The Chemisorptive Bond BASIC CONCEPTS ALFRED CLARK School of Chemical Engineering and Materials Science The University of Oklahoma Norman, Oklahoma ACADEMIC PRESS New York and London 1974 A Subsidiary of Harcourt Brace Jovanovich, Publishers COPYRIGHT © 1974, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER. ACADEMIC PRESS, INC. Ill Fifth Avenue, New York, New York 10003 United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road, London NW1 Library of Congress Cataloging in Publication Data Clark, Alfred, Date The chemisorptive bond: basic concepts. (Physical chemistry, a series of monographs, no. 32 ) Includes bibliographical references. 1. Chemisorption. 2. Chemical bonds. I. Title. II. Series. QD547.C55 547'.1'224 73-9421 ISBN 0-12-175440-5 PRINTED IN THE UNITED STATES OF AMERICA Preface This book describes the basic concepts of the chemisorptive bond on solid surfaces from the simple analogies with ordinary chemical bonds to the recent quantum-mechanical approaches that accept the difficult challenge of the solid state and the presence of a surface. The latter are usually omitted because of their mathematical complexity. Here, the blow of mathematics has been softened drastically and the physical pictures stressed. The earlier chapters are obviously relevant to chemisorption, and so to catalysis, because they present simple formulas for correlating measurable quantities. Unfortunately, these simple correlations, faithful or not, often fail to give true pictures of the mechanisms of chemisorption. Although the later chapters provide more detailed, quantum-mechanical pictures, many of them cannot be checked experimentally, because techniques have not been developed yet for measuring the variables involved. Yet, these chap- ters should not be considered irrelevant to chemisorption phenomena. Many predictions, long resistant to experimental confirmation, have ulti- mately found a place in the bedrock of science: Dalton's atomic hypothesis, for example, and, much later, nuclear fission. We believe the concepts re- viewed here are important and relevant. Some may wither under the scrutiny of future experiments; others may bloom. We hope none will be sentenced to death under a myopic edict of immediate relevancy. The author wishes to thank Mrs. Sue Ryan for skillfully and patiently typing the manuscript; the University of Oklahoma, Norman, for providing the facilities used in preparing the manuscript and figures; and Academic Press for their excellent editorial and stylistic suggestions. ix I Introduction 1.1 The Forces of Adsorption The cohesive forces in solids and the forces of adsorption are not different in principle from the forces exerted between free atoms and molecules. All interparticle forces have a common origin in the electromagnetic inter- actions of nuclei and electrons. Some of these forces are weak, like the van der Waals forces, which are considered to be made up of London dispersion forces and classical electrostatic forces. Neither transfer nor sharing of electrons occurs. Electrons may seek a new equilibrium distri- bution, but they always stay with the particle on which they entered the interaction. When an atom or molecule is bound to the surface of a solid by van der Waals forces, the phenomenon is called physical adsorption. Other forces, usually stronger, form chemical bonds and involve the transfer or sharing of electrons. When an atom or molecule is bound to a surface through overlapping of one or more of its electron orbitals, the phenomenon is called chemisorption. In physical adsorption, it is con- ventional to assume that the surface of the adsorbent is unperturbed, that its only function is to supply a potential field for the adsorbate, and that the adsorbate may be regarded as a separate thermodynamic phase. In chemisorption, these are not fruitful assumptions. The surface and the I 2 1. INTRODUCTION adsorbate should be viewed together as a new chemical entity. How many surface atoms of the adsorbent participate in the bonding of a single adsorbed particle is a question that has not been answered unequivocally for any real system and probably will not be in the near future. But theoretical and experimental attacks are making important advances. Chemisorptive bonds and conventional bonds may arise from the same electromagnetic forces, but in the surface bond these forces are far more intricately compounded. Atoms or ions in the surface of a solid must differ in some degree from their simpler, free counterparts because of the cooperative forces which bind them together, and often these forces create local perturbations so drastic that they have no free counterparts at all, such as interstitial atoms or ions and lattice vacancies. The chemisorption of a foreign atom on the perturbed centers of a solid surface, singly or collectively, presents grave, new problems, not fully encountered in the conventional bond, which demand separate treatment. 