Advances in Catalysis, Volume 41

ADVANCES IN CATALYSIS VOLUME 41

Advisory Board

G . ERTL

V. B. KAZANSKY

BerlidDahlem, Germany

J. M.THOMAS Evanston, Illinois

W. M. H. SACHTLER

Moscow, Russia

Evanston. Illinois

I? B. WEISZ State College, Pennsylvania

ADVANCES IN CATALYSIS VOLUME 41

Edited by D. D. ELEY

WERNER0. HAAG

The University Nottinghcrm. EnKland

Mohil Research and Development Corporation Princeron. New Jersey

ACADEMIC PRESS San Diego New York Boston London Sydney Tokyo Toronto

BRUCEGATES Uniwrsity of Cdforiiia Dcivis, CaliJiJrniu

Find Us on the Web! http://www.apnet.com

This book is printed on acid-free paper.

@

Copyright Q 1996 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 infonnation storage and retrieval system, without permission in writing from the publisher.

Academic Press, Inc. A Division of Harcourt Brace & Company 525 B Street, Suite 1900, San Diego, California 92101-4495 United Kingdom Edition published by Academic Press Limited 24-28 Oval Road, London NW I 7DX International Standard Serial Number: 0360-0564 International Standard Book Number: 0- 12-007841-4 PRINTED IN THE UNITED STATES OF AMERICA 96 97 9 8 9 9 00 01 Q W 9 8 7 6 5

4

3 2

I

CONTRIBLITORS ............................................................................................................................ PREFACE ..................... ....

ix xi

Vibrational Spectra of Hydrocarbons Adsorbed on Metals Part 1. Introductory Principles, Ethylene, and the Higher Acyclic Alkenes NORMANSHEPPARD AND CARLOS DE LA CRUZ

I. 11.

111.

I v. V.

VI. VII.

1 Introduction ..................................................................................... Experimental Considerations Relating to the Different Vibrational 3 Spectroscopic Techniques Available ............................................... Experimental Considerations Relating to Catalyst Preparation 7 or Sample-Handling Procedures ..................................................... Considerations Relating to the Interpretation of Vibrational 12 Spectra of Adsorbed Species .......................................................... Other Experimental Methods for Investigating the Structures of 26 Chemisorbed Hydrocarbons on Metals ........................................... Experimental Vibrational Spectroscopic Results on OxideSupported Metal Catalysts Classified by Alkene Adsorbate ........... 30 103 Conclusions ..................................................................................... 105 References .......................................................................................

Catalytic Chemistry of Heteropoly Compounds TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO I. 11. 111.

IV. V. V1.

Introduction ..................................................................................... Structure, Synthesis, Stability, and Characterization ...................... Acidic Properties ............................................................................. Acid-Catalyzed Reactions in the Liquid Phase ............................... Heterogeneous Acid-Catalyzed Reactions ...................................... Pseudoliquid Phase ......................................................................... Y

I13 1 18 139 150 I6 1 178

vi

CONTENTS

VII . VIII. IX . X. XI . XI1 .

XI11.

Redox Properties ............................................................................. Liquid-Phase Oxidation Reactions ................................................. Oxidation Catalyzed by Solid Heteropoly Compounds .................. Fine Chemicals Synthesis ............................................................... Hybrid Catalysts ............................................................................. Photocatalysis and Electrocatalysis ................................................ Conclusions ..................................................................................... References .......................................................................................

191 200 210 221 223 233 240 240

Microporous Crystalline litanium Silicates BRUNONOTARI

I. I1. 111.

I v. V. VI . VII . VIII .

Introduction ..................................................................................... Mixed Oxides of Ti and Si .............................................................. Titanium Silicates ........................................................................... Synthesis ......................................................................................... Catalytic Reactions ......................................................................... Catalytic Sites ................................................................................. Reaction Mechanism ....................................................................... Summary ......................................................................................... References .......................................................................................

253 257 267 288 293 317 318 326 327

Structural and Mechanistic Aspects of the Dehydration of Isomeric Butyl Alcohols over Porous Aluminosilicate Acid Catalysts KIRILL ILYCH ZAMARAEV AND JOHNMEURIC THOMAS I.

I1. 111.

IV. V. VI .

Introduction ..................................................................................... Characterization of Catalysts .......................................................... Kinetic Studies ................................................................................ Pathways of Butyl Alcohol Dehydration ........................................ The Nature of the +AI-O(R)-Si$ Reaction Intermediate ........... Conclusions ..................................................................................... References .......................................................................................

335 337 339 344 349 354 357

CONTENTS

vii

Thermal and Catalytic Etching Mechanisms of Metal Catalyst Reconstruction TA-CHINWE1 AND JONATHAN PHILLIPS

I.

I1. I11.

Introduction ..................................................................................... Thermal Etching ............................................................................. Catalytic Etching ............................................................................. References .......................................................................................

INDEX ............................................................................................................

359 362 383 415 423

This Page Intentionally Left Blank

Contributors Numbers in parentheses indiciite the pages on which the authors' contributions begin.

CARLOS DE LA CRUZ, Departmento de Quimica, Facultad de Ciencias, La Universidad del Zulia, Maracaibo, Venezuela ( 1 ) MAKOTOMISONO,Department of Applied Chemistry, The University of Tokyo, Tokyo 113, Japan ( 1 13) NORITAKA MIZUNO, Institute qf Industrial Science, The University of Tokyo, Roppongi. Tokyo 106, Japan ( 1 13) BRUNONOTARI, Notari Tecnologie, SnC, 20097 Sun Donato Milanese, Italy (253) TOSHIO OKUHARA, Division of Materials Science, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060, Japan (1 13) JONATHAN PHILLIPS, Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802 (359) NORMANSHEPPARD, School of Chemical Sciences, University of East Anglia. Norwich NR4 7TJ, England ( 1 ) JOHN MEURIG THOMAS, Davy Faraday Research Laboratory, Royal Institution of Great Britain, London WIX 4BS, United Kingdom (335) TA-CHINWEI, Department nf Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802 (359) KIRILLILYCH ZAMARAEV, Boreskov Institute of Catalysis, Novosibirsk 630090, Russia (335)

ix

This Page Intentionally Left Blank

With this volume of Advances in Carulysis, Herman Pines retires from the post of co-editor, which he has held since Volume 14. During this time, Herman’s popularity in the catalysis community, together with his perspicacity, has attracted many excellent contributions to Advances in Cutalysis. His own interests have been especially in the fields of organic chemicals and petroleum chemistry. It was in these areas that he achieved early and lasting fame for his work with Ipatieff, which began in 1932. Their catalytic process for producing “alkylate” for 100-octane aviation fuel played a major role in World War I1 (cf. Advances in Curulysis, Vol. I , pp. 27-28). We look forward to his continuing help as a member of the Advisory Board. At the same time we welcome our new co-editor, Bruce Gates, who will introduce the present volume. D. D. ELEY As this volume of Advances in Catalysis was going to press we received word that Herman Pines had passed away on April 10 at the age of 94. He leaves us with the legacy of his scientific work and the memory of an extraordinary man with a curious mind, boundless energy, independence of thought, human compassion, and a fine sense of humor. His early collaboration with Ipatieff began at Universal Oil Products and continued at Northwestern University where, from 1941 to 1952, he held a concurrent position as associate professor. In 1952 he became the Ipatieff Research Professor of Chemistry and director of the Ipatieff High Pressure and Catalytic Laboratory. The paraffin alkylation process mentioned above has become a major refinery process and is still practiced today on a very large scale, as is the process for the isomerization of paraffins which originated in Herman Pines’ laboratory. It was these discoveries and many others that led to his 145 patents and 265 publications, all of which were based on skillful experimental work. But at heart, Herman Pines was a theorist. As a professor he was an inspiring teacher and educator who set and demanded the highest standards. His research always aimed at new concepts and fundamental understandings. The initial work on acid-catalyzed reactions laid to rest the notion that paraffins are what their name implies, i.e., “parum affinis,” essentially inert. Imaginative studies on thermal, free-radical reactions-which led to a new manufacturing method in the perfume industry, among others-was followed by an in depth investigation of xi

xii

PREFACE

base-catalyzed hydrocarbon conversions. Most of what is known today about this subject is documented in his book, co-authored with Wayne Stalick, “Base-Catalyzed Reactions of Hydrocarbons and Related Compounds,” published by Academic Press in 1977. Few of us know that a key hydrocarbon intermediate in the manufacture of the widely used pain remedy lbuprofen is made in a basecatalyzed reaction first described by Herman Pines. The scope and depth of his work is reflected by the numerous awards he received, among them the Eugene H. Houdry Award in Applied Catalysis, the Chemical Pioneer Award, and the following American Chemical Society awards: The Fritzsche Award for his contributions to terpene chemistry, the Petroleum Chemistry Award, and the E. V. Murphree Award in Industrial and Engineering Chemistry. His colleagues, former students, and friends will miss him. W. 0. HAAG

The first chapter of this volume, by Sheppard and de la Cruz, addresses the application of vibrational spectroscopy for the characterization of adsorbed hydrocarbons. This chapter is a successor to the 1958 Advances in Catalysis chapter about infrared spectra of adsorbed species, authored by the pioneers Eischens and Pliskin. Vibrational spectroscopy continues to provide some of the most incisive techniques available for determination of adsorbate structures. The present chapter is concerned with introductory principles and spectra of adsorbed alkenes; a sequel is scheduled to appear in a subsequent volume of Advances in Catalysis. The heart of this volume consists of three chapters summarizing work on catalysts that are both industrially applied and structurally well-defined; the structural definition has allowed rapid progress in the development of relationships between structure and catalytic properties. Okuhara, Mizuno, and Misono report the catalytic properties of heteropoly compounds as exemplified by H,PW120,, and the anion [PW,,O,,,]’-. Some of these compounds are strongly acidic, and some have redox properties; the largescale applications involve acid-catalyzed reactions. The heteropoly compounds are metal oxide clusters, used as both soluble and solid catalysts. Their molecular character provides excellent opportunities for incisive structural characterization and for tailoring of the catalytic properties. Physical properties also affect catalytic performance. Catalysis sometimes occurs on the surface of the solid material, and sometimes it occurs in the swellable bulk. Notari summarizes the science and technology of catalysis by molecular sieves incorporating framework metal ions. The most important example in technology is Ti silicalite, a selective oxidation catalyst. Because the catalysts are crystalline materials, their structures are among the most well-known of any industrial catalysts, and catalytic sites such as Ti cations are identified. The work summarized in this chapter, almost all of it performed in just the preceding few years, gives an

PREFACE

...

Xlll

indication of how important molecular sieves may become as catalysts for reactions beyond those catalyzed by acids. Zamaraev and Thomas provide a concise summary of work done with a family of classic catalytic test reactions-dehydration of butyl alcohols-to probe the workings of acidic molecular sieve catalysts. This chapter echoes some of the themes stated by Pines and Manassen, who wrote about alcohol dehydration reactions catalyzed by solid acids in the 1966 volume of Advances in Catalysis. In the final chapter, Wei and Phillips tie together old and new results characterizing the processes of surface etching. They summarize evidence that chemical etching takes place by reactions of gas-phase free radicals. This subject pertains to catalyst redispersion and regeneration, and the chapter links the catalysis literature and literature less often consulted by catalytic scientists and engineers.

B.C.GATES

This Page Intentionally Left Blank

ADVANCES rh CATALYSIS. VOLUME 41

Vibrational Spectra of Hydrocarbons Adsorbed on Metals Part I. Introductory Principles, Ethylene, and the Higher Acyclic Alkenes NORMAN SHEPPARD School of Chemicul Sciences University (?f East Anglia Nonvich NR4 7TJ, England AND

CARLOS DE LA CRUZ Departmento de Quimicu Fucultad de Ciencias La Universidud del Zuliu Murucuiho. Venezuela

1.

Introduction

In 1897, Sabatier and Senderens ( I ) made a pioneering study of the use of a nickel as a catalyst for the hydrogenation of ethylene (ethene) to ethane. This investigation led to the award of the Nobel Prize to Sabatier in 1912. Since that time the importance of heterogeneous catalysts has continued to increase greatly, decade by decade, extending the boundaries of laboratory chemical researches and promoting new and more cost-effective processes within the chemical industry (2). The correct choice of a catalyst allows a desired reaction to proceed under milder conditions of temperature and pressure than would be Abbreviations: DRIFT(S)4iffuse-reflection infrared (Fourier-transform) (spectroscopy): EXAFS-xtended X-ray absorption fine structure; FTIR(S)-Fourier-transform infrared (spectroscopy); INS---inelastic neutron scattering; LEED-low-energy electron diffraction; MCT-mercury/ cadmium telluride photoconductive detector; MSSR-metal-surface selection rule; NEXAFSnear-edge X-ray absorption fine structure; NMR-nuclear magnetic resonance; PED-photoelectron infrared (spectroscopy): SERS-surface-enhanced diffraction; RAIR(S)-reflection-absorption Raman spectroscopy; STM-scanning tunneling microscopy; TPD-temperature-programmed desorption; UPS-ultraviolet photoelectron spectroscopy; VEEL(Stvibrationa1 electron energyloss (spectroscopy); XPS-X-ray photoelectron spectroscopy.

I Copyright U’ 1996 by Academic Press. Inc. All rights of reproduction in any form reserved.

2

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

the case otherwise. Often more important, catalysts make possible selectivity toward a particular desired reaction when a set of starting materials can give different sets of products. Until the 1950s the nature of the chemical surface species involved in the interactions between the surface of a catalyst and the reagents or products of a reaction-the reaction intermediates-could be discussed only as possibilities based on studies of the kinetics of the overall reaction, aided by results from isotopic variants among reactants and products ( 3 , 4 ) . Vibrational spectroscopy in the infrared region provided the first direct experimental evidence on the structures of adsorbed species themselves and has subsequently proved an excellent method for systematic investigations in both chemisorption and catalysis. Even today, vibrational spectroscopy in its various forms remains the most widely used physical method for identifying molecular surface species. It has the unique advantage that it can as readily be used to study adsorption on finely divided catalysts as on single-crystal surfaces used as simplified models. The single-crystal systems with surfaces of known atomic arrangements have the additional advantage that they can be studied by the numerous other spectroscopic and diffraction experimental techniques available in surface science. In 1954 R. P. Eischens, W. A. Pliskin, and S. A. Francis ( 5 ) of the Texaco Research Center in New York published the first infrared spectra of chemisorbed species, namely of carbon monoxide adsorbed on the silica-supported finely divided metal catalysts of Ni, Pd, Pt, and Cu. Also, in 1956, Pliskin and Eischens (6) were the first to obtain spectra of the hydrocarbons ethylene (ethene), acetylene (ethyne), and propene adsorbed on an oxide-supported metal catalyst, Ni/Si02. Eischens and his colleagues followed this up with further studies of chemisorbed n-alkenes and their surface-hydrogenation products on Ni/SiOz (7). Since these early days, many infrared studies of hydrocarbons adsorbed on oxide-supported metal catalysts have been carried out. Although much success has ultimately been achieved in identifying the chemisorbed species present, progress has tended to be slow for a number of reasons: 1. An observed spectrum is typically found to be derived from the presence

of several different adsorbed species. These arise from a plurality of different adsorption sites that can occur on particles of a given metal. Even a well crystallized particle could, for example, present sites associated with different types of facets, e.g., ( 1 1 l), (IOO), or ( I lo), and their various arrays of metal atoms (see Fig. 1 in Section IV.A.1). 2. Even when the spectrum of an adsorbed species proves to be that of only one of those anticipated, difficulties of recognition can arise because of the perturbing effects of the bonded metal atoms on the well known characteristic frequencies (wavenumbers) of hydrocarbon groups (8-10).

VIBRATIONAL SPECTRA OF HYDROCARBONS

3

3. Particular identification problems arise when unanticipated surface species are present. A case in point is the ethylidyne surface species, M3(CCH3) (M = metal atom), which has turned out to be of common occurrence from ethene adsorbed on metals near room temperature (11). About two decades after Eischens’ application of transmission infrared spectroscopy to the study of adsorption on high-area, finely divided metal catalysts, a new vibrational spectroscopic technique, termed electron energy loss spectroscopy (EELS), was developed. This technique proved to have the sensitivity required to obtain spectra from literally single monolayers of adsorbed species such as those occurring on the flat, low-area surfaces of single metal crystals. It was first applied to the study of adsorbed hydrocarbons in 1977 by H. Ibach and his colleagues, who studied ethene on Pt( 1 I 1 ) (12), and by J. C. Bertolini and colleagues, who investigated benzene on Ni( 1 1 1) and Ni(100) (13). The technique is variously described as high-resolution or vibrational electron-energyloss spectroscopy, abbreviated as HREELS or VEELS, respectively. We shall use the latter abbreviation because the EELS method is of low resolution relative to its competitor, infrared spectroscopy. The original adjective “high” relates to comparison with electronic EELS. Hitherto, in the form of reflection-absorption infrared spectroscopy (RAIRS), the infrared method had been capable of detecting single monolayers only in the exceptionally favorable (strong absorption) cases of carboxylate ions [Francis and Ellison (14)] or carbon monoxide [Chesters, Pritchard, and Sims ( 1 5 ) ] adsorbed on flat metal surfaces. The new challenge from VEELS provided the motivation for a search for improvements in RAIRS sensitivity, and this was very successfully achieved by M. A. Chesters and his colleagues through the introduction of Fourier-transform-based interferometric infrared spectroscopy (16). During the past two decades, the results obtained from hydrocarbons adsorbed on the simplified single-crystal metal surfaces by the use of VEELS or RAIRS have proved to be of crucial assistance in the interpretation of the more complex spectra obtained from finely divided metals. The VEELYRAIRS results have recently been reviewed by one of us (17); it is therefore timely to do likewise for the many results on finely divided metals now available in the literature. This is particularly so because there is a general need to revise many of the earlier interpretations owing to several improvements in our understanding of such spectra.

Ii. Experimental Considerations Relating to the Different Vibrational Spectroscopic Techniques Available A.

TRANSMISSION INFRAREDSPECTROSCOPY (18)

Infrared spectroscopy is capable of giving spectra of very good signahoise from adsorbed species on finely divided metal catalysts. Samples are usually

4

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

studied in the form of pressed disks which remain porous to gases. It is also capable of high resolution (<1 cm-I). However, such spectra are restricted in extent because very strong absorptions-near transmission blackouts-over considerable wavenumber ranges are caused by the oxide supports. Spectra of adsorbed species on such supports can in practice be obtained only above 1300cm-’ for S i 0 2 , 1050cm-’ for A1203, 1200cm-’ for TiOz, and 800 cm- for MgO. SiOz and A1203 are commercially available in the required high-area forms and are most widely used as catalyst support materials. Fortunately, the high-wavenumber-transparent regions in the spectra of these oxides contain a majority of the group-characteristic frequencies of hydrocarbon-based species (8-1 0). The development of Fourier-transform (FT) techniques and spectral-ratioing facilities in mid-infrared spectroscopy in the early 1970s opened up a new era of efficiency in the study of spectra of molecules adsorbed on finely divided materials. The FT method provided much improved sensitivity by one to two orders of magnitude. Additionally, the minicomputer/ microprocessor, incorporated in the spectrometer for the conversion of the measured interferogram into the conventional spectral form, also allowed pointby-point transmission mtioing or absorbance subtraction of the background spectrum from the adsorbent (19, 20). The available high sensitivity in turn made possible the use of such alternative sampling techniques as diffuse reflectance. which is described below.

’

B. REFLECTION-ABSORPTION INFRARED SPECTROSCOPY (RAIRS) (16) RAIRS gives spectra that, although necessarily much weaker, share with infrared transmission the capability of high resolution (<1 cm- I ) . Also, the intensity distributions within these two types of spectra are closely similar for adsorption on large metal particles or on flat metal surfaces. A metal-surface selection rule (MSSR; to be discussed below) for flat metal surfaces allows only those vibrations of adsorbed species that give rise to electric dipole changes perpendicular to the surface. This “selection rule” is also effective for adsorption on large metal particles. For small metal particles (diameter <2 nm) modes of vibration with parallel dipole changes also become observable. RAIRS spectra are routinely limited to the 4000-800-cm- I region that can be covered by a single mercury/cadmium telluride (MCT) photoconductive detector ( 16). However, alternative doped-germanium or other detectors, operable at liquid-He temperatures, will increasingly be used down to below 200 cm- I , sometimes in conjunction with synchrotron radiation sources. Photon spectroscopies in general, and infrared spectroscopy in particular, have the advantage that results can be obtained on adsorbed species in the presence of a considerable pressure of a gas phase.

VIBRATIONAL SPECTRA OF HYDROCARBONS

5

C. VIBRATIONAL ELECTRON-ENERGY-LOSS SPECTROSCOPY (VEELS) ( 2 1 ) This form of electron spectroscopy has high sensitivity for monolayer studies but much lower spectral resolution than infrared spectroscopy, normally 20-40 cm- I. Recent developments (22) hold out the possibility of experimentally reducing this down to ca. 8 cm-I, but for studies of polyatomic adsorbed species the limited resolution of VEELS remains a substantial disadvantage. However, deconvolution techniques for resolution enhancement, which utilize the band-shape of the elastic peak, and statistical noise-reduction procedures are at present markedly improving the quality of such spectra. “On-specular” spectra (for which the electron angle of incidence equals the angle of reflection) follow the same metal-surface selection rule as RAIRS; i.e., modes with perpendicular dipole changes are dominant. There is, however, a systematically different intensity distribution within such VEEL spectra relative to RAIRS (21, 24), so that usefully complementary information is provided by these two techniques. Relative intensities are much greater for the low-wavenumber features and weaker for high-wavenumber bands in VEELS compared with RAIRS; in hydrocarbon spectra the CH bond-stretching absorptions near 3000 cm can be particularly weak in VEELS. But the VEELS method has an added advantage over RAIRS in that “off-specular” spectra are excited by a different (impact) mechanism, thereby providing additional information about vibrations with dipole changes parallel to the surface or with no dipole changes at all (21). Although salient features in vibrational spectra are often adequate for the purposes of structure determination, complete mode-based analyses of spectra require the additional data available from off-specular VEELS. VEEL spectra have the advantage of being available, without change of experimental arrangements, throughout the vibrational fundamental region down to ca. 200 cm- ’, below which the elastically scattered signal becomes dominant. VEELS, as an electron spectroscopy, has the disadvantage of requiring that a high vacuum be maintained over the surface under investigation. The present-day literature contains many more spectra obtained from singlecrystal metal surfaces by VEELS than by RAIRS. However, the much higher resolution available from the more recently developed RAIRS technique and its capability of operating in the presence of a gas phase suggest that it will contribute increasingly important information in the hydrocarbon adsorption field. The three spectroscopic techniques discussed above are much the most important ones in this area, with transmission infrared spectroscopy as the predominantly useful one for work with finely divided samples. A few other vibrational spectroscopic techniques (25) have provided information on adsorbed hydrocarbons, but are at present of more limited or specialized applications. Their principal characteristics are more briefly summarized below. ~

’

6

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ D.

RAMANSPECTROSCOPY

Raman spectroscopy, alternatively described as inelastic photon scattering spectroscopy, has proved difficult to apply to oxide-supported metals because of heating effects generated by absorption of the intense incident laser beam and because of limited light scattering. Both of these restrictions are caused by the typically black or highly-colored samples. Nevertheless, useful pioneering work on hydrocarbons has been done by Krasser and his colleagues (26; see also Section 1I.E). Raman spectra of adsorbed monolayers on flat metal surfaces are extremely weak, just on the edge of attainable sensitivity. However, Campion has obtained successful spectra under favorable circumstances; his review of the situation is given in the first book cited in reference 25.

E. SURFACE-ENHANCED RAMANSPECTROSCOPY (SERS) (25) As its name implies, this technique offers very much greater sensitivity, by up to X 10' compared with normal Raman spectroscopy. This occurs on roughened metal surfaces such as electrodes and cold-evaporated films. Unfortunately, the phenomenon at its best is highly selective, limited principally to the metals Cu, Ag, and Au. On these metals excellent spectra can be obtained (27, 28). Nevertheless, a few encouraging successes have been reported for the typical group VlIl metals, such as Ni, Pd and Pt, which are of principal catalytic interest (29).

F.

DIFFUSE-REFLECTION (FOURIER-TRANSFORM) INFRARED

SPECTROSCOPY (DRIFTS) (18) In its applications to spectra on finely divided metals DRIFTS is similar to infrared transmission. Its advantage is that catalysts can be studied in powdered form, even in situ during catalytic reactions, thus avoiding some of the diffusion-limited kinetics that are encountered when pressed catalyst disks are used. G.

INELASTIC

ELECTRON-TUNNELING SPECTROSCOPY (IETS) (25)

IETS is spectroscopically similar in resolution to the infrared method except for differences in the pattern of band intensities. Unfortunately, it is limited to samples held at liquid-helium temperatures. However, in the future, in conjunction with scanning tunneling microscopy, it might yield spectra from single adsorbed molecules on particular sites (30). H.

INELASTIC NEUTRON SCATTERING (INS) (25)

INS, which is applicable to high-surface-area catalysts, applies in practice principally to vibrational modes that involve motions of hydrogen atoms. For

VIBRATIONAL SPECTRA OF HYDROCARBONS

7

hydrocarbon species, the hydrogen atoms can be involved directly in the form of CH bond-stretching or angle-bending modes, or indirectly when motions of the carbon skeleton carry hydrogen atoms with them. Unlike those of infrared and Raman spectra, relative intensities in such spectra can be calculated directly from estimated vibrational amplitudes. The chief characteristics of these various vibrational spectroscopics are summarized in Table I.

111.

Experimental Considerations Relating to Catalyst Preparation or Sample-Handling Procedures

The performance of a catalyst is well known to be sensitive to its preparation procedure. For this reason, ideally an oxide-supported metal catalyst should be subjected to a number of characterization procedures. These may include measurements of the metal loading within the overall catalyst (usually expressed in wt%), the degree of metal dispersion (the proportion of metal atoms in the particle surfaces), the mean value and the distribution of metal particle diameters, and qualitative assessments of morphology including the particle shapes and evidence for crystallinity. These properties in turn can depend on experimental variables used in the preparation, such as the choice and amounts of originating metal salts, prereduction, calcination or oxygen treatments, and the temperature and duration of hydrogen reduction procedures. Despite the number of possibly significant variables, catalysts prepared from the same metal (and sometimes from different metals) tend to give generally similar performances with the occasional significant differences (31). Not surprisingly, a similar situation prevails among the infrared spectra from a given hydrocarbon adsorbate on catalysts based on the different metal preparations. Related sets of absorption bands often denote the presence of similar surface species. However, spectra from a given catalyst can differ substantially as a function of temperature, showing interconversions from one type of adsorbed species to another. Sometimes different metals show spectra from the same adsorbed species occurring over different temperature ranges. Table I1 provides a summary of the main research groups that have contributed to the vibrational spectroscopy of adsorbed hydrocarbons on oxide-supported metal catalyst surfaces (6, 7, 26, 29, 32-68). Arranged in historical sequence, this table also identifies research groups presumed to be currently active, as judged from the literature. After a long induction period, when only a few research groups were active, recent years have seen a strong revival of interest in this research field. The entry for each research group lists the oxide-supported metal preparations used, with a summary of their typical characteristics.

TABLE 1 Characteristics of the Dixerent Vibrational Speetroseopies Used to Obtain Results from Species Adsorbed on Finely Divided or Single-Crystal Metal Surfaces

Technique

Wavenumber range (cm-')

Working resolution General (cm - I ) sensitivity

Applicability to moderate gas-phase pressures

Type of sample

Conditions for most intense bands

Degree of usage

I100"

>1

High

Finely divided metals'

Yes

Dipole change perpendicular to surface

Very great

DRIFTS

4000-ca. 1 100"

>1

High

Powdered metaVoxides Yes

Dipole change perpendicular to surface

Limited as yet

RAIRS

4000-800 (-200)

ca. 1

Low

Flat, preferably single-crystal, metal surfaces

Yes

Dipole change perpendicular to surface

Moderate

VEELS

40W200

High

Flat, preferably single-crystal, metal surfaces

No'

Dipole change perpendicular to surface (specular reflection)d. All modes'

Great

Raman spectroscopy

4W200

ca. 5

Low'

Finely divided' or flat surfaces

Yes

Polarizability change perpendicular to surface

Limited

SERS

4000-200

ea. 5

High

Mostly Cu, Ag, Au

Yes

Polarizability change perpendicular to surface

Moderate; great for Ag

IETS

4000-200

ea. 5

Moderate

Finely divided'

No

As for infrared and Raman

Limitedg

INS

4000-200

ea. 10

Moderate

Finely divided

Yes

Hydrogenic motions

Limitedh

IR transmission 4-a.

>30

"Limited by transmission of the oxide support; >I300 cm-' for silica, >I050 cm-' for alumina. 'Normally oxide-supported; see foomote a. 10 mbar. Dipolar mechanism. 'Impact mechanism. 'Enhanced sensitivity with some metals. 'Samples have to be at liquid-helium temperatures. Large samples needed, as well as access to an atomic pile as the neutron source.

' Ultra-high vacuum equipment needed; highest allowable gas-phase pressure

-

~'

VIBRATIONAL SPECTRA OF HYDROCARBONS

9

It can be seen from Table I1 that, for the most part, metal loadings within the catalysts have been rather high (-10%). Such high metal loading was because of limited infrared sensitivity for the observation of adsorbed species. However, since the development of Fourier-transform spectrometers in the early 1970s, this has no longer been necessary. For financial reasons, most industrial catalysts formed from precious metals, such as Pt and Pd, have loadings of less than 1%; and, for the hture, it is to be hoped that more samples of this type will be investigated (60). High metal loadings tend to go with large metal particle sizes, which show well-developed facets under the electron microscope. Such preparations are likely to give spectra that are simplified by the effective operation of the metal-surface selection rule (see Section 1V.B). Rather more complex spectra are to be expected from the relatively few catalysts that have particle diameters less than 2 to 3 nm. Trenary (58) has emphasized that the catalyst reduction temperature is a particularly relevant variable in causing differently profiled spectra from the same adsorbate on different preparations of a given metal catalyst. In the initial work of Eischens et al. (5, 6) the metal/oxide catalyst was studied in the form of a powdered layer on a horizontal CaFz window. Little, Sheppard, and Yates (69) instead used a tube of porous glass as the oxide support, and this could conveniently be mounted vertically in the infrared beam. [Similarly, Terenin and Roev (70) used alumina gel as a metal support to study NO adsorption.] However, the small pores in the silica glass limited the amount of metal that could be incorporated, leading to weak spectra from adsorbed species. Furthermore, porous glass retains small amounts of reactive oxides of boron and other metals that can lead to spectral contamination. Sheppard and Ward (71), and also Dunken, Schmidt, and Hobert (39), therefore decided to form disks from (salt + oxide) by high pressure from a hydraulic press, starting from pure silica in the form of Cabosil or Aerosil. These disks, which were converted to metal-on-oxide through reduction with Hz in the evacuable infrared cell, also proved to be successfully porous to gaseous adsorbates. Many infrared spectra of adsorbed hydrocarbons have been obtained using such samples. The principal limitation of this method is the difficulty of obtaining uniform temperatures over the disk surface when temperatures significantly different from room temperature are required in the absence of a substantial gas phase. This is because the poor thermal conductivity of silica makes difficult the transport of heat to or from a source in contact with the edge of the thin disk. This difficulty can be avoided by spraying a slurry of the catalyst in a volatile organic solvent onto a heated CaFz plate so as to cause the solvent to evaporate rapidly. The powder usually adheres to the CaFz plate, providing good control for work at low or high temperatures. This method, adapted from Yang and Garland ( 7 2 ) , has been successfully used by Yates (50) and Trenary ( 5 7 ) for hydrocarbon work. An alternative approach to obtaining uniform catalyst temperatures over a

10

NORMAN SHEPPARD AND CARLOS DE LA CRUZ TABLE I I Churucterizution Dutu ,for O.ride-Supported Metul Catalysts Emplqved by Research Groups Studying the Adsorption of Hydrocurhons hy Infrured (IR)or Rumun (Ru) Spectroscopy

Research group

Metal loading Spectroscopy Catalysts Precursor

(wt%)

Eischens and Pliskin

IR

Ni/Si02

Ni(NO,)?

9

Little, Sheppard, and Yates

IR

Ni/SiO? Pd/SiOz CdSi02

Ni(N03)2 Pd(N0,)2 Cu(N03)~

2.3 5.8 9

Sheppard et a/."

IR

Ni/SiOz Pd/Si02 Pt/Si02 Ir/Si02 Rh/SiO?

Ni(N03)z PdCI2 H2PtClh H21rCl,, RhCI,

Pd/Si02 Pt/Si02

PdClz H2PtC16

2.5 2.5

Ni/SiO?

Ni(N03)2 Ni(N03)2

18.7 I0 5 5 15

Dunken, Schmidt, and Hobert

IR

Erkelens ul.

IR

Nils102

el

Cu/SiOz PdJSiO? Pt/Si02 Palazov ul.

IR

Metal Reduction particle temperature size (nm) ("C) References cu. 8

<8 18 <8

9 13 16 10 II

300

6, 7

300 6Y (ofoxide)

1350

32 32.35 33,34 36.37 38

-

350

3Y

-

350 450 390 300 300

40 41 41 41 41

350 350 450 450

42

-,I0

350

43

350

44

9 9 9 9

1.5-2.6 ~

~

-

Ni/SiO* NilA1203 Pt/SiO2 PL'AI201

Ni(N03)2 Ni(NO,)? HZPtCI, H2PtCIh

IR

Pd/Si02

PdCI2

Reiicha

IR

Ni/AlzOl Ni(N01)2

23

Primet and Mathieu Blyholder

IR

Pt/AI,O,

H2PtCIc,

10

3.3

500

45

IR

Fe/Si02 Co/SiO2 NiiSiO2

FC(NO~)~ 9 Co(N03)~ 9 9 Ni(N0,)2

-

430 430 430

46

47

-

380 and 290

-

500

4X

el

Avery

Pt

ul.

Soma

Baumarten and Weinstrauch

IR

IR

Pd/A1203 PdCI? Pt/A1201 HZPtCI, Rh/A1203 RhCli

9 9 9

Pt/AI?O, HzPtCl6

10

~

-.

-

(continues)

11

VIBRATIONAL SPECTRA OF HYDROCARBONS TABLE II-(Coniinued)

Research group

Spectroscopy Catalysts

Precursor

Metal Metal Reduction loading particle temperature (wt%) size (nm) ("C) References

Krasser and Renouprcz

Ra

Ni/Si02 PdSi02

Ni(N0,): PdCI2

-

Haaland

IR

Pt/A1203

HZPtCI,

J. T . Yates cr a/."

IR

PdAl201 Pt/AI,O, Rh/A1201 Ru/AI~O~

Hexter ei ul.

Ra

Ghiotti rr ul." Hanson er ul." Ekerdt rt ul."

IR

Trenary rr ul."

cu. 2

700 400

26 ZY

10

-

365

4Y

PdClz HzPtClh RhCI, Ru(NO,),

10 10 10 10

8.4 I .9 2.8 6.6

200 200 200 200

50

Rh/SiOz

RhCl,

10

2.6

150

51

IR

CdZnO

CuO

cu. 7

230

52

Ra

Rh/A1203

Rh/NO3)> or RhC13

10

I 5011 70

53

Ni1Si02 Ni/AI2O3 Co/Si02

Ni(N03)2 Ni(N03)2 CO(NO,)~

8.3 8.3 9.5

10

400 400 400

54

14 9

Pd/AI203

Pt/A1203 Rh/A1:03 Ir/A1203

PdCl2 H,PtCI, RhCl3 HzlrClh

1.6 3 1.6 3

<3.5 I .9 1.7 <3.5

4001325 500/325 4001325 5001325

IR

Szilagi"

IR

Pt/Si02

H2PtCIh

Lavalley ri ul."

