]2-/NaY by a reaction analogous to the synthesis of the carbonyl cluster in solution.22 The resultant complex had an i.r. spectrum [ vco = 2078 (s), 2018 (w), 1958 (w), 1711 (sh) and 1744 (m) cm-'1 which resembled that of [Rh4Fe2(C0)16][TBPA]2. The residual Fe carbonyl species and the external Rh4Fe,(CO):, were extracted by washing with deoxy- genated THF under N,. HFe(CO),[Fe3(CO):;]+ Rh4(C0)12 + [Rh,Fe,(CO):;] The reduced RhFe/NaY catalyst was prepared by H2 reduction by programmed heating [373-473 K(2 h)-673 K (2 h)] after mild oxidation with 0, at 373 K of 1.0 g of [ Rh,Fe2( CO) Nay.Elementary analysis of metal contents on NaY was conducted using an inductively coupled plasma-atomic emission spectrometer (ICP) (Kyoto Kohken-UOP-2) after completely dissolving the catalyst samples in aqueous HF. The analytical data were used to normalize the catalytic performances and H/ CO chemisorption stoichiometries on the catalysts used. CO and Hydrogen Chemisorption Chemisorption measurements were performed in a static volumetric system made from Pyrex tubing, equipped with a 222AB-Baratron pressure sensor. After reduction of the catalysts (0.2 g) in flowing hydrogen, the apparatus was evacuated at 673 K and lo-, Torr for 1 h to remove chemisorbed hydrogen, before cooling to room temperature.The first isotherm was measured at a pressure of ca. 10 Torr, after 2 h had been allowed for equilibration. Subsequent isotherm points were measured at 2-3 times the initial press- ure, 2 h again being allowed for the equilibration for each point. t These abbreviations refer only to the precursors of the metal crystallites in the NaY catalysts.M. Ichikawa, L - j Rao, T. Ito and A. Fukuoka Activity Measurements 325 All the activity tests were conducted by using a flow-mode operating reactor system. The hydrogenolysis reaction was followed with a Pyrex-6 mm 0.d. glass tube containing the catalyst (0.1-0.5 g) flowing a mixture of ethane (or butane, cp grade, 99% purity) with hydrogen in a molar ratio of 1 : 20 at 373-423 K.The reactor was suspended in an electronic furnace for the hydrogenolysis, and in a water bath for benzene hydrogenation controlled within 0.5 K by the temperature controller. The temperature was internally recorded using CA thermocouple (0.3 mm 0.d.) in contact with the catalyst bed. For benzene hydrogenation pure hydrogen (1 atm, 99.9999% purity) was saturated with benzene vapour at 300k0.5 K, controlled by a glycol low-temperature bath to maintain the benzene/H, 1/20 molar ratio. Analysis was performed using a TCD-detector gas chromatograph following separation on a Porapak Q 4 m column operated at 343 K for ethane hydrogenolysis and benzene hydrogenation, and a Porapak Q (1 m, 403 K) plus alumina (4 m, 273 K) and active carbon (1 m, 273 K) column for butane hydrogenolysis. Xe N.M.R.Measurements 129 The location and electronic states of Rh, Ir, RhIr and RhFe bimetal crystallites derived from carbonyl clusters inside NaY zeolite were studied by ‘29Xe n.m.r., similarly as the previous work on Pt/NaY.*’ Samples for n.m.r. measurement were prepared to contain the same number of metal atoms (Rh + Ir) per gramme of Nay, irrespective to the metal composition of the catalysts. All the samples, in a Pyrex 10 mm 0.d. 100 mm long glass tube, were pretreated under flowing 02, with programmed heating from 300 to 673 K to remove CO and H 2 0 , followed by evacuation at Torr at room temperature. Subsequently, the sample was reduced with H2 (300 Torr) at 300-473 (2 h)-674 K (8 h), and evacuated for 2 h to remove the chemisorbed hydrogen.The H2 was purged several times and fresh H2 reintroduced into the sample. ‘29Xe n.m.r. signals on these samples were recorded under different Xe pressures by means of Brucker CXP 100 spectrometer at a frequency of 24.9 MHz and a magnetic field of 21.14 kG. EXAFS, Mossbauer and in situ F.T.I.R. Measurements Rh and Fe K-edge and Ir Ledge EXAFS measurements were carried out at 300 K at the SOR 10B line in the Photon Factory of the National Laboratory for High Energy Physics (KEK-PF). The catalyst samples were charged under He, CO or H2 in a specially designed EXAFS cell with ‘Kapton’ film windows (500 nm thick), to prevent the exposure to air and moisture.The coordination numbers (C.N.) and atomic distances were evaluated by Fourier-transform curve-fitting analysis, using the computer program PROGRAM-^,^^ as in our previous studies of Si0,-supported RhFe and PdFe bimetal cluster-derived catalysts.478 Using the same EXAFS cell, 57Fe Mossbauer spectra of RhFe/NaY were recorded at 298 K with an Austin Science-600 spectrometer. In order to determine the Mossbauer parameters, e.g. I.S. and Q.S. and relative absorption areas, the spectra were fitted by a linear combination of Lorentzians with the least-squares method.25 In situ i.r. measure- ments of the cluster formation and CO chemisorption were conducted by using a double-beam Fourier-transform i.r. spectrometer (Shimadzu F.T.1.R.-4100) at a resol- ution of 2 cm-’. Generally, 25-100 interferograms were added together to improve the signal-to-noise ratio.The catalyst sample wafer (25 mg, 20 0.d.) was mounted in a Pyrex i.r. cell and treated with a CO-H, flow by programmed heating up to 773 K. The contributions of gas-phase materials and supports were subtracted by computer process- ing using the sample and a reference cell containing an NaY wafer.326 Selective Alkane Hydrogenolysis Results and Discussion Structural Characterization of Precursor Carbonyl Clusters inside NaY [ Rh6( CO)16]/Y[ I] and Ir6( CO) I,]/ Nay[ II] showed i.r. spectra [fig. 1 (a) and (b)] consist- ing of terminal and three-centred carbonyl bands (table 1) in good agreement with those reported previously. 18319 The three-centred carbonyls of both zeolite-entrapped Rh, and Ir, carbonyls shift ca.40cm-' to a lower frequency compared with those of the free molecules.These lower-frequency shifts are interpreted as due to O-ended face bridging CO interacting with acid sites such as A13+, Na+ and H+ inside the NaY zeolite framework.18 The Fourier transform of the EXAFS oscillation k3X( k ) on sample [I] is shown in fig.2 and the results of the curve-fitting analyses are summarized in table2. The Rh K-edge EXAFS data on [I] showed good agreement with those in crystalline Rh6( CO) 16, indicative of the stoichiometric formation of hexanuclear Rh carbonyls inside NaY zeolite. There was a minor contribution ( < 5 % ) of Rh-0 bonding (C.N. = 1.9, R = 2.06 A) due to mononuclear Rh( CO), species giving the characteristic twin carbonyl bands at 2042 and 2020 cm-I in its i.r.spectrum [fig. l(a)]. Oxidation of [I] by heating from 293 to 423 K in 0, to eliminate CO to give the oxidized sample was followed with H2 reduction by programmed heating from 297 to 673 K in an H2 flow, resulting in the reduced [Rh,]/NaY [III]. The EXAFS analysis for [III], as shown in fig. 2(b) and table 2, substantiated the retention of the Rh cluster units in terms of the coordination number (C.N. = 4.6) of Rh-Rh bonds but with a slight shortening of the atomic distance ( R = 2.70 A), similar to highly dispersed Rh crystallites less than 10 A in size.,' On the other hand, the freshly reduced sample [111] gave the sharp and intense carbonyl i.r. bands [2098 (s), 1840 (s) and 1670 (w) cm-'1 in CO chemsorption as shown in fig.3(a), which resemble those of the edge-bridging Rh cluster carbonyls? inside Nay. The i.r. spectrum of the CO chemisorbed species readily developed by heating sample [IV] in CO at 400 K for 20 min, resulting in those of the face-bridging Rh,(CO)1,/NaY [2098(s) and 1670(s) cm-'1. As shown in table 3, [ Rh,]/NaY chemisorbs CO in the stoichiometry of CO/ Rh = 2.6, corresponding to that of Rh,(CO),,. Moreover, the EXAFS data [fig.2(c), table21 on [Rh,]/NaY with chemisorbed CO demonstrated that the C.N. and length of Rh-Rh bond in the resulting species were in good agreement with those of the original [Rh,(CO),,]/NaY. These results suggest that an Rh6 carbonyl cluster is regenerated by CO chemisorption, from the reduced Rh crystallites inside Nay.This is quite a contrast to the case of the highly dispersed Rh crystallites of 10 A size, supported on alumina, where metal-metal bonds are irreversibly ruptured in CO chemisorption in forming oxidized monometallic car- bony1 Rh(CO)*, as proposed by Prins arid co-~orkers.~' The remarkably different behaviour of the Rh crystallites in CO chemisorption between inside NaY and on alumina is associated with the nature of the supports. Recently, Yates2' has suggested that the oxidative cleavage of Rh-Rh bonds in CO chemisorption on Rh/A1203 is caused by the acidic OH of the alumina support. The further studies on EXAFS measurements in table 2 and CO-H2 chemisorption (table 3) demonstrated that the reduced Rh (and Ir) crystallites inside NaY frameworks are stabilized, whatever the treatment and tem- perature up to 673 K, involving several sequences of oxidation and H2 reduction, without any formation of large particles deposited on the external zeolite surface.The precursors for the series of RhIr bimetallic cluster catalysts were similarly prepared in CO-H2 at 1 atm and 473 K, from the doubly ion-exchanged N a y : [Rh3++ Ir4+]/NaY with different metal compositions. As shown in table 1 and fig. l(a)-(c), the t We propose that the main CO chemisorbed species showing i.r. bands at 2098 and 1840 cm-' is assigned to the edge-bridging isomer of Rh,(CO),,, similar to the case of Ir6(CO),6 which has two different isomer forms with different i.r. carbonyl bands assigned to the edge- and face-bridging isomers.26M. Ichikawa, L-J: Rao, T.Ito and A. Fukuoka 327 13.00 - ( a ) 10.40 - Rh-Rh - - 7.80 >z - 5 5.20- R I A 11.001 R I A Fig. 2. Fourier transform of EXAFS oscillation k3x( k ) of [ Rh6( CO) NaY [I] ( a ) and reduced [Rh,]/NaY [III] ( b ) : H2 reduction by 1 atm flowing H2 by programmed heating up to 673 K (2 h), after the mild oxidation of sample [I] (2 wt% Rh loading), and CO chemisorption on the reduced [Rh,]/NaY at 293 K (c). resulting clusters inside NaY gave carbonyl i.r. spectra characteristic of the hexanuclear carbonyl clusters, consisting of linear carbonyls (2099-2096 cm-') and face-bridging carbonyls (1756, 1752 and 1744 cm-') for Rh/Ir atomic ratios 5/1, 4/2 and 3/3, respec- tively. The face-bridging CO of the series of RhIr bimetal clusters gave sharp i.r. bands with a similar half-width, the positions of which shift systematically to lower frequency on increasing the Ir content: Rh,(CO),, (1760 cm-') to Ir6(co)16 (1734 cm-I).The results suggest that after carbonylation the samples consist of a uniform distribution of hexanuclear bimetallic Rh6-xIrx clusters, with well defined metal compositions asso- ciated with those of the starting materials, [Rh3'+Ir4+]/NaY. They are expressed as [Rh6-,IrX(CO),6]/NaY (x = 1,2 and 3), [V], [VI] and [VII], respectively. EXAFS studies on Rh K - and Ir Ledges of the resulting bimetallic carbonyl samples have been328 Selective Alkane Hydrogenolysis Table2. Results of curve-fitting analysis of Rh K-edge EXAFS data for Nay-entrapped Rh,(CO),, and reduced Rh, clusters inside NaY Rh-0, Rh-Rh Rh-CO, Rh-COB (support) sample Rh clusters C.N.R I A C.N. R I A C.N. R I A C.N. R I A [I1 r Rh,(CO),,I/NaY 3.1 2.74 1.5 1.88 1.6 2.15 1.9 2.06 [III] (1) reduced - - 0.7 2.10 - - 1.7 2.09 [RhJNaY (473 K H,) 4.6 2.70 - - (673 K H,) 4.6 2.70 - - (2) CO chemisorption [RhJNaY (473 K H7) reference sample +CO (293 K) 3.2 2.72 1.4 1.85 1.4 2.15 0.8 2.03 Rh,(CO),, in crystal 4.0 2.76 2.1 1.87 2.0 2.17 - - Estimated overall (experimental + systematic) errors were 0.02 A for atomic distance and 0.2 for coordination number on the present EXAFS data evaluation. The Debye-Waller factor ( A') was evaluated within 3.0-5.0 foi his series of catalysts. 