1.2 Approaches to the Chemisorptive Bond Concepts of the chemisorptive bond have progressed in the usual manner from the simple to the complex. Early theories essentially ignore the presence of a surface and frequently presume that the adsorbing atom seeks out and reacts with a single surface atom. Both covalent and ionic bond formation have been considered in this analogy with the diatomic molecule. Various approximations of the strength of the covalent bond between free atoms have been adapted to adsorption. For example, Eley [1] adapted Pauling's [2] approximation to the adsorption of hydrogen atoms on metals, where the metal-hydrogen bond strength is taken as the arithmetic or geometric average of the metal-metal and the hydrogen molecule bond strengths. The metal-metal bond strength is estimated from the latent heat of sublimation of the metal, taking into account the number of nearest neighbors. Corrections have been applied for polarization effects caused by the difference in electronegativities of metal and hydrogen, giving a small ionic character to the bond [3]. Formal quantum-mechanical calculations have been made for the covalent adsorption bond on the surfaces of metals [4] and heteropolar lattices [5], and even on lattice defects [6]. Many attempts have been made to correlate empty d orbitals or d-band vacancies in transition metals with the power to chemisorb, including the application of Pauling's d character of metals, which he derived from his valence-bond theory. A great number of strictly quali- tative theories representing pictorially the complexing of simple gases, 1.2 APPROACHES TO THE CHEMISORPTIVE BOND 3 olefins, and other molecules on metal and metal oxide surfaces have also been put forth, using a molecular orbital or simpler design [7]. Ionic bonds on metal surfaces have been considered in analogy with the conventional ionic bond [8]. For example, the energy change associated with the ionic mechanism for adsorption of atom A on metal M, M + A-»M-A+ may be estimated by breaking down the energy as follows: (i) remove an electron from the highest occupied level of an isolated atom of A to infinity (—le) ; (ii) transfer the electron to the Fermi level of the metal M (<f>e) ; (iii) bring A+ to its equilibrium distance z* from the surface of the metal (e2/4z*). The heat of adsorption per gram atom at zero coverage is then considered to be given by q = (-le + <t>e + e2/4z*)/N d where I and <t> are the ionization potential of A and the work function of M, respectively, N is Avogadro's number, and the last term to the right in the numerator is the classical electrostatic image energy of attraction of the ion A+ to the surface of the metal. Occasionally, the mirror-image term is replaced by a strictly interatomic potential function. Concessions to the presence of a solid surface appear in some of the approaches which we have just described; for example, in the estimation of the metal-metal bond strength from the latent heat of sublimation in Eley's adaptation to chemisorption of Pauling's method for determining covalent bond strength, in the correlation of Pauling's d character of transition metals with chemisorption, and in the use of the mirror-image force to express the strength of the ionic bond of chemisorption on metals. But in all of these approaches, the chemisorptive bond may be regarded as localized—a quasi-diatomic molecule composed of an adsorbed atom, molecule, or ion and a surface entity of the solid lattice. The analogies with conventional bonds, covalent or ionic, persist. Modern developments in solid-state theory have sparked fresh viewpoints on chemisorption. Applications began about 1950 and have gained mo- mentum ever since. With their growth, analogies between the chemi- sorptive bond and the conventional bond are fading. Yet many areas remain equivocal; for example, the empirical studies of the change in the chemisorptive and catalytic behavior of transition metals on progressive filling of d-band vacancies by metals of group IB. The studies fail to give clear evidence on the state of the chemisorptive bond, whether it is localized outside the energy limits of the d band or nonlocalized within it. In other areas, more precise models have been developed. Preeminent is the theory 4 1. INTRODUCTION of ionic chemisorption on semiconductors, called the boundary-layer theory [9], which depends on the donation of electrons by the valence band or the conduction band to the adsorbate, or the reverse, and on the building up of a potential field in the surface layers as adsorption proceeds. The maximum amount of adsorption is governed by the availability of electrons in the valence band or the conduction band, or by their capacities to accept electrons. The energy of adsorption is determined by the surface potential field, which changes with the amount adsorbed. Therefore, both quantities, amount and energy of adsorption, are sharply controlled by cooperative phenomena which have no existence in the free state. Although the conciseness and generality of'the boundary-layer theory are appealing, experiments do not always support it. Perhaps the most serious fault of the theory is that it considers only those types of adsorption that result from a transfer of charge. There are examples of systems in which ad- sorption on semiconductors occurs with no change in conductivity at all [10], thus, presumably with no transfer of charge. All approaches discussed so far have at least two serious drawbacks in common. First, each theory is built around an a priori selection of the type of chemisorptive bond, covalent or ionic, localized or nonlocalized. In a more powerful theory, the nature of the bond would evolve from the application of the theory itself. Second, it is often assumed, tacitly at least, that the forces and bonds in a solid surface and those that emerge from the surface and are available for bonding foreign atoms are not essentially different from those operating within the bulk of the solid. But the surface is a discontinuity, and this fact alone should arouse specu- lations on the existence of states peculiar to the surface. In the light of these limitations, the search for more rewarding theories based on solid- state theories continues. Full quantum-mechanical treatments of surface states and chemisorption, not mere adjuncts to narrowly defined systems, are now beginning to appear. Of course, it has been known in theory since the 1930s, from the work of Tamm [11] and Shockley [12], that states peculiar to a free surface can exist. And it has been suspected that these free-surface states could have a profound effect on our theories of chemisorption. But the complexity of the problem has deterred progress. The basic quantum- mechanical developments of Baldock [13] and Koster and Slater [14] provided an important spur. Equipped with these developments, a number of workers, notably Koutecky [15], Grimley [16], and Mark [17], started the attack on the problem in the 1960s. So far, the work deals chiefly with simple models, often one-dimensional, and employs the linear combi- nation of atomic Orbitals (LCAO) approximation of the molecular orbital 1.3 REAL SURFACES AND CHEMISORPTIVE BONDS 5 theory. No firm numerical values of energies have been calculated yet, but the results are interesting and provocative, for they have led to general conclusions which, at least, invite experiment. Ingenious experiments for investigating surface states have already been devised by Morrison [18] and Gray and co-workers [19]. The great importance of the theoretical studies is that, for the first time, criteria for localized and nonlocalized surface states have been clarified by the applications of basic principles. These criteria emerge in spite of the general nature of the theoretical solutions, dealing with arrays, often one-dimensional for mathematical tractability, of arbitrary surface atoms. The path which must be traveled to solve more realistic models is known and it is not an easy one. Grimley [20] has speculated about hydrogen chemisorption on metals, using quantum-mechanical methods similar to those for the more general cases. Mark [21] has pushed into the area of surface states and chemisorption on ionic lattices. We shall discuss these fascinating modern approaches in later chapters. The application of molecular orbital procedures just cited ignores the effects of near-neighbor atoms or ions on the nature of the d orbitals. These orbitals are split in a manner that depends on the crystallographic arrangement of the neighboring atoms or ions, and the theory which takes such splitting into account is known as crystal-field or ligand-field theory [22]. Recently, there have been qualitative and empirical attempts to apply such refinements to chemisorption [23]. It is difficult to guess where they will lead, for they have not yet passed beyond the point of initial speculation. We make no predictions beyond the crystal- or ligand- field approach to the theory of the chemisorptive bond. Yet we could travel far down this road and the real chemisorptive bond would remain uncomfortably distant. 