Studies in Surface Science and Catalysis 11 METAL-SUPPORT AND M ETAL-ADDITIVE EFFECTS IN CATALYSIS Proceedings of an International Symposium - - organized by the lnstitut de Recherches sur la Catalyse CNRS Villeurbanne and sponsored by the Centre National de la Recherche Scientifique, Ecully (Lyon), September 14-16, 1982 Editors B. Imelik, C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine lnstitut de Recherches sur la Catalyse, CNRS, 69626 Villeurbanne, France ELSEVIER SCIENTIFIC PUBLISHING COMPANY Amsterdam - Oxford - New York 1982 ELSEVIER SCIENTIFIC PUBLISHING COMPANY Molenwerf 1. P.O. Box 21 1, 1000 AE Amsterdam, The Netherlands Distributors for the United States and Canada: ELSEVIER SCIENCE PUBLISHING COMPANY INC. 52, Vanderbilt Avenue New York, NY 10017 ISBN 0-444-421 11-4 (Vol. 11) ISBN 0-444-41801-6 (Series) 0 Elsevier Scientific Publishing Company, 1982 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or other- wise, without the prior written permission of the publisher, Elsevier Scientific Publishing Company, P.O. Box 330, 1000 AH Amsterdam, The Netherlands Printed in The Netherlands IX F 0 R E W 0 R D It has long been recognized that the efficiency of metal catalysts was considerably improved by increasing the dispersion of the metal. Catalytic processes by metals have become of such increasing importance in the chemical and petroleum industries that a considerable amount of effort has been spent on the development of synthesis methods which allow highly dispersed metals to be obtained. Several methods to achieve the highest metal dispersion have been pro- posed, the most common method being tQ disperse the active metal on high surface I area support. Silica, alumina, carbon and zeolite have often been used for this purpose. The earlier views considered that the effect of the support was to sta- bilize the metal in the form of small metal particles, thus producing a large increase of the surface metal atoms/bulk metal atoms ratio, without changing the specificity of the catalyst. This view was rapidly found incapable of explaining the experimental data. It appeared that the rate and selectivity of several chemical reactions were controlled by the size of the metal particles. The interpretation which was given was that the properties of the surface metal atoms may be changed, when the metal particles decreased, because the number of edge and/or corner atoms, which obviously have a different coordination number from those in the bulk, are increasing. Several reviews have quoted evidence in favour of these structural effects on catalysis. However more recently several conflicting results and interpretations appeared in the literature and it was suggested that the support may play an important role, other than to stabilize the small metal particles, on the properties of themetal. The support effects were discussed in terms of a structural effect : the support influencing the preferential growing of a given crystallographic plan by an epitaxy phenomenon, or in terms of an electronic effect : the support inducing charge transfer. More recently, renewed interest on the effect of the support on metal catalysts appeared when it was discovered that in special experimental conditions supports like zeolites, polymers, reducible oxides may modify drastically the catalytic properties of the metals. Several questions were raised : i) why the properties of these supported metals are different from those of bulk metals, ii) how the surface of the support is involved in the interaction. Along with this scientific trend, another important aspect in catalysis by metals was the fact that addition to the metal of an inactive additive may produce a very large effect on both activity, selectivity and resistance to poisoning of the pure metal. In an effort to understand these metal-support and metal additive effects several advanced physical methods such as XPS, electron microscopy, AUGER spectroscopy, EXAFS, ESR along with chemical approach haqe been outlined resulting in a large number of contributions which still appeared conflicting in their interpretations. A rapid analysis of the published data tends to indicate that very probably metal- support interaction and metal-additive effect may not be explained by an unified theory. Thus the rapid growth of studies involving metal-support interactions which paralled the amount of work devoted to the effects of additives on metal prop- erties has prompted us to organize this symposium, sponsored by the "Centre National de la Recherche Scientifique" to clarify the new ideas on these sub- jects. The challenge is that the large amount of experimental data which will be presented at this meeting, the viewpoints, concept models which will be discussed would improve our understanding in