0 C f B s 1830 I I I I I 2000 1800 1600 25 2064 I I I I I 10 2000 1800 1600 25 2024 I I I I I 10 2000 1800 1600 wavenumber/ cm- ' Fig.3.1.r. spectra in CO chemisorption on the reduced [Rh,]/NaY: 2 wt0h Rh loading [111] ( a ) , [IrJNaY: 4 wt%Irloading [IV](b) and [Rh,Ir,]/NaY: 2.7% wt metal [Rh+Ir] loading[VIII](c), after evacuation of CO at 293 K in 10 min. P,, = 100 Torr at 293 K; The catalyst reduction was conducted by flowing H, at 673 K for 2 h. conducted, and revealed the stoichiometric forma il of bimetal RhIr clusters with different metal compositions based on those of the sf ing doubly ion-exchanged Nay. The reduced bimetal RhIr catalysts, [Rh,-,Ir,]/NaY (x = 2,3 and 4) [VIII-XI were obtained by a similar procedure to that for the reduced Rh6 and Ir, catalysts. Xe N.M.R. Studies on [Rh,-,Ir,]/NaY(x = 0.6) Catalysts 129 Fig. 4 shows the Xe concentration-dependence of the chemical shift of '29Xe n.m.r.signals observed on the series of catalysts [ Rh,-,11-~]/ NaY (x = 0,2,4 and 6), after H2 reduction at 673 K. The chemical shifts under the same pressure of Xe systematicallyM. Ichikawa, L-f: Rao, T. Ito and A. Fukuoka 329 A- - 0 lo2' 5.1020 lo2' Xe atom g-' Fig. 4. '29Xe n.m.r. chemical shift on each reduced catalyst by changing the number of Xe atoms adsorbed per gram of catalyst sample; The catalysts were reduced with H2 by programmed heating up to 673 K for 8 h and evacuated at 673 K for 2 h. x, [Rh,]/NaY[III]; 0, [Rh,Ir2]/NaY[VIII]; A, [Rh,Ir,]/NaY[X]; 0, [Ir,]/NaY[IV]. The total metal concentration of each catalyst was roughly adjusted in 4 mmol g-' Nay. Table 3. Hydrogen and CO chemisorption at 300*2 K on the series of reduced [ Rh+, It-,]/ NaY (x = 0,2,4,6) catalysts sample catalyst H/M CO/M CO/H [I111 [Rh,]NaY 0.80 2.6 3.2, [VIII] [Rh,Ir,]/NaY 0.81 1.6 2.0 [XI [ Rh21r4]/ NaY 0.87 0.84 0.96 WI [IrhllNaY 1.32 0.80 0.59 H/M and CO/M were evaluated from the total amounts of irreversibly chemisorbed H and CO on freshly reduced catalysts M: total concentra- tion of metal atoms (Rh+ Ir) of the catalyst (analyzed by ICP).increased with increasing Rh content in the cluster catalysts. The linewidths of Xe n.m.r. signals on all the samples were narrow ( CQ. 20 ppm at the Xe pressure of 517 Torr) compared with those of NaY (58 ppm). The results suggest: (1) The RhIr bimetallic crystallites have homogeneous metal compositions, and are not simply a physical mixture of Rh6 and Ir6 uniformly distributed inside Nay.(2) The '29Xe n.m.r. chemical shift, (abare) in ppm at 5 x 10'' Xe atom g-', systematically decreases across the series of clusters with increasing Ir contents, as presented in table3, with shifts of 148, 106, 85330 Selective Alkane Hydrogen olysis Table 4. Chemical shifts (ppm) of '29Xe n.m.r. signals observed on the series of catalysts [Rh,-,Ir,]/NaY (x = 0, 2, 4, 6) at a concentration of adsorbed Xe atoms of 5 x 10'' g-' NaY sample catalyst &are 8Hz satd A8 Share is the chemical shift of Xe n.m.r. signals on the reduced catalysts after evacuation at 673 K for 2 h. aH2 ratd is the chemical shift of Xe n.m.r. signals on the reduced catalysts after H2 saturated chemisorption at 299 K. A8 = $are - SH2 sntd- and 78 ppm for Rh,, Rh,Ir,, Rh,Ir, and Ir, inside Nay, respectively, as based on an isolated Xe atom.The chemical shift of [Rh,]/NaY is of a similar order of magnitude as previously reported29 for a highly dispersed Pt/NaY. In previous s t u d i e ~ ~ ~ ~ ~ ~ the Xe n.m.r. chemical shifts have been discussed based on the changing electron density of a probing Xe atom adsorbed on the metals, due to the charge-transfer non-bonding interaction with orbitals of metal clusters plus a collision factor for Xe atoms inside NaY. Since the environmental situation around each cluster inside NaY cages is essentially the same across the series of cluster catalysts with different metal composi- tions, the '29Xe n.m.r. chemical shift is mainly related to the Xe-cluster ensemble interaction of the adsorbed Xe atom.In this sense, it is likely that the Rh6 cluster inside NaY is highly electron deficient, compared with the Ir, clusters, and the electron deficiencies of the clusters systematically decrease with increasing Ir contents in the RhIr cluster catalysts. In addition, it was found that the '29Xe n.m.r. chemical shifts on the series cluster catalysts considerably decreased after the chemisorption of hydrogen and CO at 293 K, as indicated for the hydrogen chemisorption in table 4. This suggested that hydrogen/CO chemisorption results in charge-transfer from chemisorbed hydro- gen/CO to the electron-defficient Rh-rich clusters. CO and H2 Chemisorption Stoichiometries Data for both hydrogen and carbon monoxide chemisorption on a series of the reduced [Rh,-,Ir,]/NaY(x=O, 2, 4, and 6) are presented in table3.The numbers reported are the total gas uptake in strong chemisorption, extrapolated to zero pressure, evaluated in the form of adsorbed atom to metal atom ratios. The numbers of metal (Rh + Ir) atoms were calculated from the impregnated metal concentrations, which were analysed independently by ICP spectroscopy. In particular, the CO/ M values on Rh-rich cluster- derived catalysts are much higher than on Ir-rich clusters. The trends of CO chemisorp- tion amounts are in good agreement with those of carbonyl band intensities in i.r. studies across the series of Rh,, Rh,Ir2, Rh,Ir, and Ir,, as shown in fig. 2 ( a ) - ( c ) . This is systematically consistent with the decreasing order of electron deficiency on the clusters inside zeolites, which has been estimated from the '29Xe n.m.r.chemical shifts in table 4. On the other hand, the bridging CO chemisorption was markedly suppressed on the Ir-containing Rh clusters, leaving the linear CO as shown in fig. 4( c ) . This could be explained by the geometrical effect of Ir in breaking the ensembles of Rh atoms, preventing the multi-bridging CO chemisorption, similarly reported by Sachtler and co-w~rkers~' on supported alloys such as PdAg and NiCu. In contrast to CO chemisorp- tion, H/M values on the Ir-rich cluster catalysts are relatively higher (H/M = 1.0-1.4)M. Ichikawa, L-f: Rao, T. Ito and A. Fukuoka 331 Fig. 5. Turnover frequency of n-butane hydrogenolysis (mol/mol metal/103 min) at 453 K, 1 atm, C4H,,,/H2 = 1 : 20 molar ratio (0), of ethane hydrogenolysis (mol/mol metal/105 min) at 473 K, 1 atm, C,H, = 1 : 20 molar ratio (A), comparing benzene hydrogenation (mol/mol metal/103 min) at 323 K, 1 atm, benzene/H, = 1 :20 molar ratio (0) on the series catalysts of [Rh,-,Ir,]/NaY [IIII-[X](x = 0, 1 , 2 , 3 and 6 ) as a function of atom YO Rh(Rh/Rh+ Ir composition).then the Rh-rich catalysts, which are basically consistent with those (H/ Ir = 1.3- 1.8) observed on Ir black, the conventional Ir/A1203 and Ir4(C0),2/A1203 ,32 while they are H/ Rh = 0.8- 1.0 on the highly dispersed Rh/A1203 and Rh/ S O z . Consequently, the CO/H ratios on the Rh-rich clusters are markedly higher than the Ir-rich clusters. It is known that the H/Rh and CO/Ir stoichiometries are ca. 1 on the conventional Rh and Ir catalysts with a high dispersion D=0.8-1.0,32 and the values of CO/H are not too sensitive to particle size of Ir.In this sense, the Rh-rich clusters inside zeolites exhibit a unique behaviour accessible for CO, possibly due to the unusually higher electron deficiency in their orbitals, compared with those of the bulk metals. Hydrogenolysis of Butane and Ethane Hydrogenolysis of n-butane and ethane proceeded with the higher conversions on the Rh-rich cluster catalysts, but modestly on [IrJNaY at temperatures of 373-523 K. A negligible yield of isobutane (<3% of the total butane conversion) by isomerization was obtained on the series of Rh, Ir and RhIr bimetal catalysts. The turnover frequency, defined as mol [metal atoms (Rh + Ir)] min-' for ethane hydrogenolysis at 473 K and n-butane hydrogenolysis at 453 K for the catalyst series [Rh6_.Irx(CO),6]/NaY (x = 0-6) is plotted in fig.5 as a function of percentage Rh, for the cluster-derived catalysts. The reproducibility from run to run on a given catalyst332 Selective Alkane Hydrogenolysis sample was good for all reactions studied (there is ca. 10% variation for rate constants and CJC3 selectivities for one catalyst sample to the next). It is interesting to find that Ir has a very large effect on the turnover frequency of hydrogenolysis for both ethane and butane on the RhIr bimetal clusters. Because, in the catalysts derived from the zeolite-entrapped hexanuclear Rh, Ir and RhIr carbonyl clusters the particle size is well controlled and < 10 A inside NaY cages across the metal composition ranges, the activity decrease in the series of Rh6-xh-x catalysts for the hydrogenolysis of butane and ethane can be interpreted in terms of either the breaking of active Rh ensemble sites by incorporation of inactive Ir atoms, or drastic changes in the electronic states of the clusters across the metal composition range.It is noteworthy to compare the Rh6-,Irx/NaY catalysts with the Ni-Cu and Ru-Cu systems studied by Sinfelt and co-workers.12 There is a striking similarity in catalytic trend across the two bimetallic series. The suppression of hydrogenolysis is a much sharper trend on our Rh-Ir cluster catalysts than was observed for the Si0,-supported conventional Rh-Ir catalysts reported by Haller et al., where the difference is ca.an order of magnitude in the relative activity of Rh and Ir. This is due to a considerable surface enrichment of more active metal Rh in the conventional Rh-Ir catalysts. These two metals, Rh and Ir, being in the same sub-Group, VII12, intuitively might be expected to exhibit very similar behaviour; their sizes are about the same (1.34 8, for Rh CJ 1.36 8, for Ir), both metals exhibit the f.c.c. crystal structure in the bulk metal, and the electronic structures of Rh( [ Kr]4d85s1) and Ir([Xe]4f145d76s’) differ only slightly in the free atoms and even less in the bulk metal. McKee and studied the hydrogen-deuterium exchange of methane over the unsupported RhIr alloy powder, and reported no correlation other than an apparent maximum in exchange activity when the maximum number of d-band holes were present. It has been discussed in many previous papers that hydrogenolysis of alkanes is classed as a ‘structure-sensitive’ reaction in which the sites involve relatively large ensembles of metal atoms.In the cases of Ni-Cu, Ru-Cu12 and Pd-Ag” catalysts the catalytically inactive Cu and Ag atoms play a role in breaking the ensemble sizes of Ni, Ru and Pd atoms which are active for the hydrogenolysis, simply due to the geometrical ensemble size effect. In contrast to these cases, in the present RhIr bimetal cluster catalysts inside the NaY zeolite the dramatic suppression of hydrogenolysis by increasing the Ir contents is proposed to be interpreted in terms not of a simple ensemble size effect but of an electronic state associated with the electron deficiency, i.e.‘d-hole orbital’ of the clusters, as discussed for the 129Xe n.m.r. chemical shifts for the series [ Rh,-,Ir,]/NaY. The remarkable difference in hydrogenolysis activity between Rh and Ir crystallites inside NaY is based on their electron-deficient sites, which favour C-C bond scission via the alkane carbonium intermediate.34 The C2/C3 selectivity is defined as the ratio of the rates of butane conversion to ethane ( k , ) to the rates of butane conversion to methane plus propane ( k J . very slow This selectivity measured in the two reactions is compared in table 5 , where activation energies based on total conversion and C2/C3 selectivity are presented for the series ofM. Ichikawa, L-J: Rao, T. It0 and A.Fukuoka 333 Table 5. Butane hydrogenolysis selectivity and activation energies on the series catalysts of [Rh,-,Ir,]/NaY (x = 0-6) and [RhFe]/NaY catalyst conversion selectivity" 7-1 K (Yo) to ethane (YO) C2/C3h Ea" In Ad W,I/N~Y 403-453 1-20 70-73 5.1 185 58.4 [ Rh,Ir]/ NaY 403-453 0.6-22 73-76 6.8 199 60.3 [ Rh41r2]/ NaY 407-473 1-30 74-79 7.5 198 61.7 [ Rh3 Ir3]/ NaY 43 3-473 1-12 78-81 8.3 172 50.2 [ Rh21r4]/NaY 43 3-493 1-20 65-73 5.2 176 53.2 463-498 1-20 60-70 3 -2 185 51.0 [ RhFe]/ NaY 433-498 0.2-13 27-51 1.2 132 27.5 [1r61/ N ~ Y Percent of ethane in products, the balance being methane and propane. C2/C3 selectivity is the ratio of butane conversion to ethane and propane at 453 K. ' Apparent activation energy (kJ/mol-I). 