IR

PdA1203

Pd(acac);

Pt/TiOz

Pt film

Hensley and Kesmodel

VEELS

McDougall

IR

ei ul."

Ni/A1203 Ni(NO,)? Pt/A1203 Pt(NH3)&12 (EUROPT- I )

16

<5 -

55 S6 57, 5N

6.5

350

5Y

0.45

2.5

5001200

60

-

1-2

25 6.3

-

61

2

380 400

62 63-65

IR

PtiSiO,

H2PtCI,

2.4

4.1

200

66

Trunschke and Knozinger" '

IR

Rh/SiOz

RhCll

3.8

-

400

67

Szilagyi"

IR

Pt/Si02

H2PtClb

16.3

6.5

347

68

lwasawa el ul."

"

"Groups probably still active in research on hydrocarbons on metals. '(acac) VEELS.

=

acetylacetate.

wide range is to press the catalyst disk together with a thin wire mesh of a metal, such as Ta, Mo or W, which can be used for both heating and temperature measurement ( 73-75).

12

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

For kinetics studies, powdered samples are less likely to give rise to diffusionlimited regimes in the reaction rates. Even loose powdered catalysts-a further step toward “real” conditions-can now be studied using diffuse-reflection infrared techniques (62), and it is to be hoped that more studies of this type will be made. Finely divided metal samples can also be prepared in the form of evaporated films in high vacuum, usually deposited on IR-transparent alkali halide plates ( 7 6 7 8 ) . Such spectra are of interest in themselves, but tend to be much weaker than those obtained from the metal-particles-in-depth, oxide-supported catalysts. The rough surfaces of films of Cu, Ag, and Au, prepared by deposition on cold surfaces, can lead to very high-quality surface-enhanced Raman spectra (27, 28, 79, 80). The results from such experiments will be discussed in the later sections devoted to particular adsorbed hydrocarbons and metals, alongside the majority of spectra that are obtained on oxide-supported samples.

IV. Considerations Relating to the Interpretation of Vibrational Spectra of Adsorbed Species As was shown to be the case in an earlier review of the vibrational spectroscopic literature on chemisorbed CO (81), it is very profitable to consider the more complex spectroscopic results obtained on finely divided catalysts in the light of conclusions reached from work on simplified experimental situations that involve particular faces of single crystals of the metal in question (17). We therefore first consider the principles involved in the interpretation of spectra obtained under these conditions. To date, the structural interpretations of the single-crystal spectra have been best guided by comparison with the analogous spectra from possible hydrocarbon species observed as ligands on metal-cluster compounds, the structures of which have been unambiguously identified by X-ray diffraction studies (82). For the future, it is to be hoped that diffraction methods can be more directly and effectively applied to the structural identification of adsorbed species on metal surfaces themselves. So far only a few, but nevertheless important, results have been achieved from hydrocarbon layers using low-energy electron diffraction (LEED) on ordered arrays of adsorbed species. These include the identification of surface alkylidyne species from adsorbed alkenes and the study of adsorbed benzene on ( 1 11) surfaces of several metals by Somorjai, Van Hove, and their colleagues (83, 84). Additionally, the recently developed scanned-energy photoelectron diffraction method, also applicable to single-crystal surfaces, holds out the promise that more adsorbate structures will be determined on a local basis (85); in fact, this method has recently been applied to the case of ethyne on Cu( 1 1 I ) (86) and of ethyne and ethene on Ni( 1 1 1) (87, 88).

VIBRATIONAL SPECTRA OF HYDROCARBONS

13

The initial role of vibrational spectroscopy is to suggest the types of functional groups that are present. In favorable cases, this can lead on to the detection of particular surface species through the recognition of their more complete spectral patterns. Subsequently, when structural “calibration” has been achieved or confirmed by diffraction methods, vibrational spectra provide much the most efficient means of exploring the incidence of the various types of adsorbed species on a wide range of surfaces.

A. SYMMETRY CONSIDERATIONS I N THE SPECTRA OF ADSORBED SPECIES ON WELL-DEFINED SINGLE-CRYSTAL SURFACES The interpretation of vibrational spectra is dependent on a correct assessment of the symmetry properties of the adsorbed species themselves and of their vibrational modes. Several general accounts have been given of the classification of vibrations of adsorbed species in terms of the symmetry elements associated with a surface complex (89. YO).

I.

Site Svmmetries

Figure 1 shows the arrangements of metal atoms on ( 1 1 I ) , (loo), and ( 1 10) surfaces of a face-centered cubic (fcc) metal. It can readily be seen that different sites on a particular surface have different symmetry properties. For example, the top layer of atoms on sites 1, 2, 3, and 3‘ on the ( I 1 1 ) face (where the number denotes the number of metal atoms associated with the site) exhibit &fold, 2-fold, 3-fold, and 3-fold rotation axes of symmetry, respectively. At this level of discrimination the point group symmetries of those sites are Cbr,,Cl,,, C,,,, and C,,,, respectively. However, when the arrangement of atoms in the second layer is taken into account (there is another atom under only half of the 3-fold sites) the point group symmetries of the first two sites

f.c.c. (100)

f.c.c. (111)

f.c.c. (110)

FIG. I . The arrangements of surface atoms on the ( I 1 1 ), (100). and (1 10) surface planes of facecentered cubic (fcc) lattices of metals.

14

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

are reduced to C3,,and C,%,respectively and the two C3,,sites become nonequivalent. At this level there is only a 3-fold axis associated with site 1 and a 1-fold axis with site 2. In practical terms, the symmetry reductions associated with the second layer of atoms may not have measurable consequences in the observed spectra. It should be more generally noted that surface site symmetries are limited to the C,, or C,,,, types [including C,%= C,,,] because the “onesided” nature of crystal surfaces requires that no horizontal mirror-planes or axes, or centers of symmetry, can be present (89). 2. Symmetry of the Overall Adsorption Complex We consider this matter using examples of possible hydrocarbon surface complexes. Of course, the overall symmetry of an adsorption complex cannot exceed that of the surface site that it occupies. On the (1 1 1) face of Pt, Ibach and his colleagues (12, 91) have concluded that at low temperatures ethylene is adsorbed as a 1,2-di-a complex, but that at room temperature this is replaced by the ethylidyne species, structures 4 and 8, respectively in Fig. 4 (Section 1V.C); we shall use these as examples. The ethylidyne species M3(CCH3) must occur on sites 3 or 3’ of the ( 1 1 1 ) plane and have the overall symmetry of C3,,or C3 depending on whether or not the orientations of the CH bonds of the CH3 group coincide with the vertical planes of symmetry of those sites. We shall consider the symmetry of the di-a species in terms of symmetry of site 2 on which it must be adsorbed; i.e., we ignore the effect of the second layer of atoms for this purpose. If the pairs of carbon and platinum atoms are coplanar in the di-a M2(H*CCH2) species, thus forming a planar cyclic C2M2skeleton perpendicular to the surface, the adsorption complex itself will retain effective CZ,,symmetry. If, on the other hand, strain occurs in matching normal C-C, C-M, and M-M bond distances together in a 4-membered ring, and/or if substantial repulsions occur between eclipsed CH or CM bonds at either end of the central C-C bond, the effect will be to rotate the C-C bond relative to the M-M direction and to reduce the symmetry of the adsorption complex to C2. This will have the effect of coupling together considerable numbers of pairs of group vibrations-pairs that would be independent of each other under C2,,because of different symmetries. When the metal-surface selection rule is taken into consideration (see the following section), the reduced symmetry from C2,, to C2 would allow the observation of nearly twice as many fundamentals under the dipolar mechanism applicable in RAIRS and on-specular VEELS. However, if the deviation from C2,, is slight, the intensities of the additional bands would be expected to be correspondingly weak. In the less likely event that the second layer of metal atoms would have an observable effect in the spectra, the symmetries of the complexes with planar or nonplanar skeletons would be reduced to C, and C, , respectively.

15

VIBRATIONAL SPECTRA OF HYDROCARBONS

B. THEMETAL-SURFACE SELECTION RULE (MSSR)

The intensity of an infrared absorption band is proportional to the square of the magnitude of the electrical dipole change occumng between the two extremes of the vibration in question. However, the very high polarizability of the essentially free conduction electrons of a metal (a very good approximation in the infrared region) leads to an effective “image” of the opposite sign within the metal surface for any charge associated with the adsorbed species. As a result, a metal-surface selection rule (MSSR) limits the number of observable modes of vibration of the adsorbed molecule in the sense that only these modes associated with vibrational dipole changes perpendicular to the surface can be observed in RAIRS, as established by the pioneering work of Francis and Ellison (14) and Greenler (92).Parallel dipole changes are canceled out by the opposite ones within the image. The perpendicular dipole changes are enhanced by a factor of about 2 (Fig. 2). For a vibrational dipole of an adsorbed molecule that occurs at an angle to the surface, only the perpendicular component provides intensity. In general, it can be shown that it is only the vibrations of the completely symmetrical modes of the surface complex-i.e., those symmetrical with respect to all symmetry elements- that can be observed. The same MSSR applies to dipolar losses in VEELS which are dominant in on-specular spectra. We shall reserve the abbreviation MSSR solely for this dipolar-based selection rule. Features in VEEL spectra derived from the impact mechanism became more prominent in off-specular directions where the dipole-induced contributions are radically attenuated by the operation of the MSSR. However, impact-excited intensity can contribute to on-specular spectra in the form of additional strengths

5; J. a

b

C

FIG.2. The origins of the metal-surface selection role (MSSR) involving the virtual images from charges in the vicinity of the surface of a perfect metal.

16

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

for the MSSR-allowed modes, or through new impact-only features from noncompletely-symmetric modes. For the latter there are some supplementary selection rules that apply only in the on-specular direction. These are that vibrations antisymmetric to (i) a symmetry plane that coincides with the plane of incidence, (ii) a symmetry plane at right angles to the plane of incidence, or (iii) a 2-fold axis (this is necessarily perpendicular to the surface) are forbidden on-specular (21). (i) is a strict selection rule, and (ii) and (iii) are approximate rules that depend principally on the ratio of the magnitude of the vibration quantum to the energy of the incident electrons. (i) and (ii) have impact contributions effects to on-specular spectra which can depend on the direction of the plane of incidence with respect to the planes of symmetry associated with the adsorption complexes. In general, impact contributions to spectra also depend on the energies of the incident electron beam and tend to be greater for higher beam energies (21). These factors can contribute to differing relative intensities of impact-excited features in VEEL spectra from the same adsorbate/ metal system as reported from different laboratories. Differences to be discussed below between type I and type I‘ spectra from adsorbed ethene may, for example, arise for such experimental reasons rather than from differences in the adsorbed species present. The dipolar MSSR applies strictly to a flat metal surface. However, the consideration by Pearce and Sheppard (93) that adsorbed layers are typically a few angstroms (tenths of nanometers) thick in relation to the diameters of larger metal particles in catalysts (up to tens of nanometers) led to the consideration that the MSSR could have substantial effects on the intensities of infrared absorptions from adsorbed species on metal catalysts with large particles. It has been estimated that parallel modes of vibration will have their infrared absorption bands substantially attenuated at metal particle diameters of greater than 2 nm (94). This is proving to be a very important consideration in the interpretation of the infrared spectra from adsorbed hydrocarbon species on metal catalysts (20, 95, 96) and has recently become widely accepted as valid (52, 54, 57, 62, 97). The theoretical effect of the MSSR is illustrated in Fig. 3 specifically for the four vCH bond-stretching modes of a di-a MZ(H2CCH2) adsorbed species with an assumed coplanar skeleton and the symmetry C2”(19). Only the single vCH mode in which all the CH bonds expand or contract in unison is seen to give a dipole change perpendicular to the surface. For the reduced symmetry of C, associated with a nonplanar skeleton, one of the modes involving asymmetrical stretching of the CH bonds in the CH2 group would then additionally become active, with an intensity dependent on the degree of distortion from C2#,. For the ethylidyne 71 or (no)surface species (structures 2 and 3 in Fig. 4, Section IV.C), once again only the single completely symmetrical mode will be active in the vCH bond-stretching region.

17

VIBRATIONAL SPECTRA OF HYDROCARBONS

\

5

FIG.3. The application of the MSSR to the vCH bond-stretching modes of an assumed C2,,di-o adsorbed species from ethene on a metal surface (93).

As can be seen from Table 11, high metal loadings tend to lead to large metal particles for which the MSSR becomes of importance. Low loadings or special preparatior. conditions can lead to small particles that are reasonably uniform in size. In such cases the absorptions from parallel modes should be, and do, become observable. Essentially the same MSSR (that the completely symmetrical modes are active) applies to Raman spectroscopy on flat metal surfaces or large metal particles, although the relative intensities of corresponding bands are often very different and complementary to those observed in infrared spectra. However, the MSSR is not so strict in Raman spectroscopy because metals are less than perfect conductors at visible frequencies. In addition, a few extra, usually weak, features can occur in Raman spectra owing to cross terms in the polarizability, as discussed by Hexter (98)and Moscovits (27). C . GROUPVIBRATION FREQUENCIES A N D STRUCTURAL ANALYSIS: USING THE SPECTRA OF HYDROCARBON LIGANDS IN METAL-CLUSTER COMPOUNDS AS MODELSFOR SPECTRA FROM ANALOGOUS SURFACE SPECIES One of the great advantages of vibrational spectroscopy is that many hydrocarbon groupings, such as CH3, CH2, C = C , C = C , etc., have characteristic vibration frequencies/wavenumbers, many of which fall in the wavenumber ranges transmitted by oxide supports. Furthermore, for use in connection with

18

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

infrared spectra from adsorbed species on oxide supports, a very large, organized body of literature is available on such group-characteristic frequencies in the infrared and Raman spectra of organic compounds (8, 9) and, to a lesser extent, of organometallic molecules (10). When hydrocarbon groups are attached to metal atoms, substantial changes are to be expected to the wavenumers of CH, groupings directly bonded to metal. Smaller variations may also occur for groups one carbon further removed. Normally, essentially unchanged wavenumbers can be expected for groups hrther removed, although an exception to the latter expectation can occur where a bent-back carbon skeleton permits ugostic (C- H...M, hydrogen-bond-like) interactions between otherwise nonbonded CH groups and metal atoms (99). Scheme 1 and Table I11 give examples of such relationships for C I groups directly bonded to metal, in comparison with the normal hydrocarbon values (8-11, 100-108). For the metal-containing groups (M = metal), boldface type indicates those modes that would be observed on a metal surface, according to the MSSR, if the adsorbed complex assumes the most symmetrically possible relationship with respect to the metal surface site, e.g., if the CM bond of the adsorbed CH3M group is perpendicular to the metal surface. The MSSR then allows only the fully symmetric modes to be active in RAIRS or on-specular VEELS and has similar effects in the transmission infrared spectra on large metal particles. All band positions derived from the literature are quoted to the nearest 10 cm-', and larger variations can be expected for different metals. It is seen in Table Ill that metal substitution into alkane groups leads rather systematically to increases in the values of bond-stretching vCH modes, and decreases in most angle-deformation, dCH/dCH&CH3, modes. The CH3M and C-CH2M data are appropriately averaged values from the spectra of metal tetraalkyls such as (CH&3n( 100) and (CH3CH2)$n( 101), where all the ligands are the same and are expected to have little electronic influence on each other. However, a CH3M group involving, for example, the transition metal Mn with multiple CO coligands, CH3Mn(CO)5, gives virtually the same values of 2983, 2910, 1420, 1184, and 783 cm-' (109).

\

M=CH,

b

-MzCH

b

SCHEME 1. Possible C, hydrocarbon species on metal surfaces. "On metal surfaces the M atoms are usually bonded together. 'These are alternative to the upper row of structures. They are most likely to occur on d-electron-deficient metals to the left of the transition-metal periods, or on coordinatively unsaturated sites on metals to the right of these periods.

19

VIBRATIONAL SPECTRA OF HYDROCARBONS

TABLE 111 Group-Characteristic Freyuencies/Wavmumber.~ (cm I ) of Cl Hvdrocarbon Groups Direc1l.v Bonded to Metul Atoms in Comparison with Those of Analogous Unsubstifuted Hydrocarbons ~

CH, groups (CH,)C (8. '4 (CH3)M(100)"

vCH,us 2960 (s) 2990 (s)

ICH~S

BCHjas

bCH3as

2870 (ms) 2920 (s)

2900 (ms) 2780 (w)

1450 (ms) 1400 (m)

GCH,s

pCH,

1380 (s) 1050 (ms) 1180 (m) 750 (s)

CH2 groups

rCH2as

KH2s

BCHz

dCHz

CHZw

C(CHz)C (8, 9 ) C(CH2)M (101)" M(CHz)M

2920 (s) 2940 (s) 2970 (w)

2850 (ms) 2910 (s) 2925 (m)

2900 (ms) 2830 (m) 2800 (w)

1460 (m) 1420 (m) 1410 (mw)

1340 (m) 1200 (m) 960 (m)

CHzr 720 (s) 660 (s) 790 (m)

(102. 103)h.'.h

CH groups (HC)C3 (8. 9) (HC)C2M (104)" (HC)CM2 (105)hh (HC)M3 (106Fh

vCH

dCHK

2890 (m) 1340 (m) 2900 (m) 1300 (w), <1200 (s) 950 (m) 2920 (ms) 1300 (ms), 3040 (vw) 850 (s)

=CH? groups

v C H as ~

I C H s~

6CHz

C=CHz (8. 9) M=CHZ (107, /OX)','

3080 (s) 3020 (w)

2990 (ms) 2950 (m)

1415 (m) 1350 (w)

ECH groups

ICH

6CH

C=CH (8. Y) M=CH (107)'

3300 (s) 3250 (?)

635 (s) 630 (w)

I C=

CHzr

CHzw

LC=X

1280 (mw) 890 ( s ) 1650 (m) 450 (m) -600 (m) 620 (m)

x

2120 (m) 680 (mw)

Note: Boldface indicates modes allowed on a metal surface by the MSSR if the hydrocarbon group is symmetrically bonded to the surface. " M = Sn. M = 0 s . ' M = Fe. "M = I. 'M = Co. 'M = Cu. 'Two modes, doubly degenerate for (HC)C, and (HC)M,. In these compounds the metal atoms are bonded together as expected on metal surfaces. Wavenumber variations can be caused by ring strain as well as by the electronic effects of the M substituents.

In a recent survey of v C 0 values observed from transition-metal carbonyls, it was shown that CO is a relatively "neutral" coligand compared with the high electron-donating abilities of phosphorus alkyls/aryls or cyclopentadienes or with the strong electron-withdrawing capabilities of halide ligands ( I10). Hence the latter two groups of coligands are suspect when looking for the spectra of model compounds for comparison with surface species. The larger organic coligands also give complicated spectra which frequently obscure some of the bands from the hydrocarbon groupings being characterized. Fortunately, not only is CO a

20

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

relatively “neutral” coligand, but it also has absorptions that do not overlap those of hydrocarbon groups except below ca. 650 cm- where vMC or 6MCO modes can give strong bands (10). The infrared spectra of model compounds that fulfill these criteria are listed for C2 ligands in Table IV (100, 101, 105, 111-129). There we give IUPAC notations for the bonding of the ligands and our preferred alternatives which, together with ligand names, indicate more clearly the possible metallligand bonding patterns. The symbol (no)is chosen for the metallocyclopropane structure and its equivalent based on an ethyne ligand. In some well established cases, the ligand names suffice. The structures of the various model compounds are indicated in Fig. 4. In the great majority of cases, the experimental data have been obtained from model metal-cluster compounds whose structures

’,

TABLE IV Organometallic Compoitnds Used as Spectroscopic Models for Cr-Hvdrocarbon Surface Species

Formula“.h

IUPAC prefix

Prefix used

Ligand name

Model compound

ethyl

100, 101

ethene ethene ethene ethylidene

111-113 113 114 105

115

vinyl vinyl ethylidyne Mz(CCH3) 10 M=(CCH,)

ethylidyne ethylidyne

11 (HC=CH)M‘

ethyne ethyne

9

12 M(CH=CH)

References

116, 117

I! 118,IIY 117 120

121 122 123 117, 124 125 126 127 117 128 IlblIX 128

16 M,(HC=CH) 17 Mz(C=CHz)

ethyne ethyne ethyne ethyne vinylidene

I8 M*(C=CHZ)M’

vinylidene

19 M=C=CH2

vinylidene

123

20 M(C=CH) 21 M(C=CH)M;

ethynyl ethynyl

129 117. I18

13 M2(HC=CH)M’ 14 (HC=CH)M;

15 Mz(HCCH)M;

M

=

u-bonded metal atom. * M‘ = n-bonded metal atom. ‘Cp

=

~5-cyclopentadienyl

21

VIBRATIONAL SPECTRA OF HYDROCARBONS

H3 "2C

H2

I

M 1.

-CH2

\ MI

3.

H2c -CH2

/

M-

\ 4.

M

H3

C H

\I M-/ c \ M 5.

P3 I

M

M-

11.

12

15.

16

cH2

F2

H

A

i

9. HC=C

H

I : \ \ Mr,/

M-M

13.

I1

M-M

1I.

II

M 19.

il C

I

M

20

FIG.4. Structural diagrams for C2 hydrocarbon ligands on metal clusters that are available as model compounds for chemisorbed surface species, numbered in accordance with the entries in Table IV (M = metal atom).

22

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

cm-’

3200

’

3iOO

’

2800 2000

1600

1200

2ioo 2000

1600

1200

800

800

400

400

cm-1

FIG.5 . A summary of the infrared absorption bands exhibited by hydrocarbon ligands on metal atoms in various model compounds. Surface species on metals may give absorptions varying by cu. 50 cm from the band positions in the model-compound spectra in the “fingerprint” region below 1400cm- I. The patterns of band-positions and intensities are significant. H indicates MSSR-allowed modes for an analogous species on a flat surface when the adsorbed species is on a site of high symmetry; (--) indicates other absorptions that may occur for adsorption on less symmetrical sites or on small metal particles. vs-very strong; s-strong; ms-medium strong; m-medium; mw-medium weak; w-weak.

’

have been determined by X-ray crystallography. In a few cases, the hydrocarbon groupings are attached to bare metal atoms, as studied spectroscopically using low-temperature matrix isolation. Figure 5 collects together information on bands of medium (m) or strong (s) intensity expected on metal surfaces for most of the possible types of C1and C2 hydrocarbon ligands. Relationships between the latter structures are set out systematically in Scheme 2, (M = a-bonded metal atom; M’ = n-bonding to metal). In this scheme the parent adsorbate hydrocarbons are indicated by solid rectangular outlines, and dashed rectangles encompass those surface species that can be derived from the parent without breaking the CH bonds. In Fig. 5 , the symbol H indicates a probable wavenumber range associated with the MSSR-allowed completely symmetrical modes, assuming that the surface complex allows the adsorbed hydrocarbon group to retain its full symmetry. The symbol (- -) indicates other prominent, usually strong, absorptions of the ligand in the metal-cluster compound-absorptions that might also show up with

23

VIBRATIONAL SPECTRA OF HYDROCARBONS

'0

1200

800

400

24

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

\

c

M=CH(CH,) -M=CCH3-

e

M=C=CHz

e

- M=C=CHM

e

SCHEME 2. Schematic relationships between CZ hydrocarbon adsorbates and the surface species that could be derived from them. 'Possible relationships are indicated between surface species that involve not more than one additionhubtraction of H and/or M atoms. 'Strong bonding of these species to metal atoms could give rise to structures approximating to metallocyclopropanes or metallocyclopropenes, respectively. 'These structures with C=C or C= C groupings frequently occur on surfaces with additional IT bonds to further metal atoms, M'; in other cases, two additional CM bonds may replace a CC double bond. "On surfaces the metal atoms are usually bonded to each other so that the analogous molecular cluster compounds would be metallocyclopropanes or metallocyclobutanes. etc. 'See footnote to Scheme 1. 'The dashed rectangles indicate surface species that involve no CH bond breaking on adsorption of the parent hydrocarbon. These are most likely to be present under low-temperature adsorption conditions.

limited intensity under the MSSR if the hydrocarbon adsorbate occupies a less symmetrical metal site. The wavenumber ranges indicated normally cover at least 20 cm- on either side of the value found in the spectrum of the model compound in order to allow for metal variability and for the consideration that other metal atoms in the surface replace CO coligands in the cluster compounds. Intensity patterns are expected to be qualitatively similar between ligand and surface species after allowance has been made for MSSR-weakening of formally allowed absorption bands that correspond to vibrations with dipole changes at small angles with respect to the surface. The relative intensities across the complete spectra relate to expectations for RAIR rather than VEEL surface spectra (see our earlier comment in Section 1I.C). A combination of positionplus-intensity criteria is frequently of help in assessing the likely equivalence of structures of ligands and surface species. Muetterties er al. (130) published an extensive survey relating the properties of metal clusters and their ligands to analogous properties of metal surfaces and adsorbed species. Moskovits (131) has expressed caution, particularly in relating the reactivities of cluster ligands and the surface equivalents; and more recently,

'

VIBRATIONAL SPECTRA OF HYDROCARBONS

25

Johnson et al. (132) have also discussed the validity or otherwise of such analogies. Spectra from the same type of surface species that give similar vibrational spectroscopic patterns can show readily measurable frequency variations that may be much more significant in reactivity terms. However, it seems clear that in many cases the overall spectral patterns of cluster-compound ligands do provide useful and reliable “fingerprints” for identifying the structures of surface species. A number of spectral regularities can be noted from Fig. 5. In the vCH bondstretching region, it normally remains possible to distinguish between vCH absorptions of metal-substituted alkanes (3000-2800 cm- or, with agostic interactions, below 2800 cm-I), alkenes (3120-2960 cm-I, with vCH of =CH groups somewhat lowered relative to the normal hydrocarbon value), or alkynes (3320-3 100 cm-I). For C=C groups with only ri-type metal substitution, i.e., without additional n-bonding, we observe somewhat lowered vC=C wavenumbers ( 16501550 cm- I); with n-bonding only, 1600-1460 cm-I; and with a combination of ri- and n-bonding, < 1500 cm- I . As the number of bonds to surface metal atoms increases, “vCC” systematically decreases ( 1550-1 100 cm I). The so-called vCC absorptions, usually identified by a comparison of spectra from C2H, and 12 CzD, surface species, or from C2H, and I3C2Hn, usually involve coupling of the CC group vibration to a considerable extent with 6CH, 6CH2 group motions of the same symmetry. Thus, even in the case of ethene itself, considerable coupling occurs between the vC=C and 6CH2 scissors vibrations of the same symmetry ( I l l , 112). In order to assess the CC bond orders in surface n (H2CCH2)species, Stuve and Madix (133) have proposed a sum of the proportionate shifts of the analogous two absorption bands relative to the values of the parent ethene molecules, C2H4 and C2D4. They have also extended this analysis in the direction of di-a M2(H2CCH2)species. In the latter, more extreme situation, however, there would be further coupling between vCC and CH2 wagging group motions. For this reason we prefer to use the wavenumber products of these three completely symmetrical modes as a means of characterizing the CC bond order (134); moreover, we find that it can be usefully extended to include metallocyclopropane-type structures as intermediate cases (1 13). A bond order between 1 and 2 can be regarded as being principally related to the extent of back-donation from &orbitals of the metal atom into the otherwise vacant n* antibonding orbital of a C=C group. A CM a-bond in MCFCH species leaves vC=C essentially unchanged at ca. 2050 cm-I. One n-bond to metal can lower vC=C to below 1900 cm- I ; and two n-bonds, as in M(C=CH)Mi, may lower vCC to below 1600 cm-’. yCH modes of alkene or alkyne surface groups (out-of-plane CH deformation for substituted C=C) give strong to very strong absorptions in the infrared ~

26

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

spectra of the model compounds. However, in a number of the more symmetrical cases, these are weakened or forbidden on the metal surface through the operation of the MSSR. The ligand-model vibrational spectroscopy approach has contributed strongly to fairly reliable identifications on metal surfaces of C2 species of the types 1, 2 (ethene type I1 spectra) ( I 7), 3 (ethene type I’ spectra), 4 (ethene type I spectra), 8, and 13 (ethyne type B spectra); as well as to possible identifications of types 5, 7, 15 (ethyne type A spectra), 16, and 20. Approximate band positions and associated intensity distributions in the spectra from normal and perdeutero species should be considered together ( I 7). The correspondence of the infrared spectrum from 4 with type I spectra is less satisfactory for the C2D4 ligand than in most other cases. However an extra structural variable in this case is the degree of nonplanarity of the cyclic C2M2 skeleton, which may differ between the model compound and the surface species. Normally, the spectra become much more complex, and hence less definitive, for species having more than two carbon atoms. Nevertheless, some such spectra can be very usefully interpreted on the basis of assumed surface structures analogous to those identified in the corresponding C2 case. These examples will be discussed in the relevant later sections of this review, as will the more definitive spectra of highly symmetrical surface species such as benzene and cyclopentadiene.

V. Other Experimental Methods for Investigating the Structures of Chemisorbed Hydrocarbons on Metals We summarize below the types of information obtainable from the substantial number of physical methods, alternative to vibrational spectroscopy, available for investigating the structures of species on metal surfaces. A minority of these, to which we give greater attention, are applicable to work on finely divided metal catalysts. Many of the others provide direction-dependent information and are most effective when applied to adsorption studies on flat single-crystal surfaces. A. TEMPERATURE-PROGRAMMED DESORPTION (TPD) The TPD experimental technique is alternatively, but less suitably, termed thermal desorption spectroscopy (TDS). It is a very useful complement to vibrational spectroscopy and can be applied to adsorption on single-crystal or finely divided metal surfaces. TPD involves the dynamic analysis, usually by mass spectrometry, of the gases desorbed from the surface as the temperature is raised at a uniform rate, starting from a known state of adsorption. In addition to

VIBRATIONAL SPECTRA OF HYDROCARBONS

27

providing an analysis of the desorbed species-e.g., C2H4, C2H6,and Hz-from an initial dose of adsorbed ethene, the integrated result also provides a profile of the C,H,. composition of the remaining adsorbed layer as a function of temperature. This is very useful information in interpreting vibration spectra, starting from an initial state, and taken to progressively higher “annealing” temperatures. Although the timescales involved in the spectroscopic methods are necessarily longer than those applied at a given temperature during TPD-less of a kinetic and more of an equilibrium situation will be monitored by surface spectroscopy-the two techniques taken in parallel can be mutually very informative (135, 136).

B.

NUCLEARMAGNETIC RESONANCE (NMR)

NMR is a widely used and important technique for molecular structure determination as applied to bulk materials, where it competes, often advantageously, with vibrational spectroscopy. However, a lack of sensitivity has limited its application to the study of adsorption on high-area finely divided surfaces. Also, certain metals with bulk magnetic properties-e.g., Fe, Co, and Ni (but not the other group VIll transition m e t a l s t c a n n o t be studied by the technique as their magnetism causes very broad and weak resonances from adsorbed species. Physically adsorbed molecules on a finely divided oxide have considerable mobility and give NMR spectra that are just somewhat broadened versions of those of the corresponding liquids. However, chemisorbed molecules are usually much more limited in mobility, and great broadening of resonances can occur because of incompletely canceled magnetic anisotropies. Therefore, as Gay (137) has demonstrated, it is normally necessary to use the magic-anglespinning and cross-polarization techniques of high-resolution solid-state NMR as applied (in the hydrocarbon-adsorption context) to I3C nuclei. The positions of resolved I3C resonances can then be compared with those from analogous ligands on metal-cluster compounds. Although for the same structural type of ligand in metal cluster compounds, the 13C chemical shifts can be rather disadvantageously metal-dependent (117)-a situation that could be exacerbated from Knight-shift contributions on metal particles (/-?&)-this NMR method is clearly capable of further valuable development. An alternative NMR method has been to use pulsed spin-echo techniques with hydrocarbon adsorbates highly enriched in I3C. Those surface species that have 13C-’3C bonds give rise to modulated free-induction decays (FIDs), which depend on direct I3C/l3C dipolar coupling. The analysis of the modulation provides information about the C-C distances present. CI species or C2 species with 13C-’2C bonds give unmodulated FID components. Furthermore, double

28

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

resonance at the ' H frequency causes a broadening, and hence virtual disappearance, of 13C resonances involving CH groups. The intensity of the overall FID is thereby apportioned between carbon atoms with or without attached hydrogens. Slichter pioneered this technique and has also written an excellent review of the NMR contributions to the study of chemisorbed species (138).

C. THEPHOTOELECTRON SPECTROSCOPIES (UPS A N D XPS) The analysis of the energies of photoelectrons generated by monochromatic incident radiation provides information about the energy levels of the orbitals from which the electrons are ejected. Ultraviolet photoelectron spectroscopy (UPS) provides information about the outer valence orbitals, whereas the X-ray version (XPS) monitors the energies of core orbitals as a fbnction of their chemical environments (135, 136). Although techniques have been applied mainly to adsorption studies on metal single crystals, there seems to be no reason why the XPS method cannot be used to study finely divided metal catalysts under vacuum conditions. The UV-generated spectra have proved less able to provide distinctions between alternative adsorbed species than are vibrational spectra because much more spectral overlapping can occur when more than one adsorbed species is present. XPS can distinguish principally between different oxidationhybridization states of carbon atoms, but the resonances are usually too broad to allow more detailed distinctions from hydrocarbons (135). However, the XPS method has the advantage over vibrational spectroscopy that it can be quantified for the principal types of carbon environments. Angle-resolved versions of the photoelectron techniques, such as ARUPS, can be used to study the symmetries of adsorbed complexes on flat surfaces (135, 136).