this domain and so far will be instructive. It is the hope of the editors that the contributions assembled in this volume will stimulate additional works. The organizing committee wish to express their deepest gratitude to the Centre National de la Recherche Scientifique which has funded this Colloquium in the frame of "Colloques Internationaux du CNRS" which are held every year. We are grateful to Professor R. Maurel, Directeur Scientifique, who has encouraged the "Institut de Recherches sur la Catalyse, CNRS, Villeurbanne, France", to organize this colloquium. We want to thank all the authors and participants for their interest. We are also particularly indebted to Miss Lydie Badolo for her help in the preparation of this Proceedings. Thanks of the organizing committee are due to all who have contributed to the organization of this meeting and to the publication of this volume. Boris IMELIK, Claude NACCACHE, Gisele COUDURIER, H6lhe PRALIAUD, Paul MERIAUDEAU, Pierre GALLEZOT, Guy Antonin MARTIN, Jacques VEDRINE. XI P R E F A C E Au cours de ces dix dernieres annees on a assist6 a un accroissement consid6rable des recherches sur la catalyse par les metaux. Une part importante de ces etudes concerne les metaux supportes. L'intPrCt des scientifiques pour ces materiaux provient non seulement de l'importance industrielle que presentent les catalyseurs "metaux supportes" mais Bgalement du nombre de problemes fondamentaux que ces systemes posent. Parmi ceux-ci, celui qui apparait le moins bien apprehende, concerne le degre d'interaction pouvant'exister entre le metal et le support sous-jacent et les modifications des propriet6.s physichochimiques et catalytiques du metal, qu'une telle interaction peut entrainer. Au cours de ces dernieres anndes on a assist& d une explosion des travaux dans ce domaine. C'est ainsi qu'il est apparu que dans de nombreuses circons- tances, le support pouvait influencer la phase metallique de maniere d modifier la dispersion du metal, sa stabilite au frittage, la structure et la morphologie de la particule metallique, les proprietes d'adsorption du metal ainsi que ses proprietes catalytiques. Actuellement la litterature scientifique s'enrichit d'exemples nouveaux montrant l'existence d'interactions metal-support : metaux supportes sur zeolithes, metaux supportes sur oxydes basiques, metaux supportes sur oxydes reductibles. Jusqu'd present les travaux effectues dans ce domaine n'ont fait l'objet que de contributions dispersees dans des colloques et congres scientifiques. Les recherches sur les interactions metal-support et metal-ajout sont actuellement suffisamment developpees pour que le CNRS organise dans le cadre de ses colloques internationaux une reunion scientifique pour faire le point des faits experimentaux acquis ce jour dans ce domaine, confronter les idees, les hypotheses et theories developpees sur le r61e du support et sur les modifications qu'il peut produire sur les proprietes physico-chimiques et catalytiques du metal. L'organisation du Colloque a 6t6 confiee d 1'Institut de Recherches sur la Catalyse, Villeurbanne, France. XI1 Le comite d'organisation est tr6s reconnaissant au Centre National de la Recherche Scientifique qui a finance ce colloque et a son Directeur Scientifique, Monsieur le Professeur R. MAUREL pour les encouragements qu'il lUi a prodigues. 11 nous est particulierement agreable de remercier tous les auteurs des communications et tous les participants. Enfin que Madame Lydie Badolo trouve ici les remerciements des Bditeurs pour l'aide qu'elie leur a apportee 2 la realisation technique de cet ouvrage. Boris IMELIK, Claude NACCACHE, Gisele COUDURIER, Helene PRALIAUD, Paul MERIAUDEAU, Pierre GALLEZOT, Guy-Antonin,MARTIN, Jacques VEDRINE. B. Jmelik et al. (Editors), Metal-Support and Metal-Additive Effects in Catalysis 1 0 1982 Elsevier Scientific Publishing Company, Amsterdam -Printed in The Netherlands THE MODIFICATION OF CATALYTIC PROPERTIES BY METAL-SUPPORT INTERACTIONS G.C. BOND School of Chemistry, Brunel Universitv, Uxbridge LIB8 3PH, UK RESUEE Les diffgrences entre les pronrietss catalytiques de m6taux sur supnorts > divers peuvent etre dues 'a nlusieurs facteurs l'exceotion de l'interaction entre le m6tal et le sunnort; le ranport intime entre ceux-ci et les effets de taille de particule est soulign6. Les interactions sont classgescomme ou faibles (par exemple avec Si02 et A1 0 ) ou moyennes (avec zgolithes) ou fortes 2 3 (par exemple avec Ti02). Les consgauences g6ometriques et Slectroniques des interactions sont revues; les mgtaux dans les dolithes sont dgficients en 6lectrons et sur le Ti02 rCduit ils sont riches en electrons. Les effets sur le comportment chimisorptif et catalytique ne sont, dans ce dernier cas, pas comul>tement exoliquds nar les concents thsorioues