'I A (mole/mole metal lo3 min-') rates of butane conversion: exponential factor.Table 6. Catalytic performance of the series [ Rh,-,Ir,]/NaY and [ RhFe]/ NaY catalysts for the benzene hydrogenation catalysts activity (TOF)" E J ~ J mol-'h 605 1290 1048 622 108 32.8 33.8 38.2 41.1 21.6 Formation rate of cyclohexane in benzene hydrogenation at 323 K in moIe/moIe metal lo3 inin-'; flow rate = 100 cm3 min-', I atm H2 : C,H, = 20: 1 molar ratio. Apparent activation energies evalu- ated at 283-353 K. catalysts [ Rh6-xIrx]/ NaY (x = 0-6). The maximum selectivities towards the central C-C bond scission to give ethane are observed on the Rh,Ir, Rh,Ir, and Rh31r3 bimetal cluster catalysts (max. 81-75% selectivity) rather than on [ Rh,]/NaY and [ Ir6]/ NaY cluster (70 and 63%, respectively). On mechanistic considerations, the higher selectivity towards the central C-C splitting on the RhIr bimetal cluster catalysts is favourably catalysed by a 1,3-diadsorbed intermediate of b ~ t a n e , ~ ' which is possibly associated with the geometric ensemble effect where the ensemble size of active Rh atoms is decreased by Ir atoms.Benzene Hydrogenation The rates and activation energies for benzene hydrogenation on the catalyst series [Rh,-,Ir,]/NaY (x = 0-6), are presented in table 6. Both Rh and Ir cluster catalysts exhibited hydrogenation rates of a similar order of magnitude. The benzene hydrogena- tion shows a different trend across the Rh,-,Ir,/NaY catalyst series compared with hydrogenolysis of butane and ethane. As shown in fig. 6, hydrogenolysis is dramatically suppressed, while the hydrogenation of benzene is basically insensitive to (but slightly enhanced by) the metal compositions of the cluster catalysts.The activation energies for the hydrogenation are slightly increased by increasing the Ir contents in the cluster catalysts. The rate enhancement is proposed to be based on the electronic effect of the334 Selective Alkane Hy d rogenoly s is RhIr ensembles in controlling the benzene adsorption in competition with hydrogen on the clusters to promote the hydrogen activity. Butane Hydrogenolysis on [RhFe]/NaY Catalyst The catalytic properties of the RhFe catalyst derived from decomposing NaY zeolite- entrapped [ Rh4Fe2(C0),,l2- have been probed by examining the hydrogenolysis of n-butane. The resultant data are shown in table 5 . In contrast to the RhIr bimetallic catalysts, e.g.[ Rh,Ir,]/NaY and [ Rh31r3]/NaY, which displayed a high selectivity towards central C -C bond scission in butane hydrogenolysis, forming two molecules of ethane (70-85% selectivity), the [ RhFe]/ NaY catalyst showed higher selectivities for terminal C-C bond scission, to give a methane plus propane in >70% selectivity, but (25% for central C-C bond scission. This substantially different cracking pattern for the [RhFe]/NaY catalyst is strong evidence for a residual Rh-Fe interaction that modifies the character of the catalyst sites. The activities measured on [ RhFe]/ NaY were rather smaller than those on [Rh,]/NaY, but most noteworthy is that activation energies were lower for [ RhFe]/NaY catalyst.The similar trend of the selectivities towards a terminal C-C bond scission was observed by Sharpley et ~ 1 . ~ ~ on the A1203-supported Cp2W21r,( CO),-derived catalyst, compared with those on Ir4( CO) ,2/A1203. The Mossbauer spectrum of the freshly reduced [ RhFe]/ NaY gave absorption bands which could be resolved into a singlet and a pair of quadrupole doublet. They could be reasonably assigned to Feo and Fe3+, respectively. It is expected that Fe in the [RhFe]/NaY catalyst mainly coexists as Feo and Fe3+ in forming the adjacent bimetallic ensembles of RhFe' and RhFe3+. The previous EXAFS studies, coupled with Mossbauer measurements on an Si0,-supported Rh4Fe2 cluster-derived catalyst,39 revealed the direct bonding of Rh-Fe3+-0 located at the cluster/support interface.In the case of [ RhFe]/NaY, Fe3+ is probably located on the internal zeolite cages. Rodriguez-Reinaso et al.37 discussed the favourable terminal C-C bond splitting of butane hydrogenolysis on the Pt/A1203 in higher dispersion, based on a carbonium mechanism via a v-ally1 intermediate which strongly interacted with an isolated Pt atom. Accordingly, the higher selectivities towards the terminal C-C bond scission on [ RhFe]/ NaY and Ir- W/ A1203 could be associated with ionic heteroatomic adjacent sites, e.g. Rh-Fe3+ and Ir-W"+, as depicted in the following proposed mechanism: c c \*<--A / "2 C ' *\'C CH4+C3Hs. I/ Rh--Fe3+ The authors are much indebted to Prof. H. Kuroda and Dr N. Kosugi of the University of Tokyo for helpful discussions on the EXAFS evaluation and also to Prof.T. Tominaga and Dr Y. Sakai of the University of Tokyo for their Mossbauer measurements and the parameter evaluation. We also thank Dr M. Uemura and M. Kasano of the Research Center, Shin-Daikyowa Petrochemicals for their ICP analysis of the metal contents in the catalysts used. References 1 M. Ichikawa, Tailored Metal Catalysts, ed. Y. Iwasawa (D. Reidel, Dordrecht, 1985), p. 183; B. C. Gates, L. Guczi and H. Knizinger, Metal Clusters in Catalysis (Elsevier, Amsterdam, 1986), and references therein. 2 R. Psaro, R. Ugo, B. Besson, A. K. Smith and J. M. Basset, J. Organomet. Chem., 1981, 213, 215; J. M. Basset and A. Choplin, J. Mol. Catal., 1983, 21, 95.M. Ichikawa, L-f: Rao, T. Ito and A. Fukuoka 335 3 M. Ichikawa, J. Coral., 1979, 56, 127; 1979, 59, 67; A.Fukuoka, H. Matsuzaka, M. Hidai and M. Ichikawa, Chem. Lett., 1987, 941; A. Choplin, L. Huang, A. Theolier, P. Gallezot, J. M. Basset, V. Siriwardane, S. G. Shore and R. Mathieu, J. Am. Chem. Soc., 1986, 108, 4224. 4 T. Yokoyama, H. Yamazaki, N. Kosugi, H. Kuroda, M. Ichikawa and T. Fukushima, J. Chem. Soc., Chem. Commun., 1984, 962; M. Ichikawa, in Homogeneous and Heterogeneous Catalysis, ed. Yu. I . Yermakov (NVU Sci. Press, Utrecht, 1986), p.819; T. Kimura, A. Fukuoka, A. Fumagalli and M. Ichikawa, Catal. Lett., in press. 5 S. Iijima and M. Ichikawa, J. Catal., 1985, 94, 313. 6 M. Ichikawa, K. Sekizawa, K. Shkakura and M. Ichikawa, J. Mol. Catal., 1981, 11, 167; M. Kawai, M. Uda and M. Ichikawa, J. Phys. Chem., 1985, 89, 1654.7 M. Deeba, J. P. Scott, R. Barth and B. C. Gates, J. Catal., 1981, 71, 373, K. Tanaka, K. L. Waters and R. F. Howe, J. Catal., 1982, 75, 23; H. C. Foley, S. J. DeCanio, K. D. Tan, K. J. Chao, J. H. Onaferko, C. Dybowsky and B. C . Gates, J. Am. Chem. Soc., 1983, 105, 3074. 8 M. Ichikawa, Chemtech, 1982, 674; M. Ichikawa, A. Fukuoka and T. Kimura, Proc. IX Znt. Congr. Coral. (Calgary), vol. I, p. 569 (1988); A. Fukuoka, T. Kimura and M. Ichikawa, J. Chem. Soc., Chem. Commun., 1988, 428. 9 A. Fukuoka, M. Ichikawa, J . A. Hriljiac and D. F. Shriver, Inorg. Chem., 1987, 26, 3643; M. Ichikawa, Polyhedron, 1988, 7, 2351. 10 J. H. Sinfelt, C a r d Rev. Sci., 1969, 3, 175; Adu. Catal., 1973, 23, 91; C. Naccache, N. Kaufherr, M. Dufaux, J. Bandiera and B. Imelik, Molecular Sieves ZZ, ed.J. R. Katzer (ACS Symp. Ser. 538, 1977); G. C. Bond and X. Yide, J. Chem. Soc., Faraday Trans. I, 1984, 80, 969; D. J. Sajkowski, J . Y. Lee, J. Schwank. Y. Tian and J. G. Goodwin Jr, J. Catal., 1986, 97, 546. 11 J. R. Anderson and B. H. McConkez, J. Catal., 1968, 11, 54; W. A. A. van Barneveld and V. Ponec, J. Catal., 1984, 88, 382; G. A. Somorjai, Carol. Rev., 1972, 7, 87. 12 J. H. Sinfelt, J. Catal., 1973, 29, 308; Acc. Chem. Res., 1977, 10, 15; J. H. Sinfelt, J. L. Carter and J. C. Yates, J. Catal., 1972, 24, 283; R. A. Della Betta, J. A. Cusmano and J. H. Sinfelt, J. Coral., 1970,19, 343. 13 J. M. Beelen, V. Ponec and W. M. H. Sachtler, J. Catal., 1973, 28, 376. 14 T. C. Wong, L. F. Brown, G. Hailer and C . Kemball, J. Chem. Soc., Faraday Trans.I, 1981, 77, 519. 15 G. H. Via, G. Meizner, F. W. Lytle and J. H. Sinfelt, J. Chem. Phys., 1983, 79, 1527; J . H. Sinfelt and F. W. Lytle, J. Chem. Phys., 1980, 72, 4843; J. H. Sinfelt, G. H. Via and F. W. Lytle, J. Chem. Phys., 1982, 16, 2779. 16 V. Ponec and F. J. Kuijers, Appl. Surj Sci., 1978, 2, 43. 17 E. J. Rode, M. E. Davis and B. E. Hanson, J. Coral., 1985, 96, 563; N. Takahashi and M. Kobayashi, J. Catal., 1984, 85, 89; N. Takahashi, A. Mijin, T. Ishikawa and H. Suematsu, J. Chem. Soc., Faradaj. Trans. I , 1987, 83, 1958. 18 P. Gelin, Y. Ben Taarit and C. Naccache, J. Catal., 1979, 59, 357; R. Shannon, J. C. Vedrine, C. Naccache and F. Lefebvre, J. Carol., 1985, 88, 43; E. J. Rode and M. E. Davis, J. Catal., 1985, 96, 574. 19 C. Naccache, F. Lefebvre, P.Galin and Y. Ben Taarit, Chemical Reaction in Organic and Inorganic Constrained Systems, ed. R. Setton (NATO AS1 Ser. 165), (D. Reidel, Dordrecht, 1986), p. 81; F. Lefebvre, P. Calin, C. Naccache and Y. Ben Taarit, Proc. VZ Inr. Congr. Zeolite (Butterworths, London, 1984), p. 435; G. Bergeret, P. Gallezot and F. Lefebvre, Stud. Surj Sci. Caral., 1986, 28, 401. 20 M. Iwamoto and S. Kagawa, J. Phys. Chem., 1986, 90, 5244. 21 L-F. Rao, A. Fukuoka and M. Ichikawa, J. Chem. Soc., Chem. Commun., 1988,458; L-F. Rao, A. Fukuoka and M. Ichikawa, Proc. Znr. Symp. Acid-Base Catal., ed. K. Tanabe (Elsevier, Amsterdam), in press. 22 A. Ceriotti, G. Longoni, R. D. Pergola, B. T. Heaton and 0. S. Smith, J. Chem. Soc., Dalton Trans., 1983, 1433. 23 J. Fraissard and T. Ito, Zeolites, 1988, 8, 350. 24 N. Kosugi and H. Kuroda, PROGRAM EXAFS 2 1 ~ 0 3 , Res. Centr. Spectrochem., University of Tokyo 25 The program originally written by T. C. Gibb et a/. ( J . Chem. SOC. A, 1967, 1478) was rewritten in 26 L. Garlaschelli, S. M. Martinengo, P. L. Bellon, F. Demartin, M. Manassero, M. Y. Chiang, C-Yu Wei 27 H. F. J. Van’t Blik, J. B. A. Va Zon, T. Huizinga, J. C. Vis, D. C. Koningsberger and R. Prins, J. Am. 28 P. Basu, D. Panayatov and J. T. Yates Jr, J. Phys. Chem., 1987, 91, 3133. 29 T. Ito, L. C. de Menorval and J . Fraissard, J. Chim. Phys., 1983, 80, 573. 30 B. E. Nievwenhuys and W. M. H. Sachtler, J. Colloid Interface Sci., 1977, 58, 65. 31 Y. Noto-Soma and W. M. H. Sachtler, J. Catal., 1972, 32, 315; 1974, 34, 162; M. Primet, M. Matthieu and W. M. H. Sachtler, J. Catal., 1976, 44, 324. 32 N. E. Buyanova, 0. F. Zapreeva and A. P. Karnaukhor, Kinet. Katal., 1978, 19, 1196; K. Foger and J . R. Anderson, J . Catal., 1979, 59, 325. 33 D. W. McKee and F. J. Norton, J. Phys. Chem., 1964, 68, 481; Trans. Faraday SOC., 1965, 61, 2273; J. Coral., 1965, 4, 510. (1987). FORTRAN to suit HITACH-M-200H/208H. and R. Bau, J. Am. Chem. Soc., 1984, 106, 6664. Chem. Soc., 1985, 107, 3139.336 Selective Alkane Hydrogenolysis 34 D. J. C. Yates and J. H. Sinfelt, J. Cat;/., 1967,8,348; D. J. C. Yates, L. L. Murrele and E. B. Prestridge, J. Catal., 1979, 57, 41; J. T. Yates, T. M. Duncan, S . D. Worley and R. W. Vaughan, J. Chem. Phys., 1979, 70, 1219. 35 J. L. Carter, J. A. Cusmano and J. H. Sinfelt, J. Cafal, 1971, 20,223; J. R. Anderson and N. R. Avery, J. Catal., 1967, I, 315. 36 J. R. Sharpley, S . J. Hardwick, D. S. Foose and G. D. Stucky, J. Am. Chem. SOC., 1981, 103, 7383. 37 F. Rodriguez-Reinoso, 1. Rodriguez-Ramos, C. Moreno-Castilla, A. Guerrero-Ruiz and J. D. Lopez- Gonzalez, J. Cafal, 1987, 107, 1. Paper 9/00282K; Received 16th January, 1989