D. X-RAYABSORPTION (EXAFS

AND

NEXAFSmANES)

These spectra can be explored using continuum X-ray sources such as synchrotron radiation. As different chemical elements have different absorption energies, e.g., in transitions involving their K-levels, chemical analysis is possible. The continuum following the onset of an absorption edge (related to photoelectron emission with increasing energies) exhibits structure that is dependent on the chemical environment. This phenomena is termed extended X-ray absorption fine structure or EXAFS, which is caused by diffraction processes involving the photoelectrons of increasing energy. The structure can be Fourier-analyzed to provide information about the separation distances and scattering powers of atoms surrounding the element in question. Unfortunately, the interpretation of the fine structure has proved to be particularly difficult in

VIBRATIONAL SPECTRA OF HYDROCARBONS

29

the case of the lighter elements such as C, N, and 0, either in a bulk (EXAFS) or surface (SEXAFS) context. However, the X-ray spectra from the latter elements exhibit other structural features that occur just before the onset of the principal absorption edge. This near-edge X-ray absorption j n e structure (NEXAFS), alternatively described as X-ray absorption near-edge structure (XANES), is caused by the excitation of the electrons from the core orbitals to vacant antibonding 7c* and o* orbitals of the adsorbed species. These have different polarization directions and have been very effectively used to determine whether CC bonds of adsorbed hydrocarbons are parallel or perpendicular to the metal surface (136). This directional information would, of course, be lost from a finely divided metal catalyst, but the relative strengths and positions of the n*- or o*-related features might still provide some information about the general nature of the adsorbed species. To date, this possibility does not seem to have been explored, probably because the usual coexistence of several different adsorbed species on metal particles could lead to overlap and hence difficulties of interpretation. E. LOW-ENERGY ELECTRON DIFFRACTION (LEED) Because electrons are much less capable of penetrating metals than are X-rays of similar wavelength, this technique has played a very important role in studying adsorption on such surfaces. The symmetry aspects and dimensions of the diffraction patterns observed at normal incidence can provide information about the spacing and coverage of adsorbates that commonly occur in regular arrays (135, 136, 139). The arrays themselves frequently vary qualitatively in nature as a function of overall surface coverage. Vibrational spectra are often more successful in indicating the nature of, and site occupied by, the adsorbed specie(s) in question, but are best studied at coverages that give good diffraction patterns. The separation of an adsorbed layer from the metal surface can also be determined by the energy dependence of diffraction features (135, 139). LEED analysis is at its best with highly ordered adsorbed layers; however, useful information can still be obtained in poorly ordered situations, which seem to be the rule when hydrocarbons are adsorbed. F. PHOTOELECTRON DIFFRACTION (PED)

This recently developed technique (85, 86, 136) has the advantage that it can be readily applied to disordered surface layers that do not give sharp LEED patterns. What is measured is the diffraction of locally generated photoelectrons (e.g., from hydrocarbons, the 1s electrons from carbon atoms of an adsorbed species). With high-energy photoelectrons, the strongest diffraction features are caused by atoms that are aligned with the emitting atom. Hence, information can

30

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

be obtained, for example, about the direction of C M bonds with respect to the surface. Using synchrotron radiation, it is also possible to measure the diffraction intensity as a function of photoelectron energy, this provides information about the atomic separations, such as bond lengths, involved. In favorable cases, this technique allows the structures and adsorption sites of a surface species to be deduced in a very direct manner. The method merits hrther development and application. G . SCANNING TUNNELING MICROSCOPY (STM) The STM technique permits adsorbed species to be studied in situ at near atomic resolution (136),as demonstrated in a recent study of ethene adsorbed on Pt( 1 1 1) between 160 and 700 K ( 140).

VI.

Experimental Vibrational Spectroscopic Results on Oxide-Supported Metal Catalysts Classified by Alkene Adsorbate A. GENERALSTRATEGY

In this section we shall consider the results recorded in the literature that pertain to the structures of the adsorbed species. Kinetic or catalytic aspects, as could be relevant to hydrogenation, hydrogenolysis, or metathesis processes, will be treated in Part 11. Spectra of the much-investigated alkenes are discussed in detail in Part I. The spectra of the other principal types of hydrocarbon adsorbates, viz. alkynes, alkanes, cycloalkanes, and aromatics, will be analyzed in Part 11. Most results are available for the type-molecules ethene, ethyne, ethane, and benzene as well as for the metals, Pt, Pd, Ni, Rh, and Ru. Each adsorbate section commences with a summary of the surface species that have been identified or proposed on the basis of work on the simplified single-crystal systems. There may be sites, and associated adsorbed species, on the more complex and partially ordered surfaces of metal particles that are not covered by the collected results from single crystals. However, it is anticipated that many, and often the principal, surface species will be identified by this means. An extensive review of work on hydrocarbons adsorbed on metal single crystals was published in 1989 (I 7), and this review is updated here. The summary of the single-crystal results is followed by the presentation and illustration (with permission from the authors and publishers) of significant results from the literature on finely divided metals, mostly infrared spectra from adsorption on oxide-supported metal catalysts. Particular emphasis is given to results obtained at room temperature or below, where the structures present are likely to be the best-defined.

VIBRATIONAL SPECTRA OF HYDROCARBONS

31

Spectra obtained by hydrogenation will also be reviewed, although in that area relatively few results are available from single-crystal systems because the prevalent VEELS method requires the use of high vacuum over the sample. This restriction does not apply to RAIRS studies. B. 1.

ETHENE

Ethene on Platinum

a. Pr, Single-Crystal Work. Spectroscopic results are available in most detail for the (C2HJsingle-crystal Pt) system, and the summary of the principal conclusions that follows will find relevance in relation to work with most other metals. The first VEELS study of the adsorption of hydrocarbons on a metal single crystal was performed by Ibach, Hopster, and Sexton (12) and included work on C2H4/Pt(11 I). Ibach and his colleagues (141, 91) concluded that at low temperatures, ethene adsorbed in the nondissociative di-a form on Pt(1l I), a conclusion that has been generally accepted (17, 114). At room temperature a different spectrum was obtained and assigned to an ethylidene surface species. After a great deal of subsequent experimentation and discussion involving TPD (142, 91), LEED (143), comparison with the spectrum of a cluster compound ( ] I ) , further VEELS (144, 91), and latterly RAIRS work (145, 146), this species was structurally reassigned at ethylidyne (CH3C)M3. Recently, Masel and colleagues have published a valuable series of VEEL spectra from ethene on Pt(100) [(l X 1 ) and (5 X 30)] (147), Pt(ll0) [(I X 1) and (2 X l)] (148-152), and Pt(210) (151-154). This work led to the first identification of a n-adsorbed species on a clean single-crystal surface of Pt. In addition to work on clean Pt surfaces, other studies have been made of ethene adsorption on surfaces with preadsorbed H [Pt(llO) and Pt(210)l (152, 1.53, K [ P t ( l l l ) ] (155, 156), 0 [Pt(l 1 I)] (91), and Bi [Pt(lI I)] (157). Bismuth appeared to act simply as a site blocker. Preadsorption of H or K led at low temperatures to a spectrum from a new weakly bound n species here designated n*. Oxygen (91) and potassium (156) at room temperature are thought to give rise to ethylidene (105) as an intermediate in the formation of ethylidyne from di-a adsorbed ethene. In Table V we summarize the spectra of the principal adsorbed species on Pt surface planes and include more precise band positions for the di-a (158) and ethylidyne species (159) from RAIR spectra. The spectra are listed in the order of increasing temperatures for the adsorbed species. As we shall see, the patterns of spectra for the n- and n*-species and for ethylidyne vary little from metal to metal, but that of the di-a species is more variable, with other metals giving an increased wavenumber of cu. 1140 cm- for the strongest feature replacing 1040/980 cm- for Pt (17). The other surface species, except for “CCH” are rarely found on other metals.

’

’

32

NORMAN SHEPPARD AND CARLOS DE LA CRUZ TABLE V On-Specular VEEL Spectra Associated with the Structures of Adsorbed Species ,from Ethene Adsorbed on Single-Crystal PI Surfaces"' Surface species

n-Species

Wavenumbers (cm and intensities

I)

Surfaces

3080-3020 (m), 1530-1500 (w), Pt(210) Pt(l10) ( I x I ) and (2 X I )

Temperature range (K) References

< 100-300 148-155. I13

n*-Species (weakly bound)

3060-3040 (m), 1620 (w), 1360 (m). 960 (s)

K/Pt(I I I ) H/Pt( I 10)

ca. 100

Di-o species' MdCzH4)

cu. 2940 (ms), ca. 1420 (m), 1040/980 (s), 460 (ms)

All Pt surfaces

Ethylidyne'

cu. 2890 (ms), ca. 1340 (s), ca. I130 (s), ca. 450 ( m )

Pt(I I I ) and Pt(100) (5 x 20)

260-450 91, 11

(C H3C )M3

Vinylidene M2(C=CH2)?

[ca. 2940 (m)], 1585 (m), 1040 (m) [440 (m)]

Pt(100) ( I x I )

270-350

(Di-o/n) M2(CH=CH)M'?

ca. 2980 (m)], I130 (vs), ca. 800 (mw), [450 (m)]

Pt(100)(1 x I )

300-450 147. 124.

Ethylidene (CHICHM

ca. 2970 (m), 1420 (ms, bd), 1040 (w), 480 (s)

O/Pt(lIl)

3 00

Ethylylidyne? M3(CCH2)M

ca. 2940 (s), 1420 (ms). 980 (ms). ca. 470 (m)

Pt(210) Pt( 110)

"CCH"?

cu. 3050 (mw), ca. 860 (s)

Pt(l10) (2 x I )

c-c

ca. 2200 (m)

Pt(I 10) Pt(2 10)

<10&300

152, 155. 156 91, 147, 157. 114

147, 117

117

K/Pt( 1 I I )

91, 156.

I05 290-350

148. 154

450-500

151

>600

151

"Where more than one spectrum of the same system is recorded in the literature, estimated average values are recorded in this table. *In the majority of cases, data for adsorbed C2D4 are also available. "More precise band positions as recorded by RAIRS are 2903, 1048, and 991 cm- I (158). 'More precise band positions recorded by RAIRS are 2884, 2795, 1339, I 1 18 (145. 146).

b. Finely Divided Pt-Low- Temperuture Spectra. Low-temperature spectra from ethene on oxide-supported Pt collected from the literature are shown in Fig. 6. The original spectrum for ethene adsorbed on Pt/Si02 at 128 K by Morrow and Sheppard (32) is shown in Fig. 6A. It covered on1 the vCH region and showed a single dominant broad absorption at 2907 cm and a considerable weaker one at ca. 3020 cm-'. A more recent version of this spectrum by De La Cruz and Sheppard (96) is illustrated in Fig. 6B. Taken at 189 K under dynamic conditions of warming up from lower temperatures, it includes the lower wavenumber region down to the SiOz blackout near 1300 cm-'. The following spectrum at 187 K by Soma (47, 160) on Pt/A1203 (Fig. 6C) was the first to show the advantage of using the better transmission of A1203 down to

-7

33

VIBRATIONAL SPECTRA OF HYDROCARBONS cm-’ 3000

2800

1600

I

3000

2800

1400

I

1600

I

1400

1200

I

‘

Ptl

1200

cm-’

FIG. 6. Infrared spectra from ethene adsorbed on metal-particle samples of Pt at low temperatures: (A) WSi02 at 128 K (32); (B) WSi02 at 195 K (96);(C) Pt/AI203 at 187 K (160); (D) PtlA1203 at 180 K, reprinted with permission from (57), copyright 1988 American Chemical Society.

’.

1 100 cm- A fourth spectrum (Fig. 6D) recently measured at 180 K on Pt/A1203 by Mohsin, Trenary and Robota (57) is in good general agreement with Soma’s work. Collectively, these spectra very clearly show the coexistence of the 7~ complex, as originally identified spectroscopically by Prentice, Lesiunas, and Sheppard (20) and independently by Soma (160), and of the di-a complex as identified by De La Cruz and Sheppard (95) and independently by Mohsin, Trenary, and Robota (57). Absorptions from small amounts of ethylidyne are indicated by asterisks in Figs. 6C and 6D (see below). Indirect experimental evidence implying the presence of a x-bond-containing complexes from ethene on metal catalysts was first produced by Shopov and Palazov and their colleagues (161-163). They pointed out that coadsorption of ethene and CO led to a low-wavenumber shift of the vC0 absorption and interpreted this in terms of the effects of n-donation by ethene. Subsequently, they suggested that the complex in question was a (di-ah) ethyne surface

34

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

species ( I 63) so as to account for the lack of prominent v=CH absorptions from a n complex above 3000 cm-I. It is now appreciated that the operation of the MSSR can account for the weakness of the v=CH absorptions. However, experiments involving kinetic studies of the H2/C2H4 reaction over Pt/SiO2 suggest that alkylidyne species can also play a role in shifting vC0 to lower wavenumbers ( I 64). In each of the spectra illustrated in Fig. 6, absorptions at ca. 3020 (v=CH2 s) and 1500 cm-' (vC=C/GCH2), together with that at 1200 cm-' (GCH~/VC=C) as observed on Pt/A1203, represent the expected completely symmetrical modes of a n-complex that occur above the alumina cutoff at ca. 1 100 cm- I . The relative intensities of the 1500- and 1200-cm-' absorptions suggest that this surface species has some metallocyclopropane (na) character. However, we shall retain the n symbolism here as it is widely used in the literature for such spectra. On the larger metal particles (5-1 5 nm in diameter) applicable to Figs. 6A and 68, these are the only modes expected on flat surfaces according to the MSSR. The spectra from the small-particle catalysts, Figs. 6C and 6D (ca. 1.9 nm for Figs. 6D; see Table II), could show relaxations of the MSSR. Additional absorptions near 3070 cm-l (v=CH2 as), 2995 (v=CH2 s), and a possible contribution to the band near 1425 cm-l (6=CH2) (57, 58) can hence be attributed to the less symmetrical modes of the 72 complex. Absorptions near 2910 and 1420 cm- have been assigned to the di-0 complex (96, 5 3 , probably adsorbed on (1 1 1) facets because of its transformation at higher temperatures into ethylidyne, (CH3C)M3 (see below). Other modes of the di-0 complex on ( I 1 I ) (Table V) are not observed because of the alumina cutoff. It is of interest that, unlike the case of the n-complex, no additional absorptions for the di-a complex are observed in the small-particle spectra due to relaxation of the MSSR. However, the n complexes may well be adsorbed on step or comer sites as this species is only found on the corrugated ( 1 10) and ( 2 10) surfaces on single-crystal Pt ( I 47-I 54). For the di-a species, probably on flat ( I 1 1 ) terraces, the operation of the MSSR would be expected to be more pronounced for a given mean particle size. Figure 6B shows an additional absorption near 2920 c m - ' which is not subsequently involved in conversion to ethylidyne. This probably represents the presence of a di-o type of complex on non-(1 1 1) sites. Although it could alternatively be associated with the dissociatively adsorbed ethylylidyne species, M3(CCH2)M (Table V), which also contains M(CCH2) groups and possibly occurs on (1 10) and (210) planes (150,154, a nondissociative di-a type of structure seems more likely as the species is stable at the low temperature of 189 K (96). We shall therefore refer to this species henceforth as di-a*. c. Finely Divided Pt-Spectra at Room Temperature and Above. Figure 7 collects together, in historical sequence, the now considerable number of spectra

VIBRATIONAL SPECTRA OF HYDROCARBONS

35

in the literature for ethene adsorbed at room temperature on PtISiO2 or Pt/A1203 catalysts. Derived from work using a very wide range of catalyst preparations, these spectra have a gratifying number of features in common. We have omitted only some earlier pioneering spectra of ethene on Pt (and other metals) catalysts supported on porous silica glass (69, 165). Although resembling those illustrated here, these spectra are of relatively poor quality because the thickness of the tubes of porous glass that were used gave rise to highly curved backgrounds even in the vCH region. De la Cruz and Sheppard (96) and independently Mohsin, Trenary, and Robota (57) studied the changes in the low-temperature spectrum (Figs. 6B and 6D) as the sample was raised to room temperature. They found that the ca. 29 10 cm- absorption decreased in proportion to the increases in intensity of absorptions at ca. 2885 cm- and 1340 cm- '. Analogous spectral changes with temperature, under low resolution, had been earlier recorded on Pt( 1 1 1 ) by Ibach et al. (141) using VEELS. After the reinterpretation of Ibach's room-temperature spectrum from ethylidene to ethylidyne (I I, 91) with the help of RAIRS spectra (145, 146), Sheppard et al. (166, 95, 167) and independently Yates et al. (97, 50, 168) identified the 2885- and 1340-cm-' absorptions as arising from ethylidyne groups adsorbed on (111) facets of the metal particles. Morrow and Sheppard (32) in the vCH region, and Soma in the 1700-1 100 cm-' region ( 4 9 , had earlier studied the growth of the ca. 2885and 1340-cm-' bands as a hnction of temperature without knowing (at the time) the identity of the species giving the new absorptions. Soma had interpreted the ca. 1500-, 1420-, and 1200-cm-' bands as originating in the n-complex. These all diminished in intensity as the new bands grew; hence it was concluded that the n-complex had been transformed into the new species. However, in the single vCH spectrum recorded at 187 K (160) a band at 2908 cm-' is clearly seen (Fig. 6C) that is now attributed by Sheppard and by Trenary to 6CH2 s of the di-u species; the ca. 1420-cm- I band could be its 6CH2 companion. As the n-complex spectrum independently exhibits a high temperature sensitivity (see below, 146), it is concluded that Soma's spectra are also best interpreted in terms of the di-o to ethylidyne transformation. With the help of the above considerations and the single-crystal data of Table V, it is now possible to relate all the prominent features in the room-temperature spectra (Fig. 7) to surface species as follows:

'

'

n-Complex (as at low temperatures): 3015 (vCH2s), 1500 (vC=CIGCH2), and ca. 1200 cm-l (GCH,IvC=C); plus on small-particle catalysts: ca. 3075 (vCH2as), 2995 cm-' (vCH2 s), and possibly 1425 cm-' (6CH2). Ethylidyne: 2885 (vCH3s), 2800 (2 X 6CH3 as), 1340 (6CH3 s), and 1125 cm- (vC-C), all allowed according to the MSSR on large-particle

'

36

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

2800

3000

A

A;j 1

1600

“-’

1400

1

I

I

1200 I

I

Pt

N

ri

N

B f - t 1

I

Pt

m m N m

FIG. 7. Infrared spectra from ethene on Pt near room temperature: ( A ) Pt/SiO: (.<.?); (13) PtiAlzOl ( 4 7 ) ;(C) Pt/AI2O3(171); (D) WSi02 (EUROPT-I) (63); (E) Pt/SiOz (167);( F ) P1/A1203. reprinted with permission from (50), copyright 1987 American Chemical Society; ( G ) Pt/Si02 (YO); (1-1) Pt/AI2O3, reprinted with permission from (58). copyright 1991 American Chemical Society; ( I ) Pt/SiOz (EUROPT-I). DRIFT spectrum (62); ( J ) Pt/SiOz (68).

’

catalysts; plus 2940 (vCH3 as) and 1410 cm- (h’CH3a s ) additionally allowed by MSSR relaxation on small-particle catalysts (50, 58, 63). Di-o*: ca. 2925 (vCH2 s) and 1425 cm- (dCH2).

’

All the room-temperature spectra have in common strong absorption bands from ethylidyne, which is therefore shown to be a particularly stable species, probably associated primarily with ( 1 1 1 ) facets on the nictal particles. The room-temperature di-a* species is most consistently identifiable on large Pt particles (Figs. 7A, C, E, G , I, and J) where, because the MSSK operates, thc similar-frequency parallel modes of the ethylidyne species are not expected to

37

VIBRATIONAL SPECTRA OF HYDROCARBONS

3000

2800

1 7 7

F

I

-

I

' m

2

F. m N ( I)

1

I

Pt 0

1

gg

' " 5

0

m N

m

I 2800 1600

3000

L cm 1400 -l

1200

FIG.7+Conrinued)

be present. On the small-particle catalysts, absorption bands at the somewhat different positions near 2940 and 1410 cm- (Fig. 7F) can be attributed to the latter modes of ethylidyne, a conclusion supported by a careful coveragedependent analysis of the spectrum on EUROPT-1 (63). The spectrum from EUROPT-1 obtained by DRIFTS shows the same absorptions, but with somewhat different relative intensities [Fig. 71 (62)]. More variability is associated with absorption bands from the n-complex. On the larger metal particles a single absorption near 3015 cm- occurs, as expected from the MSSR. In one small-particle spectrum [Fig. 7H, 1.9 nm (58)] the n-complex spectrum is strong with additional absorptions resulting from the relaxation of the MSSR. But on another such spectrum [Fig. 7F, also 1.9 nm (SO)] no n-complex absorptions are observed at all. Trenary et al. (58) have suggested

'

38

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

that the significant experimental difference between these two cases is the final evacuation temperature of the catalyst after reduction. These were respectively 623 K (350°C) and 475 K (202°C). They consider that the higher evacuation temperatures remove increasing proportions of strongly held and unreactive hydrogen from the metal surfaces, as originally suggested by Eischens (6, 7) in the context of Ni. The presence of hydrogen is thought to inhibit n-complex formation at room temperature. Although considerable hydrogen derived from the dissociation of ethene desorbs near 300 K on Pt( 1 1 1) (142, 1 4 4 , the “rough” sites on metal particles, where ethene probably adsorbs as the n complex, may well retain hydrogen at high temperatures. Trenary el al. also point to a hightemperature sensitivity for n-complex formation on the “hydrogen-depleted’’ surfaces. These ideas find some support in experiments by Masel et al. on clean and hydrogen-covered ( 1 10) and (210) single-crystal faces of Pt (151, 152). On these hydrogen-free surfaces (151), a rapid reduction in the concentration of surface n-species with increasing temperature occurred just below 300 K. On deliberately hydrogen-covered single-crystal surfaces (152), both the normal and a more weakly perturbed n species were observed at low temperatures on (2 X 1 ) (1 10) and on (2 10). However, both forms were transformed on the surfaces, or eliminated, at temperatures as low as 250 and 215 K, respectively. The spectrum by Szilagyi (68) shown in Fig. 73 differs from others of a similar profile in having more prominent absorptions at 2995 and 1420 cm- I . The author preferred to attribute those absorptions, together with others at 3020 and 1505 cm-I, to a vinyl surface species rather than to a n-complex. This assignment was based on the grounds that the vC=C mode would involve atomic motions parallel to the metal surface and that the 1505 cm-l band has considerable intensity. However, no account was taken of the consideration that a change in the CC distance during vibration is likely to affect the degree of donationhack-donation between the adsorbed n species and the metal atom to which it is attached. Hence dipole moment change perpendicular to the surface is generated, as expected on symmetry groups for a filly symmetrical mode of a C,,, surface species. Morrow and Sheppard (32) studied the absorption of ethene on their largeparticle catalysts at several temperatures up to 473 K. The absorptions at ca. 2885 and 2800 cm- I , nowadays attributed to ethylidyne, slowly decreased in intensity while an absorption near 2920 cm- increased and broadened. A weaker band at ca. 2960 cm- I was also strengthened. The increased absorptions can probably be attributed to v C H ~and v C H ~modes of a dimerized surface species (see below), a conclusion also supported by an analysis of the spectrum on EUROPT-1 (Fig. 7D) (63). A number of spectra have been recorded from CzD4 adsorbed at room temperature on oxide-supported Pt (47, 57, 168, 1 6 9 . The n species is considered to absorb at 2310, 2213, 2177, and 1320cm-’ (57, 160, 169), the 2310 and

‘

39

VIBRATIONAL SPECTRA OF HYDROCARBONS

2 177 cm modes being the non-completely symmetrical modes allowed on small-particle catalysts. The di-o* species adsorbs at 2140 cm- (57, 169) and the ethylidyne species at 2079, 2035 cm- I, and 1147 cm-’ (169). The direct relationship between the spectra of ethene on Pt( 1 1 1) and on Pt/SiO2 has been explored (170). Hensley and Kesmodel have recorded VEEL spectra of ethene on Pt/A1203 and Pt/Ti02 (61). Palazov, Shopov, and colleagues have coadsorbed ethene and CO on Pt/AI2O3 (171). From the substantial downward shift of the vCO frequency, they concluded that electron donation was occurring from n-adsorbed ethene even though the weak ethene spectrum (Fig. 7C) showed no sign of absorptions from this species. However, it should be borne in mind that, C-H bond for C-H bond, ethylenic vCH in hydrocarbons absorb about five times more weakly than alkyl ones ( 1 72). Yoshitake and lwasawa (66) obtained vCH spectra from ethene and Pt/AI,O, very similar to those published by Trenary et al. (58), but showed that doping with sodium in the form of Na2C03 followed by calcination shifted the ethylidyne 2888-cm- band to 2869 cm- without affecting the vCH absorption from the n-bonded species. Ito and Suetaka (78, 173) obtained a spectrum from ethene adsorbed at room temperature on a Pt film evaporated on a quartz plate. They assigned absorptions at 3300 and 3200 cm- to adsorbed ethyne (acetylene) obtained by dissociative adsorption, and broad bands at ca. 2900 and 2725 c m - ‘ to saturated adsorbed species. The 2900-cm-’ band could be from di-o* or ethylidyne. The low wavenumber of 2725 cm- (a “soft” mode) probably implies an end-on interaction of CH bonds with surface metal atoms.

’

’

d. Finely Divided Pt-Hydrogenation. Morrow and Sheppard ( 3 2 ) showed that the surface species from ethene adsorbed on Pt/Si02 were virtually eliminated on hydrogenation at room temperature to give ethane in the gas phase. At higher temperatures, ethane remained the principal product although increasing amounts of n-butane also appeared. Over the same temperature ranges, increasing absorption bands of a residual surface species appeared. The latter spectrum had a profile that resembled that expected from a n-butyl group, as originally suggested by Eischens for ethene on Ni/Si02 (3,and so surface-dimerization had clearly occurred. This spectral pattern, which occurs after hydrogenation on a number of catalysts of different metals, will be discussed in more detail in Section VI.B.2.d concerned with finely divided Pd. Soma was the first to obtain a spectrum in the 1700-1300 cm- region attributable to an ethyl group after hydrogenation of ethene on Pt/AI,O, at the lower temperature of 229 K (47). Using a similar Pt/Si02 catalyst, de la Cruz and Sheppard ( 1 74) showed that a very small dose of hydrogen at 294 K led to a weak but well-defined spectrum with features at 2960, 2930, 2875, 2860, 1470, and 1380 cm-I from a possible ethyl surface species. Larger doses of hydrogen again give ethane as the product. The ethyl structural assignment has

40

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

recently been supported by a similar infrared spectrum obtained from the photochemical reaction of ethyl chloride on Pt/SiO2 (1 75). Trenary et al. (57) have shown that at 140 K successive doses of hydrogen lead to the selective elimination of both the di-a and IC species at comparable speeds. At room temperature (58), when only IC and ethylidyne species were present on their catalysts, hydrogenation led to rapid elimination of the n species but much slower removal of ethylidyne. Soma obtained similar results at 195 K. Kinetic hydrogenation studies (to be discussed in more detail in Part 11) by Avery on Pt/SiO2 (164, 176) and by Beebe and Yates on Pt/A1203 (177) have also shown that ethylidyne is the ethene-derived surface species to be hydrogenated most slowly and that overall hydrogenation activity can be high in the absence of absorption bands from surface ethylidyne. Soma has also discussed qualitatively (47) and quantitatively (1 78) the hydrogenation of IC surface species via a PtH intermediate. e. Finely Divided Pt: A Historical Perspective on the Spectroscopic Identification of the Adsorbed Species from Ethene. Figures 6A and 7A (32) were the first published spectra from ethene adsorbed on oxide-supported Pt. At that time (the 1960s), the most probable surface species were considered to be the di-a and/or n nondissociatively adsorbed species and the M(CH=CH)M or M2(CH-CH)M2 dissociatively adsorbed species. The room-temperature spectrum (Fig. 7A) showed the presence of unsaturated (3010 cm-') and saturated (2920, 2880, and 2795 cm-I) surface hydrocarbon species. The 3010 cm-l band was deliberately not assigned to the presence of the IC complex; rather, it was attributed M(CH=CH)M owing to the absence of the stronger absorption near 3070cm-' normally characteristic of C=CH2 group in the spectra of organic compounds (8).The 2920-, 2880-, and 2795-cm- I absorptions were all principally assigned to the expected di-a M(CH2-CH2)M species, which was expected to have at least three active vCH fundamentals and a fourth inactive one (see Fig. 3). It was, however, suggested that the weak 2795-cm-' absorption was probably from an overtone of a SCH2 fundamental in Fermi resonance with vCH2s. It was further thought, from variable-temperature work, that an absorption from another species, M2(CH-CH)M2, might be overlapping at 2920 cm-I. However, the low-temperature spectrum (Fig. 6A) provided a puzzle as it had only a single, if broad, absorption in the saturated region at 2907 cm-I. It was expected that the nondissociated di-a species, i.e., without CH bond breaking, would be the principal saturated one at low temperatures. How could a species with four CH bonds give only a single vCH absorption? With recorded misgivings (32), the 2907-cm - absorption was therefore attributed to the less CH-rich M2(CH-CH)M2 species. The arrival of Fourier-transform infrared spectrometers in the early 1970s afforded much better spectra in the dCH3/SCH2/vC=C region, partly because of

'

41

VlBRATIONAL SPECTRA OF HYDROCARBONS

the greatly improved sensitivity and partly because point-by-point computer ratioing of transmission spectra from the catalyst, with or without adsorbed species, allowed even strong background Si02 absorptions to be effectively eliminated down to the blackout at ca. 1300 cm-I. Typical resulting spectra are shown in Figs. 7D, 7E, and 7G. Immediately in evidence was a ca. 1500 cm-l absorption as expected for a n-complex (20). Why had the vCH spectrum been found to be inconsistent with this? Also, surprisingly strong absorption occurred at 1340 cm-I, which was difficult to relate to any then-expected surface species (38, 47). A further experiment in which CO was coadsorbed on Pt/Si02 after ethene adsorption gave the spectrum (after CO adsorptionhefore CO adsorption) shown in ratioed form in Fig. 8 (179, 180). The most notable changes concerned the strong absorptions near 2885cm-' and 1340cm-'. The first of these suffered a major reduction in intensity, but the latter revealed only a small frequency shift, essentially without change of intensity. This behavior, which had first been observed on Rh/Si02 by Pearce (38), led to the conclusion that these two absorptions (surprisingly, in view of their mutual strengths) arose from different surface species. As the ca. 2920-cm- I absorption was much less affected by CO adsorption, it was also concluded that this should be attributed to a different species. It was, however, a major dilemma that each of the three

-0

0-

I2 m N

3200

3000

2800

2600

1600

1400

cm -1 Fic. 8. The ratio of spectra from ethene on Pt/SiOl at room temperature (afterhefore) CO postadsorption ( I N @ .

42

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

absorptions in the vCH region thereby appeared to originate in a different surface species, whereas at least two of them were expected to have four CH bonds! This conundrum led to the hypothesis that on these large metal particles the MSSR, known to apply to flat metal surfaces (14, 92), could be responsible for these drastic spectral simplifications (93). This possibility reconciled the 30 10-cm- band with the n-complex absorption at 1500 cm - I , but left open the structural attributions of the 2920- and 2885-cm- I bands. The next experiments involved the adsorption of H2CCD2, HDCCHD, and C2D4on Pt/Si02 (179, 180 169). The C2HzDz experiments led to the conclusion that H/D isotopic scrambling occurred for the species giving the 2885-cm-’ absorption and hence to the tentative conclusion that it contained a CH3 group. Partially deuterium-substituted species CH2D and CD2H could account for absorptions at 2900 and 29 15 cm- I, respectively ( 179). This was an extremely unexpected conclusion as no previous suggestion had been made of the presence of CH3C groups from the absorption of ethene. We therefore did not publish this conclusion until much later when further evidence was available (169). Hydrogenation of all adsorbed species to ethane at room temperature has since shown that only C2 adsorbed species are present, so the CH3 species cannot be of type CH3M. In the preceding few years, electron-energy-loss spectroscopy had been developed as a means of obtaining vibrational spectra of single monolayers on flat single-crystal surfaces. As mentioned earlier, Ibach and colleagues (12) had applied this technique to ethene adsorbed on Pt( 1 1 1) and concluded that, at room temperature, the spectrum corresponded to an ethylidene species (CH3CH). They interpreted a strong band at 1360 cm-’ as the 6CH3 s mode of this species. This corresponded well, within VEELS experimental error, to the 1340-cm-’ absorption in the spectrum from Pt/Si02. On the one hand, this corroborated the possibility of CCH3-containing species on Pt; but on the other hand, we had concluded from our CO experiment that the 2885- and 1340-cm I absorptions were from different species! Could there be fwo CH3-containing species present? “Most improbable!” we thought. In due course, in a now wellknown story (see reference 17, pp. 604-606), the surface species on Pt( II 1 ) was identified as ethylidyne rather than ethylidene. But this did not immediately solve our dilemma, as the vCH regions of VEEL spectra are weak and difficult to interpret. Eventually, however, analogous RAIRS work on ethene on Pt( 1 1 1 ) at room temperature (145. 146) showed quite clearly that the 2885- and 1340-cm- I bands are after all from the same ethylidyne species. Consequently, we had to conclude that our CO experimental result simply denoted a drastic effect of adsorbed CO in an immediately adjacent site on the intensity of the vCH3s mode. Probably, as is the case for coadsorption on Rh( 1 1 1 ) ( 181), a superlattice is formed with CO in adjacent 3-fold sites. Without this experimental “red hemng” from CO coadsorption, the ethylidyne structural

’

-

VIBRATIONAL SPECTRA OF HYDROCARBONS

43

assignment would probably have been achieved a decade earlier. Subsequently, as already described in Section V1.B. 1 .c, variable-temperature work on Pt/SiO, (96) and Pt/AI2O3(57) identified the surface di-a species as that responsible for the 2910- and 1420-cm-' absorptions and hence, after all, for the original lowtemperature spectrum on Pt/Si02 (Fig. 6A). Morrow and Sheppard's original spectroscopic assignments were very wide of the mark, particularly because the strongest absorptions in both the vCH and dCH2/dCH3 regions proved to have originated in the quite unexpected ethylidyne species! It took three experimental advances (FTIR, VEELS, and RAIRS) and one new theoretical understanding (MSSR) to reach convincing conclusions. We shall see later that the identification of the ethylidyne species also greatly improved our understanding of the spectra of chemisorbed higher alkenes on finely divided metal surfaces. 2.

Ethene on Palladium

a. Pd, Single-Crystal Work. The spectra/species correlation for the (1 1 1 ), (loo), and (1 10) single-crystal faces of Pd are summarized in Table VIA (133, 182-188). At low temperatures, unlike the case of Pt, even the very flat (1 11) and (100) faces of Pd prefer n-complex formation. Nishijima et al. ( 186, 187) have interpreted the more complex low-temperature spectrum on ( 1 10) as also from n complexes; but in Table VI we have preferred the alternative view of Chesters et al. (185),that n and di-a species coexist. This is because the strongest feature that gives additional complexity to the spectra on ( 1 10) compared with the (1 1 1) and (100) faces is just as expected for a di-a species (17). The work on Pd( 1 10) includes studies of the effects of preadsorbed Cs (187) and H (188, 189) on the ethene spectrum. In the latter case, ethane was desorbed at low temperatures and in much larger quantities. Ethene desorption was also detected at lower temperatures. b. Finely Divided Pd-Low- Temperature Spectra. Figure 9A shows the spectrum of ethene of Pd/A1203 at 195 K obtained by Soma (47).Figure 9B adds a previously unpublished spectrum on Pd/Si02 taken at 233 K by James (190). Both clearly show the presence of n-complexes by virtue of absorptions near 1510 cm- with companions at 2975 cm-l (a somewhat lower value than for the analogous spectrum on Pt) and at 1241 cm-I. Absorptions from ethylidyne occur at 2872 and ca. 1325 cm-', and from the di-a species at 2927 and 1420 cm-I. Soma identified the latter absorption with the n-complex, but the spectrum in the 3000-cm-' region in Fig. 9B suggests the alternative assignment. It is notable that the ethylidyne species occurs appreciably at 195 K and substantially at 233 K, much lower temperatures than with Pt, ca. 300 K. It persists to above room temperature.