1.3 Real Surfaces and Chemisorptive Bonds Real surfaces have many complexities that will probably never get incorporated into theories, and these complexities are transmitted to the chemisorptive bond. Solid surfaces are not necessarily at equilibrium. They may be "frozen" into a nonequilibrium state, and theories of nonequi- librium states tend to be arbitrary. Whether at equilibrium or not, solid surfaces may be perturbed and irregular, unlike the uniform arrays as- sumed in theoretical studies. Perturbations and irregularities can take many different forms. They can lead to a spectrum of adsorption-site energies and an arbitrary geographical distribution of sites. They can lead 6 1. INTRODUCTION to the presence of drastically different types of sites, such as lattice va cancies and interstitial atoms or ions. Recently, it has been shown that surfaces can be completely reconstructed by the process of chemisorption [24J, especially in the chemisorption of oxygen .dissociatively on metals, where metal atoms are drastically rearranged upon interaction with ad sorbate, so that the surface is not a neat'array ofmetal atoms with oxygen atoms above them, but a jumble of both of them. Finally, chemisorbed particles may interact with each other, causing obscure changes in the chemisorptive bond. Although theories may never conquer the real surface and chemisorptive bond, there is plenty of room for optimism short of this goal. REFERENCES 1. D. D. Eley, Discuss. Faraday Soc. 8, 34 (1950). 2. L. Pauling, "The Nature of the Chemical Bond." Cornell Univ. Press, Ithaca, New York, 1939. 3. M. Malone, J. Chern. Phys. 1, 197(1933). 4. S. M. Kogan and V. B. Sandomirsky, Bull. Acad. Sci. USSR Sere Chem. 1681 (1959). 5. Th. Wolkenstein and V. L. Bonch-Bruevich, J. Exptl. Theoret. Phys. (USSR) 20, 624 (1950). 6. V. L. Bonch-Bruevich, J. Phys. Chern. (USSR) 27,...662, 960 (1953). 7. W. E. Garnerand F. J. Veal, J. Chem. Soc. 1487 (1935); P. Cossee, J. Catal. 3, 80 (1964); J. J. Rooney, ibid. 2, 53 (1963). 8. R. C. L. Bosworth, Proc. Roy. Soc. (London) Sere A 162, 32 (1937). 9. K. Hauffe and H. J. Engell, Z. Elektrochem. 56, 366 (1952). 10. Y. Kubokawa andO. Toyama, J. Phys. Che'm. 60, 833(1956). 11. 1. 1."amm, Phys. Z. Sowjet I, 733 (1932). 12. W. Shockley, Phys. Rev. 56, 317 (1939). 13. G. R. Baldock, Proc. Cambridge Phil. Soc. 48, 457 (1952). 14. G. F. Koster and J. C. Slater, Phys. Rev. 95, 1167 (1954). 15. J. Koutecky, Advan. Chem. Phys. 9, 85(1965). 16. T. B. Grimley, Advan. Catal. 12, 1(1960). 17. P. Mark, Catal. Rev. 1, 165(1967). 18. S. R. Morrison, Private communication (1966). 19. T. J. Gray and P. Amigues, Surface Sci. 13, 209 (1969); T. J. Gray and R. S. Cichowski, Ph.D. Thesis, N. Y. StateCollegeofCeram. at AlfredUniv. (1968). 20. 1'. B. Grimley, "Chemisorption" (W. E. Garner ed.), Butterworth, London and Washington D.C., 1957. 21. P. Mark, J. Phys. Chern. Solids 29, 689 (1968). 22. L. E. Orgel, "An Introductionto rrransition-MetalChemistry." Wiley, New York, 1960. 23. G. C. Bond, Discuss. Faraday Soc. 41, 200 (1966). 24. J. W. May, Surface Sci. 18, 431 (1969). II Analogies with Simple Chemical Bonds-The Covalent Bond The pure covalent bond and the pure ionic bond are limits that the real chemisorptive bond never reaches. In this chapter, we shall discuss those theories that assume a predominantly covalent bond and make only small corrections for ionic character. In the following chapter, we shall discuss the ionic bond lying at the opposite end of the spectrum. Intermediate bonds with significant covalent and ionic character are difficult to fit into the simple models of this chapter and the next, which make a priori se lections of bond type. Further, we shall assume that these bonds are analogous to the bonds in diatomic molecules. This does not mean that concessionsto thepresenceofasurfacearenotmade. Forexample, electro negativities, from which the ionic character of a covalent bond is de ternlined, are related to the properties ofthe solid or its surface. Similarly, the character of the d orbitals of transition metals, often linked in theory to chemisorption, is based on the metallic state and not on the isolated atom. Yet the chemisorptive bonds considered in this chapter and the next do preserve the analogy with the 'localized, simple chemical bond. Each bond type links an adsorbed atom, molecule, or ion to a surface entity of the solid lattice despite that entity's association with other surface atoms, ions, or defects. We shall consider first Eley's extension of Pauling's rules for bond 7