ISSN:0301-7249    

DOI:10.1039/DC9898700321

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 337-344 Butane Hydrogenolysis over Single-crystal Rhodium Catalysts Abhaya K. Datye* and Bernard F. Hegarty Department of Chemical and Nuclear Engineering, University of New Mexico, Albuquerque, New Mexico 87131, U.S.A. D. Wayne Goodman* Department of Chemistry, Texas A&M University, College Station, Texas 77843, U.S. A . Hydrogenolysis of n-butane has been studied over the (1 10) and ( 1 11) surfaces of Rhodium. On Rh( 1 lo), the products of the hydrogenolysis consist of methane > ethane > propane. The hydrogenolysis reaction exhibits a good fit to Arrhenius behaviour for reaction temperatures up to 500 K. At higher temperatures the reaction rate tends to ‘roll over’ due to insufficient surface coverage of hydrogen. The ‘roll over’ affects product distribution, yielding more complete hydrogenolysis to methane.The behaviour is qualitatively similar on Rh( 11 l ) , with ‘roll over’ occurring at a lower temperature, namely 475 K. However, the hydrogenolysis is more selective on Rh( 11 l ) , yielding 50 mol O/O ethane. The high ethane selectivity seen in previous work on Ir( 110) is not seen on the Rh( 110) surface presumably because the Rh surface does not exhibit the (1 x 2) ‘missing-row’ reconstruction that is stable on Ir( 110) surfaces under reaction conditions. The hydrogenolysis selectivity of the Rh single-crystal surfaces correlates well with supported Rh metal particles subjected to oxidation-reduction cycling. An important lingering question in catalysis is the relationship between the structure and composition of a catalytic surface and the reactivity and selectivity demonstrated by that surface.The use of orientated single crystals has been shown to be particularly informative regarding the assessment of the effects of surface composition and geometry.’ For example, in our laboratories, the activity for ethane’ and butane hydrogen~lysis~ to methane on nickel has been shown to depend critically on the particular geometry of the surface, the more open (100) plane being far more reactive than the closed-packed (1 11) plane. The latter surfaces are encountered more prevalently in f.c.c. materials as the particle size is increased via successively higher annealing temperature^.^ These results then are consistent with rate measurements on supported nickel catalysts,’ which show hydrogenolysis activity to be a strong function of particle size, the larger particles exhibiting the lower rates.In a related study, the selectivity for ethane production from the hydrogenolysis of n-butane over Ir single crystals has been demonstrated to scale with the concentration of low-coordination-number metal surface atoms. The Ir( 110)-(1 x 2) surface, which has a stable ‘missing-row’ structure, has been found to produce ethane very selectively. This contrasts with the results for the close-packed I r ( l l 1 ) surface, where only the statistical scission of C-C bonds has been observed.‘ The results of this study‘ correlate qualitatively with the observations made previously for selective hydrogenolysis of n-butane to ethane on supported Ir catalysts’ as a function of Ir particle size.As shown in fig. 1, the results for Ir( 1 lo)-( 1 x 2) model very well the small-particle limit whereas the results for Ir( 11 1) relate more closely to the data for the corresponding large particles ( > l o nm). By assuming particle shapes the general behaviour of declining selectivity with larger particle size can be accurately modelled,‘ as illustrated in fig. 1. 337338 20 0 Butane Hydrogenolysis - I I I I side view Fig. ( a ) Schematic representation of the (110)-(1 x 2) and (111) surfaces of Ir. The z axis is perpendicular to the plane of the metal surface. C,, designates the coordination numbers of the metal surface atoms. ( b ) Selectivity for C2H, production (mol o/o total products) for n-butane hydrogenolysis on Ir single crystals‘ and supported Ir catalysts’ at 475 K. The effective particle size for the single-crystal surfaces is based on the specified geometrical shapes.A, lrA1203; W, Ir/Si02. The stoichiometry of the surface intermediate leading t o high ethane selectivity, based on kinetics and surface carbon coverages subsequent to reaction, is suggested to be a metallocycle pentane.6 Based on analogous chemistry reported in the organometallic literature,’ the mechanism responsible for ethane is postulated to be the reversible cleavage of the central C-C bond in this metallocycle intermediate. This is a 1,4- diadsorbed hydrocarbon species on the sterically unhindered and coordinatively unsaturated ‘C7’9 sites on the corrugated Ir( 1 lo)-( 1 x 2 ) surface. These studies have shown that it is now possible using single-crystal surfaces to construct in a systematic fashion uniform surfaces with specific reaction sites.This then enables a rather precise approach to detailing at the atomic level the origins of site specificity in surface-catalysed reactions. In this paper we report a study of n-butane hydrogenolysis on Rh( 110) and Rh( 11 1) single crystals. This work, together with the previous data for Ir,‘ strongly suggest that a key factor in determining the selective catalytic chemistry of Ir and Rh for butane hydrogenolysis to ethane is the availability of an unhindered (coordinatively unsaturated) atom site. Such sites are available only at high metal dispersions on supported metal catalysts or on the Ir( 1 lo)-( 1 x 2 ) surface.On the Rh( 110) surface, where the (1 x 2 ) reconstruction does not occur, and for the close-packed Rh(ll1) surfaces, a second mechanism, presumably involving an intermediate coordinated to multiple surface sites, may be operative. Experimental The experiments were performed in a stainless-steel, dual-chambered apparatus which has been described in detail elsewhere.’ The chambers are linked uia a gate valve andA. K . Datye, B. F. Hegarty and D. W. Goodman 339 0.001 1.6 1.7 lo3 K/ T 2 Fig. 2. Specific reaction rate for the hydrogenolysis of n-butane over Rh( 110) (n, = 9.6 x 10'' sites crn-*). Pbutglne = 10 Torr and PHI = 200 Tom. 0, Methane; +, ethane; A, propane. each can be evacuated to <lo-'' Torr.? Crystals were mounted on a retraction bellows and translated vertically between the analysis chamber and the reaction chamber. The analytical chamber is equipped with a cylindrical mirror analyser (CMA) for Auger spectroscopy (AES) and a quadrupole mass analyser for thermal desorption spectroscopy (TDS).The reaction chamber, which can be pressurized to several atmospheres,$ was operated as a batch microreactor. The crystal temperature was monitored by either a chromel/alumel thermocouple (for Rh) or a W-5% Re/W-26'/0 Re thermocouple (for Ir) spot-welded to the back of the crystal. The temperature of the sample was maintained during reaction by an RHK temperature programmer to *1 K. Rhodium crystals were cleaned in the analysis chamber by annealing at 1100 K for 10 min followed by oxidation at 5 x lo-* Torr O2 at 700 K for 1 min and then a second anneal to 1000 K until no surface impurities were detectable by AES.Reaction products were analysed by gas chromatography. Absolute reaction rates on the Rh crystals were calculated from the reactor volume (600cm'), duration and temperature of reaction, the measured surface area of the crystals and the known atomic density of each Rh surface.'" Butane (Scientific Gas Products, nominally 99.99% ) was degassed repeatedly at 80 K. Impurity ethane and propane were removed by multiple distillations from a liquid-solid hexane bath. Hydrogen (99.99%) was supplied by Matheson and used without further purification. Results The specific turnover frequencies for the production of methane, ethane and propane from 20 : 1 H,-n-butane on an Rh( 110) single crystal in the temperature range 450-620 K are plotted in fig.2. At all temperatures the order of the abundance of products is methane > ethane > propane. Below 500 K the activation energy for all three products was 135 * 6 kJ mol-'. Above 500 K, the product distribution shifted to reflect more complete hydrogenolysis. t 1 Torr = 101 325/760 Pa $ 1 atm = 101 325 Pa.3 40 Butane Hydrogenolysis o . o o : . ~ 0 100 ZOO 300 400 500 600 PH Torr I0 Fig. 3. Hydrogen pressure dependence for n-butane hydrogenolysis over Rh( 110). T = 500 K, Pbulane = 10 Torr. 0, Methane; *, ethane; A, propane. Fig. 4. Specific 0, reaction rate for the hydrogenolysis of n-butane over Rh( 11 1) (n, = 1.6 x Methane; *, ethane; A, propane.Phu,.,ne = 10 Torr. PH> = 200 Torr. The variation in the reaction rate with change in the partial pressure of hydrocarbon and hydrogen was studied at 500 K. A change in the butane partial pressure was observed to produce little variation in activity and selectivity. In contrast, variations in the partial pressure of hydrogen at this temperature (fig. 3) show a ‘volcano’ curve. Above 350 Torr, the hydrogen dependence is of negative order. Below 350 Torr, the reactivity decreases with decreasing hydrogen pressure, and the selectivity towards ethane and propane is reduced. Fig. 4 and 5 show the results for kinetic studies of the butane-hydrogen reaction on the Rh(ll1) surface. The Arrhenius plot of fig. 4 shows the same trend towards moreA.K . Datye, B. F. Hegarty and D. W. Goodman 341 0 200 400 600 800 11 00 P,,/Torr Fig. 5. Hydrogen pressure dependence for n-butane hydrogenolysis over Rh( 11 1). T = 475 K, Pbutene = 10 Torr. 0, Methane; +, ethane; A, propane. complete hydrogenolysis at higher temperatures as was observed for the Rh( 110) surface. The onset of ‘roll over’ is at a lower temperature, 475 K. At temperatures below 475 K the yield of ethane is consistently higher than that of methane by a factor of ca. 1.55. The hydrocarbon partial pressure dependence for the Rh(ll1) surface at 475 K is slightly positive, whereas the selectivity is independent of butane pressure. An increase in the partial pressure of hydrogen (fig. 5) up to 200Torr results in an increase in catalytic activity. Above 200 Torr the reaction is negative order in hydrogen; below 200 Torr the selectivity to ethane and propane increases with increasing hydrogen pressure. Discussion The literature reports of butane hydrogenolysis on supported Rh are less definitive than the results for Ir.’ Wong et al.” reported that at 454 K in a pulse reactor, 1.5 nm Rh particles on silica gave an ethane/propane ratio of 7.8, while 5.7nm particles gave a ratio of 3 .Bernard et aZ.” also report high ethane selectivity on Rh/alumina. In related studies, central scission of n-pentane (to ethane and propane) has also been seen on highly dispersed Rh on alumina,13 and on highly-dispersed Rh/Si02 .I4 A key issue in the present studies is the suitability of the corrugated Rh(ll0) for modelling highly dispersed, supported Rh catalysts as is the case for Ir( 1 lo)-( 1 x 2 ) and supported Ir catalysts.6 Likewise, if the correlation between Ir and Rh holds for the ( 1 10) surfaces, it is expected that Rh( 11 1) will demonstrate the chemistry of a relatively large (> 100 A) supported Rh particle.For both Rh surfaces, the extent to which hydrogenolysis proceeds increases with increasing reaction temperature (fig. 2 and 4). This is in keeping with the general trend for increased cracking at higher temperatures for alkane reactions. The term ‘roll over’ has been used to describe the fall in overall activity at the high temperatures which leads to a decrease in the selectivity for the production of ethane and propane on Ir crystals.6 For Ir, decreasing the partial pressure of H2 at the temperature of onset of roll over induces the same selectivity change as observed for an increase in reaction temperature.The same correlation was observed for Rh in the present study. The origin342 Butane Hydrogenolysis of this effect is believed to be the same for both Ir and Rh, viz. as the reaction temperature is raised beyond a critical temperature, defined primarily by the hydrogen partial pressure, the hydrogen surface coverage falls below a saturation or critical coverage. The lower hydrogen coverage then reduces the efficiency of the hydrogenation of surface hydrocarbon fragments. The higher temperature of onset of roll over on the more open Rh surface was also observed for Ir. This behaviour correlates with the higher binding energy of hydrogen adatoms on the I r ( l l 0 ) - ( 1 x 2 ) surface and suggests that the source of the reactive hydrogen is the metal surface rather than, for example, an ‘active’ carbonaceous over- layer.The possibility of adsorbed hydrocarbon species acting as transfer agents of hydrogen has been suggested for various alkane reactions.15 The absence of high ethane selectivity on Rh(ll0) very likely relates to stability of the Rh( 110) surface towards the (1 x 2) reconstruction. The corresponding Ir( 110) surface undergoes a reconstruction (stable under reaction conditions), described as Ir( 1 lo)-( 1 x 2) or ‘missing-row’ structure, resulting in rows of the highly coordinatively unsaturated C, sites. These sites can form the metallocyclopentane species which has been proposed as an intermediate in the central scission of butane to ethane.No analogous sites exist on the unreconstructed Rh( 110) surface. Although the absence of C, sites is the most likely explanation for the observed selectivity in hydrogenolysis products, there are noteworthy differences in the organometallic chemistry of Ir and Rh. Specifically, the rhodiacyclopentane analogue of the iridiacyclopentane complex has been shown to be more difficult to synthesize and, once made, decomposes more readily.’‘ Butane hydrogenolysis on Rh( 11 1) appears to operate via the same mechanism as it does on the Ir( 11 1) surface. First, dissociative chemisorption of butane and hydrogen occurs followed by irreversible cleavage of the terminal carbon-carbon bond of the adsorbed hydrocarbon. Further C-C bond cleavage prior to product desorption leads to the methane and ethane observed as initial products.The reaction kinetics on the Rh( 11 1) surface (fig. 4 and 5) follow the same trend as those on the Rh(ll0) surface. The roll over in the Arrhenius plots sets in at a lower temperature (ca. 475 K) but has the same overall effect, namely, decreasing ethane and propane selectivity (fig. 4). The reaction is slighly positive order in hydrocarbon partial pressure. A decrease in the partial pressure of hydrogen at 475 K (fig. 5 ) has the same effect on product selectivity as has an increase in temperature. Likewise for this surface, the rollover in the Arrhenius plots at ca. 475 K can be ascribed to a decrease in the surface concentration of hydrogen.The primary difference then between the Rh(ll0) and R h ( l l 1 ) surfaces is the relatively large ethane selectivity for the close-packed Rh( 11 1) surface. Although the ethane selectivity is not as distinctive on Rh( 11 1) as for the Ir( 1 lo)-( 1 x 2) surface,‘ the selectivity towards its production is consistently higher (by a factor of 1.55) at reaction temperatures below 475 K. Braunschweig et al.” have used high-resolution transmission electron microscopy (HRTEM) to correlate changes in Rh particle morphology, induced by oxidation- reduction cycles, with change in butane hydrogenolysis selectivity. Ethane selectivity was lower for Rh particles that were oxidized and subsequently reduced under mild conditions, leading to a ‘roughened’ surface structure.” When the Rh was reduced at higher temperatures, where the more stable ( 1 11) surface facets would be exposed, the ethane selectivity increased.The oxidation-reduction cycling of Rh/Si02 by Gao and Schmidt” also shows similar trends. Hence, the dominance of ethane production on the close-packed Rh( 11 1) surface is in qualitative agreement with the results on supported Rh metal particles that were 5.0 nm in diameter” or larger.” The selective route to butane hydrogenolysis seen on highly dispersed Ir7 and on the single-crystal Ir( 1 lo)-( 1 x 2) surface‘ is not evident on any of the Rh surfaces studiedA. K. Datye, B. F. Hegarty and D. W. Goodman 343 here. Other reports on Rh,’t-’4 however, do imply that highly dispersed Rh catalysts are indeed selective towards central bond scission. For instance, Yao et a/.” reported that central bond scission of n-pentane was the only hydrogenolysis mode on the highly dispersed so-called ‘delta’ phase on Rh/AI2O3.This selectivity was lost when the metal loading was increased, presumably due to formation of ‘metallic’ Rh surfaces. The data imply that the mechanism involving the 1,4-diadsorbed n-butane which may occur only on low-coordinated C, sites is absent on the Rh( 110) and Rh( 11 1) surfaces studied in this work. Conclusions Single-crystal metal surfaces allow us to study in a systematic fashion the role of surface structure on catalytic activity and selectivity. We have found that the selectivity in n-butane hydrogenolysis is markedly affected by surface structure.Ir( 110) surfaces which have a high concentration of C, low-coordination sites show marked propensity to central bond scission. This selective hydrogenolysis route may involve adsorption of the n-butane as a metallocycle pentane and subsequent cleavage at the central carbon- carbon bond. Similar surface chemistry is exhibited by highly dispersed Ir and Rh supported metal catalysts. On larger metal particles, and on Ir( 11 1) surfaces, a non- selective hydrogenolysis is observed, yielding equal amounts of ethane, methane and propane. On Rh( 1 lo), which is not stable towards the (1 x 2 ) reconstruction, we find a rather non-selective hydrogenolysis, but on Rh( 11 1) a slightly higher proportion of ethane is observed in the products. This implies that for the close-packed Rh(ll1) surfaces, a second mechanism, presumably involving an intermediate coordinated to multiple surface sites may be operative.We acknowledge with pleasure the partial support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, and by Sandia National Laboratories, Albuquerque. References 1 13. W. Goodman, J. Vac. Sci. Techno/., 1982, 20, 522; D. W. Goodman, Ace. Chem. Res., 1984, 17, 194; D. U’. Goodman, Annu. Rev. P/ij,s. C/7em., 1986, 37, 425; D. W. Goodman and J. E. Houston, Science, 1987,236, 403; A. G. Sault and D. W. Goodman, in Molecule-Surface Interactions, ed. K. Lawley (John Wiley, Chichester, 1988). 2 D. W. Goodman, Surf Sci., 1982, 123, L679. 3 D. W. Goodman, Proc. 8th I n t .Congr. Catal., Berlin, July, 1984 (Verlag Chemie, Berlin, 1984). 4 J . E. A. Clark and J. J. Rooney, Adc. Caral., 1976, 25, 125. 5 J . T. Carter, J . A. Cusumano and J. H. Sinfelt, J. Phys. Chem., 1966, 70, 2257; D. J. C. Yates and J. H. infelt, J. Catal., 1967, 8, 348; G. A. Martin, J. Catal., 1979, 60, 452. 6 J . R. Engstrom, D. W. Goodman, and W, H . Weinberg, J. Am. Chem. Soc., 1986, 108, 4653; J. R. Engstrom, D. W. Goodman and U’. H. Weinberg, J. Am. Chem. Soc., 1988, 110, 8305. 7 K. Foger and J . R. Anderson, J. Catal., 1979, 59, 325 8 For example, R. H. Grubbs and A. Miyashita, J. Am. Chem. Soc., 1978, 100, 1300; R. H. Grubbs, A. Miqashita, M. Liu and P. Burk, J . A m C’hem. So(,., 1978, 100, 2418. 9 R. \ a n Hardeveld and F. Hartog, Adc. Caral., 1972. 22, 75. 10 T. W. Root, L. D. Schmidt and G . B. Fisher, Surf: Sci., 1985, 150, 173. 11 T. C. Wong, L. C . Chang, G. L. Haller, J. A. Oliver, N . R. Scaife and C. Kemball, J. Caral., 1984, 87, 389; T. C. Wong, L. F. Brown, G. L. Haller and C. Kemball, J. Chem. Soc., Faradax Trans. 1, 1981, 71, 519. 12 J . R. Bernard, J. B o u q u e t and P. Turlier, Proc. 7th Int. Congr. Catal., paper A7, Tokko (1980). 13 H. C . Yao, Y . F. Y . Yao and K. Otto, J. Cafal., 1979, 56, 21. 14 J . K. A. Clarke, K. M. G. Rooney and T. Baird, J. Catal., 1988, 111, 374. 15 M. W. Vogelzang and V. Ponec, J. Catal., 1988. 111, 77; S. M. Davis, Zaera and G. A. Somorjai, J. Caral., 1985, 84, 206; 1977, 82, 439.344 Butane Hydrogenolysis 16 A. Cuccuru, P. Diversi, G. Ingrosso and A. Lucherini, J. Organomer. Chem., 1981, 204, 123. 17 E. J. Braunschweig, S. Chakraborti, A. D. Logan and A. K. Datye, Proc. 9fh Int. Congr. CafaL, ed. M. J. Phillips and M. Ternan (The Chemical Institute of Canada, 1988), vol. 111, p. 1122. 18 S. Chakraborti, A. K. Datye and N. J. Long, J. CafaL, 1987, 108, 444. 19 S. Gao and L. D. Schmidt, J. Cafal., 1988, 111, 210. Paper 8/05055D; Received 19fh December, 1988