actuels. ABSTRACT Differences between catalytic pronerties of metals on various sunnorts can be caused by a number of factors anart from metal-sunnort interaction; the close connection between this and particle size effects is emphasised. Interactions are classified as either weak (e.g. with Si02 and A1203), medium (with zeolites) or strong (e.g. with Ti02). Geometric and electronic consequences of the interactions are reviewed; metals in zeolites are electron-deficient and on reduced Ti02 are electron-rich. The effects on chemisorntive and catalytic behaviour are in this last case not fully explicable by current theoretical concepts. INTRODUCTION Highly-dispersed metal Particles on a suitable sunnort find major annlications in a range of technicallv significant processes. Among the more obvious nractical advantages of this kind of catalyst structure are ease of handling, suitability for use in fixed or fluidised bed flow reactors, and high thermal stability. The traditional view of the suvuort was that it was an inert material which merely carried the metal oarticles, well senarated from each f other, in a physically convenient form. The first radical departure from this concept was the discovery of bifunctionality in about 1950: the use of acidic suooorts such as amorohous Si02-A1203, later disolaced by halided A1 0 2 3’ permitted the cooperative exercise of carbonium-ion-catalysed and metal- catalysed transformations which constitute the basis for petroleum reforming. In the following quarter of a century, the feeling arose that the suanort should be expected to exert some influence on the metal. This feeling was given substance by a number of scattered and fragmentary observations (refs. 1, 2), some of which might have been exolicable by Darticle size effects, but others of which concerned specific or areal rate differences from one support to another (ref. 2). The first deliberate attempts to explore metal-supnort interactions, and to manipulate them, were made by Schwab (ref. 3) and by Solymosi (ref. 4) who showed that changing the semiconducting orouerties of the suuuort by altervalent ion doping affected the activation energies of a number of reactions. The value of this early work is limited by the unavailability at the time it was oerformed of analytical methods now routinely applied, but it deserves careful re-examination. It is essential from time to time to insoect the older literature, for otherwise we run the risk of rediscovering the wheel. Subsequent developments, including the recognition of a strong metal-suaport interaction (SMSI), suggest that we are still in the infancy of our ability to control, modify and direct the catalytic oronerties through the chemical environment of the SuoDort. CLASS IF1C AT ION OF PETAL- SUPP ORT INTERACT IOV S The Duroose of this aaoer is to construct a framework within which the mutual interactions between metal and sumart can be logically and systematically explored, and their effects classified. These interactions are many in number and varied in nature. Some have a trivial origin, but may be nonetheless significant: others have a deep significance, but to identify and understand these the trivial have first to be recognised and eliminated. I take it as axiomatic that it is the modification of the intrinsic catalytic properties of the metal by the other variables of the system which is of interest. This being so, we need some standard of reference to define the unmodified state of the metal, and to establish whether in a given instance there is any interaction or not. Highly disoersed metals can be ureoared in the absence of a support, as colloidal dispersions, aerosols or matrix-isolated clusters (ref. 5), but their catalytic applications are limited (ref. 6). AS soon as we seek to form small metal oarticles on a sumort by one or other of the conventional methods, we immediately encounter a fundamental difficulty. Let us imagine we set out to comuare the Properties of metal A on suoports X and 3 Y, and that we use the same salt of A, the same method of application and that we aim for the same metal content, either on a weight or a surface concentration basis. This hypothetical experiment typifies much recently published work. Now if there are significant differences between X and Y in terms of surface area, porosity, pore structure and surface chemistry, it is unlikely that either the average size or the size distribution of the metal particles or their location within the grains of the suwort will be the same. We have therefore first to establish whether any observed differences are attributable to a specific particle size effect or to the exercise of nore diffusion limitation. Anxieties of this kind are the greater when comnaring zeolites with conventional supports. If