44

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

c. Finely Divided Pd-Room-Temperature Spectra. The substantial number of these spectra in the literature are collected together in Fig. 10 (191; 20 and 167; 190, 192 and 193; 50,97 and 177; 58; 60; 190 and 193). With the exception of Fig. 10F (to be discussed below), the spectra show many similarities among themselves and to the room-temperature spectra on Pt (Fig. 7). The ethylidyne species clearly dominates the spectra (ca. 2930, 2870, 2780, ca. 1400, 1325, 1090 cm-'), but both n complexes (3080, ca. 3020, ca. 2975, 15 10, 1240 cm - ) and di-a* complexes (2920, 141 5 cm- ') are also present in most cases. The ethylidyne species are probably once again located on (1 1 1) facets of the Pd particles. As was the case with finely divided Pt, Figs. 9B and 1OC imply that, on warming, it is the di-a rather than the n complex that transforms into ethylidyne. This is despite the fact that on Pd( 111) the n species was reported to transform into ethylidyne without any other intermediate being observed ( 182, 183). Spectrum 10F by Binet et al. (60) is of particular interest as it is obtained on an exceptionally highly dispersed (51%) catalyst of low loading (0.45% Pd) at 295 K. It is seen that, although once again all three surface species are present, in this case the n complex gives the dominant spectrum. Figure 10G, obtained by James on PdSi02 at room temperature, shows a remarkably simple spectrum with an entirely different profile from the rest (190, 193). The standard procedure for preparing catalysts that give the typical high-particle-size spectrum of Fig. 10B involves a reduction of PdC12 on SiOz in hydrogen at 625 K, followed by an overnight annealing process of about 12

'

TABLE VI On-Specular VEEL Spectra Associated with Structures of Adsorbed Species j r o m Ethene Adsorhed on Pd and Ni Single-Crystul Surfaces ~~

Surface species

Wavenumbers and intensities

Surfaces

Temperature range ( K )

References

A. Palladium 3080 (m), 2980 (ms), 2780 (ms)," cu. 1510 (mw), ca. 1230 (mw), cu. 910 (vs)

(111), (100).

90-260

182-1 89

2980 (s), 1455 (ms), I135 (s), 920 (s)

(100)

80-275

133. 184

ca. I125 (s), 390 (s)

( I lo)*

80-275

185-189

(w),1334 (s), 1080 (ms), 914 (m),780 (m)

(Ill)

ca. 300

182, 183

ca. 2950

and ( I 10)'

(continues)

45

VIBRATIONAL SPECTRA OF HYDROCARBONS TABLE VI-(Conrinued) Surface species

Wavenumbers and intensities

Surfaces

Temperature range (K) References

?CCH

ca. 2950 (w), ca. 1260 (w, bd), ca. 930 (s)L'

(100). ( I 10)

300400

133. 187

?CH

ca. 2940 (m), ca. 800 (s)

(1 11). (1 10)

400-500

183, 187

I50

167 I95

B. Nickel n-Species (CrHdM'

3070 (m), 2990 (m), 1520 (ms), 1255 (s), 900 (s)

( 1 I O/C)

Di-o species Mz(C2H.d

2995-2950 (m). ca. 2700 (w, bd),'

( I 1 I ),' ( I 10)

[5(111) X ( I lo)]"

(loop

80- I 50

195-200 167

230-300

195-197

146Cb1400 (mw), 117&1100 (s). 890 (m), 450-400 (vs) ?M2(CH=CH)Mi di-ddi-n

2925 (s), 1220 ( s ) , 880 (ms)

nn-Vinyl M(CH=CH*)M'

3090 (m), 3025 (s), [2920 (m)], ( l 0 O f h 1425 (m), [ I 150 (m)], 930 (s). 620 (ms)

I75

199

?CCH

2990 (m), 1290 (mw), 890 (s),' 465 (ms)

(1 10)

300

200, 195

2920 (m), 1350 (s), 960 (mw). 480 (mw)

(100)

275

I99

?Mz(CHCH,)

(111)

[XI I I ) x ( I lo)]'

"The 2708-cmcm-' absorption occurs on Pd(l I I ) only. hNishijima et a/. (186. 187) have interpreted the low-temperature spectrum from ethene on Pd( 110) as arising entirely from the n species; following Chesters et al. (18.9, we prefer to interpret the spectrum as from overlapping contributions from n and di-n species because the spectrum changes with coverage, leading to an additional absorption, just as expected for the di-n species. "This strongest absorption band is a CH-deformation mode as shown by the isotopic shift to CD. 'We interpret the low-temperature spectrum from ethene on Ni[5(1 I I ) X (IIO)] (195) as from a (no) plus a vinyl species. ' N o t observed on Ni( I 10). 'We interpret the 90 K spectrum from ethene on Ni( 100) as probably from overlapping contributions from a type B ethyne and a di-o ethene species as there are coveragedependent relative variations of band intensities; Zaera and Hall (199)interpret this spectrum as that of a n-species of ethene. "'The additional 620/430-cm I absorptions from C2H4/CZD4suggest outof-plane yCH absorptions of a type B adsorbed ethyne. An alternative possibility is surface hydrogen (see text). In agreement with Zaera and Hall ( 1 9 9 , we interpret the high-coverage 175 K spectrum from C2H4/Ni(100) as from a vinyl group plus some undissociated ethene (2920 and I I50 cm I), The vinyl assignment i s supported by a recent spectrum of this group on Pt( I I I ) ( 2 0 0 ~ )The . strength of the yCH modes implies in both instances that the plane of the vinyl group is not perpendicular to the surface, i.e., that it is o-bonded to one metal atom and n-bonded to another. ' A RAIRS study of the di-o species at 110 K on Ni(l I I ) gives the more precise band positions of 2937, 1422, and 1084 cm- (201). ~

'

46

NORMAN SHEPPARD AND CARLOS DE LA CRUZ cm-‘

3000

2800

1400

1600

I

1200

1

I

m

N c

e

N

Pd

m

I

1600

1400

1200

c m-‘ FIG.9. Infrared spectra from ethene on Pd at low temperatures: ( A ) Pd/Al2O3 at 195 K (47); (B) Pd/SiOz at 233 K (/YO).

hours’ duration. The spectrum shown in Fig. 10G was obtained after reduction followed by a much shorter annealing time of 1 hour and had absorptions at 2940 (ms), 1415 (vs), and 1345 cm-’ (mw). It occurred in concurrence with two absorptions at 1745 and 1910 cm- from adsorbed hydrogen, as described below. Both adsorbed hydrogen and CO also gave unusual spectra on these unannealed catalysts. With H 2 , additional extra absorptions appeared in sequence with increasing hydrogen adsorption at 1745, 1900, and 2052 cm-l (190). With adsorbed CO, a much greater-than-normal proportion of linearly adsorbed CO was present; also, the 2-fold bridged species saturated at a different position (1970 cm- I ) compared with the fully annealed catalyst (1990 cm- I ) . Such spectra could be readily reproduced by this revised procedure from newly reduced catalysts, but gradually reverted to the more normal type of spectrum after repeated reduction cycles. The integrated CO absorption intensities were very similar for the two types of catalyst preparation, implying that it is the detailed arrangement of surface metal atoms that is different rather than the metal particle sizes of cu. 20 nm. Spectra of CO on annealed Pd/Si02 have been interpreted in terms of the presence of well-defined ( 100)- and (1 I 1 )-type facets, which vary in proportion with catalyst “break-in” (81, 194). It is probable that the spectra from the unannealed catalysts imply the presence of much rougher surfaces, with sites that would be more typical of edges or comers. The spectrum of ethene on the unannealed samples was further investigated by use of the isotopic variants I2C2H4,C2D4, CHlCD2, and l2CH2I3CH2.The strongest absorption band appeared at 1415, 1400, 1395, and 1387 cm-l, respectively, showing that it corresponds to a vibrational mode of largely vCC character (190, 193). The wavenumber range implies a bond order considerably

’

47

VIBRATIONAL SPECTRA OF HYDROCARBONS cm

2800

3000 I

I

I

-1

1600 I

1LOO 1

1

1200 1

1

I 1

Pd

cm

-'

FIG. 10. Infrared spectra from ethene on Pd near room temperature: (A) Pd/SiOz ( I Y I ) ; (B) Pd/SiOz (167); (C) Pd/SiOz (190); (D) Pd/A1203. reprinted with permission from ( 5 0 ) , copyright I987 American Chemical Society; (E) Pd/A1203, reprinted with permission from (5X), copyright 1991 American Chemical Society; (F) PdA1203 (60); (G)Pd/Si02, unannealed catalyst (190).

closer to 2 than to 1. The absorptions in the vCH region were at 2940 (C2H4), 2940 (CH2CD2), and 2930 cm-I, respectively; and the second band in the low wavenumber range occurred at 1345, 1345, and 1338 cm - I , respectively. Clearly the latter band is of the type of a CH deformation mode. For CH2CD2 and C2D4, exchange occurred between CD/CH and silica OH/OD groups, implying dissociation of CH/CD bonds on adsorption. Although the 1345-cm absorption from C2H4 is not far from the position of the ethylidyne 6CH3s mode at 1325 cm-', there was no trace of a corresponding v C H s~ absorption in the higher wavenumber region. Hence it is clearly derived from the single surface species that is present. The simplicity of the spectrum in Fig. IOG,the high value of vCC, and the parallel production of absorptions from adsorbed hydrogen all suggest that the originating surface species is dissociatively adsorbed. Inspection of the spectral data from model compounds summarized in Fig. 4 gives the (CH=CH)M; (14

'

48

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

cm-'

F

r-r m c

3000

2800

1600 cm

-1

1400

1200

FIG. IO-+Conrinued)

in Table IV) as the best fit for the strong 1415-cm-' vCC mode. Unfortunately, this species has its vCH mode (ca. 3100) at a much higher value than 2940 cm- ',as recorded in Fig. 10G. M2(C=CH2)M' (18) gives a better overall fit in the vCH and vCC regions, but in each case the absorptions of the model compound (2990 and 1470 cm-I) occur at rather higher values. However, the spectrum of Fig. 1OG agrees well with the VEEL spectrum of ethyne on Pd( 1 1 1) (vCC = 1400 cm-I, vCH = 2990 cm-I) ( 1 9 4 ~ that ) has been classified as a type B spectrum corresponding to a di-oh adsorbed (HCCH) species. The VEEL vCH value is somewhat higher than is recorded for the finely divided catalyst. However, VEEL spectra are of low resolution, and often include in the vCH region contributions from impact-excited modes that are expected at higher wavenumbers. It would be of interest to study spectra from ethene adsorbed on newly reduced but relatively unannealed catalysts of other metals. The marked changes in surface sites implied in going from Fig. 1OC to Fig. 10G are analogous to those expected during catalyst break-in phenomena ( I 94). , d. Finely Divided Pd-Hydrogenation. At 195 K, adsorbed species from ethene adsorbed on Pd/SiOz are completely hydrogenated off the surface as ethane (191). At room temperature, the addition of hydrogen to Pd/A1203

VIBRATIONAL SPECTRA OF HYDROCARBONS

49

removes the n complex very rapidly, but ethylidyne much more slowly (58). Typical spectra of surface species retained on Pd/Si02 after room-temperature hydrogenation are shown in Figs. 11A and 11B. The gas phase consisted of butane and a little ethane (191). Retained spectra of similar profiles were early observed on Ni/Si02 (7) and have been attributed in both cases to the presence of surface n-butyl groups. Also in both cases (as shown in Fig. 11B) the spectrum in the vCH region is greatly reduced in intensity on evacuation, but virtually restored on readdition of gas-phase hydrogen. This phenomenon will be discussed below (Section VI.C.2.c) in more detail in connection with adsorbed butenes. The spectrum itself implies the occurrence of surface dimerization either before or during hydrogenation. The “n-butyl” type of spectra can vary somewhat in profile, particularly near 2870 cm-’. A probable explanation for this is given in Fig. 1 lC, where only partial hydrogenation has led to the retention of some ethylidyne species (1338 cm-l and 2870 cm-I). It might be asked to what extent the MSSR could cm-’

3000

1600

2800

E E,

x

1400

0.4 E

3000

1600

2800

1400

cm-1

FIG.1 I . Infrared spectra from ethene on Pd after hydrogenation: + Hz, hydrogen added; f Hz, hydrogen added followed by evacuation. (A) PdSiO2 (191); (€3) PdSiOz (192); (C) Pd/SiOz (1Y0); (D) Pd/SiOz, unannealed catalyst (190).

50

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

cause changes in the profile of a spectrum from an alkyl group such as n-butyl. This topic will be considered in more detail later in connection with the spectra of chemisorbed alkanes and related surface species in Part 11. For now, however, we note that the conformational flexibility of such an n-butyl group will reduce the selectivity of intensity charges caused by the operation of the MSSR. Partial hydrogenation of the species present on the short-time unannealed Pd/Si02 catalyst (Fig. 11D) leads to a retained species giving a more methyl-rich special profile (2950- compared with 29 15-cm- absorptions), possibly from surface ethyl groups or, more probably, from physically adsorbed n-butane. On Pt/Si02, complete hydrogenation leads to complete removal of adsorbed species at room temperature to give exclusively gas-phase ethane; but at higher temperatures, ca. 370 K, once again n-butane appears in the gas phase and n-butyl surface spectra are retained. The dimerization process is seen to occur at some 100 K higher temperatures on Pt than on Pd.

'

3. Ethene on Nickel a. Ni, Single-Crystal Work. Table VIB summarizes the spectroscopic results on different crystal faces (167, 195-201). There are notably more dissociatively adsorbed species occurring on Ni compared with Pd or Pt, and those that are common to the latter metals tend to occur at lower temperatures. Ni( 100) is the only flat ( 1 1 1) or (100) surface of any metal that shows evidence of a dissociatively adsorbed species at the lowest temperature studied, in this case at low coverage at 90 K (199). However, as virtually the same spectrum also occurs on the stepped [5(111) X (loo)] surface at 150 K (195). It would be worthwhile to reinvestigate the ( 1 00) face to check for the presence of crystal defects. There is evidence that the di-a species is present on Ni( 100) at higher coverages at 95 K [(198, 199); see also note f in Table VI]. However, the spectrum shows a strong additional absorption at 620 cm- (ca. 420 cm from C2D4) which is attributable to some form of H/D-dependent vibration. This is an unusually low value for a CH deformation mode. One possibility is a cisdisubstituted species with di-a bonding to two metal atoms and n-bonding to a third one, such as occurs in type B spectra from adsorbed ethynes on certain metal faces including Ni( 110) [see Fig. 9 of reference (131.An alternative possibility might be to assign the 620-cm-' absorption to the vHNi, mode of surface H atoms formed by adsorption on a C4 site of (100). Karlsson et al. have shown that the adsorption of hydrogen alone on a Ni( 100) surface gives absorption bands at ca. 530 cm-' at low coverage, or 630 cm- at high coverage (202). Ni( 11 1 ) shows a clear di-a spectrum at low temperatures (195-197, 201) but is exceptional among fcc (1 11) surfaces in not showing a conversion to ethylidyne at higher temperatures. The spectrum observed at 300 K is clearly from a (di-ah) (CH=CH) complex instead of ethylidyne (195, 197). On the other hand, as will be seen in the next section, the ethylidyne species is readily ~

'

VIBRATIONAL SPECTRA OF HYDROCARBONS

51

observed on finely divided Ni/SiOz or Ni/Al2O3 catalysts. Zhu and White (203) have confirmed by surface secondary-ion mass spectrometry that on Ni( 1 1 l), adsorbed ethene does convert directly into adsorbed ethyne; only at high coverages did they observe slight conversion to ethylidyne. Lapinski and Ekerdt (204) have discussed possible reasons for the production of ethylidyne which presumably occur on (1 1 1) facets of the finely divided catalysts. One possibility is that hydrogen present at the time of ethene adsorption may open up an alternative pathway to ethylidyne via ethyl surface species. On Ni( 110) (167, 200) and the also-ridged [5(111) X (1 lo)] surface (195) ethene dissociation at room temperature leads to some form of “CCH” species bonded to multiple metal atoms. However, this is not the (o/di-n) species of 21 in Table IV/Fig. 4, as both the vCH and vCC wavenumbers of the adsorbed species (2990 and ca. 1290 cm-’) are much lower in value than these of the model compound (3160 and 1530 cm-I) (117). b. Finely Divided Ni-Low-Temperature Spectra. Infrared spectra from ethene adsorbed on Ni/SiOz and Ni/Al2O3 are collected together in Fig. 12 (32, 205, 55, 55, 204). Lapinski and Ekerdt (55, 204) have carried out detailed and comprehensive studies in the temperature range 180-228 K. At 180 K they showed that, in addition to physically adsorbed ethene and ethane, two chemisorbed n complexes were present with absorptions at 1547/1227 and 15251 1255 cm- together with a third spectroscopically dominant species with absorptions at 1470 and 1182 cm-’. They described the latter as a species with hybridization between n-bonded ethylene (sp’) and di-o-bonded ethylene (sp3) and gave it the symbol of on. They pointed out that this spectrum was of the type I’ pattern found on a number of single-crystal metal surfaces (206). This species has more recently been identified with metallocyclopropane-like species (3 in Table IV and Fig. 4) for which we have used the symbol (no). As the temperature was raised from 180 to 228 K (Figs. 12A, B, D, and E), a new spectrum with absorptions at 2870, 2795, 1340, and 1125 cm-l gradually grew into dominance which, as Lapinski and Ekerdt were the first show on Ni, corresponds to the ethylidyne species. At yet higher temperatures (Fig. 12B), a fifth spectrum grew at the expense of that from ethylidyne, the possible nature of which will be explored in the next section. It is very remarkable that none of the four low-temperature spectra found, with their related species, on the oxide-supported Ni catalysts has been identified on single-crystal Ni planes (Table VI). The two n species were assigned by Lapinski and Ekerdt ( 5 5 ) to ethene, probably adsorbed close to the Ni/oxide interface (1 547/1227 cm - I ) and on clean Ni sites (1525/1255 cm-I). They are envisaged as probably adsorbed on rough parts of the Ni particles, e.g., on corner or kink types of sites which are not present on the single crystals studied to date. No explanation was offered for

’,

52

NORMAN SHEPPARD AND CARLOS DE LA CRUZ cm-1

3000

2800

16pO

,

14PO

12pO

I

Ni~

IA

3000

2800

cm - 1

FIG.12. Infrared spectra from ethene on Ni at low temperatures: (A) Ni/SiOl at 195 K (32); ( B ) Ni/SiOz at 195 K, solid curve, and after warming, dashed curve (205); (C) Ni/Si02 (lower spectrum) and Ni/A12O1 (upper spectrum) at 180K. dotted peaks are from physically adsorbed ethene, reprinted with permission from (55), copyright 1990 American Chemical Society; (D) NVSiO2 and Ni/AIZO1 at 208 K, reprinted with permission from (55). copyright 1990 American Chemical Society; (E) Ni/AI20] at 228 K, reprinted with permission from (55, 204). copyright 1988 and 1990 American Chemical Society.

an absorption at 2905 cm-’ (Fig. 12C) in the same spectrum. For this, a likely attribution is to a proportion of the di-a species that are prevalent on several single-crystal faces at low temperatures. The 2905-cm- absorption could, in this context, be associated with another weak absorption at 1415 cm- I , expected for a di-a group. The molecular origin of the additional 2853-cm- absorption in Fig. 12C remains unknown. The firm identification of the alkylidyne group was particularly unexpected. These are almost certainly adsorbed on triangular 3-fold bridging sites on (1 1 1)

’

’

VIBRATIONAL SPECTRA OF HYDROCARBONS

53

facets of Ni particles, despite the fact that in the corresponding temperature regime (1 1 1) single crystals themselves give the unambiguous type A spectrum of an ethyne-derived species (17). Studies using surface secondary-ion mass spectrometry have shown that under normal conditions the low-temperature di-a species on Ni( 1 1 1) transforms directly into the more hydrogen-deficient ethyne species without any intermediate (203). However, that study did show the presence of a minority of ethylidyne species at high coverage. Lapinski and Ekerdt invoke the possibilities of the saturation-coverage situation present with the Ni catalysts, or of surface-adsorbed hydrogen which is difficult to eliminate from such samples, to account for the ethylidyne formation. A recent detailed study of ethene adsorbed on Pt( 11 1) using STM techniques has shown that the transformation from di-a to ethylidyne species on that surface occurs in a collective manner; i.e., it grows outward from edges of di-a islands (140). Possibly the high-coverage situation on the Ni metal particles triggers off a similar collective transformation to ethylidyne. However, the alternative suggestion, that residual hydrogen on the Ni particles (204) could be responsible for ethylidyne formation, also needs to be taken seriously. C2H2 species on Pt( 1 11) and Pd( 1 1 1) surfaces have been converted to ethylidyne by hydrogen (141, 182, 183). It was shown that in the temperature range 205-238 K, ethylidyne species did not undergo H/D isotopic exchange after they had been formed (55). It was also noted that the 2870/1340 cm- intensity ratio in the spectrum of ethylidyne was somewhat reduced on “H-free” Ni/A1203 in comparison with the “H-covered” catalyst. This recalls the sensitivity of this intensity ratio to CO coadsorption on Pt/Si02 (Section V1.B. 1 .e). The most conspicuous absorptions in the spectra on Ni particles that were reduced in intensity as the ethylidyne absorptions grew were those of the (no) species. Lapinski and Ekerdt suggested that this, too, was adsorbed on (1 1 1 ) facets at the lower temperature. Perhaps this alternative to the expected di-a species on (1 11) sites is also produced by high-coverage compression of the monolayer. However, quantitative measurements showed that species other than (no) must be involved in the early stages of ethylidyne formation, which occurred with limited loss of intensity from the (no) spectrum. It was suggested that the n species were the other ones contributing to ethylidyne formation. However, if the 2905 cm-I absorption does correlate with the presence of di-a species, this is an alternative and probable precursor, as on Pt or Pd. Hydrogenation at 210 K led, as is usual on other metals, to the rapid removal of the ( m ) species, but a much slower removal of ethylidyne (55). In this temperature regime the gas phase was ethane with a trace of n-butane (55, 32). Condensation of butane may have been responsible for a weak temperaturedependent, methyl-rich spectrum of surface species after hydrogenation at 195 K (32).

’

54

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

c. Finely Divided Ni-Room- Temperature and Higher; also Hydrogenation. Figure 13 collects together spectra at or near room temperature from ethene adsorbed on Ni/Si02 or Ni/A1203 as published by a number of groups (6, 7, 40, 32, 207, 62). In several cases the spectra obtained after the addition of hydrogen are included as inserts. Spectra similar to Figs. 13B and 13C, before and after hydrogenation, were also recorded by Blyholder et al. (208). Figure 13A presents the pioneering early spectra obtained by Pliskin and Eischens (6, 7). The full-line spectra were obtained on “hydrogen-covered” samples produced by cooling the reduced Ni/Si02 sample in hydrogen

FIG.13. Infrared spectra from ethene on Ni near room temperature, with insets after the addition of H2: +H2, hydrogen added; f H 2 , hydrogen added followed by evacuation. (A) Ni/Si02: hydrogen-covered catalyst, solid curve; hydrogen-free catalyst, dashed curve (6, 7). (B) Ni/Si02 (40). (C) Ni/Si02 (32). (D) Ni/Si02, hydrogen-covered and hydrogen-free catalysts, reprinted with permission from (.?On, copyright 1956 and 1958 American Chemical Society. (E) Ni/AI20, (62).

VIBRATIONAL SPECTRA OF HYDROCARBONS

55

followed by evacuation at 308 K; the dashed-line spectra were obtained on “bare” catalyst samples obtained by the evacuation of hydrogen at the reduction temperature of 623 K before cooling to 308 K. It is seen that the hydrogen-covered sample gave the stronger vCH absorptions after adsorption, but the bare samples gave the stronger spectrum after hydrogenation. Pliskin and Eischens concluded that after initial adsorption on the hydrogen-covered sample, the surface species was principally of a saturated substituted alkane nature (they suggested MCHzCH2M or MzCH-CHM2) with a smaller unsaturated fraction. On the bare sample, unsaturated species were much more in evidence, particularly when account was taken of the much greater infrared absorption, CH-bond for CH-bond, from saturated than from ethylenic species (172). They also showed that the surface species obtained after hydrogenation (they suggested a surface ethyl group on the hydrogen-covered surface) could be dehydrogenated or rehydrogenated on evacuation ( - H2) or replacement of gas-phase H2 ( + Hz), respectively. Dehydrogenation occurred most extensively on partially covered Ni surfaces and each f H 2 cycle led to a minor proportionate removal of adsorbed species. Although some later investigations (40, 32) did not produce marked spectral differences between adsorption on hydrogen-covered or bare Ni/SiOz catalysts, Primet and Sheppard (207) did again obtain similar results to those of Pliskin and Eischens. It was concluded that most of the spectra in the literature, and the strongest features in the vCH region between 2880 and 2890 cm-l (Fig. 13B-E), are from at least partially hydrogen-covered surfaces (207). We therefore prefer to use the term “hydrogen-depleted” rather than “hydrogen-free’’ or “bare” for catalysts that have been evacuated at the reduction temperature. In some cases (Figs. 13C and 13E) the strongest absorption was resolved into two components near 2870 and 2890 cm-I, with the latter growing at the expense of the former as a function of time. Where resolution was not possible, post-adsorption of CO led to great reductions in intensity of the ethylidyne absorptions (207). The 2870-to-2890-cm - transformation is mostly clearly observed in detail in Figs. 12B and 13E (205, 62). In the latter case-Fig. 13E, which provides an extended wavenumber region through the use of Ni/A1203it was clearly evident that the 2872-, 1337-, and 1 120-cm- I absorptions identified in the previous section as from ethylidyne species, decrease in intensity as new ones grow at 2957,2935,2887, 1455, 1425, and 1354 cm-‘ (62, 209). The 2957/2887- and 1455/1354-cm-I pairs of absorptions are as expected from vCH3 as/vCH3 s and 6CH3 as/6CH3 s modes of CH3C groups. Under the MSSR, the clear detection of both the as and s modes implies that the C-C bond is at a considerable angle to the surface normal. Possibilities include an ethylidene (CH3CH) surface species or a dimerization of ethylidyne to give a (CH3CCCH3) species presumably bonded to the surface as a di-metal-substituted cis-but2-ene. We have assumed above that a single new species grows at the expense

’

56

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

of ethylidyne. It is not clear from the experimental evidence cited (62) why Holmes et al. postulated three different new species. In fact, a species of formula CH3CCCH3 has already been identified on a Pt(l11) surface by adsorption of cis- or truns-but-2-ene at room temperature. It is a (di-ah) (CH3CCCH3)species with its molecular plane tilted with respect to the metal surface ( 2 f O ) .On Pt( 11 1) this has absorptions at ca. 2960,2885, 1430, and 1354cm-’, which are close to the principal extra absorptions shown in Figs. 12B and 13E. We prefer this structural assignment. Hydrogenation at room temperature on Ni/Si02 led to a restricted amount of solely n-butane in the gas phase (32) plus a substantial spectrum from residual adsorbed species (inset in Fig. 13C). After evacuation to remove physically adsorbed butane, the spectrum of the residual adsorbed species had the n-butyl profile discussed above in relation to ethene adsorption on Pd catalysts. The spectrum before evacuation (insets to Figs. 13C and 13D) were more methylrich because of the presence of physically adsorbed n-butane. After hydrogenation, then, dimerization to C4 species had clearly occurred, as shown by both the surface and gas-phase products. Morrow and Sheppard ( 3 2 ) showed that long-term standing of an Ni/Si02 catalyst in ethene led to increasingly intense spectra from adsorbed species, which gradually generated n-butyl-like profiles. It may therefore be concluded that dimerization can occur on the surface before hydrogenation, possibly via the di-cr adsorbed (CH3CCCH3) species discussed above as intermediate. At 432K, Morrow and Sheppard (32) found that only very weak initial spectra were obtained from the adsorption of ethene on Ni/SiO2, but that hydrogenation led to strong, broad, n-butyl-type spectra from surface species together with methane in the gas phase. The latter observations show that C-C bond scission can occur on this catalyst at elevated temperatures. The great increase in intensity from adsorbed species on hydrogenation implies that very hydrogen-deficient surface-carbide-like spectra were formed on initial adsorption at 423 K. This was accompanied by the generation of ethane in the gas phase through a self-hydrogenation process. Ito et al. (173) have observed weak spectra from ethene adsorbed on evaporated Ni films, including vCH “soft-modes” at 2700 cm-‘ and what could be an absorption band from a 71 complex at ca. 1000 cm-’. Partial Raman spectra have been observed from ethene adsorbed on Ni/SiOl at 180 K and at roomtemperature (26, 211). Krasser et al. (26) appear to have interpreted the spectrum taken at 180 K predominantly in terms of a di-a species. However, the vCH region with bands at 3000, 2924, and 2860 cm- suggest the presence of both saturated and unsaturated hydrocarbon groupings. Krasser et al. interpret the ca. 990 cm- I band as from the vCC mode of the di-a species, but the EELS data on Ni( 11 1) suggest that the vCC mode should be near 1100 cm-I, possibly that shown in the Raman spectrum at CQ. 1070 cm-I. Indeed, the set of Raman bands

VIBRATIONAL SPECTRA OF HYDROCARBONS

57

at 2924, 1456, ca. 1070, 910, and 428 cm-' correlate rather well with those found in the VEEL and RAIR spectra of the di-a complex on Ni( 1 1 1) (195, 196, 201). The low temperature Raman data, with their capability of penetrating the region that is obscured in the infrared region by blackout absorptions from the oxide support, are therefore very encouraging. It would be particularly advantageous to obtain a similar spectrum from adsorbed C2D4 at the same temperature. At room temperature, Raman spectra were obtained from adsorbed C2D4 and CzD4. Although some of the higher wavenumber bands find correspondences with those of the surface ethyne species that is found in the 1R spectra of Ni/Si02 at this temperature, there are numerous additional Raman bands of which only a proportion can be attributed to specific hydrocarbon surface species such as benzene.

4. Ethene on Rhodium a. Rh, Single-Crystal Work. VEELS studies have been made of ethene adsorbed on Rh( 11 1) (212, 213), Rh( 100) (214), and Rh( 100) with coadsorbed CO (215). On Rh( 1 1 1) at 3 10 K, a well-defined spectrum from the ethylidyne species was obtained (2880, 1337, 1121 cm- '). A similar spectrum with somewhat different band positions (2905, 1350, 1010 cm- ') was obtained at room temperature from ethene adsorbed on Rh( 100) in the presence of coadsorbed CO (215); a somewhat less well-defined spectrum was subsequently obtained on clean Rh( 100) at 310 K (214). Rh( 100) appears to be the only surface on which the ethylidyne species is bonded to a 4-fold site. There is a cluster analog (216). At low coverage at 325 K, the spectrum on Rh( 100) was of the type that has been associated with "C2H" surface species [3065-2985 (s), 1345-1305 (w), 830-805 cm- (s)]. The exchange of ethylidyne species with gas-phase D2 has been studied on Rh( 11 1) (212).

'

b. Finely Divided Rh. Figure 14 shows collected spectra from the literature from ethene adsorbed on Rh/Si02 or Rh/A1203 ( 19, 38, 93. 47, 50, 58, 58, 67). Figures 14B and 14E are low-temperature spectra and very clearly show the presence of (no) and ethylidyne species. It is also considered probable by the reviewers that the 2939- and 1406-cm- absorptions in Fig. 14E are from the di-a species. The absorptions were assigned by the original authors to the vCH3 as and 6CH3 as modes of the ethylidyne species adsorbed on small metal particles, but the 1406-cm-' absorption seems to be too strong for this to be a likely choice. Truschke and Knozinger (67) observed reversible hydrogenation of the spectrum in Fig. 14F to give a methyl-containing species, which they suggest is surface ethyl; as ethylidyne is involved, another possibility is perhaps ethylidene. The band positions associated with the (na), di-a*, and ethylidyne species are listed in Table VII in comparison with those from other metals.

'

58

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

FIG,14. Infrared spectra from ethene on Rh ncar room temperature except where indicated: ( A ) Rh/Si02 ( I Y . 38); (B) Rh/A1203, 195 K, dotted band from physical adsorption (47); (C) Rh/Al2O3, reprinted with permission from (50), copyright 1987 American Chemical Society; (D) IUuA1203, reprinted with permission from (58), copyright 1991 American Chcmical Society; (F) Rh/SiOz (67).

TABLE VII Infrared Spectral Data for the Species Formed from Ethene Adsorption on a Range of Oxide-Supported Metals Metal

7[ --t

co

(nu)

di-ddi-u*

Ethylidyne

Not observed

1470 1185 NA" 2975? 1525 1255 NA 1547 1227 NA

Ni

1545 1280 895

NI'

CUh

Ru

-

2905?

3080 2966

-

1505

Pd

3080 2970

-

1510 1240 NA

NI

Ag"

1220 NA

2980

Pt

3075 3020 2955

1495 1200 NA

NI

1535 1270 935

Au" NA-xperimental

1505

1190

1420 1347

56

2875 2795

1410 1340 1125

32. 40. 55 62. 204, 207

23 I

2890 2810

1350 1130

50

2940?

1405

2945 2885 2800

1410 1340 1115

38, 47, 50. 58, 67

2930(di-u) 2920 (di-u*)

l4Io

2930 2870 2780

1400 1325 1090

50, 58 20, 167

-

1580 1315 955

Ir

2880 2810

Not observed

Not observed

Rh

References

Not observed

NA 2910(di-u) 2920 (di-u*)

data not accessible. SERS data. ' NI-not

1420 1425

2947 2895 2815

1420 1350 1155

58

2935 2885 2800

1410 1340 1130

32. 47, 50, 58. 68, I 6 7

Not observed investigated.

22 7

22 7

60

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

3000

2800

1600

cm-1

1200

1400

co

c U

z

v

r 3000

2800

Ru

1+9201vr1 I

2

1600 cm-, 1400

1200

FIG. 15. Infrared spectra from ethene on various metals near room temperature unless otherwise indicated: (A) Co/Si02, hydrogen-covered (208.21 7); (B) Co/SiO2, 228 K (56); (C) Ru/AI2OI, reprinted with permission from (50). copyright 1987 American Chemical Society; (D) Ir/AI20,, reprinted with permission from (58). copyright 1991 American Chemical Society; (E) CdZuO (52).