ISSN:0301-7249    

DOI:10.1039/DC9898700337

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1989, 87, 345-355 GENERAL DISCUSSION Prof. P. B. Wells (University of Hull) said: I wish to address the matter of the dependence of selectivity on metal particle size contained in fig. 6 of Prof. Ponec's paper. In Faraday Discussion 72,' I reported an increase in selectivity in iridium- catalysed butadiene hydrogenation as a function of metal loading and hence of iridium particle size. Since the selectivities in ethyne hydrogenation and butadiene hydrogena- tion over a given metal are comparable, I conclude that my earlier results and the present results of Prof. Ponec concur. It is interesting, however, to compare our interpretations. The present paper proposes that intermediates having a high degree of ligancy with the surface are partly responsible for ethane formation, and thus their formation is suppressed by diminishing the mean particle size and by alloying.In the paper at this meeting by Datye et al. it is proposed that low coordination-number surface atoms as found in the Ir( 1 lo)-( 1 x 2) structure, typify surface atoms in highly dispersed supported iridium catalysts, whereas surface atoms in Ir( 11 1) typify those in the surfaces of larger iridium crystallites. The surface site for ethylidyne is the three-fold site well provided by the Ir( 11 1)-surface. Your claim that fewer ethylidyne and ethylidene intermediates are formed on the smaller metal particles is consistent with the work of Datye et al. Thus, the surface science studies are mutually supportive. My interpretation of selectivity is based on quite different models and mechanisms.First, my isotope tracer work has demonstrated the participation of adsorbed alkylidenes in alkane formation only in the case where catastrophic breakdown of selectivity occurs over palladium.2 For the other five platinum metals and rhenium, selectivity is well interpreted by my hydrogen occlusion modeL3 In that paper I demonstrated that the extent of hydrogen occlusion in these metals (as powders or as supported metals) is related to the state of disorder within the metals, as prepared by reduction at temperatures below their Huttig temperatures. My theory provides a good correlation between the extent of hydrogen occlusion, the Huttig temperature, and the selectivity for six elements. The tests of this theory were: (i) that selectivity should increase with decreasing particle size (because with that decrease, the amount of metal in the bulk decreases and the extent of occlusion must decrease) and (ii) that selectivity should markedly increase if catalysts were annealed at very high temperatures (because the inherent disorder and hence the hydrogen occlusion capacity, should decrease).These experimental tests have been reported previously,' and both predictions were obeyed. Indeed, iridium catalysts annealed at 1000 "C provided selectivities of 95%, on a par with that given by palladium. I do not think we can marry our interpretations. We are both measuring the extents of disorder (variously defined) in the bulk or in the surface and are looking for the consequences that ensue.I claim that selectivity is high at well ordered regions of the surface and low over disordered region^.^ If I understand correctly, you claim the reverse: that alkylidene is an intermediate in alkane formation and is less abundant on a more disordered surface, so you require selectivity to be low over well ordered regions of the surface and high over disordered regions. I propose that the critical experiment is provided by the effect on selectivity of annealing metal particles. My model predicts an increase, and that is my reported experimental result.' I would welcome your comments. 1 A. G. Burden, J . Grant, J . Martos, R. B. Moyes, and P. B. Wells, Faraday Discuss. Chem. SOC. 1981, 72, 95. 2 R. G. Oliver and P. B. Wells, J. Catal., 1977, 27, 364.3 P. B. Wells, J. Caral., 1978, 52, 498. 3453 46 General Discussion Prof. Ponec responded: I feel quite reassured by two facts you mention in your remark: (1) that there is so far full agreement between the experimental data available; ( 2 ) that surface science studies are mutually supportive, in other words our interpretation is supported by the report by Datye et al. in these proceedings. What remains as a point of discussion is your idea of the ‘reversibly occluded H2’ and its role in the lowering of the selectivity to ethene. I am glad that you recall Discussion 72 since it was there where I first had the opportunity to express my doubts about the occlusion model. Since then, no new experimental evidence has appeared proving either the existence of reversibly occluded hydrogens or the role of it in hydrogenation of alkynes into alkanes.Therefore, I keep my doubts. Why is Ir so exceptional in occluding hydrogen? Why is the ‘reversibly occluded hydrogen’ (H2, obviously, since Ir is not forming hydrides easily) so good for the non-selective hydrogenation? The six metal elements tested by your laboratory also differ in other parameters and not only in those which you relate to the occluded hydrogen. For example, in their electronic structure, and due to this difference the metals also differ in their property to form various complexes, complexes bound by multiple bonds as ethylidyne or ethylidene (like) complexes. Also, after your repeated expression of your confidence in the ‘occlusion model’ I am still not convinced that the chemical bond aspects of the differences between the metals (or with a given metal, between the particles of different sizes) are less important than the rather hypothetical ‘occluded hydrogen’ and its hypothetical role in the non-selective hydroge- nation. Nevertheless, the experiments that you suggest should be done and I hope that we shall perform them.It would be the greatest pleasure and privilege if I had the opportunity to continue our discussion at the next Faraday Discussion. However, the existing average frequency of the Discussions devoted to catalysis does not make me optimistic in this respect. Prof. P. B. Wells communicated: It is important to appreciate that the hydrogen occluded in iridium and neighbouring transition elements is atomic not molecular, is stabilised in disordered structures, and exceeds in quantities that can be dissolved in ordered structures (i.e.in annealed metals). Dr P. A. Sermon (Brunel University, Uxbridge) (communicated): Do you have any evidence that co-impregnation of porous silica by solutions of two metal salts produces well defined and homogeneous bimetallic particles? In previous work you have emphasised the importance of metals in positive oxidation states; was this also the case here? Prof. Ponec replied: Evidence exists that Ir-Pt when coimpregnated, do form homogeneous alloys.’ There is no such evidence for Ir-Au and Ir-Cu systems, which do not form solutions; the Ib metal is just blocking the surface of Ir. There is no information available how the Ib metals are exactly distributed; some data (EXAFS) are known for systems (Ir-Cu/Si02) studied by Meizner et al.’ Please note that in this paper not a pellet, but rather a small-particle SiOz powder has been impregnated. The role of M“+ has not been studied.1 M. J. Dees and V. Ponec, J. Cnrul., 1989, 115, 347. 2 G. Meizner, G. H. Via, F. W. Lytle and J. H. Sinfelt, J. Chem. Phys., 1985, 83, 353. Dr G . McDougall (University of Edinburgh) then commented: I am concerned that all the EELS spectra reported were recorded in the specular direction, that is to say that the angle of incidence of the electron beam is equal to the collection angle of the scattered electrons. In this configuration both dipole and impact scattering may con- tribute to the observed loss intensity.The relative contribution from the two scattering mechanisms is strongly dependent on the degree of order in the adsorbate overlayer.General Discussion 3 47 In your system, where there are co-adsorbed species, the long-range order is likely to be low so almost all losses would be expected to show both a dipole and a substantial impact contribution. In these circumstances, the relative intensities of modes, even of the same symmetry type, may vary disproportionately as a function of coverage. I feel this factor may account for some of the apparent complexity in the spectra without resorting to assignment of the 1340 and 1090 cm-' ( P # 1) losses to ethylidene. Prof. Ponec responded: Thank you very much for suggesting this to us; it is indeed, an idea to be considered carefully.We expected the difference between the intensities of the dipole- and impact-scattering to be so large that a variation in their contributions would not lead to P # 1 ratios in the intensities in question. With transmission i.r. the ratio of intensities, P, is almost 1 for the model ethylidyne compound, but P Z 1 for small Pd and Pt particles, although the mechanism of absorption and selection rules are here different from those of EELS and reflection i.r. On the other hand, the i.r. transmission allow an alternative interpretation that ethylidene is also being formed and seen. 1 S. B. Mohsin, M. Trenham and H. J. Robota, J. Phys. Chem., 1988, 92, 5229. 2 J. Card., 1985, 96. 3 T. P. Beebe Jr and J. T. Yates Jr, J Am. Chem. Soc., 1986, 108, 663.Prof. J. B.Nagy (FUNDP, Namur, Belgium) remarked: I3C-n.m.r. of adsorbed species on polycrystalline samples could help to distinguish between ethylidene (=CH-CH) and ethylidyne (_C-CH,). For example, in spin-echo I3C-'H double-resonance (SEDOR) flipping the H nucleus changes the precession frequency of the I3C nucleus which is bound to an H atom and can destroy the echo of that I3C nucleus.' 1 P-K. Wang, C. P. Slichter and J . H. Sinfelt, J. Phys. Chem., 1985, 89, 3606. Prof. W. Palczewska (Polish Academy of Science, Warsaw, Poland) said: P. B. Wells, in his introductory lecture, mentioned the mechanism of the CzHz-CzH4-CzH6 hydro- genation proposed by Thomson and Webb' and supported by experimental evidence presented by other authors as well. The mechanism emphasizes, that C,H, adspecies, formed on a metal catalyst surface, together with metal free active sites, represent a complex field of action, where the hydrogenation is taking place in the over-layer.My question to Prof. Ponec concerns his opinion on that mechanism, taking into account the presented results and their interpretation. 1 S . J. Thomson and T. Webb, J. Chem. SOC., Chem. Commun, 1976, 526. Prof. Ponec replied: Hattori and Burwell' tested the idea of Thomson and Webb by following the reaction of cyclopropane hydrogenation to propane using a pulse- reactor. The result obtained did not confirm the mechanism you mention. Since the possibility exists that ethyne behaves differently from cyclopropane, we decided to study the ethyne hydrogenation in a pulse-reactor. So far, with Ir our results were the same as obtained by Hattori and Burwell, i.e.at variance with the ideas of Thomson and Webb. Palladium and platinum catalysts are presently under study. 1 T. Hattori and R. L. Burwell Jr, J. Chem. Soc., Cliem. Commun., 1978, 127. Prof. A. Zecchina ( University of Turin, Italy) began the discussion of Prof. Ichikawa's paper: There is a remarkable difference between the spectra of Rl,(Co) generated in situ [fig. l ( a ) ] or generated by adsorption of CO [fig. 3(a)]. In particular the frequencies of bridged CO differ by ca. 100 cm-'. Their relative intensity with respect to linear CO bands is also quite different. Can you add further comment to this? Prof. Ichikawa replied: We propose the stoichiometric formation ('Ship-in-bottle synthesis') of hexanuclear Rh carbonyl clusters by CO chemisorption on the reduced348 General Discussion Rh particles inside Nay.They have been characterized with CO chemisorption stoichiometries and Rh K-edge EXAFS for C,N, and Rh-Rh distances, which are in good agreement with those of Rh6(CO)16/NaY. The only remarkable difference is found in i.r. spectra between [Rh,]/NaY+CO and Rh (C0)16/NaY, showing the bridged CO bands at 1830cm-' and 1760cm-', respectively. There is no reference for the edge-bridged type Rh6(C0),' by the conventional synthesis in solution, where only the stable face-bridged Rh,(CO),, is formed. We suggest that the Rh carbonyl species formed by CO chemisorption would be an edge-bridged Rh6(C0)16 inside NaY as the analogous formation of unusual Pd' carbonyl cluster 'Pd13(CO)x/ NaY' recently reported by Sachtler et al.We have two isomers of k6(co)l6 e.g. edge- and face-bridged types, showing 1830 cm-' and 1780 cm-', respectively. It is interesting to find that the edge- bridged Rh6( CO)16 species generated by CO chemisorption at room temperature was easily converted under CO/or CO-H2 atmosphere at 50-100 "C to the face-bridged Rh6(co)16 having the characteristic band at 1760 cm-'. Dr J. C. Vedrine (Institut de Recherches sur la Catalyse, CNRS, Villeurbanne, France) commented: You have prepared small alloy particles RhIr or RhFe which you say are inside NaY cavities and you relate catalytic properties to particle size. However, the particles occupy the whole void volume of the cavities, i.e.the reactant molecules can only react via the window openings. On the other hand, acidic sites of the zeolite