a difference in the metal size distribution is indeed confirmed between the A/X and A/Y systems, there is the additional uossibility that the system showing the lower average size will demonstrate the greater metal-support interaction. A further complication is that the average particle size may either increase (due to sintering) or decrease (e.g. due to B-PdH formation and decomuosition (ref. 7)) during use. It is no easy task to disentangle particle size effects and metal-support interactions. It might be thought uossible to overcome this uroblem by applying pre-formed metal particles (e.g. as colloids or by metal vapour denosition) so that the same size of metal particle could be obtained on a number of different supports This indeed is technically feasible (ref. 8), and such routes to sumorted metals deserve deeper study; but the mode of attachment of ure-formed uarticles to the support would probably be different in kind and weaker in strength than that resulting from the use of normal chemical methods. The techniques of impregnation, adsorption preciuitation, co-precipitation and ion exchange, and the methods pioneered by Yermakov (ref. 9), are preferable if the closest chemical contact between metal and suuport is desired. The idea of using pre-formed particles removes certain difficulties, but introduces others, and is not a perfect solution to the problem. A further problem, usually ignored, is whether different catalysts have equally clean surfaces. The potentially harmful effects of C1- are well-known, but catalysts having different metal contents and different particle sizes can appear to show a suecific particle size effect which is really caused by different C1- contents (ref. 10). Impurities from the sunport can also uoison the metal, and the different extents to which this can hanuen with various supports can also lead to spurious results which can be attributed to metal- support interactions (ref. 11). In addition to supplying the acid function in bifunctional catalysts, the support provides a suitable vehicle for incorporating a wide range of promoters.* The discovery of the beneficial effects of adding Re, Sn and Ge (as well as many other metals) to Pt/A1203 reforming catalysts has occasioned much research, 4 which is beyond the scope of this paper to attemot to summarise. The deliberate addition of reducible promoters to improve the performance of the whole catalyst does not constitute a metal-support interaction as normally understood. However the incorporation of dopants in the suuport and their consequential effects on the metal's behaviour (refs. 3, 4) are valid considerations; for example, the selectivity of the Ag-catalysed oxidation of ethylene has recently been imoroved by adding MgO and Ce02 to the A1 0 suonort (ref. 12). The inclusion of Grouu 2 3 IIA oxides facilitates the reactivation of metals under oxidising conditions (ref. 13). It would be a mistake to think that a supported metal always consists of only one kind of particle. There is now considerable evidence to suggest that very highly (oerhaos atomically) dispersed metal can sometimes co-exist with larger particles, and it is easy to fall into the trap of seeing only what is easily visible (e.g. by TEM). Supported metal catalysts can be structurally comp 1e x. Although we are primarily concerned with the effect of the supoort on the metal, it is equally true that the suooort is modified by the metal. Indeed of necessity the modifying effects must be mutual. It has been known for some years that small metal particles in zeolites both affect the crystal's acid strength and are affected by it (ref. 14). The chemical constitution of the support will influence the electronic structure of metal oarticles provided the contact is sufficiently close, so that the means of delicately tuning the metal's properties are to hand. A further important kind of metal-support interaction is that which involves the migration or soillover of active species from metal to suooort (ref. 15) and vice versa (ref. 16). The elegant work of Teichner has demonstrated the validity of the effect under reducing conditions, and catalytically useful spillover has also been strongly indicated in CO oxidation over Pd/Sn02 (ref. 17). Reaction may also occur at the oeriphery of metal oarticles by cooperation between adjacent metal and support sites; this idea, first mentioned in the context of petroleum reforming more than 25 years ago (ref. 18), wrongly as it turned out, has recently been revived (refs. 19, 20). We therefore see that apparent particle size effects can be due to partial poisoning, and apparent metal-support interactions to particle size effects or spillover catalysis. The unambiguous identification of a metal's modification by the support is not always simole.