5 . Ethene on Other MetaIs: Co, Ru, Ir, Cu, Ag, and Au

Figure 15 shows infrared spectra from ethene adsorbed on a number of other oxide-supported metals [Co (21 7, 208, 56); Ru (50); Ir (58); Cu (52)l. Co/Si02 at 228 K (56) clearly shows a spectrum, Fig. 15B, dominated by ethylidyne species. At room temperature (208, 21 7) a spectrum is obtained (Fig. 15A) rather similar to that obtained from Ni/Si02, after ethylidyne has been substantially converted to what we have postulated to be the adsorbed C4 (CH3CCCH3)species. However, a weak adsorption at 3000 cm- suggests the

'

VIBRATIONAL SPECTRA OF HYDROCARBONS

61

presence of a proportion of n species and an additional shoulder occurs at 2850 cm-’, suggesting the occurrence of alkyl CH2 groups. On hydrogenation the n-butyl surface species is observed after light initial evacuation, as was the case on Ni/Si02 surfaces. Additional work on Ru(0001) to that previously reviewed ( 1 7 ) includes a VEELS study on C2H4 and C2D4 by Sakakini et al. (218,218a),which shows a type I’ spectrum at 170 or 130 K, either representing a mixture of di-a and n species (two absorptions at 850 and 945 cm-’ for C2H4 raise the possibility of two adsorbed species) or a (no) metallocyclopropane structure. The spectrum is clearly that of ethylidyne at 220 K, as shown by VEELS (2890, 1360, 1120, 420cm-’) (218), and at 240K by RAIRS (2877 and 1340cm-I) ( 2 1 9 ~ ) . Similar VEEL spectra have been obtained in the presence of coadsorbed CO (220). The 300 K spectrum on Ru/A1203 (2888, 1350/1344, 1131 cm-’) is equally clearly from ethylidyne. Studies by VEELS have explored the adsorption of ethene on Ir( 11 1 ) (221, 222) and on Ir( 110) (223). On the (1 11) surface, ethene adsorbs as the ethylidyne species (2940, 1400, 1165 cm-I), possibly in conjunction with some R complexes (ca. 1500 tail, 986 cm- I ) at as low a temperature as 180 K. By 300 K a second, possibly “C2H,” species with a prominent absorption at 825 cm-’ occurs at low coverage and coexists with ethylidyne at high coverage. The “C2H” species dominates the spectrum at all coverages at 500 K (3028, 1260, 838 cm- ’). The same species were obtained on the Ir( 1 10) surface over similar temperature ranges, but the ethylidyne spectrum at 170 K has stronger features from the vibrations with dipole changes parallel to the surface (particularly CH3 rocking and a probable additional intensity contribution in the 1400-cmp region from dCH3 as), as would be expected for such species with CC bonds perpendicular to the (1 1 1) step-like facets, which are at an angle to the surface normal. Adsorption of ethene on the Ir( 110) surface with preadsorbed CO reduces the overall intensity of the hydrocarbon adsorptions and raises the decomposition temperatures of the ethylidyne and “C2H” species by about 50 K. The room-temperature spectrum on the metal particles of Ir/A1203 (58) clearly shows the presence of the ethylidyne groups (2947, 2986, 2816, 1418, 1350, 1156 cm-I) and of the (no) species (2978, 1504, and 1188 cm-I) on the small-particle catalyst. A di-a species may also contribute to the 2947and 1418-cm-’ absorptions as the latter is rather strong in relation to the 1350-cm- I dCH3 absorptions. The restricted wavenumber range available to the Ir/A1203 catalyst does not allow a check to be made on the presence of the “CCH” species at room temperature. Ethene adsorption on Cu( 100) at 80 K (224)or on Cu( 1 1 1) at 91 K (225) give the classic spectra of n species (3000, 1556, 1290, 903; and 3075, 1557, 1285, 990 cm- I , respectively) with the CQ. 1556- and 1290-cm- I absorptions of similar intensities and weak relative to that of the out-of-plane CH wagging

62

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

'

modes in the 1000-900 cm region. In strong contrast, ethene on Cu( 1 10) at 100 K (226, 226a) has given a spectrum in which bands at 1522 and 126 1 cm are dominant with no absorption observed in the CH2-wagging region. This has been interpreted under the MSSR as representing a n-bonded species with the CC bond tilted to a high angle with respect to the copper surface rather than in the usual parallel orientation. At 175 K, more normal spectra are obtained. Ethene adsorbed on CdZnO at 300 K (52), Fig. 15E, gives closely similar wavenumbers and relative intensities to the spectrum on Cu( 100) with absorptions at 1550, 1290, and 920 cm- '-so much so that the spectrum may imply the dominance of (100) facets on the Cu particles. Ethene adsorbed on colddeposited Cu films studied by SERS gives bands at 1544, 1278, and 896 cm(227).A 3-monolayer covering of Cu on Ru(0001) gives a similar spectrum with somewhat shifted band positions at 1504, 1248, and 880 cm- (218). Infrared spectra do not appear to have been successfuly recorded on oxidesupported silver or gold, but a RAIRS investigation on Ag( 100) has given bands at 3075 and 940 cm- (228).The 3075-cm- band occurs only at high coverage at the same time as the 950-cm-' absorption diminishes. These changes are taken to denote tilting of the ethene molecule about the CC axis in order to remove steric strain. However, SERS on cold-evaporated films at 40 K clearly denotes the presence of n complexes on Ag at 1582, 1318, and 956 cm- and on Au at 1536, 127 I , and 934 cm-' (227).Similar SER spectra on cold-evaporated Ag have been obtained by other authors (229-233), and also on Ag colloids (227). SER spectra from n-complexes on roughened Au electrodes give bands . are somewhat near 1545/1535, 1288/1278, 900, and 425 cm-' ( 2 3 3 ~ )These potential-dependent with the higher-wavenumber bands growing in intensity at low potentials. A 1/12 monolayer of Au on Ru(0001) gives the somewhat different band positions of 1480, 1260, and 970 cm- (218), but again clearly from a n-complex. In the SER spectra the first two bands are strong and the third weak which, as is common in Raman spectra, is the opposite of the intensity pattern in VEEL or infrared spectra. However the same modes of vibration are active as predicted under the operation of the MSSR. Wolkow and Moskowits (233) have shown, through a study of the benzene/Ag system using both SERS and VEELS, that SERS is distinctly selective with respect to one of two different adsorbed species. However, in the cases of Cu, Ag, and Au, no vibrational spectroscopic technique has yet found other than n-complex formation. ~

'

'

'

'

'

'

'

6. Ethene on Metals-An Overall Perspective a. Single-Ctystal Metals. As emphasized in an earlier review (19, the understanding of the vibrational spectra from hydrocarbons on metal single crystals provides an essential information base for the interpretation of such

VIBRATIONAL SPECTRA OF HYDROCARBONS

63

spectra obtained from finely divided metal catalysts. The simplified spectra on the single crystals often provide reliable one-at-a-time identification of adsorbed species. In addition other experimental techniques such as LEED, photoelectron diffraction, STM, NEXAFS, etc. (Section V) can be applied in single-crystal work but not to finely divided catalysts. These can provide important information about the orientation of CC bonds with respect to the metal surface and often approximate values for CC bond lengths. By this means, for example, the CC bonds of n and di-o species have been confirmed to be parallel to the metal surface, and that of ethylidyne has been shown to be perpendicular to the surfaces of ( 1 1 1) or (100) planes [see (17) for a summary]. More recently, the CC bond of the di-o species from ethene on Ni(ll1) has been shown by photoelectron diffraction to be approximately parallel to a surface metal-metal bond (88), again as anticipated from vibrational spectroscopy. Single-crystal studies have been made of ethene on selected crystal faces of all the group VIIl and group IB (groups 8-11 on recent IUPAC numbering) metals except for cobalt and osmium. It is interesting that the latter metal has proved to be most effective in the preparation of inorganic metal cluster compounds which have formed the basis of a wide range of hydrocarbon ligands available for vibrational analysis (Table IV). At sufficiently low temperature, the breaking of CH bonds is likely to be inhibited, leading to nondissociatively adsorbed (CH2CH2) species in the form of di-o (type I spectra), (no)(type I’ spectra) ( I I 3 ) ,or n species (type I1 spectra) ( I 7). For the relatively close-packed flat surfaces [( 1 1 1 ) or (1 00) of fcc metals; (0001) of hcp metals], this is normally found to be the case experimentally. The exceptions are Ni( 100) at 150 K ( I 99) [where it seems probable that an ethyne species shares the surface with the di-o one (Section VI.B.3.a)l and for Ir( 1 I 1 ) (221, 222) (where ethylidyne has already formed at 180 K, possibly in conjunction with “CCH”). These two surfaces merit hrther study at lower temperatures. The type I spectrum of the di-a species has been found at low temperatures on most investigated crystal planes of Fe (234, 235), Ni, Ru, and Pt (Tables V and VI). The intermediate type I’ spectrum, which we have related to the (no) species, occurs on Fe( 1 1 l), Rh( 1 1 l), Ru(0001), and Pd( 100). In the Ru(0001) and Pd( 100) cases, however, different spectra have been obtained in different laboratories (13, leading to some uncertainty as to whether these type I’ patterns may alternatively arise from superimposed component spectra from di-o and n species ( I 7), or to experimentally variable contributions from impactexcited features (Section 1V.B). Such species could therefore profitably be reinvestigated, and the higher resolution of RAIRS would be particularly valuable in this respect. Of the group VIII metals (IUPAC Groups 8-10), only Pd exhibits a strong preference for n-bonding (type I1 spectra) over di-o, and in all three ( 1 1 l),

64

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

(loo), and (1 10) planes at low temperatures (Table VI)..This surface species is predominant on Cu(100) and Cu(ll1) and also, from SERS evidence, very probably in Ag and Au as well. All these metals have in common filled d’’ orbitals, Pd being the only group VIII metal with this electron configuration. The extra stability of the filled d ” orbitals would be expected to inhibit the substantial d-to-n* back-bonding that would be necessary to form di-a species. Type I1 spectra are also found on a number of other group VIII metals when 0 or C are present as coadsorbates (17). The latter probably attract electrons and hence once again reduce the degree of d-to-n* back-bonding. Preadsorption of H or K leads to spectra from the particularly weakly bonded n* species on Pt( 1 10) and Pt( 1 1 1j, respectively (Table V). Although a separately resolved n* spectrum has not been obtained, particularly weakly bonded ethene has also been adsorbed on Pd( 1 10) after H preadsorption (188, 189), together with the ready formation of quantities of ethane. Type I and type I1 spectra appear to coexist on the corrugated Pt( 110) and Pt(210) faces, as shown by Masel and his colleagues using VEELS (Table Vj. These systems could profitably be reinvestigated using the higher resolution of RAIRS, particularly because the abnormally low wavenumbers for the absorption from the coupled vCCICH2-wagging modes of the di-a species on Pt overlap the 1000-900-cm- I region where the most prominent absorptions of the n species are expected. When these types of experiments were first carried out, the question was of the type “Is undissociatively adsorbed ethene in the form of di-a or n complexes?’ The finding is that on a given type of surface such as (1 1 l), the answer varies with the metal. Thus for (1 1 1 ) surfaces the di-a form occurs on Ni and Pt but the n complex forms on Pd. Also, for a given metal, different planes give different species; thus, on corrugated planes, e.g., Pt(210), Pt( 1 lo), and Pd( 1 lo), TL and di-a complexes can coexist, presumably on the different sites presented by the top rows of metal atoms and by the inclined facets. It must therefore be concluded that the adsorption energies of these two types of species are not greatly different. In enthalpy terms, the stronger CC bond strength in the n complexes is probably rather well compensated by stronger carbon-to-metal bonding in the di-a case. At the extremes, the spectra of the di-a and 7~ complexes are clearly distinguishable as types I and 11, respectively. The type I’ spectra are intermediate in form; thus we have suggested that these may form a continuous series with type 11 spectra for C2H4 bonded to a single metal atom. If this is the case, in due course other examples may be found to bridge the remaining gap between type I’ and type I spectra, i.e., with unsymmetrically bridged species. However, there remains the possibility that in some cases the observation of type 1‘ rather than type I spectra is due to experimentally variable impact contributions (Section 1V.B).

65

VIBRATIONAL SPECTRA OF HYDROCARBONS

Moving to spectra at higher temperatures, with one recorded exception, all the low-temperature ethene di-a or n species on (1 11) planes first transform to ethylidyne. The exception is Ni( 1 1 1) where the new spectrum is that of ethyne, type A, corresponding to di-ddi-n surface bonding. This latter transformation, in which the di-a CC bond changes orientation from approximately parallel to perpendicular to surface M-M bonds, has been confirmed direction by photoelectron diffraction (87). In general, as the temperature is raised, increasing amounts of hydrogen are desorbed while the carbon atoms are retained by the surface. With the help of temperature-programmed desorption, it is possible to follow the evolution of the overall mean C : H ratio of the residual species. Different surfaces of the same metal give different spectra at higher temperatures, and these in turn require different postulates for the surface species. Different species of composition C2Hz have been variously proposed to be di-0-vinylidene on Pt( 100) (1 X l), di-ah-vinylidene on Pd( 1 1 1) (derived from ethylidyne), l,l, 1,2 tetra-a-vinylidene (ethylylidine) on the corrugated surfaces Pt(210) and Pt( 1 lo), di-aldi-n-ethyne on Ni( 1 1 l), di-n-ethyne on Ni( loo), and 1,1,2,2 tetra-a-ethyne on Pt(100) (1 X 1) (see Tables V and VI). These structural assignments take into account the arrangements of surface metal atoms (assuming no surface relaxation) as well as the positions and intensities of spectral features. The spectrum attributed to di-oh-vinylidene has also been observed on Pt( I 1 1) at 350 K after decomposition of ethylidyne generated from ethyne (acetylene) (141). CH (methylidyne) species are usually identified as giving strong absorptions in the 800-cm-’ region from the 6CH mode. Similar features with the addition of broad but weaker absorptions attributable to vCC in the 1350- 1 100-cm- region are attributed to “CzH” species. Because of the operation of the MSSR, the 6CH mode of methylidyne species is expected to show up in this manner only on corrugated or other nonflat surfaces. An alternative CH possibility is di-a bonding to two metal atoms and n-bonding to a third on (1 1 1 ) faces (210), as was first suggested by Demuth and Ibach (235a). The ‘‘C2H” species has been identified on several (1 10) and (1 1 1) surfaces (see Tables V and VI and earlier text). On flat surfaces the residual carbon forms graphitic layers at high temperatures. CZ species without hydrogen derived from ethene adsorption have instead been observed on corrugated Pt surfaces, in the vicinity of 600K, presumably bonded within the grooves. The same species clearly occurs from ethyne adsorbed on the nonflat Ni[5 (1 11) X (1 lo)] surface after adsorption at the very low temperature of 150 K (195). In a number of cases-viz. Ni( 1 1 l), Ni( loo), Pd( 1 11) (Table VI), Fe( 11 1) (234) and Fe( 100) (235)-broad and low-wavenumber vCH absorptions occur in low-temperature spectra. These are attributed to agostic (hydrogen-bond-like) interactions with surface metal atoms (99), leading to weakened and more

’

66

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

reactive CH bonds. These bands are described as from soft CH modes and tend to be stronger in off-specular VEEL spectra, possibly because the C-H...M interactions are at a shallow angle with respect to the surface. The case of cyclohexane (to be discussed in Part 11) is an exception as the CH bonds concerned are perpendicular to the surface. The soft modes then occur in both VEEL and RAIR spectra. It needs to be determined whether these features are associated with the predominant surface species tilted toward the metal surface, or whether they arise from adsorption on step or kink defects on the otherwise flat surfaces where such interactions would be geometrically easy. Such “hydrogen-bonding” vXH interactions are normally expected to give rise to particularly strong integrated absorptions per CH bond. For the predominant II, II*,or di-a associatively adsorbed ethene species, the first two (which are based on bonding to a single metal atom) would seem to have more flexibility for tilting toward the surface. This could occur either about the CC axis or about an axis perpendicular to CC but parallel to the surface. For the di-o species, the latter degree of freedom is unlikely to be available, and nonplanarity of the dimetallocyclobutane skeleton would have to be extreme (beyond the staggered configuration) in order to force CH bonds toward the surface. C-H...M interactions at defects are likely to occur at low coverage, whereas tilting on flat surface may be enhanced by steric effects at high coverage. In order to distinguish between the various possibilities, it would appear to be most profitable to make high-sensitivity on- and off-specular VEELS experiment in several cases using selectively deuterium-substituted ethenes as a hnction of coverage. The choice of Ni( 1 1 1) would involve a di-a species and that of Pd( 1 1 1 ), a II species. This topic will be discussed further below (Section VI.D.1.a) in the context of spectra from adsorbed 2-methylpropene. In other cases-e.g., Ni( 100) (199), Ir(l11) (221), Ir( 110) (223), or Rh( 100) (211)--low-coverage spectra differ substantially from these near saturation coverage. This also suggests possible strong initial adsorption on a minority of surface defects, giving dissociatively adsorbed species or, alternatively, simply greater degrees of dissociation on low-coverage surfaces with a multiplicity of adjacent reaction sites, possibly leading to surface reconstruction. Experiments with exceptionally flat and perfect surfaces as monitored by STM (140) would help to distinguish between these possibilities. b. Finely Divided Metal Catalysts. The spectroscopic data on different oxide-supported metals for the II to (na), di-o, and ethylidyne species are shown in Table VII. In general, the band positions for a given metal (often available down to the alumina cutoff at ca. 1100 cm-’) agree very well with those of the same species occurring on single-crystal planes. (i) Ethylidyne. The agreement between flat surfaces and supported particles particularly well exemplified by the data for the ethylidyne species, which on the

VIBRATIONAL SPECTRA OF HYDROCARBONS

67

finely divided metals are expected to be specific to ( 1 1 1) facets. That this is so has been nicely demonstrated in a recent study by Paul et al. (2356) on Hi3C= I2CH2 adsorbed on Pt/A1203 which clearly showed dipolar coupling of the 1345-cm- I absorption of ethylidyne, as expected for side-by-side adsorption on flat surface facets. Table VIl shows that for ethylidyne, the most environment-sensitive band is that from the vCC vibration which occurs in the 1090-1 155-cm-' region. These bands, observed on the (1 11) single crystals for the metals Ru, Rh, Pd, Ir, and Pt, occur at 1 120, 1 121, 1080, 1 165, and 1 130 cm- I , respectively; the corresponding values on the oxide-supported metals are (Table VIl) 1130, 1 1 15, 1090, 1155, and 1130 cm-I, the latter values being notably more precise because of the higher infrared resolution. Other principal absorptions in the 1340 (6CH3 s) and 2880 cm- (CH3 s) regions vary in a parallel manner. The higher precision of the infrared band positions and the extensive data recorded in the right-hand section of Table Vll provide a good opportunity to compare the relative strengths of interactions between ethylidyne and the different metal surfaces. It will be assumed that a lower vCC value goes with a lower CC bond order and stronger binding to the surface. It is seen that, within a given transition-element period, vCC consistently drops (and we infer that the binding to surface increases) on passing from lower to higher atomic numbers, i.e., from Co to Ni, Ru to Pd, and lr to Pt. The same behavior is observed for the ca. 1340- and 2880-cm-' absorptions. However, on going down the groups (e.g., Ni to Pd to Pt), it is the second transition element period that consistently give the lowest vCC values.

'

(ii) The R species. Good correspondences are also found between the spectral data for TI species on finely divided metals (left-hand section of Table V11) and those recorded on single crystals of Pt (Table V), Pd and Ni (Table VI), and Cu and Ag (Section VI.B.5). There are, however, species observed on the catalysts that have not been observed on single crystals [e.g. the (no)species on Ni/A1203] and vice-versa [e.g. the (no)species on Pd( loo)]. The R species occur on flat single-crystal faces, (100) or ( 1 1 I), of Cu, Ag, and Pd; but they also occur on cormgated (210) or (1 10) etc. planes of the group VIIl metals Pt, Pd, and Ni. The latter observations correlate with the fact that n species and ethylidyne frequently occur side-by-side on many of the group VIIl finely divided metals, suggesting that the R species interacts with the rougher sections of the surfaces of the metal particles. It is common practice to assess the strength of interaction of R species with the metal concerned by evaluating the reductions in wavenumber of the coupled vC=C/6CH2 modes from their values of 1623 and 1342cm-l for an unperturbed ethene molecule. By this criterion, adsorption on the group lb (IUPAC group 11) metals causes the least perturbations of the adsorbate

68

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

molecules, as would generally be expected. Within the group VIII metals and using the value of the 1 18&1260-cm- mode as the most sensitive parameter, Ir and the (no) species on Ni are assessed to be most strongly bound and Pd and a second n species on Ni most weakly bound. Ignoring the latter Ni species, which occurs only in minor amounts on Ni/A1203compared with the (no)species, the bonding pattern relative to metal is qualitatively closely similar to that discussed earlier for ethylidyne; i.e., smooth gradation obtains along the sequence of elements in a particular period, but within individual groups it is in each case the second-period element that has the most extreme value. However, the pattern is largely the inverse of that observed for ethylidyne. For the n species it is Pd that gives the least strong binding to a group VIII metal, whereas in the ethylidyne case this metal was inferred to give the most strongly bound species.

'

(iii) Theoretical considerations. In view of the experimental finding that Pd prefers n to di-o bonding, it would have been much less surprising if Pd had been shown to have the strongest rather than the weakest n bonding! The implication is that di-o bonding on Pd is quite exceptionally weak. There is, however, an alternative kinetically controlled explanation for the apparent preference of Pd(ll1) for the n complex. The n complex appears to transform directly to the ethylidyne species at higher temperature. But it has recently been shown on Pt( 11 1) that the n species makes this transformation via the intermediate di-o species (236). Perhaps the Pd(l11) situation is analogous, but the activation energy for transformation to di-a is high (because of the weakness of the n bonding) such that at the appropriate temperature the di-a species formed immediately converts to ethylidyne. It should be recorded that Wong and Hoffmann (236) have given theoretical consideration to the n versus di-a preferences of Ni, Pd, and Pt, using extended Huckel calculations within a tight-binding formalism. Their theoretical results are consistent with the experimental finding that Pd( 1 1 1) has greater preference for n relative to di-o bonding in comparison with Ni( 111) and Pt( 1 1 1). They show that s and p as well as d metal orbitals are involved in bonding with the n and n* orbitals of ethene. For Ni(ll1) and Pt(1 l l ) , this donation to n* contribution is high, leading to a preference for a 2-fold site, i.e., di-o bonding. With Pd(l1 l), on the other hand, the donation to n* is much weaker, a factor probably influenced by the fact that the Fermilevel on Pd ( - 1 1.5 eV) is at lower energy than those of Ni ( - 9.8 eV) or Pt ( - 10.8 eV). The same theoretical methods have been applied to ethylidyne absorbed in the Rh( 100) surface, with the conclusion, as experimentally found, that this species will be stable on the 4-fold site (237). However, the extended-Huckel treatment, as well as the atom-superposition and electron-delocalization molecular orbital (ASED-MO) method, predict that the CC bond length will be shorter, and hence the vCC mode higher in wavenumber on the 4-fold site, whereas experimentally

VIBRATIONAL SPECTRA OF HYDROCARBONS

the reverse is the case, namely 1 120 cmRh( loo), respectively (212, 214).

'

for Rh( 1 1 1) and 1015 cm-

69

' for

(iv) The di-a and di-a* species. It is more difficult reliably to pick out the absorption bands of di-a or di-a* species at the higher temperatures when ethylidyne is also present. This is because the vCH2 s and 6CH2 modes of this species absorb very close the vCH3 as and 6CH3as modes of ethylidyne on small metal particles, where the MSSR exerts little influence. Such data as are available are listed in the central section of Table VII. In the case of Pt/Si02 or Pt/A1203,the vCH2s mode of the di-a species on (1 1 1) facets can be picked out because of the one-to-one inverse intensity relationship to the formation of ethylidyne. The di-a* absorption in that case, which persists after ethylidyne formation, is probably also reliably identified on this large-particle catalyst. The di-a and di-a* identifications are probably also reliable on Pd, but on other metals the lower wavenumbers associated with the ethylidyne absorptions create difficulties. It would be valuable to fill some of the di-a/di-a* gaps in Table VII. It would also be of interest to extend the data from n-complex and ethylidyne species on the missing metals Fe and Os, and from n complexes alone for Co and Ru. Lowtemperature measurements are likely to be particularly fruitful so that some spectra could be obtained without overlap from ethylidyne features. (v) Ethyl surface species. Only limited success has been achieved in obtaining spectra of the intermediate species, presumably ethyl, on hydrogenating (CH2CH2) surface species to ethane. This probably because the rate-determining step is the conversion of C2H4 to C2H5, followed by very rapid conversion to ethane. Theoretical calculations for C2H4-derived species on Pt( 11 1) have suggested that the ethyl group is very unstable in the presence of coadsorbed hydrogen (238). More low-temperature hydrogenation experiments would be very valuable for this purpose. However, it is likely that the expected ethyl species has been detected on Pt/Si02, Pt/A1203 and possibly on Rh/Si02 (174, 47, 67). (vi) Temperature-dependent sequences of adsorbed species. The various adsorption characteristics of different metals can also be explored by comparing the temperature ranges during which a given adsorbed species displays its spectroscopic signature. At low temperatures most oxide-supported catalysts show the mutual presence of n or (no)species, probably on rough surface sites, and/or of di-a* and di-a species, of which the latter occurs on (1 1 1) facets. On raising the temperature on Pt and Pd, the di-a species clearly converts to ethylidyne, while the di-o*, where present, is retained. On Ni it seems that it is a (no)species that disappears as ethylidyne is formed at ca. 190 K. This is the more surprising because on Ni single-crystal surfaces it is the di-a species that is dominant. However, ethyl formed on a Pt(ll1) surface by photochemical

70

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

dissociation of ethyl chloride (239) has been shown by VEELS to transform at increasing temperatures, first to the R species, next to di-a, and then to ethylidyne. In the Ni case it is therefore possible that the (no) species transforms to ethylidyne via di-a, with the di-r7 to ethylidyne step being very rapid so that di-0 is not detected spectroscopically. Again, further low-temperature work is called for. Except for the case of Ni, R species are usually retained (sometimes in decreasing amounts) up to room temperature and above as coadsorbed species with ethylidyne. On Co/SiOz and Ni/Si02, proportions of ethylidyne species have also been converted into another methyl-containing species at room temperature. As hydrogenation at this temperature gives considerable amounts of n-butane in the gas phase, and n-butyl groups are retained on the surface, the new species is probably one involving surface dimerization. In only a few cases have the spectra of adsorbed ethene been explored much above room temperature, in contrast to the situation on single-crystal metals where measurements up to 600 K are commonplace. The metal-dependent temperature for some surface transformations are summarized in Table VIII, in comparison with similar reactions of the linear butenes (Section VI.C.2.c). At sufficiently high temperatures, e.g., ca. 430 K on Ni/SiO2 (32) or at ca. 500 K on Pt/Si02 (240),the CC bond of ethylidyne is broken to give CH4. TABLE VIlI Approximate Temperatures (K)at Which Several Reactions Involving Ethene (Cd or the Linear Buienes (C4) Occur on Metal Surfaces

Alkylidyne formation

Alkylidyne decomposition"

c 2

c4

cz

Ni

<230 230

<300 <200

280 250

Ru Rh Pd

<300 200 250

(<230)d (<220)" 6200

250

Co

Ir Pt

(
Hydrogenation to surface alkyls rather than to gasphase alkanes

c:

C2h

c 4

300

<300 250

<300 250

>300

<300

(200f <200

<300 240

<300

<300 <300

(6300)"

330

270 340

c 4

<300 400

Dehydrogenation of surface alkyls by evacuation

400

c 4

CH4 formation by hydrogenolysis

cz

c4

400

300 350

430

(500)'

400 600

Note. All temperatures are estimates for the substantial completion of the reaction in question. Usually, they involve interpolation between measurements at intervals of 100 K or more. "To C4 species for ethylidyne but to shorter chain species for higher alkylidynes. 'n-Butyl species from the dimerization of ethylidyne; the Cz ethyl species are not retained on hydrogenation. ' C 6 . "c3. 'c~,c~.'I~(I~~).

VIBRATIONAL SPECTRA OF HYDROCARBONS

71

(vii) Hydrogenation phenomena. At low temperatures, the addition of hydrogen on different metals leads in general to the gradual elimination of all adsorbed species, with the generation of ethane in the gas phase. This continues into the temperature range where ethylidyne is also present. In the latter situation, as might be expected, the nondissociatively adsorbed 7r and di-a species are eliminated from the surface by hydrogen much more rapidly than ethylidyne (20, 58). The rate of removal (and H/D exchange) with ethylidyne depends on the number of locally free sites, i.e., inversely on surface coverage (213).The relative rates of hydrogenation of 7~ and di-a species have not yet been evaluated. Ethylidyne is not strictly a “spectator” species because it is also reduced to ethane, but its proportionate contribution to the catalytic process must be very low. On Pt/Si02 (176, 164) and Pd/AI2O3 (177) with flowing reaction systems, it has been shown that rapid hydrogenation occurs in the absence of spectra from ethylidyne or other adsorbed species. However, on Pt/Si02 it was shown that at higher ethene/H2 ratios, first the ethylidyne and then the di-a* species began to accumulate on the surface, gradually displacing the ca. 2130-cm- vPtH absorption which has been found to be involved in the hydrogenation process ( I 78). These findings do not support the suggestion by Thomson and Webb (241) that the similarities in hydrogenation kinetic parameters during catalysis by a wide range of different transition metals could be understood if the reaction involved the participation of similar hydrocarbon residues in all cases. On the other hand, it is clear that at higher hydrocarbon/H2 ratios, relatively unreactive hydrocarbon species (mostly ethylidyne) do indeed accumulate on the surface in all cases. At higher temperatures increasing proportions of n-butane join ethane as gasphase hydrogenation products. On lowering the temperature to room temperature for spectroscopic measurements, physically adsorbed n-butane increasingly contributes to the spectra of adsorbed species. Further temperature rises lead to increasing amounts of chemisorbed n-butyl species in all cases, and finally to gas-phase methane through C -C bond scission. In early studies of adsorbed ethenes and other alkenes, it was noted that considerable intensity enhancements from adsorbed species occurred in the vCH region on hydrogenation (7, 32, 41 ). Partly, this could be accounted for by the replacement of weakly absorbing =CH bonds by those of an alkane nature (7) and by the generation of additional alkane-type CH bonds on hydrogenation. However, the degree of intensity enhancement was such as additionally to suggest the initial presence of large proportions of dissociatively adsorbed species (32). But this interpretation of these experimental results was seen to become unreliable on the later realization of the effect of the MSSR in reducing the intensities of spectra of adsorbed species relative to those of gas-phase products. It is probably only the adsorbed molecules around the edge of the

’

72

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

profile of a metal particle in the infrared beam that can effectively absorb radiation at near-grazing angles, and then only with perpendicular vibrational dipoles. At the other extreme, molecules adsorbed on facets perpendicular to the beam will be effectively invisible (94), as there will then be no electric vector of the incident light that is perpendicular to the metal surface, but these will also contribute gas-phase intensity on hydrogenation. Even when the hydrogenated species are retained on the surface, so that the MSSR consideration applies equally to each case, there can be remarkable structure-selectiveintensity effects caused by the selection rule. One effect is to cause hrther reductions in intensities from the v=CH modes from x-bonded species, which are essentially parallel to the surface, compared with those from adsorbed alkyl species that replace them. Another example of an intensity effect is provided by ethylidyne, where the MSSR eliminates the intrinsically strongly absorbing vCH3 as modes in favor of the usually weaker-adsorbing vCH3 s ones. Reflecting these factors, a marked adsorbed-species to gas-phase intensity ratio was found for ethene adsorbed on Pt/Si02 at room temperature (32). Hydrogenation led principally to gas-phase ethane and a vCH intensity ratio (afterhefore) hydrogenation was estimated to be ca. 10. On the other hand, for ethene on Ni/Si02, the hydrogenated species was largely retained on the surface as n-butyl, but there were also some n-butane gas-phase contributions; in this case, the intensity increase was by the lower factor of ca. 3 (32). (viii) Spectral variations related to differences in sample preparation. Differences in spectra that relate to the metal concerned, or to the temperature of measurement, have been discussed already. A number of other experimental variables could be of significance. These include (1) the choice of oxide used as the support; (2) the extent to which the catalyst sample, before ethene adsorption, is hydrogen-covered or hydrogen-depleted; (3) whether measurements are made on pressed or deposited films (by transmission spectroscopy) or alternatively on loose powders (usually by DRIFTS); (4) variations in the mean metal particle sizes; and ( 5 ) the extent to which the reduced metal particles are annealed. For four metals (Pt, Pd, Ni, and Rh) closely similar spectra from ethene have been obtained from samples with either SiO2 or A1203supports, in the region available to both, as shown in earlier figures. Although alumina is well known to be a more acidic oxide than silica, the difference does not seem to affect appreciably the spectra of species on the metal particles. Eischens and Pliskin (7) originally obtained different spectra on Ni/Si02, depending upon whether the catalyst sample was evacuated at the reduction temperature or at room temperature after cooling down in an atmosphere of hydrogen. They termed these samples “hydrogen-free” or “hydrogen-covered,” respectively. We prefer the term “hydrogen-depleted’’ for the former. Others have

VIBRATIONAL SPECTRA OF HYDROCARBONS

73

found it difficult to consistently reproduce these differences, even for Ni/Si02 ( 2 0 3 , and much less pronounced differences have been observed on other metals. It has, however, been suggested that variable proportions of n species in spectra from different preparations may be correlated with the temperature of hydrogen reduction and thereby with different degrees of hydrogen retention by the metal particles. The retention of hydrogen by finely divided samples may also lead to the presence of species different from those present in single-crystal planes, e.g., the occurrence of ethylidyne on Ni/Si02 but not on Ni( 1 11). No systematic differences appear to exist between spectra taken with presseddisk or deposited-film catalysts samples. As yet, measurements based on diffuse reflection from powders have been few. Such spectra on EUROPT-1 in Fig. 71 and Ni/Si02 in Fig. 13E exhibit the same absorptions as those of pressed-disk samples, but with somewhat different relative intensities. Systematic DRIFTS experiments would be of interest. One substantial difference between spectra from ethene adsorption on small or large particles is the occurrence in the former of prominent absorptions that are much weaker or missing in the latter. This is primarily a spectroscopic rather than a structural phenomenon, based on the operation of the MSSR. However, when comparisons can be made for similar temperature-reduction and other experimental conditions, it is observed that smaller metal particles tend to have stronger MSSR-allowed absorptions from n species and weaker ones from ethylidyne. As n species appear to be more abundant on corrugated single-crystal surfaces, this correlates with the expectation that small particles will have greater proportions of rough areas for n species adsorption and smaller amounts of flat (1 1 1) facets. This pattern is particularly prominent in the spectrum from a Pd/A1203 catalyst with a very low metal loading (Fig. 1OF). Through the high sensitivity of presentday FTIR, we now have the capability to obtain good spectra from such lowloading catalysts that are used in actual industrial processes. Only in the case of Pd/Si02 catalysts have experiments been made with samples that are left relatively unannealed after reduction. However, the spectroscopic differences proved to be dramatic in this case (Figs. 1OG and 1OC). Although clearly annealed and “broken-in” metal samples are of more interest from the catalytic point of view, it would be of interest to explore this phenomenon further using other metals. The major spectral differences in the Pd case imply very different surface-atomic arrangements on newly prepared catalysts, including the virtual absence of (1 11) facets.

C. THEHIGHERLINEARALKENES

In this and the following sections, we shall discuss the collected results obtained on the surfaces of different metals.

74 1.