created during the exchange process of Na+ by Rh3+ or Ir4+ chlorides in aqueous solution, as well known, may play an important role. Comparison with other acidic catalysts such as silica, alumina or large-pore acidic zeolites (mordenite, offretite, etc) for catalytic properties is necessary. Could you comment on the reactant accessibility (activity should decrease) and role of acidity? Prof. Ichikawa replied: After removal of carbonyls from cluster precursors such as Rh6-,IrX(CO),6/NaY(X = 0-6) by mild oxidation, we predict that the resulting H2- reduced RhIr particles less than 8 A in size (which was estimated from EXAFS data: see table 2 ) leave a free space available for butane accessibility inside NaY cavities of 13 A diameter.The reduced Rh particles inside NaY chemisorbed 16CO to give Rh6(CO),,, which occupies the whole void volume of the cavities. As a similar answer to the comments of Prof. Ponec, we consider that the acidity of zeolites is not the decisive parameter for the markedly different activities and selectivities in butane hydrogenolysis on a series of Rh,-,Ir,/NaY catalysts, compared with the ensemble size-effects by changing Rh/ Ir metal compositions. Prof. R. A. van Santen (University of Technology, Eindhoven, The Netherlands) asked: Can the author inform us on the stability of his catalysts. Did he do the physical characterization only on the catalysts as prepared or does he also have information on the physical properties of the spent catalyst? Prof.Ichikawa answered: (1) The zeolite-entrapped RhIr and RhFe catalysts showed fairly stable activities and selectivities in butane hydrogenolysis and benzene hydrogena- tion for a time-on-stream of 40-100 h. We have conducted the EXAFS evaluation for the particle size and metal composition for the Rh,Ir2/NaY and Rh6/NaY catalysts after the prolonged butane hydrogenolysis at 220 "C. The results suggest that the particle sizes and metal compositions were not appreciably changed in terms of C.N. for Rh-Rh and Rh-Ir bonds before and after the catalytic reactions. Prof. V. Ponec (Leiden University, The Nrtherlands) then commented: In your paper all variations in activity and selectivity are explained by differences in metal particles.However, in your catalyst preparation, steps are involved which can change the composi- tion (Al) and acidity (Na) of the zeolite. Impregnation by e.g. RhC13 releases HCl,General Discussion 3 49 which can leach out Na, Al, or both. You find most pronounced differences in butane hydrogenolysis (much less in benzene hydrogenation) and you also find much larger differences between Ir and Rh, than found usually with other supports. These two facts strengthen my suspicion expressed above. Prof. Ichikawa replied: We have prepared a series of Rh,-,Ir,/NaY (x = 0-6) catalysts by using different precursors, e.g. RhC13, Rh(NH3)5C12+, IrCl; and Ir(NH3)&12+ for the carbonylation. We found that the activities and selectivities towards C-C bond scission in butane hydrogenolysis are not appreciably influenced by the type of precursor, but depend remarkably upon the Rh/Ir metal compositions.Additionally, the i.r. study in pyridine titration demonstrated no effective difference for the Lewis and BrBnsted acidity on Rh,(CO) 16/ Nay-derived catalysts prepared by using RhC13 and Rh( NH3)5C12+ as the precursors. Accordingly, we suggest that the marked activity differences in butane hydrogenolysis are associated with the Rh ensemble sizes of RhIr bimetal cluster catalysts, not with the acidity of zeolites in the catalyst preparation. Dr A. F. Masters (University of Sydney, New South Wales, Australia) said: The synthetic strategies employed by Prof. Ichikawa, and the materials produced, are quite important.Equally, it is essential to ensure that they are adequately characterised . This requirement is underscored by the many characterisation techniques Prof. Ichikawa has employed in an effort to characterise the species produced. However, the syntheses of metal carbonyl cluster species are generally non-trivial, and can be quite experimentally demanding. The preparations often involve complex separations of many products with similar properties, and the target compound is often obtained only in low yield. Thus, for example, genuine [Rh4Fe2(C0),6]2- is only obtained in 50% yield by the method reported in ref. (22). This synthetic method does not involve the reaction of [HFe,(CO),,]- and [Rh,(CO),,] as reported in the present work. Moreover, the infrared spectrum in the carbonyl-stretching region of [Rh,Fe2(CO),6][NMe3(CH2Ph)]2 in CH3CN as reported in ref.(22) (2048 w, 2005 vs, 1986 s, 1961 m, 1948 m, 1904 w, 1728 sh, 1710 ms cm-') is different from that reported in the present work (2078 s, 2018 w, 1711 sh, 1744 m cm-I). Similarly, although the syntheses of zeolite-entrapped Rh6-xIrx(CO)16 (x = 0-6) are claimed, the infrared spectra on which these claims are based (table 1 of paper) are not significantly different. The difficulties associated with the interpretation of EXAFS data have been discussed earlier. The differences in the Rh-Rh distances reported in table 2 of the paper are not statistically significant. Similarly, the coordination numbers reported are not necessarily indicative of Rh6 (presumably pseudo-octahedral) clusters.Other geometries are equally well accommodated by these coordination numbers. Thus, it is far from clear that these synthetic procedures have produced materials each with a single well characterized species within the zeolite. In making these comments, I do not wish to detract from, but rather to re-emphasise the significance of, Prof. Ichikawa's work. He has prepared a set of unusual materials with interesting catalytic properties. The challenge now is to attempt to identify the natures of these species. This characterisation will not be an easy task. Prof. Ichikawa addressed Dr Masters comments: As for the synthesis of Rh,Fe2(CO):; we used HFe3(CO),I/NaY with Rh,(CO),, as the starting materials. We found by in situ i.r. studies that HFe,(CO);, inside NaY was converted partially into Fe3(CO):; and Rh4(C0)12 was decomposed to Rh(C0)2 species by heating to 70-100 "C, where the Rh4Fe2(CO):, species was formed in the analogous reaction of Fe,(CO):; and [Rh(CO),CI] in THF solution.23350 General Discussion The carbonyl stretching bands of the resulting RhFe species formed by this procedure are systematically shifted, compared with those of a free Rh,Fe,(CO):; in solution.This is reasonably interpreted as due to the intrazeolitic interaction between the RhFe bimetal cluster and Lewis-acid sites of zeolite wall, in a similar way to Rh,(CO),,/NaY. We predict from the i.r. characterization, as shown in table 1 and fig. 1 of the paper that RhIr bimetal carbonyl clusters with octahedral metal frameworks are homogeneously synthesized inside Nay.According to the EXAFS data for Rh,(Co),,/NaY we suggest that Rh, clusters interact with zeolite oxygen and possibly exist in a pseudo-octahedron framework. The evaluated coordination number of Rh-Rh bonds, smaller than four, is associated with the minor contribution of the Rh-0 bond due to the Rh(CO)? species, as discussed in the paper. Prof. H. Knozinger (Uniuersity of Munich, Federal Republic of Germany) then remarked: In their paper, the authors report synthesis of carbonyl clusters in zeolite cages. The metals involved are known to form well defined molecular carbonyl cluster compounds. In contrast, carbonyl clusters of palladium are not known. We wish to mention that we recently succeeded in stabilizing Pd clusters in the supercages of NaY zeolites; the cluster presumably contains 13 atoms and after adsorption of carbon monoxide the carbonyl infrared spectra clearly indicate formation of a molecular carbonyl cluster.The naked Pd,3 cluster would be entrapped in the supercage since its diameter is slightly larger than that of the window and the carbonylated cluster would still fit into the supercage.' 1 L. L. Sheu, H. Knozinger and H. M. H . Sachter, Cam/. Leu., 1989. 2, 129. Prof. Ichikawa, in response, said: We appreciate the finding of Pd13 carbonyl clusters by CO chemisorption on Pd/NaY catalyst. At the same time, I would like to suggest the 'ship-in-bottle' formation of a structurally unstable hexanuclear Rh carbonyl cluster by CO chemisorption on the reduced Rh particle inside Nay.Dr J. Evans (Southampton University) said: Characterisation of cluster species within the pores of a zeolite is difficult. The EXAFS technique used as part of this study presents both an opportunity and a difficulty. We have analysed the Rh K-edge EXAFS of [Rh,(CO),,],' the molecule central to the discussions of this paper. As show? in fig. 1, there are three main peaks in the Fourier transform of these data. Around 4 A there is a peak due to the non-bonded Rh-Rh distance across the cluster octahedron. This feature is indicative of such a cluster skeleton, so my first question to Prof. Ichikawa is whether such features were clearly identifiable in his data. My second point relates to the large peak in the Fourier transform near 3 A. This is labelled in the paper as being due to Rh-Rh back-scattering.Our analysis of the data on the pure compound, using multiple scattering procedures,? suggests that this is a composite of three components. In addition to the metal-metal back-scattering, the terminal and face-bridging carbonyl oxygen atoms also contribute substantially; these are separable because of the large difference in the Rh-C-0 bond angle, giving radically different multiple scattering contributions which are highly sensitive to the M-C-0 angle. I would like to ask how these effects were treated in the analysis of the EXAFS data on the intra-zeolite cluster. 1 R. J . Price, Ph.D. Thesis (University of Southampton, 1987). 2 N. Binsted, S. L. Cook, J. Evans, G. N . Greaves, and R. J . Price, J. Am.Chem. Soc., 1987, 109, 3669. Prof. Ichikawa replied: (1) Fig. 2 in the paper shows three main peaks in the Fourier transform of Rh K-edge EXAFS for the Rh,(CO),,/NaY sample, involving an additional peak at 3.65 A (3.8, A after the phase-shift correction), which might be dueGeneral Discussion I 351 1 2 3 5 R I A Fig. 1. Rh K-edge EXAFS of [RhJCO),,] (upper) and the Fourier transform (lower) of Fourier- filtered experimental (-) and calculated ( - - - ) data. to the non-bonded Rh-Rh distance across the cluster octahedron as Prof. Evans suggested in his comment. We expect from the EXAFS parameters that there is some deformation of cluster framework inside Nay, compared with a free Rh,(CO),, which has been evaluated by him. (2) We have estimated the EXAFS parameters, coordination number and interatomic distance, R, for Rh-Rh, Rh-C-0 (linear) and Rh,=C-0 (bridged) by the curve- fitting procedures for Rh,(Co),,/NaY using the reference EXAFS data of Rh,(cO),,, Rh4(C0),2, Fe(CO)S and Rh film.I appreciate Prof. Evans suggestion that we have to check the angular dependence of the multi-back-scattering contribution for the Rh-C-0 bondings in our EXAFS evaluation. Dr C. Dossi (Northwestern University, Euanston, Illinois, U.S.A.) had two questions for Prof. Ichikawa: (1) Why did you extract the in situ prepared carbonyl clusters only when supported on the external surface of the zeolite crystal? (2) Could YOU propose a mechanism for explaining the selective formation of Rh,Ir6-x(co)~6 clusters with x depending on the initial Rh"/1r4+ ratio? Prof.Ichikawa answered: ( 1 ) We have synthesized Rh,(CO),, on the external NaY zeolite by using Rh,(CO),, deposited on NaY at 70-120 "C, which gave the face-bridged352 General Discussion CO band at 1805 cm-', in contrast to that of the intrazeolitic Rh,(CO),, from Rh3'/NaY which show the band at 1760 cm-I. In the in situ synthesis of Rh6(C0),, using Rh3'/NaY in CO-H2 or CO-H20 the carbonyl clusters are formed exclusively inside the NaY cages. (2) We propose a mechanism of hexanuclear RhIr bimetallic carbonyl clusters as follows. A mixture of Rh" and 1r4+ impregnated inside a unit of the supercage with a homogeneous distribution is partially reduced and carbonylated to make Rh(CO)? and Ir(CO), species. The mobile carbonyl species inside a unit of cavities come together to build up the tetrahedral carbonyl intermediates Rh,_,Ir,(CO)lz (.x = 0-4), which react further with Rh(CO)? and Ir(CO), converting to the final and stable hexanuclear clusters of R h 6 ~ ~ ~ r y ( c o ) , 6 .All cluster formation occurs inside the cavities, retaining their initial metal compositions. The successive propagation process of cluster formation has been observed by the in-situ FTIR studies on Rh'++Ir4'/NaY zeolite under a CO-H2 atmosphere, and the results essentially support the above-mentioned mechanism of bimetallic cluster formation. Prof. J. B.Nagy (FUNDP, Namur, Belgium) provided the final questions on Prof. Ichikawa's paper: (1) Are you sure of the localisation of the metal particles in supercages? Following the "'Xe-n.m.r.results, the Ir, particles would not be in the supercages because the reference 'empty' NaY gives the same results (see fig. 4 of paper). The particles could better occupy microcavities formed by defect sites. (2) What are the size and shape of the particles? Are they octahedra as you suggest or are they of 