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

Propene

a. Single-Crystal Results. Compared with the case of ethene, there are relatively few experimental results from propene adsorbed on metals in either the single-crystal or finely divided forms. VEEL spectra have been obtained from propene adsorbed on the singlecrystal planes Pt(l11) (210), Rh(ll1) (242, 243), and Ru(0001) (244) over substantial temperature ranges. A RAIR spectrum on Pt( 1 1 1) at 340 K (170, 245, 246) gives more precise values for the propylidyne band positions and is otherwise in excellent agreement with the corresponding VEEL spectrum down to the cutoff of the infrared detector near 900 cm- I . A RAIR spectrum of propene on Ru(000 1) at 230 K also shows the characteristic pattern of bands for . 16 summarizes schematically both propylidyne in the vCH region ( 2 1 9 ~ )Figure the low-temperature spectra, which are assigned to the di-o species by analogy with the ethene case, and intermediate-temperature spectra, which are assigned to propylidyne (see below). The similarities between the spectra within each group provide evidence for the structural relationships on the different metals. It should be noted that the propylidyne assignment applies to spectra on Pt( 1 1 1 ) at considerably higher temperatures than on Rh( 11 1) or Ru(0001). LEED studies by Koestner et al. (83) confirm the presence of propylidyne species at the appropriate temperatures on both Rh( 1 1 1) and Pt( 1 1 1). At higher temperatures the propylidyne species on Rh( 11 1 ) and Ru(0001) is clearly shown from spectroscopic evidence to decompose to ethylidyne at ca. 210 K and by 293 K, respectively. There is evidence from Pt/Si02 (next section) that such decomposition does not occur on Pt until ca. 400 K. Studies on Rh( 1 11) using CH2=CHCD3 as the adsorbate (242) demonstrate that it is the propylidyne C-C bond attached to metal that is broken during ethylidyne formation. Compared with the spectra of ethylidyne itself, the decomposition spectra on Rh(ll1) and Ru(0001) have additional absorptions at 740 and ca. 3005 cm-l for Rh( 1 1 I ) and at 790 cm-' for Ru(0001). These imply the presence of an additional alkene-like surface species from the breaking of the C-C bond of propylidyne. It is not clear whether this is from an initially formed C, species or from a subsequently polymerized species. It has been shown that in the Rh( 1 1 1) case the presence of surface hydrogen facilitates the propylidyne-to-ethylidyneconversion. Although no likely mechanism for this transformation has yet been suggested, it must involve multiple steps and probably either an initial isomerization or H addition at the a position, as propylidyne itself does not have CH bonds near the surface. It could be profitable to research for metastable intermediates, possibly allyl-like species that have been postulated for the butenes (next section). Experiments involving deuterium-substituted propenes, analogous to those carried out by Dent and Kokes (247) to investigate the adsorption of propene on ZnO, would be of

VIBRATIONAL SPECTRA OF HYDROCARBONS 3000

cm-1 2800 1600 1400 1200 1000

800

75

600

FIG. 16. A schematic summary of VEEL or RAIR spectra or propene on metal surfaces at different specified temperatures: (A) Cu film, 165K (2Jf); (B) P t ( l I l ) , 170 K (210); ( C ) Rh(l1 I ) , 100 K (242); (D) Ru(0001), 153 K (244); (E) Pt( I 1 I), 340 K (210); (F) Rh( I I I), 220 K (242); ( G ) Ru(0001). 233 K (244).The spectra are grouped according to the assigned structures, i.e.. (A) the n complex; (B, C, D) the di-rr complex; and (E, F, G ) the propylidyne complex.

interest, particularly using the high resolution of RAIRS. At higher temperatures again spectra are interpreted in terms of the presence of ‘‘C2H” and/or CH species on all three surfaces. A di-a species formed across the original double bond in propene would have no symmetry elements so that, in principle, all 27 modes internal to the hydrocarbon groupings could be active in the VEEL or RAIR spectra. In view of the limited resolution of the available VEEL spectra, it would be unrealistic to attempt a comprehensive vibrational analysis at this stage. However, the spectra observed are clearly from a substituted alkane surface species, and they exhibit all the strong features expected from a di-a species, viz. v C H ~ I V C H ~ ,

76

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

2970-2920; 6CH3 as/6CH2 s, 1450-1400; CH2 wag, 1240-1200; and CH3 rock and vCC modes, 1 100-870 cm- I . A detailed analysis is more worthwhile in the propylidyne case because a high-resolution RAIR spectrum is available on Pt( 1 1 1) and because the MSSR forbids out-of-plane modes with vibrational dipoles parallel to the surface in the RAIR spectrum. Furthermore, with the first CC bond of the M3(CCH2CH3) group essentially perpendicular to the surface and an assumed tetrahedral CCC angle, it is to be expected that v C H as ~ and 6CH3 as modes should be particularly strong because they are associated with motions more nearly perpendicular to the surface. The reverse is the case for the symmetrical counterparts. These factors have been considered in detail for the Pt case by Shahid and Sheppard (248) and by Chesters et al. (170, 245, 246). The following very satisfactory assignments have been made for the a' modes: 2961 (vCH3 as), 2921 (vCH2 s), 2865 (vCH3 s), 1450 (6CH3 as), 1407 (6CH2 scissors), 1303 (CH2 wag), 1103 (vCC), 1039 (CH3 rock), 929cm-' (vCC). The principal features in the spectra from Rh( 1 1 1) and Ru(0001) can be given analogous interpretations. Additional (m) or (ms) features between 825 and 765 cm-' in the spectra may be associated with the early onset of CH/"C2H" decomposition products. b. Finely Divided Metals-Spectra of the Initially Adsorbed Species. Figure 17 collects together spectra from the literature from propene adsorbed on finely divided metals (Pt 170, 248, 179; Pd 43; Ni 6, 71, 249). Palazov et al. ( I 71) have shown a similar spectrum on Pt/A1203to that in Fig. 17A, but did not list the peak wavenumbers. Earlier, Clark and Sheppard obtained less well defined spectra on Pt supported on porous silica glass (165, 250). Most of the spectra in Fig. 17 show their most prominent absorption in the vCH region at ca. 2960 cm- I , closely coincident with that from propylidyne on Pt(ll1) (see also Fig. 22B below). The next most prominent band in the propylidyne spectrum is at 1450 cm-', and this too is clearly present in the spectrum on Pt/Si02, Fig. 17A. As a result of a similar spectrum from propane on Pt/SiOz, which shows an even closer resemblance to that from propylidyne, Shahid and Sheppard (248) and Chesters et al. ( I 70) confidently concluded that this species is present in considerable proportion after the adsorption of each hydrocarbon on Pt/Si02 near room temperature. Features additional to those of propylidyne occur on Pt/Si02 in the vCH region in the form of a prominent band near 2920 cm- and a medium-strength band near 2880 cm- (Figs. 17A, B, D, and E). In the 6CH2/6CH3region, which has been investigated on Pt/SiO2 and Ni/A1203 (Figs. 17A and E), an additional prominent absorption occurs near 1350 cm-I, and there is probably intensity additional to that attributable to propylidyne near 1450 cm-'. All of these additional features can reasonably be accounted for in terms of the presence of

'

'

VIBRATIONAL SPECTRA OF HYDROCARBONS 3000

3000

2800

1600

1400

1200

2800

1600

c m - l 1400

1200

77

FIG. 17. Infrared spectra from propene on various metals near room temperature: (A) Pt/SiOz (179. 248); (B)Pd/SiO* (43); (C)Ni/SiOl (6,7); (D) Ni/Si02 (7)); (E) Ni/SiOz (249).

a di-a species as follows, even though we do not have a high-resolution RAIR spectrum of this species for comparison: 2920, vCHzMs (by analogy with ethene on Pt/Si02); 2880 cm-I, vCH3 s; 1450 cm-l, 6CH3 as; 1350 cm- I , 6CH3s. Marshall et af. (249)have suggested that the 1350-cm- I absorption may be partially attributable to the vCH3 s mode of a CH3 group attached to metal, resulting from the decomposition of propylidyne (see below). However, the characteristic wavenumbers for such a mode of CH3M species (Table 111) do not support this assignment; an alternative possibility would be the 6CH3 as mode of a surface CH3 group, but the intensity is unusually high under the MSSR for this alternative assignment.

78

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

If a n species from propene were present, this too could be contributing to some of the extra absorptions discussed in the previous paragraph. The weakness of the v=CH absorption above 2980 cm-' in the spectrum on Pt/Si02 (Fig. 17A) led Shahid and Sheppard to suggest that this is probably the least abundant of the three probable species. On the other hand, Palazov et al. ( I 71) have attributed the low-wavenumber shift of the vC0 absorption from carbon monoxide when coadsorbed with propene to the presence of some n-adsorbed species, and we have seen above that the MSSR reduces the intensity of v=CH modes. Marshall et al. (249) obtained very good-quality spectra from propene on Ni/Si02 (Fig. I7E). which clearly indicated the presence of strong contributions of propylidyne on initial adsorption. On standing, however, the absorptions attributable to propylidyne (indicated by I) gradually diminished in intensity while those, equally clearly from ethylidyne, grew (T). Such a transformation was earlier observed (see the previous section) by VEELS on Rh(ll1) and Ru(0001). Figure 17C shows a pioneering spectrum obtained by Pliskin and Eischens on Ni/SiO2 (6) under low resolution. Its profile is different from that of Fig. 17E, but it may represent a situation in which transformation to ethylidyne has proceeded much further (compare Fig. 13E from ethene on Ni/Si02). The spectrum also implies the presence of a high proportion of surface-alkene species than is indicated in the other spectra in Fig. 17. More recently, Munro and Ravel (251) have obtained a RAIR spectrum from propene adsorbed on an evaporated Cu film (Fig. 16A); this spectrum is clearly in large measure from a n complex and can therefore be used as a reference for searching for n spectral features in the spectra of Fig. 17. For a n-bonded species whose C3 skeleton is roughly parallel to the surface, it is the out-of-plane modes that are likely to be strongest under the MSSR. The strongest features in the spectrum should therefore be the adsorbate counterparts of the very strong yCHI yCH2 bands of propene that occur at 991 and 91 1 cm- I . These are clearly present as strong bands with the expected relative intensities at 959 and 887 cm-' in the spectrum of the adsorbed species. More detailed assignments can be made as follows: 3080, 3020, v=CH2/v=CH; ca. 2950, vCH3 as; 2916, vCH3 s?; 2847, 2 X 1440 in Fermi resonance with vCH3 s; 1557, vC=C/GCHz ; 1440, 6CH3 as in-plane; 1420, 6CH3 as, out-of-plane; 1370, 6CH3 s; 1267, GCH2/uC=C; 959, yCH=CH; 887 cm-I, yC=CH2. There is, however, one intensity anomaly with respect to MSSR expectations. The band assigned above, with query, to vCH3 s at 2916 had a strong intensity despite the quasi-in-plane nature of the vibration involved. According to the MSSR, it might have been anticipated that vCH3 as would be stronger than vCH3 s, rather than vice-versa. It is probably for the latter reason that Munro and Raval preferred the alternative assignment of the 2916-cm- I absorption to vCH3us. However, in propene itself the vCH3s absorption occurs at 2934 cm- and is exceptionally strong, whereas vCH3 us is weaker and expected to have a higher value near 2960 cm-I.

VIBRATIONAL SPECTRA OF HYDROCARBONS

79

Another spectrum from a n complex has been observed by SERS following the adsorption of propene on cold-evaporated Ag (229). The intensity distribution, as expected for a Raman spectrum, is different from that in the infrared spectrum, but it does show readily observed features at 1612 and 1300 cmfrom the coupled vC=C and 6=CH2 modes (1557 and 1267 cm-l on Cu). As the corresponding absorptions on Pt or Ni are expected to be at notably lower wavenumbers than on Cu or Ag, and as no absorptions are discernible between 1550 and 1460cm-l in the spectra on the group VIII metal surfaces, this negative result reinforces the view that n species are not abundant on the latter catalysts. A recent theoretical study by Delbecq and Sautet (252) has, in agreement, concluded that for propene on Pt( 111) the K species is less stable with respect to di-o than is the case for ethene; on Pd(ll1) the difference was much less marked. More low- and room-temperature studies of propene on several A1203-supported metals are needed to obtain reliable information about the nonpropylidyne species. Avery (43) and Ward ( 7 1 ) have shown that at room temperature closely similar spectra are obtained by the adsorption of cyclopropane or propene on PdSi02 and Ni/Si02, respectively. Isomerization of cyclopropane to propene most probably occurs as an intermediate situation. In the case of propene on Pt/Si02, spectra have been studied over the range from room temperature to 573 K (248). By 433 K, the propylidyne absorptions have weakened and those of ethylidyne have appeared. At 473 K, most of the surface alkane groups have decomposed, and at 573 K a single broad, weak absorption occurs near 3060 cm-’ in the whole range down to the SiOz cutoff at 1300 cm-’. This could be the vCH mode of the “C2H”-type spectrum obtained on Pt( 11 1) in this higher temperature range.

’

c. Finely Divided Metals-Hydrogenation of the Adsorbed Species. Hydrogenation on Pt/Si02 at room temperature led to the production of gas-phase and physically adsorbed propane. When this propane was removed by evacuation, only a very weak alkyl-like spectrum was left from residually chemisorbed species. This was little changed by the reintroduction of hydrogen (248). A methyl-rich spectrum from propene adsorption had been earlier observed after hydrogenation during work on Pt supported on porous glass (250). The marked contrast between this spectrum and that from a n-propyl group chemisorbed on a Ni catalyst had led to the hypothesis that the former arose from chemisorbed 2-propyl species. It was subsequently shown to be caused instead by physically adsorbed propane that was present because of the greater propensity for complete hydrogenation of the initial surface species on Pt. Hydrogenation at 373 K again led to propane and a very weak spectrum on evacuation (248). In this case, however, readmission of H2 at 473 K led to a more substantial CH2-rich alkyl-type spectrum that was completely removed by

80

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

evacuation at that temperature. Readdition of H2 at 573 K gave a somewhat weaker alkyl spectrum but with the addition of methane in the gas phase. The n-alkyl-type spectrum suggests the occurrence of a degree of polymerization at the higher temperature (>373 K) followed by partial C-C bond scission by hydrogenolysis at 573 K. Similar features occur in analogous spectra from but1-ene (see Fig. 21 below). The addition of hydrogen at room temperature to propene on Pd/Si02 (43) also led to a spectrum from physically and chemically adsorbed species. Evacuation led to complete removal of the spectrum, but readmission of Hz gave the n-propyl spectrum once again. The phenomena were repeated during several fH2 cycles, and the nature of the processes involved will be discussed in more detail in connection with analogous hydrogenation spectra from the adsorbed butenes. Much more of the adsorbed species at room temperature were removed by hydrogenation, as propane, from Pt compared with Pd or Ni. d. A x-Ally1 &$ace Species. Although derived from the thermal decomposition of 1-chloroprop-2-ene (ally1 chloride) on Ag( 110) at 180 K, rather than from propene itself, it is of interest that a surface a-ally1 species, q3-C3H5, appears to have been characterized (252a). At 300 K this species dimerizes to give a-adsorbed hexa- 1,5-diene. 2. Linear Butenes a. Single-Crystal Results. Studies of the vibrational spectra of the linear butenes (but-1-ene and cis- and trans-but-2-enes) have been made on the Pt( 11 1) surfaces by Avery and Sheppard (210, 253) using VEELS over a wide temperature range, and by Chesters, de la Cruz et al. (170) using RAIRS in the vicinity of room temperature. Chester, Horn et al. (254) have studied but-1-ene on Ru(OOO1) at 150 K by RAIRS and have also measured the 300 K spectrum. It was concluded that the VEEL spectra for each of these butenes on Pt( 1 1 1) at the temperature of ca. 170 K could be satisfactorily interpreted in terms of the presence of di-0 adsorbed species, different in structure for each adsorbed isomer as expected. At 300 K on Pt( 11 1) adsorbed but-1-ene was concluded from the VEEL spectrum to be very probably in the form of a n-butylidyne species corresponding to a loss of one H atom, as shown by TPD. At that temperature, however, cis- and trans-but-2-ene at full coverage gave spectra that were identical among themselves but very different from that from but- 1-ene. As the TPD evidence implied a loss of two H atoms, and because the spectrum was dominated by methyl absorptions, it was suggested that the structure of the surface species was (CH3CCCH3). Because no absorption band was observed above 1500 cm-', and arguing by structural analogy with the bonding of ethyne itself to the Pt( 11 1) surface (17), it was suggested that this substituted ethyne is

81

VIBRATIONAL SPECTRA OF HYDROCARBONS

di-ah-bonded to the surface. The same conclusion was reached from the RAIR spectrum of trans-but-2-ene on Ru(0001) at ca. 225 K (254~). By 450 K the VEEL spectra derived from all three linear butenes on Pt( 11 1) were the same, exhibiting a strong and sharp band at 720 cm- and medium-strength bands at 3070, 11 10, and 1410 cm- ’. A cis-disubstituted C-CH=CH-C skeleton was suggested for the structure in order to account for the prominent 720-cm- feature and for the C/H ratio determined by TPD. A more definite identification of this species would aid greatly in understanding the decomposition pathway of the linear butenes; similar experiments starting from perdeuterobutenes would also be valuable. The same intermediate spectrum was obtained at 385 K from adsorbed butadiene (253). At higher temperatures again, 500-600 K, the VEEL spectra implied the presence of more hydrogen-deficient species such as CH and ‘‘C2H’ in all cases. The RAIRS study on Pt( 11 1) at high coverage near room temperature gave much higher resolution spectra and in each case showed the same prominent absorptions obtained at 300 K by VEELS. The identical nature of the spectra from the cis- and trans-but-2-enes was confirmed. It was also shown that the adsorption of but-2-yne (dimethylacetylene) gave the same spectrum, thereby confirming the hydrocarbon structure of the adsorbed species. This spectrum finds very ready interpretation in terms of the proposed structure, viz. ca. 2930 (m,bd), vCH3 as; 2886 (vs), vCH3 s; ca. 1430 (m) 6CH3 as; 1354 (s), 6CH3 s; 1036 (ms) vCCICH3 rock. No assignment is given for the vC=C mode, but it should be noted that a closely similar spectrum from but-2-yne absorbed on Ni( 1 1 1 ) does give a weak additional band at 1580 cm- that could be from vC=C (255). If this is correct, it reopens the question as to whether the hydrocarbon species is di-ah- or only di-a-bonded to the Pt( 1 1 1) surface. Two metal substituents on C=C might be adequate to lower vC=C to ca. 1580 cm- I without invoking additional 7c bonding. A di-a form of binding with the carbon skeleton in a plane perpendicular to the surface would be quite consistent with the observed spectrum. Indeed, a surface-inclined d i d n species in principle has the additional allowed fundamentals of the vCH3 as, dCH3 as, and particularly of the CH3 rocking type, none of which has been observed in VEELS or RAIRS. In the VEELS study of but-I-ene on Pt( I 1 I ) at 300 K, it was suggested that the n-butylidyne species, which in principle can have trans and gauche conformational isomers, seemed to be predominantly in the more symmetrical trans form because of the relative lack of complexity of the spectrum. This was confirmed at higher coverage by RAIRS. Additional evidence was provided by the observation that particularly strong bands occurred at the positions expected for the vCH3 s and 6CH3 s modes whose vibrational dipoles should be strictly perpendicular to the metal surface, assuming tetradedral angles in the PtC4 skeleton. The spectrum of the presumed trans form can very readily be

’

’

82

NORMAN SHEPPARD AND CARLOS D E LA CRUZ

interpreted on this basis as follows: 2934 (ms), vCH2M s; 2874 (s), vCH3 s; 1379 (ms), 6CH3s; 1331 (mw), CH2 wag; 11 10 (m), vCC. The RAIRS study showed that at lower coverage on Pt( 111) additional absorptions appeared at ca. 2885, ca. 1430, 1354, and 1035 cm-', which were interpreted as arising from a proportion of gauche isomers of n-butylidyne. This assignment was supported by an apparent reversibility of the spectrum as a function of coverage. However, the present reviewers noticed, with surprise, that the additional absorptions-in terms of their precise positions, relative intensities, and bandwidths-are virtually identical with those listed above for the (CH3CCCH3)surface species from the adsorbed but-2-enes. It is also significant that the lack of symmetry of a gauche n-butylidyne species leads to the expectation of a considerably more complex spectrum, particularly in the low-wavenumber fingerprint region. This reinterpretation of the spectra implies that, although no spectroscopic evidence had previously been found for changes from but- 1-ene-derived to but2-ene-derived surface species on Pt( 1 1 1) at this temperature, such isomerism can nevertheless occur at lower coverage. As the (CH3CCCH3) surface species might be expected to take up more space on the surface than trans-n-butylidyne (it also involves an additional surface H atom), the driving force for the transformation could be a reduction of steric hindrance as the coverage is decreased. The transformation itself is a most remarkable one and would once again seem to initially require either internal isomerization of the n-butylidyne or (CH3CCCH3) species, or initiation by the addition of an H atom to the butylidyne. Possibly, a methyl-substituted ally1 species may be involved (see next section). As similar component spectra have been observed on Pt/Si02 and Ni/A1203(next section), the conditions for such a transformation could best be pursued on ( 1 1 1) planes of these metals using a range of deuterium-substituted butenes. In reverse, it should be noted that Koestner et al. (83) had earlier suggested from TPD and LEED evidence that the but-2-enes might occur on Pt( 11 1) in the form of butylidyne at ca. 300 K, as has been found spectroscopically on Pt/Si02 (see the next section). The VEELS study of but-1-ene on Ru(0001) gave a spectrum that is clearly from a n-butylidyne species at 150 K, with a single additional absorption at 2969 cm-I (vCH~ as) attributed to a residual proportion of the di-0 species. At 300 K the prominent absorption at 1350 cm-' was assigned to the breakdown of the Cq chain to give ethylidyne (254). A recent study of trans-but-2-ene on Ru(0001) (254a) gives a poorly defined spectrum at low temperature, but one at 250 K is clearly from the di-aln (CH3CCCH3) species. b. Finely Divided Metals-Spectra of the Initially Adsorbed Species. Figure 18 shows low-temperature spectra at 195 K from but-1-ene adsorbed on several metals [Pt (256), Pd (191), Ni (2501. The spectra on Pt and Ni are closely similar at this temperature. That on Pt is essentially unchanged at

83

VIBRATIONAL SPECTRA OF HYDROCARBONS

FIG.18. Infrared spectra from but-I-ene on various metals at low temperatures: ( A ) Pt/Si02, 195 K (256); (B) Pd/SiO2, 195 K (191); (C) Ni/Si02, 195 K (256).

room temperature while that on Ni is notably different on the Si02-supported catalysts (256). The interpretation of the spectra will be discussed below in relation to the room-temperature results, as will the rather different spectra on Pd/Si02. The latter spectrum shows a notably stronger v=CH absorption at 195 K. Figure 19 collects together many of the spectra in the literature from but- 1-ene adsorbed on oxide-supported metals near room temperature [Ni (7, 256, 54, 249), Pd (19 4 , Pt (71, 256, 257), Co (56), Ir (36)].Figure 19A is the pioneering spectrum from but-I-ene on Ni/Si02 obtained by Eischens and Pliskin (7). These authors were the first to point out that the spectra are virtually identical for all the linear butenes, i.e., that alkene isomerization is facilitated by this catalyst near 300 K. This observation has since been generalized to apply to the spectra of virtually all the oxide-supported metal catalysts near room temperature. For this reason, we have illustrated only the available spectra on the different metals from adsorbed but- 1-ene. Those from the but-2-enes sometimes differ in detail (2.53, particularly in exhibiting stronger absorptions at ca. 2890 and 1350 cm-I, which we have seen are from the (CH3CCCH3) surface species, and possibly from the coverage-dependent incomplete conversion of this species to n-butylidyne. Many of the spectra in the vCH region in Fig. 19 are dominated by three absorptions at ca. 2960, 2930, and 2875 cm-'. It is noteworthy that, whenever the ca. 2875 cm- absorption is weak, that at ca. 2930 cm- is also weakened. There is a consistent relationship between the strength of the ca. 2875-cm-' absorption and others at ca. 1380 (m) and 1330cm-' (w). The latter four absorptions, and their relative contributions, correlate extremely well with the RAIR spectrum from but-I-ene on Pt(1 l l ) , attributed to the trans form of n-butylidyne (see, for example, Fig. 22C below). The same applies to Ni/A1203 at room temperature, Fig. 19D, where Marshall et al. (249) show that a set of absorptions decrease together as a function of time, thus correlating extremely well with the n-butylidyne spectrum. A new absorption at 1334 cm- denotes the growth of ethylidyne. It was pointed out that, since propylidyne decomposes on Ni/A1203 to ethylidyne, the ethylidyne derived from n-butylidyne could

'

'

84

NORMAN SHEPPARD AND CARLOS DE LA CRUZ 3000

2800

1600

cm-l

1400

1200

1

I

I

I

I

I

I

I

I

Ni

I

x Ni

2

Ni N

,

In

1 1400

1600

s

-3

p 1200

cm-1

FIG.19. Infrared spectra from but-I-ene on various metals near room temperature: (A) Ni/SiOz (7); ( B ) Ni/SiOL (256); ( C ) Ni/SiOz (54); (D) Ni/AI2O3 (249); (E) Pd/Si02 (191); (F) Pt/SiOz ( 7 1 , 257). the dotted line is probably from H on Pt; ( G ) Co/Si02 (56). the dashed line indicates a variable-intensity feature; ( H ) Ir/SiOz, reprinted with permission from (36). copyright I972 American Chemical Society.

logically be formed via a propylidyne intermediate, i.e., n-butylidyne .+ propylidyne --* ethylidyne. Unfortunately, it is difficult to confirm the presence of the propylidyne intermediates as there are already strong bands from the butenes near 2960 and 1450cm-', which are the positions of the most prominent absorptions of propylidyne. An alternative possible source of ethylidyne is from a splitting of the central C-C bond of the (CH3CCCH3) species (the reverse of the dimerization of ethylidyne on Ni/Si02 discussed in Section VI.B.3.c), but this is less likely because the absorptions at 2886 and 1356cm-', which are characteristic of (CH3CCCH3)itself, appear unchanged while the n-butylidyne absorptions diminish. After allowances are made for the component spectra from n-butylidyne and the (CH3CCCH3) surface species, the room-temperature spectrum from butI-ene on Pt/Si02 shows that at least one additional surface species must be present for attribution to prominent absorptions at ca. 2965, 2930, and

VIBRATIONAL SPECTRA OF HYDROCARBONS

85

crn-l

3000

2800

1600

1400

1600

1400

'A.I V

)

N

N

e m

cm-'

FIG. 19+Continzred)

1450 cm- I . This could well be a di-a, or di-a* species adsorbed on non-( 1 1 1 ) facets. Weak absorptions at ca. 3025 and 3005 cm-l also indicate the presence of an alkene species, probably n-adsorbed. This could also contribute to the above three alkyl-group absorptions. It must be mentioned that, prior to the VEELS and RAIRS studies on Pt( 1 1 1) that emphasized the presence of propylidyne and n-butylidyne surface species derived from the corresponding alkenes, Campione and Ekerdt (54) were the first to apply MSSR reasoning to the spectra from the linear butenes on Ni/Si02. They had proposed an alternative trans (CH3CH-CHCH3) di-a surface species to account for the strong absorptions at 2875 and 1380 cm- Reasons for preferring the more recent alkylidyne-based structural assignments have been given (254). The realization that the closely similar spectra from but-I-ene and the but2-ene denoted isomerization on Ni/Si02 also led Campione and Ekerdt (54) to search for evidence of surface intermediates. Before pumping off the but- 1-ene gas phase, they observed three weak absorptions at 1560, 1545, and 1535 cm -

'.

'

86

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

in a clear region of the spectrum; these they attributed to allylic-type intermediates. These absorptions were from labile species, which were readily removed by pumping. Whether these absorptions correspond to ally1 or to particular II species would be best investigated on ( 1 11) surfaces. Our recent conclusion that effective isomerization can occur even on the Pt( 11 1) plane shows that it is no longer necessary to invoke diffusion between different types of particle facets (170, 257) in order to bring this about. In Fig. 19A the original spectrum on Ni/SiOz obtained by Eischens and Pliskin (7) is dominated by a broad and strong absorption centered at ca. 2890 cm-I, suggesting that on this metal the (CH3CCCH3) species may be more dominant than in the spectrum shown in Fig. 19D, which is from a catalyst with high Ni loading. The spectra on Pd/Si02, Figs. 18B and 19E, are different from most of the others. The former is, however, similar to one obtained on Pt/SiOz at 403 K (Fig. 20C below), suggesting that dehydrogenation occurs on Pd/SiOz even below room temperature, possibly helped by the high capacity of Pd to absorb hydrogen. On the other hand, a spectrum from pent-1-ene on Pd/Si02 is essentially “normal.” It would be advantageous to reinvestigate but1 -ene on Pd/SiOZ and/or Pt/AlzO3 over a range of temperatures, coverages, and annealing conditions. Shahid and Sheppard (257) have made a study of the spectra from the adsorbed linear butenes on Pt/SiOz between room temperature and 573 K. The spectra are illustrated in Fig. 20. Initial heating to 373 K indicates some removal of the prominent CH3 absorptions, from n-butylidyne or from (CH3CCCH3), and the growth of absorptions at ca. 2920 and ca. 1510 cm- I . These changes suggest that the first dehydrogenation process is to convert a proportion of CH3 groups to CHzPt. By 403 K most of the original alkyl-type absorptions have disappeared, to be replaced by broad and ill-defined absorptions in the 3000-2800-cm- I region of about half the integrated intensity of the originals. By 573 K the only remaining absorption is a broad band of medium strength centered at 3060 cm- I , which clearly originates from v=CH or v=CHZ groups. Campione and Ekerdt (54) have studied the spectra on Ni/SiOz between room temperature and 343 K. At 313 K a spectral profile in the vCH region was obtained that was similar to that of Fig. 19D. A similar spectrum was also observed in the 150&1300-cm- I region except for an additional substantial broad absorption at 1415 cm-I. The latter, which continued to grow up to 343 K, was suggested to be caused by “severely dehydrogenated“ surface species. We suggest this may be from -CHzNi groups. TPD studies over the same temperature range showed that as the gas-phase butenes declined in concentration, they were gradually replaced by n-butane, presumably formed by self-hydrogenation on the metal surface. Anderson and Ekerdt (56) on Co/SiOz also showed that the n-butylidyne absorption at 2872 cm- was selectively weakened in raising the temperature

’

87

VIBRATIONAL SPECTRA OF HYDROCARBONS

I

J

I 1600 1500 1400 320

1700

FIG.20. Infrared spectra of but-I-ene adsorbed on Pt/SiO, near room temperature followed by evacuation (A) and then after heating successively to the temperatures indicated: (B) 373 K; (C) 403 K; (D) 473 K; and (E) 573 K. At each temperature the gas phase was evacuated and the spectra measured after cooling down to room temperature. The spectra were ratioed against that of the clean catalyst (257).

from 293 to 355 K. In the gas phase the butene was again replaced by n-butane up to 373 K. Thereafter, it was gradually replaced by propane, probably via a propylidyne surface intermediate. A SER spectrum of but-1-ene on cold-evaporated silver at 60 K gives, as expected, a spectrum from a R complex with a 36-cm-’ lowering of vC=C (257a).

c. Finely Divided Metals-Hydrogenation and Subsequent Dehydrogenation of the Adsorbed Species. Eischens and Pliskin (7) showed that the hydrogenation of the adsorbed linear butenes on Ni/Si02 led to spectra as expected from the presence of chemisorbed n-butyl groups, i.e., the same pattern of spectrum that was described above as arising from a dimerized and then hydrogenated form of adsorbed ethene. Campione and Ekerdt obtained closely similar results from hydrogenation on Ni/Si02 at 301 K, and they added spectra in the 6CH2/6CH3 region (54). Similar spectra were obtained near room temperature on Co/Si02 (56) and Ir/Si02 (36). However, hydrogenation on Ni/Si02

88

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

at the lower temperature of 195 K (2.50, and on Ir/Si02 at 173 K, gave principally gas-phase or physically adsorbed n-butane and few residual chemisorbed species. Pt/SiO2, even at room temperature, gave principally n-butane and only weak residual alkyl species. Pd/Si02 at room temperature also gives mostly n-butane on hydrogenation of the adsorbed species. The general situation seems to be the occurrence of n-butane production at low temperatures and chemisorbed n-butyl at higher temperatures, with metals differing principally in the temperature range over which the change from butane to surface butyl occurs. For Ir, Co, and Ni this change occurs below, and for Pd and Pt above, room temperature. On hydrogenation of but-1-ene at 393 K, Pt/Si02 gave a spectrum indicating less n-butane [Fig. 21B (2501 but which, in both the vCH3/CH2 and 6CH3/CH2 regions, was essentially eliminated by evacuation at that temperature (Fig. 2 1C). On the other hand, the readdition of hydrogen at 523 K led to a substantial n-alkyl spectrum with a ratio of the vCH31vCH2absorption bands showing the

L-2

1700

1600 1500 1400 1

20

Rc. 21. Infrared spectra of but-I-ene adsorbed on Pt/SiOz near room temperature and after evacuation (A) followed successively by (9)addition of H2 at 393 K (C) evacuation at 393 K; (D) addition of H2at 523 K;(E) evacuation at 523 K; (F) addition of H2 at 593 K;and ( G ) evacuation at 593 K. All spectra were measured after cooling down to room temperature and were ratioed against that of the clean catalyst (257).

VIBRATIONAL SPECTRA OF HYDROCARBONS

89

retention of adsorbed species, possibly with longer-than-C4 alkyl chains. Evacuation at 523 K (Fig. 21E) again eliminated the spectrum, which was restored, however, by readdition of hydrogen at 593 K. At this temperature the presence of some gas-phase methane (a weak absorption at 301 9 cm- ’) showed that a small proportion of hydrogenolysis had occurred. The CH3/CH2 spectral eliminationhestoration phenomenon is a very remarkable one that requires further consideration. Early on, Eischens and Pliskin (7) had shown that the intensity of the n-butyl spectrum on Ni/SiO2 was reduced again after pumping H2 from the cell, but was in large measure restored by reintroducing H2. Only partial coverage of the surface by n-butyl gave more effective intensity reduction on evacuation. They concluded that the hydrogen was removed via vacant metal sites rather than directly from the adsorbed hydrocarbon groupings. Avery (43) had also earlier showed, from the study of propene and pent- 1-ene adsorbed even at room temperature, that Pd/Si02 catalysts are particularly effective in the complete removal of spectra in the vCH region after evacuation of H2, with full reinstatement on hydrogen reintroduction. He attributed this to the great capability of Pd to absorb hydrogen. In a quantitative study of the spectra after rehydrogenation following the adsorption of pent- 1-ene, he also established that the spectra of complete alkyl chains were removed and reinstated essentially simultaneously in a rapid “zip-fastener” manner (258). Because of the complete nature of the spectrum removal on evacuation, Avery suggested that the alkyl chains lay along steps in the metal surface so that each carbon atom in the chain could form two metal-carbon r7 bonds (or three at the end) when hydrogen was totally removed by evaluation. Because of the ease and generality of the process, Shahid and Sheppard (257) later suggested an alternative model in which the removal of one hydrogen per carbon converts the alkyl chain into a completely conjugated polyene that could form multiple n bonds parallel to any surface, such as ( 1 1 l), ( 1 10) or (loo), with straight chains of metal atoms. The great weakness of the resulting spectrum could partly be accounted for by the fact that, CH bond for CH bond, alkene vCH absorptions are about 5 times weaker than alkane ones ( I 72), and partly by the consideration that the MSSR dictates that v=CH absorption bands from CH bonds that are essentially parallel to the surface be further greatly weakened. When we proposed the conjugated poiyene model for the nonabsorbing species in the vCH region after removal of H2, we were not aware that the same model had been proposed much earlier by Shopov, Andreev, and Palazov (162) in connection with the adsorption of hex-1-ene on Ni/SiOz. They also favored a limited degree of n conjugation in the initially adsorbed species and had at that time no reason to postulate alkylidyne species. Also, they suggested that the ratio of absorption strengths for CH3/CH2 groups as a function of chain length was not always consistent with the presence of an n-alkyl group after

90

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

hydrogenation. They preferred the presence of more than one C-M bond in the hydrogenated (they termed it the “half-hydrogenated”) state. It is consistent with the conjugated-polyene interpretation after dehydrogenation that the adsorption of buta-1,3-diene, which is particularly likely to be n-bonded on Pd, gives negligible absorptions in the whole vCH region on initial adsorption on Pd/Si02 at room temperature, but strong n-butyl-like spectra on the addition of hydrogen (43). In hydrogenatioddehydrogenation reactions involving n-butyl groups, derived from the hydrogenation of buta- 1,3-diene on Rh/A1203, Basu and Yates (2.58~)showed that if CO was adsorbed after hydrogenation, then dehydrogenation was no longer possible. They concluded that the CO blocked the vacant sites needed to receive hydrogen atoms from the dehydrogenation reaction. Below the vCH region, long all-trans conjugated chains give no other infrared active modes until below the 1300-cm-’ cutoff of the silica, or the 1100-cm- ' cutoff of alumina, supports (259). Taking into account the MSSR, a particularly strong absorption would be expected from an out-of-plane CH mode near 10I5 cm - I , and this should be readily detectable in a RAIRS study on a Pd( 1 I 1) plane. n Adsorption of such a conjugated system could also be fairly flexible in relation to the spacings between the metal atoms in an underlying row. There is a reasonable match between the separation of adjacent C=C bonds of such a species (ca. 0.25 nm) and the metal-metal bonding distances (0.254.28 nm). Eischens and Mertens (260) carried out a joint magnetization and gravimetric analysis of the adsorption of but-1 -ene on Ni/Si02 at room temperature. Measurements were made after initial adsorption, after saturation with H2,and after the evacuation of H2, the last cycle being repeated more than once. By gravimetric means they deduced that the removal of the alkyl spectrum by the evacuation of hydrogen corresponded to the loss of a single hydrogen per carbon atom. This is in agreement with the above hypothesis of the conversion of the alkyl chain to a conjugated alkene. To determine the number of sites, i.e., bonds to the surface, occupied by the adsorbed hydrocarbon species (the number of the latter being determined gravimetrically) Eischens and Mertens used an experimental criterion developed by Selwood (261). This assumes that either a NiH or a NiC bond to the surface will localize one electron in the nickel surface and thereby reduce the magnetization by the same amount. They therefore concluded that each initially adsorbed but-1-ene molecule made on average 6.6 bonds to the surface, partly in the form of NiC and partly NiH, caused by dissociative adsorption. This figure agrees reasonably well with the presence of n-butylidyne (3 + 1 bonds) or the (di-oh) (CH3CCCH3)species (3 2 bonds). The magnetization did not change on introducing hydrogen, presumably because H atoms occupied the site freed by the conversion of the above species to n-butyl. After subsequent removal of hydrogen from the system, the demagnetization was reduced to the equivalent of ca. 4 bonds per original adsorbed butene. This is a

+

VIBRATIONAL SPECTRA OF HYDROCARBONS

91

reasonable expectation for a n-conjugated system (2 electrons donated per C=C group for pairing with uncoupled d electrons). For adsorbed ethene the number of bonds per adsorbed molecule was estimated at 4.6 in comparison with (3 + 1 ) for the ethylidyne species. 3.

>C4 Linear Alkenes

a. Single Crystal Results. Pent-1-ene seems to be the only molecule in this class that has been studied for adsorption on a metal single crystal. In their VEELS study on Pt( 1 1 l), Avery and Sheppard (210) qualitatively interpreted spectra obtained at 200 and 300 K as representing di-0 and pentylidyne adsorbed species, respectively. The relative simplicity of the 300 K spectrum suggested that, at the high coverage used, the pentylidyne surface species had adopted the planar zigzag conformation. The alternating relative intensities of the 6CH3s mode from ethylidyne (s), propylidyne (w), butylidyne (s), and pentylidyne (w) supported this supposition and, as will be discussed below, finds confirmation in the spectra on finely divided metal catalysts. At higher temperatures, decomposition pathways appeared to have common features with those from but- 1 -ene discussed above. b. Finely Divided Metals. Figure 22 compares the room-temperature alkylidyne spectra from ethene, propene, and but-1-ene on Pt/Si02 and from pent1-ene on Pd/Si02 (Fig. 2 of reference 257). This very clearly shows the alternation in intensities of the 6CH3 s mode noted on Pt( 1 1 1 ). Because of the much beffer resolution of these infrared spectra in comparison with VEELS, the analogous expected intensity alternation of the 6CH3 s and 6CH3 as modes is also evident in this figure. Figures 23A and B show the vCH regions of the near-room-temperature spectra of pent-1 -ene adsorbed on Pd/Si02 before and after hydrogenation ( 4 3 ) , and on Ni/Si02 after hydrogenation only (7). Another spectrum on Pt/Si02 with coadsorbed CO has been recorded by Palazov et al. ( I 71) but without a listing of band positions. The strongest absorption at 2968 cm- obtained after initial adsorption on Pd/Si02 is as expected for the planar zigzag pentylidyne species; but, unfortunately, we do not have a RAIR spectrum on Pd( 1 1 1) with which to determine which of the other absorption bands are associated with that species. The absorption at 3020 cm- I , however, clearly shows that some surface alkene groups are present, probably the n-adsorbed species. The spectrum after hydrogenation on Ni/Si02 exhibits a reasonable n-pentyl profile when the limited resolution available at that time is taken into account. On Pd/Si02, the initial addition of hydrogen led to a spectrum with a higher ratio of the 2945-cm- I compared with the 29 12-cm- I absorption, characteristic of CH3as and CH2 as modes, respectively, than would be expected for n-pentyl.

'

92

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

cm-’ I

1500

2800

3000 I

1300

I

t

c

.-0 ul .-ul

E

ul

c

0

L

I-

i

3000

2800

I0

cm-’ FIG.22. A comparison of the infrared spectra, all near room temperature for (A) ethene, (B) propene, (C) but-I-ene, all on WSiO,; and (D) for pent-I-ene on Pd/SiO2. The spectra show a clear alternation of intensities of the vCH&CH3s (0)and vCH3as/GCH3 us absorptions (0) as a function of the number of carbon atoms. The shaded spectra were obtained by RAIRS for the adsorption of the same hydrocarbons on Pt( I I I ) and attributed to alkylidyne surface species with planar zigzag carbon skeletons (257, 43. 170).

This is probably because of the physical adsorption of some n-pentane molecules with their two methyl groups. After evacuation and then readdition of hydrogen, a n-pentyl profile was again obtained and retained through repeated fH2 cycles. As discussed in the butene examples, the spectrum of the surface alkyl groups was much reduced (eliminated on the PdSi02 case) in evacuation.

VIBRATIONAL SPECTRA OF HYDROCARBONS

0 N

::z - m N

3000

mnN

g"N

N

2800

3000

cm-,

93

'H2 2800

FIG. 23. Infrared spectra from pent-I-ene adsorbed near room temperature before and after hydrogenation on (A) Ni/Si02 (7) and (B) Pd/SiOz (43).

It was with pent-1-ene on Pd/Si02 that Avery demonstrated the zip-fastener relationship of the hydrogenatioddehydrogenation processes on surface n-alkyls (258). He also pointed out that hydrogenation led to a wavenumber lowering from 2967 to 2945 cm-I, and a broadening, of the absorption for the vCH3 as mode on hydrogenation of the initially adsorbed species (43). He attributed this effect, which seems to be at its most pronounced on Pd, to interaction of the CH3 group with the surface after hydrogenation. It is noted that, of the transition metals, it is Pd that most strongly absorbs hydrogen. It was the virtually identical infrared spectra from the hex- 1 -ene, hex-2-ene, and hex-3-ene isomers when adsorbed on Ni/Si02 (7) that first led Eischens and Pliskin to the conclusion that the finely divided metal catalysts involve isomerization of the monohexenes. Their mutual spectrum and that obtained after hydrogenation are shown in Fig. 24B. Similar spectra at higher resolution were obtained by Shopov, Andreev, and Palazov (162) and by Erkelens and Liefkens (262);these are illustrated in Figs. 24C and D, respectively. The spectrum by Shopov et al. shows rather more absorption above 3000 cm- attributable to alkene-type surface species; however, because probably all the spectra in Fig. 24 were initially recorded on substantially sloping backgrounds from the metal catalysts themselves, it is easy to over- or underestimate the strengths of weak absorptions in this region. Erkelens and Liefkens (Fig. 24D) obtained somewhat different spectra on H-covered and H-depleted surfaces. Weaker absorptions occur on the H-free surface at 2926 and 2851 cm-I, the v C H ~as and vCHzs absorptions of alkyl chains; but the vCH3as and vCH3s absorptions at 2963 and 2876cm-' are relatively unchanged. It seems that some additional nonterminal attachment of the C6 chain to the surface occurs on the H-free surface.

'

94

NORMAN SHEPPARD AND CARLOS DE LA CRUZ cm-1

3000

2800

3000

2800

cm-1 FIG. 24. Infrared spectra from hex-I-ene adsorbed near room temperature before and after hydrogenation on (A) Pt/A1203 ( / 7 / ) ; (B) Ni/Si02 (7); (C) Ni/Si02 (162); and (D) Ni/SiOz on a hydrogen-covered surface (upper spectrum) and on a hydrogen-depleted metal surface (lower spectrum) (262).

After hydrogenation and the pumping-off of any resulting n-hexane, a vCH band profile is retained, closely similar to that expected from a surface n-hexyl group. Erkelens and Liefiens (262) made the interesting observation that attempted “hydrogenation” with deuterium led initially to a large increase in intensity in the vCH region and a generation of only a weak broad absorption in the vCD region. The addition of hydrogen to the hex-I-ene initially absorbed on a D-covered Ni/Si02 surface also led to a large absorption increase in the vCH region to give a spectrum virtually identical with that in Fig. 24D obtained after hydrogenation. It was concluded that the formation of new CH bonds by hydrogenation was more efficient that that of CD bonds. It is possible that, in the experiment involving the addition of gas-phase D2, it is the surface hydrogen atoms originally formed from dissociate adsorption of the hexene that remain immediately adjacent so that, under pressure from more adsorbed D2 or H2, they

VIBRATIONAL SPECTRA OF HYDROCARBONS

95

are most rapidly transferred back to the carbon skeleton. The hydrogeddeuterium exchange and addition processes are very slow, involving reaction for more than 20 hours. Similar spectra were obtained from hex- 1 -ene adsorbed on Pt/Si02 at 305 K coadsorbed with CO [Fig. 24A (I 7f)]or on Pt/AI203 at 473 K (48), although in neither case were precise band positions recorded. Spectra on Pt/A1203 at the higher temperature of 473 K by Baumarten and Weinstrausch (48) also showed absorption in the 175O-1400-cm-' region. After initial adsorption, followed by flushing with helium to remove physically adsorbed species, the spectrum showed (in addition to the usual three absorptions in the vCH region) a substantial absorption tail above 3000 cm-' and a broad absorption at ca. 1560 cm-I. These may be from n-adsorbed or dehydrogenated surface species. Additionally, this spectrum showed expected absorptions at ca. 1460 (ms) and 1380 cm-' from the 6CH&CH3 as and 6CH3 s modes, respectively. At room temperature on Ni/Si02 the corresponding absorptions were listed at 1466 and 1376 cm-', but not illustrated. Spectra from hept- 1-ene, hept-2-ene, and hept-3-ene have also been obtained by Shopov et al. (161, f 6 2 ) on Ni/SiOz. 4.

Linear Dienes

a. Buta-1.3-diene. It seems that the only study of buta-1,3-diene on a single-crystal surface has been on Pt( 1 11) investigated by VEELS (253). Two overlapping sets of absorption bands varied in relative intensity between 170 and 300 K. One set was well assigned to vinyl and the other to an alkane group. It was suggested that at 170 K the surface species involved di-a bonding across one double bond, with the other remaining as a vinyl group. At 300 K it was considered that a greater proportion of a di-(di-a)-bonded, i.e., 1,2,3,4-a-bonded (but still nondissociatively adsorbed) species was present. At 385 and 450 K, spectra were shown similar to the analogous spectra from the linear butenes. Among the finely divided catalysts, Ward studied buta- 1,3-diene adsorbed on Pt/SiO2 and Ni/SiOz (71); Avery on Pd/Si02 (43); Soma on Pd, Ni, and Co/A1203 (263); and Basu and Yates on Rh/A1203 (264). The results are somewhat different on the different metals, and more work needs to be done before definitive structural assignments can be made. However, some regularities are worth noting at this stage. The room-temperature spectra are most easily discussed. On Ni and Co (7f,263), it is clear that considerable polymerization has occurred to give lengthening polymethylene chains evincing the usual strong absorptions at ca. 2925, vCH2 as; 2854, vCH2 s; and 1450 cm- 6CH2 in the region down to the 1300-cm-' SiOz cutoff. In addition, there is a set of absorptions on Ni characteristic of non-n-bonded vinyl groups (3075, v=CH2; 30 10, v=CH; 1650, vC=C; 1407, 6=CH2; 1308,6=CH) with analogous bands on Co/Si02.

'

96

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

Hydrogenation led to substantial additional absorption from -CH2 - groups and the elimination of the vinyl absorptions. Weak bands at ca. 2960 and 1370 cm- from methyl groups also occurred. On PdSiO2 (43, 263) at room temperature no discernible spectra were obtained after adsorption; but after hydrogenation, strong absorptions occurred from surface-attached n-alkyl groups of limited chain length. As was mentioned above in the context of the adsorbed linear butenes (Section VI.C.2.c), it seems that in the absence of hydrogen in the gas phase, the strong capability of Pd to absorb hydrogen leads to dehydrogenation to give linear polyenes who spectra are very weak for MSSR reasons. Buta-l,3-diene on Pt/Si02 at room temperature ( 7 1 ) shows a poorly defined spectrum in the vCH region with alkene- and alkane-type components. On hydrogenation, well-defined n-butyl spectra appear (as also from but- 1-ene), showing that on this metal no polymerization has occurred. Rh/A1203 also showed no polymerization, but in fact the reverse (264). At room temperature the spectra did not differ greatly from those of adsorbed but1-ene in the vCH region; but absorptions at ca. 2880 and 1340 cm-', which grew in intensity up to 350 K, showed the presence of a growing fraction of ethylidyne, i.e., C2 surface species. Between the lower temperatures of 200 and 250 K, absorptions from this catalyst, such as those listed above, denoted the presence of non-n-bonded vinyl groups. In the same temperature range, additional absorptions at ca. 1580 (ms), 1430 (ms), and perhaps 1380cm-' (m) occurred at very similar positions to analogous absorptions in the spectra on Pd/A1203 and Ni/A1203at 241 and 238 K, respectively. A band at 1320 cm- in the spectrum on Rh/A1203at 225 K, associated with the other three by Basu and Yates (264), did not appear in the spectra on Pd or Ni. Because of the 1580-cm- absorption, it is most probable that the above trio of bands shows the presence of vinyl groups that are n-bonded to the surface. In their second paper on buta-1,3-diene adsorbed on Rh/AI203, Basu and Yates ( 2 5 8 ~ )concluded that self-hydrogenationldehydrogenation reactions occurred on the surface above 250 K, presumably involving different adsorption sites. They also showed that added hydrogen led to adsorbed n-butyl groups but only above 230 K. We infer from the spectra that the first self-hydrogenation product was adsorbed but- 1-ene and that the methyl-rich spectrum first observed on adding H2 to the system at 230 K was from physically adsorbed n-butane. SER spectra of butadiene on cold-evaporated Ag at 60 K (257a) have been assigned to a mixture of R- and di-n-bonded surface species. Hence a number of possible surface structures for adsorbed buta- 1,3-diene have been proposed, including mono-n or di-a adsorption involving one vinyl group, di-n or di-(di-a) (i.e., 1,2,3,4-tetra-a) absorption involving both vinyl groups, and possibly a metallocyclopentane after dehydrogenation which could lead to the formation of polymethylene chains. To these should be added a

'

'

'

VIBRATIONAL SPECTRA OF HYDROCARBONS

97

CH2=CH-CH2-CM3 structure, i.e., an unsaturated analog of butylidyne, the possible source of the absorptions from non-n-bonded vinyl listed above for Ni/Si02 and Co/Si02. Further progress in structural interpretations probably requires spectra on simplified single-crystal systems. Such low- and room-temperature spectra using the higher resolution of RAIRS would be particularly valuable on three of the metals, Pt, Ni or Pd, and Rh, which give different results at room temperature on the finely divided metals. b. Hexa-l.5-diene. Spectra from this nonconjugated diene have been reported on Ni/SiOz by Shopov et al. (162) and by Erkelens and Liefiens (262). The spectra at room temperature were said to correspond closely to the same spectra obtained from each of the hexenes, with the presence of methyl absorptions proving evidence for isomerization. However, a well defined spectrum from the di-n-bonded form of hexa-1,Sdiene has been observed on Ag( 110) at 300 K ( 2 5 2 ~ ) .

D. BRANCHED-CHAIN ALKENES 1. 2-Methylpropene (Isobutene) This first of the branched alkenes is capable of giving symmetrical adsorbed species and hence spectra that are not too complex. Its spectra therefore merit individual attention. a. Single-Ciystal Results. 2-Methylpropene has been studied by VEEL on Pt( 1 1 1 ) between 170 and 420 K (210) and on Ni( 11 1) between 80 and 180 K (265). It has also been investigated by RAIRS on Pt( 1 1 1 ) at 90 K and room temperature (266) and on Ru(0001) between 90 and 300 K (254). The on-specular VEEL spectra at 170 K on Pt( 1 1 1) and at 80 K on Ni( 1 1 I ) have similar patterns of prominent bands [2920 (m), 1470 (m), 1090 (s), 800 (m), and 460 cm-I (vs) on Pt(ll1); 2910 (m), 1450 (m), 1055 (s), 760 (m), and 445 cm- (vs) on Ni( 11 l)]. The RAIR spectrum at 90 K on Pt( 1 1 1) has better resolution than in VEELS and has corresponding prominent features at 2910 (s), 1426 (mw), and 1062 cm-l (s) in the accessible region down to ca. 850 cm-l (266).That on Ru(0001) at 90 K has a closely similar pattern of features [2886 (s), 1046 (s)]. It seems probable that the spectra on all three metals at these low temperatures are derived from the same type of adsorbed species. By analogy with the cases of adsorbed ethene on Pt( 1 1 1) and Ni( 1 1 l), and by taking into account the relative simplicity of the present spectra, it was suggested by Avery and Sheppard (210) that this is the di-o species. Other authors have followed this assignment, but Hammer et al. (265) have pointed out, with the help of

'

98

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

analogous spectra from (CD3)2C=CH2, that both alkene CH bonds appear to exhibit “soft” vCH modes which show up prominently in off-specular spectra. “Soft” modes are associated with “hydrogen-bond-like” interactions with metal atoms and the spectra from (CD3)2C=CH2 show no features from “free” vCH modes. Hammer et al. conclude that in some way the CCH2 of the molecule must be tilted downward toward the surface. It is difficult to envisage this occurring sufficiently from a di-a on the flat (1 11) surface, at least without surface reconstruction. Such tilting could more readily occur about the single metal atom to which a R- or dimethyl-substituted metallocyclopropane species could be coordinated. In the present context, the vCH frequencies of the terminal CH2 group seem to be remarkably low, at 2580/2660 cm- to be from a n complex. Furthermore, no absorption has been observed in the 1650-1450 cm region for attribution to a a-bonded C=C group. Steric effects between the surface metal atoms and the bulky methyl groups could cause some tilting of the C=CH2 group toward the surface, but seem unlikely to be sufficiently strong to cause such a strong lowering of the soft K H modes. Also in this connection it should be recalled that the spectrum of ethene itself on Ni( 1 11) shows such soft modes. Could there be a coexistence of metallocyclopropane and di-a species from ethene on this surface? Or could type I spectra-of which that from ethene on Ni( 111) is a typical example-be simply an extreme form of the type I’ spectra that we have associated with metallocyclopropane structures, and not from a di-a structure after all? We recall the earlier ambiguities in relating the spectra of the model di-a or metallocyclopropane osmium-based model compounds to typical type I spectra (Section IV.C, p. 26). At the higher temperature of 300 K on Pt( 1 1 1) a different, but still relatively uncomplicated, spectrum was obtained that Avery and Sheppard found to be consistent with the presence of the 2-methylpropylidyne (isobutylidyne) structure. This species has no conformational isomers associated with the carbon skeleton and the prominent absorption bands (at 2970, v C H as; ~ 1460,6CH3 as; and 1010 cm-I, CH3 rocking) are as expected according to the MSSR for the alkylidyne. The RAIR spectrum on Pt( 1 11) at room temperature was in agreement. Similar spectral features were observed on Ru(0001) and Ni(ll1) at the lower temperature of 180 K, particularly the dominant CH3 as absorption in the RAIR spectrum of Ru(000 1). However, the VEEL spectrum at this temperature on Ni(ll1) lacked the strong feature at 800cm-’ observed in the 300 K spectrum on Pt( 11 1). Hammer et al. doubted the retention of any vCH absorption in the VEEL spectrum from (CD3)2C=CH2 at 180 K and observed a weak band at 1595 cm- I , off-specular. They therefore preferred the interpretation of the spectrum on Ni( 11 1 ) in terms of a d i - a h [(CH3)2C=C] surface species. The RAIR spectrum on Ru(0001) at 300 K exhibited a single prominent absorption at 1366 cm- I , suggesting the occurrence of skeletal decomposition

’,

~

’

VIBRATIONAL SPECTRA OF HYDROCARBONS

99

to the ubiquitous ethylidyne. The spectrum from 2-methylpropene on Pt( 1 1 1 ) at 420 K was tentatively interpreted, in view of a TPD-estimated overall surface composition of C4H4.5, in terms of a CH(CHM2)3 species with 3-fold axial symmetry. b. Finely Divided Catalysts. Figure 25 shows collected spectra from 2-methylpropene adsorbed on SO2-supported catalysts of several metals [Pt (71, 269, Pd (43), Ni (71), Ir (36)]. Shahid and Sheppard (267) have most recently discussed the spectrum on Pt/SiO2 in detail. The room-temperature spectrum is shown in Fig. 25A. On heating to 373 K, this spectrum substantially simplified and gave dominant hydrocarbon absorptions at 2960 and 1450 cm- I , as expected under the MSSR for a 2-methylpropylidyne species. The 373 K spectrum on Pt/SiO2 is also closely similar to those obtained at room temperature on Pd/Si02 and Ni/Si02 (Figs. 25B and C). Additional absorptions on 2800

3000

N

3000

cm-'

1600

1400

"N"8

%

N

m

tH2 2800

1600

1400

cm-'

FIG.25. Infrared spectra from 2-methylpropene (isobutene) adsorbed near room temperature. and with hydrogenation spectra inset, on (A) Pt/Si02 (267); (B) Pd/SiOz (43);(C) Ni/Si02 (71); and (D) Ir/Si02, reprinted with permission from (36).copyright 1972 American Chemical Society.

100

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

Pt/Si02 at room temperature at 2970, 2925, 1045, and 1375 c m - ' were attributed to the presence of R- and di-a-bonded adsorbed species. The strong 6CH3 s feature at 1375 cm-' is particularly as expected for the di-a species, with its CH3C groups oriented at a high angle with respect to the surface. It may have been the conversion of di-a to the alkylidyne species, such as occurs for ethene on Pt/Si02 somewhat below room temperature, that accounts for the intensification of the 2960-cm- ' absorption from the alkylidyne on heating to 373 K. The profile of the spectrum at room temperature from Ir/SiO2 (Fig. 25D) resembles that from Pt/Si02 except for a more prominent vCH absorption at 2920 cm- ', possibly from the presence of a higher proportion of di-a species. As in the case for adsorbed but-1-ene on Pt/SiO2, the broad absorption at 1600cm-' in Fig. 25A is attributed to the presence of chemisorbed bridged hydrogen. It intensifies on heating up to above 473 K, and this would be consistent with additional surface hydrogen from thermal decomposition of the hydrocarbon species. At the same time, the alkyl absorptions become weaker and broader while a broad absorption from v=CH and/or v=CH2 grows to become the strongest spectral feature at 573 K. The inset spectra in Fig. 25 were observed at room temperature in the vCH region after hydrogenation ( + H2), followed by evacuation ( - H2) in order to remove physically adsorbed 2-methylpropane. Unlike the case of the linear butenes, considerable intensity is retained after the 4zH2 procedure and the species in question is clearly still a methyl-rich one, probably the alkyl group 2-methylpropyl. On Pt/Si02, hydrogenation was also studied at 373 K and above. In this case, the residual adsorbed species after 2-methylpropane removal at 373 K was weak and no longer methyl-rich. Readdition of hydrogen at 473 K led to a substantial spectrum of an unbranched n-alkyl nature, indicating the occurrence this time of skeletal surface isomerization. The same type of spectral removal ( - H2) and restoration ( + H2) thereafter occurred as for the adsorbed linear alkenes. Once again, some methane production, implying C -C hydrogenolysis, occurred at 573 K and above. 2.

Other Brunched-Chain Alkenes

The only higher branched alkene that appears to have been studied on a single-crystal plane is 2,3-dimethylbut-2-ene on Ni( 1 1 l), for which the VEEL spectrum has been measured at 80 K (255). At monolayer coverage, it gives a strong methyl-rich spectrum with absorptions at ca. 2870 (ms), 1470 (s), ca. 1400 (ms), 1060 (s), 690 (w), cu. 360 (s), and 320 cm-' (s). The first three prominent absorptions are those of the methyl groups and the strong 1060-cm absorption is probably from a coupled CH3 rocklvCC mode. It seems most probable that this is from the di-a species, although steric strain between vicinal methyl groups is more likely to give a surface species of C2 rather than C2,, symmetry.

'

VIBRATIONAL SPECTRA OF HYDROCARBONS

101

The adsorption of 2-methylbut-2-ene on Ni/SiO2 and PtISiO2 (71) and on Ir/Si02 (36), of 3-methylbut-1 -ene on Ni/SiO2, and of 4,4-dimethylpent-l -ene on Pd/SiOz (43) have been recorded, principally in the vCH region. For the most part, these give spectra dominated by absorptions from constituent CH3 groups, and it is difficult to draw specific structural conclusions from the data. Hydrogenation leads to slow desorption of the corresponding alkanes.

E. CYCLICALKENES The spectra of these will be discussed in Part I1 because of their intermediate status between cycloalkanes and aromatic hydrocarbons.

F. COMPARATIVE REACTIVITIES OF HYDROCARBON SPECIES ADSORBED ON DIFFERENT METALSURFACES Many of the same reactions of adsorbed hydrocarbon species occur on the surfaces of different metals but over differing temperature ranges. A lower temperature of completion of a given reaction implies a higher reactivity toward the product in question. In Table VIII are collected together the approximate completion temperatures of six different reactions on oxide-supported metals. A comparison is made between the analogous reactions involving C2 species derived from the adsorption of ethene and those involving C4 species from the linear butenes. In a few cases, analogous data are used from C3, Cs, or C6 species where the C4 data are incomplete. The experimental data are fragmentary and only approximate; nevertheless, some interesting trends can be discerned. It is satisfactory to note, for example, that the quoted temperatures for particular reactions involving C2 or C4 species are usually closely similar within the typical estimated uncertainties of f30 K. This implies that the reactivity on a particular metal depends principally on the functional group attaching the hydrocarbon species to the surface, as in M3CCH2R’ for the alkylidynes or MCH2R’ for the surface alkyls (R’ = alkyl). In the case of alkylidyne decomposition, it should be noted that different reactions are involved for the C2 and C4 species. It is apparent that Pt is rather generally the least reactive of the group VIII (IUPAC 8-1 0) metals, as its reaction-completion temperatures are substantially the highest. Only the temperature of alkylidyne formation is an exception to this generalization, where Pt is not notably different from the other metals. Palladium seems to be a particularly effective dehydrogenation metal in alkyl dehydrogenation on evacuation or, for the C4 species, alkylidyne formation from

102

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

di-a species. It has been suggested that this is related to the unique capacity of Pd to absorb, as well as to adsorb, hydrogen.

G . HIGHERALKENES ON METALS-AN OVERALL PERSPECTIVE Because relatively few VEEL or RAIR spectroscopic studies have been made for the higher alkenes (substituted ethenes) adsorbed on metal single crystals, the situation for the identification of the chemisorbed surface species is much less complete than is the case for ethene as the adsorbate. Furthermore, the number of conceivable surface structures is much greater than in the ethene case including, for >C, species, the effects of isomerization of the adsorbing species. Nevertheless, substantial progress has been made. The VEELS and RAIRS studies of propene on several metal surfaces, albeit all close-packed ones [more work on fcc (1 10) and (100) planes would be valuable], seem to have established sound criteria for identifying alkylidyne species on single-crystal or finely divided metals. A similar study of a series of alkenes on Pt( 1 I l), again by VEELS and RAIRS, also helped to generalize the conclusions from propene to the longer-chain alkenes. However, as hydrocarbon species on Pt seem to exhibit reactivity at higher temperatures than on most other metals (Section F above), similar studies on several single-crystal planes of other metals such as Ni and Pd would be very worthwhile. Nickel is likely to illustrate greater general reactivity, and Pd a greater propensity for dehydrogenation processes. At lower temperatures, spectra taken on these three metals, Pt, Pd, and Ni, might also permit the identification of the several possible substituted (CH2CH2) types of surface species, i.e., of the n, metallocyclopropane, or di-a types. Our earlier discussion of the case of 2-methylpropene adsorbed on Ni( 1 1 1) highlighted the continuing spectroscopic uncertainties in distinguishing between these species derived from the higher alkenes. For this purpose the capacity of both VEELS and RAIRS to give results in the fingerprint region below 1300 cm- is of importance. For finely divided metals the use of the more transparent A1203 support would be advantageous. For these further explorations the new higher-resolution techniques of VEELS (although unlikely to compete with RAIRS in this one respect) are of complementary importance for sensitivity reasons, and because off-specular studies can provide information about the modes with vibrational dipole changes more parallel than perpendicular to this surface. The off-specular studies are also more widely useful for the identification of vCH “soft” modes, with their implications about the nature and orientation of species with respect to this surface. A careful choice of the adsorbates to be studied could accelerate our understanding of this research field. Propene and 2-methylpropene have the advantages that they offer no spectroscopic complications from conformational isomerism within the hydrocarbon part of the adsorbed species, except perhaps

’

VIBRATIONAL SPECTRA OF HYDROCARBONS

103

in relation to nonplanarity of the CzMz dimetallocyclobutane skeleton in the case of the di-o species. In general, the spectra become more difficult to interpret as the number of carbon atoms increases, and even the linear butenes bring into play conformational mobility and/or isomerization reactions. Clearly, the most interpretable results are likely to be obtained from studies of the linear butenes, in normal and selectively deuterium-substituted forms. Although at sufficiently low temperatures all the individual linear alkenes preserve their identity as chemisorbed species, one of the clear-cut conclusions from work to date is that-even for the relatively unreactive Pt surface-at room temperature the finely divided metals cause virtually complete isomerization between but-1-ene and the cis- and trans-but-2-enes. The extent to which this can even occur on the close-packed Pt( 1 1 1) face at lower coverages at 300 K is a current question. Attempts to identify intermediates in the isomerization process, possibly of an allylic nature, would be a matter of priority on single crystals or finely divided metals. The conditions under which the hydrogenation of the higher alkenes leads to gas-phase alkanes or to surface-anchored alkyl species remain to be identified. High coverage may be a factor in inhibiting the replacement of the last carbonmetal bond by CH in order to give the gas-phase product, but this needs to be investigated further. Also on single-crystal surfaces, using RAIRS or VEELS, it should be possible to confirm or deny the alkane-to-polyalkene transformation that has been postulated as a means of accounting for the experimentally wellestablished, and highly-efficient, zip-fastener type of dehydrogenation of surface n-alkyl groups on the removal of gas-phase hydrogen.

VII.

Conclusions

In this article (Part I) we have comprehensively reviewed the structural implications of the vibrational spectroscopic results from the adsorption of ethene and the higher alkenes on different metal surfaces. Alkenes were chosen for first review because the spectra of their adsorbed species have been investigated in most detail. It was to be expected that principles elucidated during their analysis would be applicable elsewhere. The emphasis has been on an exploration of the structures of the temperature-dependent chemisorbed species on different metal surfaces. Particular attention has been directed to the spectra obtained on finely divided (oxide-supported) metal catalysts as these have not been the subject of review for a long time. An opportunity has, however, also been taken to update an earlier review of the single-crystal results from adsorbed hydrocarbons by one of us (N.S.) (I 7). Similar reviews of the fewer spectra from other families of adsorbed hydrocarbons, i.e., the alkynes, the alkanes (acyclic and cyclic), and aromatic hydrocarbons, will be presented in Part 11.