'radeau-type' as Boudart suggested previously? (3) Did you use "'Xe n.m.r. to probe the number of H and CO molecules adsorbed on the metal particles? (4) Did you determine the heat of adsorption of benzene, because it is also included in the apparent activation energies of table 6 of your paper? Prof. Ichikawa replied: (1) The Ir,/ NaY sample showed the smallest chemical shift in "'Xe-n.m.r., compared with Nay. This is probably due to the electron saturation of Ir particles having their 'raft' structure inside NaY cavities. From the H/CO chemisorp- tion stoichiometries we consider that Ir particles are located inside Nay, not outside.Additionally, it is of interest to find that Ir6(CO),6 was formed by CO chemisorption at higher temperatures such as 120-200 "C on the Ir,/NaY sample. (2) From the CO/Rh ratio and Rh-K ed e EXAFS (coordination number) we suggest that the Rh particle sizes are less than 8 1 and exist in a pseudo-octahedron shape, being deformed due to metal-support interaction with Rh-0 bonding. (3) We have measured the '"Xe-n.m.r. chemical shifts varying with the H/CO uptake. We are in the process of estimating the number of H atoms and CO molecules adsorbed on the RhIr particles inside Nay. (4) We have not yet measured the heat of benzene adsorption on a series of RhIr bimetallic particles, which might be related to their hydrogenation activities for benzene.Dr B. E. Nieuwenhuys (Leiden University, The Nerherlands) made the first comment on Prof. Datye's paper: It was stated in the introduction that the (111) surfaces are encountered more prevalently in f.c.c. materials as the particle size is increased. Such behaviour may be true for Rh, Ni etc. However, it may be different for Pt and Ir with their reconstructed (100) surfaces. It has been shown by Schmidt et al.' that Pt forms cubes with exclusively (100) faces if grown in a hydrogen atmosphere at 650°C. The particular preference of Pt to exhibit (100) planes on its surface may be related to the reconstruction of the Pt (100) surfaces.It has been suggested that the surface free-energy of the reconstructed Pt (100) surface is lower than that of the Pt (1 11) surface.' For Rh, hexagonal shapes were found as expected for predominantly (1 11) surfaces.' According to the authors the high ethane selectivity of Ir ( 1 10) was not found for Rh (110), presumably because Rh (110) does not exhibit the (1 x 2 ) missing rowGeneral Discussion 3 5 3 reconstruction with C7 sites. My first comment is that unreconstructed f.c.c. (110) surfaces also have a high concentration of C7 sites. Moreover, in addition to Professor King's comments I would like to say that most of the unreconstructed (110) surfaces (like those of Rh and Ni) reconstruct in the presence of hydrogen.Hence, it is likely that both Ir and Rh exhibit a similar reconstructed surface structure under the high hydrogen pressures used for the hydrogenolysis studies. In my opinion, it is more likely that the different selectivities of Rh and Ir are caused by the intrinsic properties of Ir and Rh and not by differences in surface structures. 1 T. P. Chojnacki and L. D. Schmidt, J. Card., 1989, 115, 473. Prof. Datye replied: The equilibrium shapes of small particles in heterogeneous catalysts are influenced by temperature, adsorbates and the oxide support. The adsorbate can cause changes in equilibrium shape provided the surface atoms have sufficient mobility. Hence, the transformation of rounded Pt particles to cubes mentioned in your question occurs at 650 "C, a temperature that leads to considerable sintering of supported Pt/SiO,.Most catalysts are generally reduced at 400 "C and the smaller particles exhibit rounded shapes, with the larger particles showing well defined (1 11) and, to a lesser extent, (100) facets. This is true of both Pt' and Rh2 catalysts. Hence it is reasonable to expect that the (1 11) surfaces will be encountered more prevalently as one goes to the larger particles in f.c.c. metals. In response to your second question, we would expect the behaviour of Rh and Ir surfaces to be similar owing to their comparable reactivities in hydrogenolysis reactions and the similar particle-size effects, i.e. highly dispersed Rh and Ir catalysts show a high selectivity to ethane from n-butane, whereas the larger particles of both metals yield non-selective cracking to methane, ethane and propane.Hence, the key difference between Rh (110) and Ir (110) must be the stable (1 x2) reconstruction on Ir (110) surfaces. 1 M. L. Sattler and P. N. Ross, Ulrramicroscopy, 1986, 20, 21. 2 A. D. Logan, E. J. Braunschweig, A. K. Datye and D. J. Smith, Ultramicroscopy, in press. Prof. D. A. King (University of Cambridge) said: In the paper by Datye et al. it is suggested that a key factor in the selective catalytic chemistry for butane hydrogenolysis is the availability of low coordination atom sites. In particular, the selectivity of Ir { 1 lo} is attributed to the corrugated missing row (1 x 2 ) surface. However, although this is the stable structure of the clean (110) surface, it is well known' that electronegative adsorbates tend to lift the (1 x 2 ) reconstruction of the clean surface, returning it to a (1 x 1 ) structure.Under reaction conditions it therefore seems very likely that at least that part of the surface fractionally covered by adsorbate, and hence responsible for the reaction, is Ir { 110) (1 x 1). 1 For example: S. R. Bare, P. Hofmann and D. A. King, Surf: Sci., 1984, 144, 347. Prof. Datye replied: Previous work' has shown that on Ir (110) surfaces, the (1 x2) reconstruction is not lifted by the presence of hydrocarbon residues. In addition, adsorption of H2 on Ir (110)-(1 x 2 ) did not lift the (1 x 2 ) surface reconstruction.* 1 W. H. Weinberg, Suru. frog. Chem., 1983, 10, 1. 2 D. E. Ibbottson, T. S. Wittrig, and W. H. Weinberg, J.Chem. Phys., 1980, 72, 4885. Prof. C. Kemball ( University of Edinburgh) then commented: Rhodium is normally an active metal for hydrogenolysis but many of the results in the present paper are for temperatures above 500 K because of the small areas of the catalysts. Have the authors tested a catalyst at lower temperature after it has been used above 500 K to ascertain whether or not irreversible changes occur at the higher temperatures?3 54 General Discussion If irreversible changes do take place, they may result from the formation of strongly adsorbed hydrocarbon residues at the higher temperatures. On the other hand, reversible behaviour would indicate a change in the rate-determining step possibly from C-C bond rupture to product desorption as the temperature is raised.We’ have evidence that the rate-controlling step for the hydrogenolysis of 2,2-dimenthylpropane over supported rhodium catalysts changes from C-C bond rupture below 450 K to desorption of methane at higher temperatures. 1 J. A. Oliver and C. Kemball, to be published. Prof. Datye responded: The turnover frequencies for n-butane hydrogenolysis on the Rh single crystals are in very good agreement with those measured on supported Rh particles [ref. (17) in the paper]. Reaction kinetics are usually measured at lower temperatures on supported catalysts because of the intrusion of diffusion limitations at elevated temperatures. Since this limitation is not encountered on single crystals, it is possible to measure kinetics over a much wider range of temperatures as pointed out recently.’ Post-reaction analysis of surface carbonaceous residues by Auger electron spectro- scopy shows that the fractional coverage is independent of reaction conditions,2 i.e.pressure and temperature. In recent work on Rh(100)3 we have observed that the selectivity changes seen at higher temperatures are reversed when the reaction is carried out at temperatures below the point where ‘rollover’ occurs. This would suggest that the changes in the catalyst surface that are responsible for the ‘rollover’ are reversible and not caused by the accumulation of carbonaceous residues. Further evidence is provided by the detailed kinetic modelling of the reaction steps as performed for the hydrogenolysis of n-butane on Ir.’ The onset of ‘rollover’ is accompanied by a change in the rate-determining step from C-C bond rupture to that of product desorption.The rate of desorption of the alkanes is affected by the presence of hydrogen adatoms on the surface and consequently, ‘rollover’ is associated with a depletion in surface hydrogen at elevated temperatures. The temperature at which ‘rollover’ occurs is related to the hydrogen binding-energy on the surface, with ‘rollover’ occurring at a lower temperature on the (1 11) surface where H2 is more weakly bound. Hence, our results are in agreement with the data on hydrogenolysis of 2,2-dimethylpropane cited by you. 1 D. W. Goodman, Ace. Chem. Rex, 1984, 17, 194. 2 J. R. Engstrom, D. W. Goodman and W. H. Weinberg, J. Am. Chem. SOC., 1988, 110, 8305 3 A.D. Logan, A. K. Datye and D. W. Goodman, to be published. Prof. G. C . Bond (Brunel University, Uxbridge) said: The terminology applied to surface structures is purely arbitrary and does not necessarily reflect the perception a colliding molecule has of what it is about to meet. In fact both the reconstructed and the unreconstructed (110) surfaces of a f.c.c. metal can both be represented as stepped (1 11) surfaces, and in each case the top parallel rows consist of atoms having sevenfold coordination, the latter having twice as many such rows as the former. It is therefore difficult to see how high ethane selectivity can be specifically correlated with the existence of a reconstructed surface. Dr Datye and his co-authors represent the hydrogenolysis of n-butane to methane, ethane and propane as if three parallel and unrelated processes were responsible for their formation.This is very unlikely to be the case, and, as has been recently emphasised’ the reaction is better regarded as proceeding by a rake mechanism, which permits a much closer insight into the effect of variables on product yields. This point can be illustrated by reference to the effect of hydrogen pressure variation (see fig. 3 and 5 of Dr Datye’s paper). The observation that methane selectivity increases as hydrogen pressure is decreased has been made by others, including ourselves, but is more clearlyGeneral Discussion 355 0.2 1 i 0.1 L/ 1 s2 0 0 0.2 0.4 0.6 0.8 1.0 Fig. 2. Hydrogenolysis of propane over Ni/MgO at 563 K: effect of hydrogen pressure on product selectivities of propane (pressure, 0.042 atrn),.appreciated from straightforward selectivity plots, such as those shown in fig. 2. of this discussion than from the rate plots in the figures I have referred to. The fact that methane is the preferred product at low hydrogen pressure implies that fewer hydrogen atoms are required to convert the adsorbed C , intermediate into C, species than to effect its desorption as ethane. More quantitative information can be derived from a fuller analysis of the reaction scheme. 1 G. C. Bond, J. Catal., 1989, 115, 286. 2 S. P. Sarsarn, Ph.D. Thesis (Brunel University, 1984) Prof. Datye, in reply, said: It has been postulated in previous work' that the probable intermediate in the selective hydrogenolysis of n-butane to ethane is a 1,4-diadsorbed butane, i.e. a metallacycle pentane. Decomposition of the metallacycle pentane ligands gives rise to two ethylene ligands [ref. (8) in our paper] and may account for the high C, selectivity on Ir (110)-(1 x 2) surfaces. Utilizing the bond lengths and bond angles for Ir complexes containing metallacycle pentane ligands, one finds that significant repulsion is expected between the a-hydrogens and the adjacent Ir atoms on a (1 11) surface. However, no such repulsion is expected if one coordinates the ligand about the low coordination number C , atoms on the (1 lo)-( 1 x 2 ) surface. This is because even though both the reconstructed and unreconstructed (1 10) surfaces consist of stepped (1 11) surfaces having exposed C, atoms, the steps in the (1 lo)-( 1 x 2 ) surface are twice as wide. Hence, the steric hindrance around the C7 sites is less pronounced in case of the reconstructed Ir (1 lo)-( 1 x 2) surface. At low reaction temperatures, the hydrogenolysis product distributions suggest the existence of two separate pathways; i.e. central-bond scission C, + 2Cz and terminal- bond scission C4 -+ C3 + C , . This is evident in the nearly equal reaction rates for C, and C3 from n-butane on Ir (1 lo)-( 1 x 2) or of C, and C, over highly dispersed Rh in the hydrogenolysis of n-pentane [ref. (13) in our paper]. Under these conditions, C-C bond rupture appears to be the rate-limiting step. At elevated temperatures, or at low hydrogen pressures, alkane desorption becomes rate-limiting and hence the reactive356 General Discussion intermediate can undergo additional C-C bond breaking, giving rise to the rake mechanism cited above. Under these conditions it is of course incorrect to visualise the process as occurring solely through parallel reaction pathways, and the Kempling- Anderson’ analysis allows proper estimation of the kinetic parameters. 1 J. R. Engstrom, D. W. Goodman and W. H. Weinberg, J. Am. Chem. SOC., 1988, 110, 8305. 2 J. C. Kempling and R. B. Anderson, Ind. Eng. Chem. Prod. Res. Den, 1970, 9, 116.

ISSN:0301-7249    

DOI:10.1039/DC9898700345

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

L I S T O F P O S T E R S Adsorption on high-area porous catalysts under uhv conditions M. Bowker, University of Liverpool Synergy in methanol synthesis over Cu catalysts R. Burch, R. J. Chappell and S. E. Golunski, University of Reading FTIR detection of selective oxidation of benzene and C4 linear hydrocarbons on vanadia-titania 'monolayer' catalysts Hydrogen-oxygen reaction over Re/ y-Al,O, catalysts in a vacuum microbalance G. W. Chadzynski, Polish Academy of Sciences, Warsaw, Poland The catalytic activity of chromia/zirconia in the hydrogenation of propene A. Cimino, S. De R o d , S. Febbraro, G. Ferraris, D. Gazzoli, V. Indovina and M. Valigi, Universita La Sapienza, Rome, Italy Characterisation of electrocatalysis in electrolytic 0, and H, evolution through determination of the adsorption behaviour of the kinetically involved intermediates B.E. Conway, T. C. Liu and G. Jerkiewicz, University of Ottawa, Canada Surface characterisation of promoted iron catalysts during Fischer-Tropsch synthesis P. M. Loggenberg, R. G. Copperthwaite and J. P. F. Sellschop, University of the Witwatersrand, Johannes- burg, South Africa Oxidative dimerizations of methane: correlations with surface properties of doped alkaline-earth- metal oxides D. McNamara," J. Cunningham," W. Hirschwald,b B. K. Hodnett' and M. Jauch,' "University College, Cork, Ireland, 'Freie Universitat, Berlin, FRG, 'National Institute of Higher Education, Limerick, Ireland Observation of the surface structure of small metal particles using model heterogeneous catalysts A.D. Logan, E. J. Braunschweig and A. K. Datye, University of New Mexico, Albuquerque, USA The use of EXAFS in identifying the Ni sites of a nickel-exchanged zeolite Y for the trimerization of acetylene P. J. Maddox, E. Dooryhee, A. T. Steele, J. M. Thomas, C. R. A. Catlow, N. G. Greaves and R. P. Townsend, 73e Royal Institution, London Platinum-rhenium catalysts supported on NaY and NaHY zeolites: high methane selectivity in n-heptane hydrogenolysis C. Dossi,'*' R. Psaro,b R. Ugob and W. M. H. Sachtler," "Northwestern University, Evanston, USA, University of Milan, Italy Molybdenum oxide modified p e n t a d zeolites I. M. Harris,' J. Dyer," A. A. Garforth," C. H. McAteerb and W. J. Ball,' "UMIST, hBP, Sunburv-on-names Preparation and characterisation of molybdenum metathesis catalysts J.Evans and J. F. W. Mosselmans, University of Southampton Genesis and characterisation by laser Raman spectroscopy and high-resolution electron micro- scopy of alumina-supported MoS, crystallites Edmond Payen," Slavik Kasztelan,".' Sabine Houssenbay,' Raymond Szymanski' and Jean Grimblot," Uniuersiti des Sciences et Tech- niques de Lille Flandres-Artois, Villeneuve d' Ascq, France, hInstitute Franfais du PrCtrt?le, Mal- maison, France Cobalt manganese CO hydrogenation catalysts: solid state and surface characterisation R. G. Copperthwaite," G. J. Hutchings,b M. van der Riet," S. Colley" and M. Betts," 'University of the Witwatersrand, Johannesburg, South Africa, University of Liverpool The effects of ageing and moisture on the decomposition-retention of hydrogen cyanide by a copper chromate catalyst impregnated on activated charcoal cloth P.3. Brown,a G. G. Jayson," G. Thompson" and M. C. Wilkinson,' ' Liuerpool Polytechnic, hChemical Defence Establishment, Porton Down G. Busca, G . Ramis and V. Lorenzelli, Universita di Genova, Ita1.v 3573 5 8 List of Posters The interaction of carbon monoxide and synthesis gas with small rhodium particles P. Johnston, R. W. Joyner and P. D. A. Pudney, University of Liverpool Ir4(CO),2/Si02: A selective catalyst for the isomerisation of 1,5-cyclo-octadiene to the 1,4- isomer Jan Kaspar,' A. Trovarelli,b G. Dolcetti' and M. Graziani," a Universitu di Trieste, Italy, IR spectroscopy of surface electromagnetic waves in studies of adsorption and catalysis P.A. Shafranovskii, M. Yu Sinev, M. A. Moskaleva, G. N. Zhizhin, B. R. Shub and 0. V. Krylov, USSR Academy of Sciences, Moscow, USSR and Troitsk, USSR Supported metal cluster compounds as precursors of Fischer-Tropsch catalysts R. S. Armstrong," T. Bell," A. L. Chafee,' V. W. L. Chin," H. J. Loeh,' A. B. J. Lucchese," A. F. Masters" and M. A. Williams,".' University of Sydney, Australia, hCSIRO Division of Fuel Technology, Menai, Australia Nickel-catalysed olefin oligomerisation S. J. Brown," D. J. Byrum," L. M. Clutterbuck," C. M. Lindall," A. F. Masters," J. I. Sachinides,b P. A. Tregloanb and M. Vender,' "University ofsydney, Australia, University of Melbourne, Australia Promotion on supported platinum The reactions of C 0 2 on Cu surfaces: catalysis of the methanol synthesis and water-gas shift reactions R.G. Copperthwaite," M. A. Morris," M. W. Roberts," R. A. Ryder," G. C. Chinchen,' M. S. Spencerb and D. A. Whan,b "University College, Cardifl blCI Chemicals and Polymers Ltd, Billingham Non-intrusive temperature measurement of the components of a working catalyst by neutron resonance radiography J. C. Frost," P. Meehan," S. R. Morris," R. C. Ward" and J. Mayers,b "BP Research International, Sunbury-on-names, bSERC, Chilton Studies by EXAFS, XP and UV visible spectroscopies of the state of nickel in oxidised Ni)Ji02 and .Ni/Si02 A. R. Gonzalez-Elipe, J. P. Espinos and G. Munuera, Instituto de Ciencia de Materiales de Sevilla, Sevilla, Spain Calcination and pretreatment of supported commercial vanadia-titania catalyst for partial oxida- tion of o-xylene to phthalic anhydride V.Nikolov, A. Anastasov and C. Zareva, Bulgarian Academy of Sciences, Soja, Bulgaria Pd(100)-B as model catalyst for partial hydrogenation of alkynes M. Krawczyk and W. Palczewska, Polish Academy of Sciences, Warsaw, Poland Catalytic oxidation of carbon monoxide and structure sensitivity Sivanandi Rajadurai and J. J. Carberry, University of Notre Dame, USA Some new insight into the reactivity of molecules in catalysis through coadsorption at single crystal surfaces A. F. Carley, P. R. Davies, M. W. Roberts and Song Yan, University of Wales College of Cardi' Activity and selectivity of copper in a positive oxidation state intercalated within well defined sites in Cu,W03 in CO hydrogenation Studies of catalytic decomposition of N 2 0 on solid solution series M,CO~_,~O, R. Sundarajan and V. Srinivasan, Indian Institute of Technology, Madras, India Effect of valence state on catalytic activity of copper-containing ternary oxides J. Christopher and C. S. Swamy, Indian Institute of Technology, Madras, India Structure and reactivity of some supported platinum catalysts G. D. McLellan," G. Webb," R. B. Moyes,b s. Simpson,b P. B. Wells,b S. D. Jackson' and R. Whyman,' "University of Glasgow, 'University of Hull, 'ICI Chemicals and Polymers Lid, Billingham Universita di Udine, Italy G. McDougall, University of Edinburgh M. S. W. Vong and P. A. Sermon, Brunei University

ISSN:0301-7249    

DOI:10.1039/DC9898700357

出版商:RSC    

年代:1989

数据来源: RSC