104

NORMAN SHEPPARD A N D CARLOS DE LA CRUZ

Knowledge of the chemisorbed species present on a surface provides an essential database for the investigation of catalytic and other reaction mechanisms in which they are involved. The elucidation of reaction mechanisms in turn requires studies of chemical kinetics. An increasing number of such kinetic studies are now being made which involve vibrational spectroscopy and hydrocarbon adsorbates, and these will also be reviewed in Part 11. Overall perspectives of the results from ethene and the higher alkenes have been attempted in Sections VI.B.6 and V1.G. What has become clear, particularly in the context of hydrocarbon adsorption, is that the study of spectra on single-crystal surfaces is of great assistance in finding the correct interpretation of the more complex multispecies spectra obtained from finely divided metal catalysts. This has only become possible by the development of VEELS and RAIRS, the latter allied with the Fourier-transform methods that have also transformed the quality of the spectra from metal-particle catalysts obtained by transmission infrared spectroscopy. The use of RAIRS in turn has emphasized the general significance of the MSSR. The majority of identifications of the structures of adsorbed species have to date been made by vibrational spectroscopy. This is because, once spectral patterns have been assigned to particular species, the analysis is very rapid and can be carried out in the presence of more than one type of surface complex. The assignment of spectral patterns has in the past been made using the infrared spectra of ligands of known structures on metal-cluster model compounds. This method finds its ultimate origin in the use of diffraction methods, in these cases usually X-ray diffraction. It would be even more satisfactory if model spectra could be assigned from structures that are directly determined on metal surfaces. So far, relatively few such hydrocarbon structures have been determined in this way by LEED (83, 84). However recent advances in diffraction techniques, including tensor-LEED and photoelectron diffraction (PED), are likely to greatly improve the situation in the near future. The newer PED technique is particularly welcome as it provides local-site information and is not dependent on the occurrence of regular arrays of adsorbates. It is some forty years since Eischens, Pliskin, and Francis in 1954 made the break-through of obtaining spectra from chemisorbed monolayers on metal/ silica catalysts. The subsequent incorporation of three experimental developments (FTIR, VEELS, and RAIRS) and a theoretical understanding (MSSR) has led in the intervening years to tremendous advances in the field of spectroscopic research that they pioneered-so much so that a suggestion in 1954 that we might have reached our present capability by 1995 would almost certainly have been met with disbelief. But much remains to be done in the vibrational spectroscopy of chemisorbed hydrocarbons because Nature has the capacity to match scientific advances by revealing sophisticated new phenomena for investigation. In this review, we have been at pains not only to summarize what has

VIBRATIONAL SPECTRA OF HYDROCARBONS

105

been achieved but also to point to some of the more promising pathways ahead. ACKNOWLEDGMENTS The authors are very grateful to the following persons who have given permission to us, for this review, to reproduce spectra from their publications or theses redrawn to a uniform format for comparison purposes: Dr. N. R. Avery, Prof. E. Baumgarten, Prof. G. Blyholder, Prof. R. P. Eischens, Prof. J. G. Ekerdt, (the late) Dr. J. Erkelens, Dr. G. Ghiotti, Prof. H. Knozinger, Dr. D. 1. James, Dr. G. S. McDougall, Prof. B. A. Morrow, Dr. J. D. Prentice, Dr. M. Primet, Mrs. A. Lesiunas, Prof. D. Shopov, Dr. Y. Soma, Prof. M. Trenary, Dr. J. W. Ward, and Prof. J. T. Yates, Jr. We are also indebted to the following copyright holders of the publications concerned for their permission to reprint the spectra in this review with references cited in figure captions: Academic Press (Figs. 7B, 9A, 13A. 13B, 14B, 15A, 15B, 178, 17C, 19A, 19C, 23A, 23B, 24B, 24D, 25D); Akademiai Kiado, Hungary (Figs. IC and 24A); the American Chemical Society (Figs. 6D, 7F, 7H. IOD, IOE, 12C, 12D, 12E. 13D, 14C, 14D, 14E, 15C, 15D, 19H, 25D); the American Institute of Physics (Fig. 17C); Baltzer Scientific Publishing Co. (Figs. 14F, 17E, 19D); Bulgarian Chemical Communications (Fig. 24C); Elsevier Science (Figs. 71, 75, 13E, 15E, 17A); Journal de chimie physique (Fig. IOF); the National Research Council of Canada (Fig. 25A); the Royal Society of London (Figs. 6A, 7A, 7E, IOB, 12A, 13C, 18A, 18C, 19B); and the Royal Society of Chemistry, London (Figs. 6B, 6C, 7G, 9A, 19F, 20, and 21). One of us (N.S.) thanks the U.K. Science and Engineering Research Council for a series of research grants that supported the work of his laboratories at the Universities of Cambridge and East Anglia in this research area. We are both very grateful to the Royal Society, London, and to the Consign0 Nacional de lnvestigaciones Cientificas y Tecnologicas of Venezuela for supporting two transatlantic exchange visits which assisted greatly in the writing of this review article. REFERENCES 1. Sabatier, P., and Senderens, J. B., Compf. Rend. 1358 (1897). 2. Bond, G. C., “Heterogeneous Catalysis; Principles and Applications,” Oxford Univ. Press, Oxford, 1974. 3. Kemball, C., Ah: Cafal. 11, 223, (1959). 4. Burwell, R. L. Jr., Acc. Chem. Res. 2, 289 (1969); Catal. Left 5, 237 (1990). 5. Eischens, R. P., Pliskin, W. A,, and Francis, S. A,, J. Chem. Phys. 22, 1786 (1954). 6. Pliskin, W. A,, and Eischens, R. P., J. Chem. Phys. 24, 482 (1956). 7. Eischens, R. P., and Pliskin, W. A,, Adv. Catal. 10, 1 (1958). 8. Bellamy, L. J., “Infrared Spectra of Complex Molecules,” Vol. I , 3rd Ed. Chapman & Hall, London, 1975. 9. Lin-Vien, D., Colthup, N. B., Fateley, W. G., and Grasselli, J. G., “Infrared and Raman Characteristic Frequencies of Organic Molecules,” Academic Press, New YorWLondon, 1991. 10. Nakamoto, K., “Infrared and Ramam Spectra of Inorganic and Coordination Compounds,” 4th Ed. Wiley-Interscience, New York, 1986. I / . Skinner, P., Howard, M. W., Oxton, 1. A,, Kettle, S. F. A,, Powell, D. B., and Sheppard, N., J. Chem. Soc. Faraday Trans. 2, 77, 1203 (1981). 12. Ibach, H., Hopster, H., and Sexton, B., Appl. Sur- Sci. I, I (1977). 13. Bertolini, J. C., Dalmai-lmelik, G., and Rousseau, J., Stir- Sci 67, 478 (1977). 14. Francis, S. A,, and Ellison, A. H., J. Opt. SOC.Amer. 49, 131 (1959). 15. Chesters, M. A,. Pritchard, J., and Sims, M. L., J. Chem. SOC.Chem. Commun. 1454 (1970). 16. Chesters, M. A,, J. Electron Spectrosc. Relat. Phenom. 58, 123 (1986).

106

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

f7. Sheppard, N., Ann. Rev. Phys. Chem. 39, 589 (1988). f8. Willis, H. A., Van Der Maas. J. H., and Miller, R. G. J. “Laboratory Methods in Vibrational

Spectroscopy,” Wiley, Chichester, 1988. 19. Sheppard, N.. Chenery, D. H.. Lesiunas, A,, Prentice. J. D., Pearce, H. A., and Primet, M., in “Molecular Spectroscopy of Condensed Phases” (P. Grosmann, S. G. Elkomosa and T . Ringeissen, Eds.), p. 345. Elsevier, Amsterdam, 1975. 20. Prentice, J. D.. Lesiunas. A.. and Sheppard, N., J. Chem. Soc. Chem. Communn. 76 (1976). 21. Ibach, H., and Mills, D. L., “Electron Energy Loss and Molecular Vibrations.” Academic Press, New York. 1982. 22. Ibach, H., J. Electron Spectrosc. Relaf. Phenom. 64/65, 819 (1993). 23. Frederick, B. G., Nyberg. G. L., and Richardson, N. V., J. Electron Spectrosc. Reluf. Phenom. 64/65, 825 (1993); Frederick, B. G., and Richardson, N. V., J. Elecfron Spectrosc. Relut. Phenom. 73, 149 (1995). 24. Chesters, M. A., and Sheppard, N., in “Spectroscopy of Surfaces” (R. J. H. Clark and R. E. Hexter, Eds.) p. 409. Wiley, Chichester, 1988. 25. See the appropriate chapters in (i) “Vibrational Spectroscopy of Molecules on Surfaces” (J. T. Yates, Jr., and T. E. Madey, Eds.) Plenum. New York, 1987; (ii) “Vibrational Spectroscopy of Adsorbates” (R. F. Willis, Ed.). Springer-Verlag. Berlin, 1980 or (iii) “Vibrational Spectroscopy for Adsorbed Species” (A. T . Bell and M. L. Hair, Eds.). Amer. Chem. SOC. Washington, D. C., 1980. 26. Krasser, W., Fadini, A., Rozenmuller, E., and Renouprez, A. J., J. Mof. Sfrucf. 66, 135 (1980). 27. Moskovits, M., and DiLella, D. P., J. Chem. Phys. 73, 6068 (1980). 28. Pockrand, I., Pettenkofer, C., and Otto, A.. J. Electron Specfrosc. Relat. Phenom. 29, 409 ( I 986). 2Y. Krasser, W., Ervens, H., Fadini, A., Renouprez, A. J., J. Raman Spectrosc. 9, 80 (1980); Krasser, W., and Renouprez, A. J., Solid Sfafe Commun. 41, 231 (1982). 30. Sheppard, N., Specfrochirn. Acta A45 (supplement), 149 (1990). 31. Bond, G. C., “Catalysis by Metals.” Academic Press, New York, 1962. 32. Morrow, B. A.. and Sheppard, N., Proc. R. Soc. London A31 1,391 (1969); J. Phys. Chem. 70, 2406 ( 1966). 33. Kember, D. R., and Sheppard. N., J. Chem. Soc. Farada-v Trans. 2 77, 1309 (1981). 34. Kember, D. R., Ph.D. Thesis, University of East Anglia, 1976. 35. De La Cruz, C., and Sheppard, N., Specfrochim. Acfa. 50A, 271 (1994). 36. Ravi, A., and Sheppard, N., J. Phys. Chem. 76,2699 (1972). 37. Ravi, A., and Sheppard. N., J. Catul. 32, 289 (1971). 38. Pearce, H. A., Ph.D. Thesis, University of East Anglia, 1974. 3Y. Dunken, H., Schmidt, K.. and Hobert. H. H., Z. Chem. 4, 312 (1964); Z. Chem. 3, 398 (1963). 40. Erkelens, J., and Liefkens, Th. J., J. Catal. 8, 36 (1967). 41. Erkelens, J., and Eggink-Du Burck, S. H.. J. Catal. 15, 62 (1969). 42. Palazov, A., Harson, M. S.. and Shopov. D. M., J. Catal. 36, 251 (1975). 43. Avery, N. R., J. C a r d 19, I5 (1970). 44. Reiicha, R., Collect. Czech. Chem. Commun. 37, 1607 (1972). 45. Primet, M., and Mathieu, M-V., J. Chim. Phys. 72, 659 (1975). 46. Blyholder, G.. Shihabi, D., Wyatt, W. V., and Bartlett, J., J. Cafal 43, 122 (1976). 47. Soma, Y., J. Catal. 59, 239 (1979). 48. Baumarten, E., and Weinstrausch, F., Specfrochim. Acra 35A, 1323 (1979). 49. Haaland, D. M., SurJ Sci. 102,405 (1981). 50. Beebe, T. P. Jr., and Yates, J. T., Jr., J. Phys. Chem. 91, 254 (1987).

VlBRATfONAL SPECTRA OF HYDROCARBONS

107

Parker. W. L., Siedle. A. R., and Hexter, R. M., J. Amer. Chem. Soc. 107, 4584 (1985). Ghiotti, G., Boccuzzi, F., and Chiorino. A,, SurL Sci. 178, 553 (1986). Hanson, D. M., Udagawa, Y., and Tohji, K., J. Amer. Chem. Sac. 108, 3884 (1986). Campione, J. J.. and Ekerdt, J . G., J. Catal. 102, 64 (1986). Lapinski, M. P., and Ekerdt, J. G., J. fhys. Chem 94, 4599 (1990). Anderson, K. G., and Ekerdt, J. G., J. Catal. 116, 556 (1 989). Mohsin, S. B., Trenary, M., and Robota, H. J., J. fhys. Chem. 92, 5229 (1988). Mohsin, S. B., Trenary, M., and Robota, H. J., J. Phys. Chem. 95, 6657 (1991). Szilagyi, T., J. Mol. Struct. 174, 395 (1988). Binet, C., Jadi, A,, and Lavalley, J.-C., J. chim. fhys. 86,471 (1989). Hensley, D. A., and Kesmodel, L. L., Sur-.’Sci. 231, 361 (1990); J. fh.v.s. Chem. 95, 1368 ( I 991 ). 62. Holmes, P. D., McDougall, G. S., Wilcock, 1. C., and Waugh, K. C., Catal. Today 9, 15 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61.

( 1991 ).

63. McDougall, G. S., Chesters, M. A., Sheppard, N., and De La Cruz C., in preparation. 64. Bond, G. C., and Wells, P. B., Appl. Catal. 18, 225 (1985). 65. Geus, J. W.,and Wells. P. B., Appl. Catal. 18, 231 (1985). 66. Yoshitake. H., and Iwasawa, Y., J. Catal. 131, 276 (1991). 67. Trunschke, A,, and Knozinger, H., Card Lett. 17, 295 (1993). 68. Szilagyi. T., Anal. Chim. Acta 253, 29 (1 99 I ); Vibr. Spectrosc. 2, 29 ( 199 I ). 69. Little, L. H., Sheppard, N., and Yates, D. J. C., froc. R. Sac. London A259, 242 (1960). 70. Terenin, A., and Roev, L., Spectrochim. Acta 15, 946 (1959). 71. Ward, J. W., Ph.D. thesis, Cambridge University, 1962; Sheppard, N., Discuss. Furadav Sac. 41,254 (1966). 72. 73. 74. 75. 76. 77. 78. 7Y. 80. 81. 82. 83. 84.

85. 86. 87. 88. 89. YO.

Yang, A. C., and Garland, C. W., J. fhys. Chem. 61, 1504 (1957). Jin, T., Zhou. Y., Mains, G. J., and White, J. M.. J. fhys. Chem. 91, 5931 (1987). Ratajczykowa, I., and Stanasiuk, Z., J. Ph.vs. E 21, 906 (1988). Hughes, M., Bourget, L., and Greenler, R. G., J. Catal. 139,691 (1993). Nash, G. P.. and De Sieno. R. P., J. fhys. Chem. 69, 2139 (1965). Wojtczak, J., Queau, R.. and Poilblanc, R., J. Catal. 37, 391 (1975). Ito, M., and Suetaka, W., SurJ Sci. 62, 308 (1977). Pockrand, I., “Surface Enhanced Raman Vibrational Spectroscopy at SolidKas Interfaces” (Springer Tracts in Modem Physics, Vol. 140). Springer, Berlin, 1984. Bodrov, A. V., Kimel’Fel’D, Ja. M.. and Mostovaya, L. M., J. Mol. Struc. 60, 43 1 ( I 980). Sheppard, N., and Nguyen, T. T.. in “Advances in Infrared and Raman Spectroscopy,” Vol. 5 (R. J. H. Clark and R. E. Hester, Eds.) pp. 67-148. Heyden, London, 1978. Sheppard, N., and De La Cruz, C., React. Kinet. Catal. Lett. 35, 2 I ( I 987). Koestner, R. J., Van Hove, M. A., and Somorjai, G. A., J. Phys. Chem. 87, 203 (1983). Lin. R. F., Blackman, G. S., Van Hove, M. A,, and Somojai, G. A., Acta Crystallogr. B43,368 (1987); Somorjai, G. A,, Van Hove, M. A,, and Bent, B. E., J. fhys. Chem. 92, 973 (1988). Fritzsche, V., and Woodruff, D. P., Phys. Rev. 846, 16128 (1992). Bao, S., Schindler, K-M., Hofmann, Ph., Fritzsche, V., Bradshaw, A. M., and Woodruff, D. P., Surf: Sci. 291, 295 (1993). Bao, S., Hofmann, Ph., Schindler, K-M., Fritzsche, V., Bradshaw, A. M., Woodruff, D. P., Casado. C., and Asensio, M. C., SurJ Sci. 307-309, 722 (1994). Bao, S., Hofmann, Ph., Schindler, K-M., Fritzsche, V., Bradshaw, A. M., Woodruff, D. P., Casado, C., and Asensio, M. C., J. fhys. Condens. Matter 6, L93 (1994). Sheppard, N., and Erkelens, J., Appl. Specrrosc. 38,471 (1984). Richardson, N. V., and Sheppard, N., in “Vibrational Spectroscopy of Molecules on Surfaces” (J. T. Yates, Jr. and T. E. Madey, Eds.), pp. 1 4 8 . Plenum, New York, 1987.

108

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

91. Steininger, H., Ibach, H., and Lehwald, S.. Sui$ Sci. 117, 685 (1982).

92. 93. 94. 95. 96.

97. 98. 99. 100.

101.

102. 103. 104. 105.

106. 107. 108.

109. 110. 111.

112.

113. 114.

115. 116.

I 17. 118. 119.

120. 121. 122. 123.

Greenler, R. G.. J. Chem. fhys. 44, 310 (1966). Pearce, H. A., and Sheppard, N., SurJ Sci. 59, 205 (1976). Greenler, R. G., Snider, D. R., Witt, D., and Sorbello, R. S., SurJ Sci. 118, 415 (1982) Sheppard, N., James, D. I., Lesiunas, A., and Prentice, J. D., Commum. Depf. Chem.. Eulg. Acad. Sci. 17,95 (1984). De La Cruz, C., and Sheppard, N., J. Chem. SOC.Chem. Commun. 1854 (1987). Beebe, T. P. Jr., Albert, M. R., and Yates, J. T., Jr., J. Cafal. 96, 1 (1985). Nichols, H., and Hexter, R. M.. J. Chem. fhys. 73,965 (1980). Brookhart, M.. Green, M. H. L., and Wong, L., frog. Inorg. Chem. 36, I (1988). Taimsalu, P., and Wood, J. L., Trans. Faraday. SOC.59, 1754 (1963). Schrader, B., “RarnadInfrared Atlas of Organic Compounds” (2nd ed., 1989), VCH Verlagsgesesellschafl, Weinheim, (Spectrum M-SN-02). Oxton, 1. A,, Powell, D. B., Sheppard, N., Burgess, K., Johnson, B. F. G., and Lewis, J., J. Chem. SOC.Chem. Commun. 719 (1982). Chang, S-C., Kafafi, Z. H., Hauge, R. H., Whitmore, K. H., Billups, W. E., and Margrave, J. L., Inorg. Chem. 25, 4530 (1986). Sheppard, N.. Trans. Furaday. SOC.46, 533 (1950). Anson, C. E., Sheppard, N., Powell, D. B., Norton, J. R., Fischer, W., Keiter, R. L., Johnson, B. F. G.,Lewis, J., Bhattachanya, A. K.. Knox, S. A. R., and Turner, M. L., J. Amer. Chem. SOC.116,3058 (1994). Howard, M. W., Kettle, S. F. A., Oxton, 1. A,, Powell, D. B., Sheppard, N., and Skinner, P., J. Chem. SOC.Faraday. Trans. 2 77,397 (1981). Chang, S-C.. Hauge, R. H., Kafafi, Z. H., Margrave, J. L., and Billups, W. E., J. Amer. Chem. SOC.110, 7975 (1988). Chang, S-C., Hauge, R. H., Kafafi, Z. H., Margrave, J. L., and Billups, W. E., J. Amer. Chem. SOC.109,4508 (1987). Andrews, M. A., Ekert, J., Goldstone, J. A.. Passel, L., and Swanson, B., J. Amer. Chem. SOC. 105,2262 (1983). De La Cruz, C.. and Sheppard, N., J. Mol. Sfrucr. 224, 141 (1990). Powell, D. B., Scott, J. G. V., and Sheppard, N., Specfrochim. Acfa. 28A, 327 (1972). Hiraishi, J., Specfrochim. Acfa. 25A, 749 (1969). Anson, C. E., Sheppard, N., Powell, D. B., Bender, B. R., and Norton, J. R., J. Chem. Soc. Faraday. Truns. 90, 1449 (1994). Anson, C. E., Johnson, B. F. G., Lewis, J., Powell, D. B., Sheppard, N., Bhattacharyya, A. K., Bender, B. R., Bullock, R. M., Hembre, R. T., and Norton, J. R., J. Chem. SOC. Chem. Commun. 703 (1989). Masetti, G., and Zerbi, G. Specfrochim. Acra. 26A, 1891 (1970). Andrew, J. R.. Kettle, S. F. A,, Powell, D. B., and Sheppard, N., Inorg. Chem. 21, 2874 ( 1982). Evans, J., and McNulty, G. S., J. Chem. SOC.Dalfon Trans. 639 (1983); 79 (1984). Coffer, J. L., Drickamer, H. G., Park, J. T., Roginski, R. T. and Shapley, J. R., J. fhys. Chem. 94, 1981 (1990). Anson, C. E., Ph.D. thesis, University of East Anglia, 1987. Foulet-Fonseca, G. P., Jouan, M., Doa, N. Q., Huy, N. T., and Fischer, E. O., J. chim. fhys. 87, 13 (1990). Keiller, B. T., Ph.D. thesis, University of East Anglia, 1988. Kersting, M., El-Kholi, A., Miiller, U., and Dehnicke, K., Chem. Ber. 122, 279 (1989). Kline, E. S., Kafafi, Z. H., Hauge, R. H., and Margrave, J. L., J. Amer. Chem. SOC.109, 2402 (1987).

VIBRATIONAL SPECTRA OF HYDROCARBONS

109

124. Anson, C. E., Keiller, B. T., Oxton, 1. A,, Powell, D. B., and Sheppard, N., J. Chem. SOC.

Chem. Commum. 470 (1983). Iwashita, Y., Tamura, F., and Nakamura, A,, Inorg. Chem. 8, 1179 (1969). Gervasio, G., Rossetti, R., and Stanghellini, P-L., Organomefallics 4, 1612 (1985). Stanghellini, P-L., and Rossetti, R., fnorg. Chem. 29, 2047 (1990). Diana, A,, Gambino, O., Rossetti, R., Stanghellini, P-L., Albiez, T., Bernhardt, W., and Vahrenkamp, H., Spectrochim. Acfa. 49A, 1247 (1993). 129. Sucher, R. E., Lemmon, D. H., and Miller, F. A,, Specrrochim. Acru 23A, I169 (1967). 130. Muetterties, E. L., Rodin, T. N., Band, E., Boucher, C. F., and Pretzner, W., Chem. Rev. 79,91 ( 1979). 131. Moskovits, M., Acc. Chem. Res. 12, 229 (1979). 132. Johnson, B. F. G., Gallup, M., and Roberts, Y. V., J. Mol. Cufal.86, 51 (1993). 133. Stuve, E. M., and Madix, R. J., J. Phys. Chem. 89, 105 (1985). 134. Anson, C. E., unpublished work. 135. Roberts, M. W., and McKee, C. S., “Chemistry of the M e t a l a a s Interface.” Oxford Univ. Press, Oxford, 1978. 136. Zangwill, A., “Physics at Surfaces,” Cambridge Univ. Press, Cambridgemew York, 1988. 137. Gay, I . D., J. Mugn. Reson. 58, 413 (1984). 138. Slichter, C. P., Ann. Rev. Phys. Chem. 37, 25 (1986). 139. Somorjai, G. A,, “Chemistry in Two Dimensions,” Cornell Univ. Press, London, 1981. 140. Land, T. A,, Michely, T.. Behm, R. J., Hemminger, J. C., and Comsa, G. J. Chem. Phys. 97, 6774 (1 992). 141. Ibach, H., and Lehwald, S., J. Vac. Sci. Technol. 15, 407 (1978). 142. Demuth, J. E., SurJ Sci. 80, 367 (1980). 143. Kesmodel, L. L., Dubois, L. H., and Somorjai, G. A,, J. Chem. Phys. 70, 2180 (1979). 144. Baro, A. M., and Ibach. H.. J. Chem. Ph.vs. 74,4194 (1981). 145. Chesters, M. A,, and McCash, E. M., SurJ Sci. 187, L639 (1987). 146. Malik, 1. J., Brubaker, M. E., Mohsin, S. B., and Trenary, M., J. Chem. Phys. 87, 5554 (1987). 147. Hatzikos, G. H., and Masel, R. J., SurJ Sci. 185, 479 (1987). 148. Yagasaki, B.. and Masel, R. I., SurJ Sci. 222,430 (1989). 149. Yagasaki, B., and Masel, R. I., J. Amer. Chem. SOC.112, 8746 (1990). 150. Yagasaki, B., and Masel, R. I., J. Phys. Chem. 94, 1066 (1990). 151. Yagasaki, B., Backman, A. L., and Masel, R. I., J. Vac. Sci. Techno/. A8, 2610 (1990). 152. Yagasaki, B., Backman, A. L., Chen, B., and Masel, R. I., J. Vac. Sci. Technol. A8, 2616 (1 990). 153. Yagasaki, B., and Masel, R. I., SurJ Sci. 226, 5 I ( I 990). 154. Backman, A. L., and Masel, R. I., J. Phys. Chem. 94, 5300 (1990). 155. Windham, R. G., Bartram, M. E., and Koel, B. E., J. Phys. Chem. 92, 2862 (1988). 156. Windham, R. G., and Koel, B. E., J. Phys. Chem., 94, 1489 (1990). 157. Windham, R. G., Koel, B. E., and Paffett, M. T., Langmuir 4, 1 1 13 (1988). 158. Fan, J., and Trenary, M., Langmuir 10, 3649 (1994) 159. Malik, 1. J., Brubaker, M. E., Mohsin, S. B., and Trenary, M., J. Chem. Phys. 67, 5554 (1 987). 160. Soma, Y., J. Chem. SOC.Chem. Commun. 1004 (1976). 161. Shopov, D., and Palazov, A,, Kinef. Kafal. 8, 862 (1967). 162. Shopov, D., Andreev, A,, and Palazov, A., Comm. Depr. Chem., Bulg. Acud. Sci. 2,321 (1969). 163. Palazov, A., Bonev, Ch., Shopov, D., Kharson, M. S., Fialkova, 1. M., and Kiperman, S. L., in “Proceedings of the Fourth International Symposium on Heterogeneous Catalysis,” Varna (D. Shopov, A. Andreev, A. Palazov, and L. Petrov, Eds.). Bulg. Acad. Sci., Part I, p. 3 19 ( 1979). 125. 126. 127. 128.

110

NORMAN SHEPPARD AND CARLOS DE LA CRUZ

164. Avery, N. R., unpublished work. 165. Clark, M., Ph.D. thesis, University of Cambridge, 1960. 166. Anson, C. E., Bandy, B. J., Chesters, M. A,, Keiller, B., Oxton, I. A,, and Sheppard, N., J. Electron Spectrosc Relar. Phenom. 29, 3 I5 (1 983). 167. Bandy, B. J., Chesters, M. A., James, D. I., McDougall, G. S., Pemble, M. E., and Sheppard, N., Philos. Trans. R . Sac. London A318, 141 (1986). 168. Beebe, T. P. Jr., and Yates, J. T. Jr., Surj.’ Sci. 173, L606 (1986). 169. Chesters, M. A,, De La Cruz, C., Gardner, P., McCash, E. M., Prentice, J. D., and Sheppard,

N.,J. Electron Specrrosc. Relar. Phenom. 54/55,739 (1990). 170. Chesters, M. A., De La Cruz, C., Gardner, P., McCash, E. M., Pudney, P.. Shahid, G., and

Sheppard, N., J. Chem. Soc. Faraday Trans. 86, 2757 (1990). 171. Palazov, A., Sbrkany, A., Bonev, Ch., and Shopov, D., Acla Chim. Acad. Sci.Hung. 108, 343

(1981). Francis, S . A,, J. Chem. Phys. 18, 861 (1950). Ito, M., Mori, Y., Kato. T., and Suetaka, W., Appl. Surf: Sci. 2, 543 (1979). De La Cruz, C., and Sheppard, N., J. Mol. Srruct. 247, 25 (1991). Grassian, V. H., personal communication. Sheppard, N., Avery, N. R., Morrow, B. A,, and Young, R. P., in “Chemisorption and Catalysis” (P. Hepple. Ed.), p. 135. Institute of Petroleum, London, 1971. 177. Beebe, T. P. Jr., and Yates, J. T. Jr.. J. Amer. Chem. Soc. 108, 663 (1986). 178. Soma, Y., J. Catal. 75, 267 ( I 982). 179. Prentice, J. D., Ph.D. thesis, University of East Anglia, 1977. 180. De La Cruz, C., Ph.D. thesis, University of East Anglia, 1987. 181. Blackman. G. S., Kao, C. T., Bent, B. E., Mate, C. M., Van Hove, M. A,, and Somorjai, G. A., Surf: Sci. 207. 66 (1 988). 182. Gates, J. A,, and Kesmodel, L. L., SurJ Sci 120, L461 (1982). 183. Gates, J. A., and Kesmodel, L. L., Surj: Sci. 124, 68 (1983). 184. (a) Tardy, B., and Bertolini, J. C., J. chim. Phys. 82,407 (1985); (b) See also Heitzinger, J. M., Gebhard, S. C., and Koel, B. E. J. Phys. Chem. 97, 5327 (1993) for Pd on Mo (100). 185. Chesters, M. A,, McDougall, G. S., Pemble, M. E., and Sheppard, N., Appl. SurJ Sci. 22/23, 369 (1985). 186. Nishijima, M., Yoshinobu, J., Sekitani, T., and Onchi, M.. J. Chem. Phys. 90, 51 14 (1989). 187. Sekitani, T., Yoshinobu, J., Onchi, M., and Nishijima, M., J. Phys. Chem. 94, 6847 (1990). 188. Nishijima, M., Sekitani, T., Takaoka, T., and Fujisawa, M., J. Mol. Catal. 74, 163 (1992). 189. Sekitani, T., Takaoka, T., Fujisawa, M., and Nishijima, M., J. Phys. Chem. 96, 8462 (1992). 190. James, D. I., Ph.D. thesis, University of East Anglia, 1983. 191. Morrow, B. A., Ph.D. thesis, University of Cambridge, 1965. 192. Lesiunas, A,, MSc. thesis, University of East Anglia, 1974. 193. James, D. I., and Sheppard, N., J. Mol. Strucr. 80, 175 (1982). 194. Baddour, R. F., Modell. M., and Goldsmith, R. L., J. Phys. Chem. 74, 1787 (1970). 194a. Timbrell, P. Y . , Gellman, A. J., Lambert, R. M., and Willis, R. F., Surface Sci. 206, 339 (1988). 195. Lehwald, S., and Ibach, H., SurJ Sci. 89,425 (1979). 196. Hammer, L., Hertlein, T., and Miiller, K., SurJ Sci. 178, 693 (1986). 197. Bertolini, J. C.. and Rousseau, J., Surf: Sci. 83, 531 (1979). 198. Lehwald, S.. Ibach, H.. and Steininger, H., Surf: Sci. 117, 342 (1982). 199. Zaera, F., and Hall, R. B., Surf Sci. 180, L123 (1987); J. Phys. Chem. 91, 4318 (1987). 200. Stroscio, J. A., Bare, S . R., and Ho, W., Surf: Sci 148, 499 (I 984). 200a. Liu, Z.-M., Zhou, X-L., Buchanan, D. A,, Kiss, J., and White, J. M., J. Amer. Chem. Sac. 114, 203 I (1992). 172. 173. 174. 175. 176.

VIBRATIONAL SPECTRA OF HYDROCARBONS

111

201. Cooper, E., and Raval, R, Surj.' Sci.. 331-333,94 (1995). 202. Karlsson, P. A,, Martensson, A-S., Andersson, S., and Nordlander, P., Sur$ Sci. 175, L759 ( 1986). 203. Zhu, X.-Y., and White, J. M., Surf: Sci., 214, 240 (1989). 204. Lapinski. M. P., and Ekerdt, J. G., J. Phys. Chem. 92, 1708 (1988). 205. Primet, M.. and Sheppard, N., unpublished work. 206. Sheppard, N., J. Electron Spectrosc. Relat. Phenom. 38, 175 (1986). 207. Primet, M., and Sheppard, N., J. Catal. 41,258 (1976). 208. Blyholder, G., Shihari, D., Wyatt, W. V., and Bartlett, R., J. Catal. 43, 122 (1976). 209. McDougall, G. A., personal communication. 210. Avery, N. R., and Sheppard, N., Proc. R. Sac. London A405, I (1986). 21 1. Krasser, W., and Renouprez, A. J., J. Raman Spectrosc. 1 I, 425 (1981 ). 212. Dubois, L. H., Castner, D. G., and Somojai, G. A,. J. Chem. Phys. 72, 5234 (1980). 213. Koel, B. E., Bent, B. E., and Somorjai, G. A,, Surf: Sci. 146,211 (1984). 214. Slavin, A. J., Bent, B. E., Kao, C-T., and Somorjai, G. A., Surf: Sci. 206, 124 (1988). 215. Slavin, A. J., Bent, B. E., Kao, C-T., and Somorjai, G. A,, Surf: Sci. 202, 388 (1988). 216. Eady, C. R., Fernandez, J. M., Johnson, B. F. G., Lewis, J., Raithby, P. R., and Sheldrick, G. M., J. Chem. Sac. Chem. Commun. 421 (1978). 217. Blyholder, G., and Wyatt, W. V., J. Phys. Chem. 78, 618 (1974). 218. Sakakini, B., Swift, A. J., Vickerman, J. C., Harendt, C., and Christmann. K., J. Chem. SOC. Faraday Trans. 1 83, 1975 (1987). 2 1 8 ~ .Sakakini, B., Havendt, C., and Vickerman, 1. C., Spectrochim. Acta 43A, 1613 (1987). 219. Mims, C. A,, Weisel, M. D., Hoffmann, F. M., Sinfelt, J. H., and White, J. M., J. Phys. Chem. 97, 12656 (1993). 219a. Ransley, I. A., Ilharco, L. M., Bateman, J. E., Sakakini, B. H., Vickerman, J. C., and Chesters, M. A,, Surf: Sci. 298, 187 (1993). 220. Henderson, M. A., Mitchell, G. E., and White, J. M., Surf: Sci. 203, 378 (1988). 221. Marinova, T. S., and Kostov, K. L., Surf: Sci. 181, 573 (1987). 222. Marinova, T. S., and Chakarov, D. V., Surf Sci. 192, 275 (1987). 223. Chakarov. D. V., and Marinova, T. S., S u r - Sci. 227,297 (1990). 224. Nyberg, C., Tengstal, C. G., Andersson, S., and Holmes, M. W., Phys. Chem. Lett. 87, 87 (1982). 225. McCash, E. M., Vacuum 40, 423 (1990). 226. Jenks, C. J.. Bent., D. E., Bemstein, N., and Zaera, F., Surf Sci. 277, L89 (1992). 226a. Kubota, J., Kondo, J. N., Domen, K., and Hirose, C., J. Phys. Chem. 98, 7653 (1994). 227. Ertiirk, U., and Otto, A., Surf: Sci.. 179, 163 (1987). 228. Slater, D. A,, Hollins, P., and Chesters, M. A., Surf: Sci. 306, 155 (1994); J. Electron Spectrosc. Relat. Phenom. 64165, 95 ( 1993). 229. Moskovits, M., and DiLella, D. P., Chem. Phys. Lett. 73, 500 (1980). 230. Wood, T. H., and Zwemer, D. A,, J. Vac. Sci. Technol. 18,649 (1981). 231. Ertiirk, U., and Otto, A,, Chem. Phys. Lett. 149, 284 (1988). 232. Brings, R., Mrozek, I., and Otto, A,, J. Roman Spectrosc. 22, 1 19 (1991). 233. Wolkow, R. A,, and Moskovits, M., J. Chem. Phys. 96, 3966 (1992). 233a. Patterson, M. L.. and Weaver, M. J., J. Phys. Chem. 89, 1331 (1985). 234. Seip, U., Tsai, M.-C., Kiippers, J., and Ertl, G., Surf: Sci. 147, 65 (1984). 235. Bernasek, S. L., Ann. Rev. Phys. Chem. 44, 265 (1993). 2 3 5 ~ .Demuth, J. E., and Ibach, H., SurJ Sci. 78, L238 (1978). 235h. Paul, D. K., Beebe, T. P., Jr., Uram, K. J., and Yates, J. T., Jr., J. Amer. Chem. Sac. 114, 1949 ( 1992).

112 236. 237. 238. 239. 240. 241. 242.

NORMAN SHEPPARD AND CARLOS DE LA CRUZ Wong, Y.-T., and Hoffmann, R., J. Chem. Sac. Furaduy Trans. 86,4083 (1990). Schiott. B., Hoffmann, R., Awad, M. K., and Anderson, A. B.. Langmuir 6, 806 (1990). Ditlevsen, P. D., Van Hove, M. A., and Somorjai, G. A,, Surf Sci. 292, 267 (1993). Lloyd, K . G., Roop, B., Campion, A., and White, J. M., SurJ Sci. 214, 227 (1989). De La Cruz, C., and Sheppard, N., J. Catal. 127,445 (1991). Thomson, S. J., and Webb, G., J. Chem. Soc. Chem. Comrnun. 526 (1976). Bent, B. E., Mate, C. M., Crowell, J. E.. Koel, B. E., and Somoqai, G. A., J. Phys. Chem. 91, 1493 (1987).

243. Wang, D., Wu, K., Cao, Y., Zhai, R., and Guo, X . , Surf Sci. 223, L927 (1989). 244. Sakakini. B. H., Ransley, I. A., Odnoza, C. F., Vickerman, J. C., and Chesters, M. A,, SurJ Sci. 271, 227 (1992).

245. Gardner, P., Ph.D. thesis, University of East Anglia, 1988. 246. McCash, E. M., Ph.D. thesis, University of East Anglia, 1987. 247. Dent, A. L., and Kokes, R. J., J. Amer. Chem. Sac. 92, 6709 (1970). 248. Shahid, G., and Sheppard, N., Spectrochim. Acta 46A, 999 (1990). 249. Marshall, P. R., McDougall, G. S., and Haddon, R. A,, Top. Catal. I, 9 (1994). 250. Sheppard, N., Pure Appl. Chem. 4, 71 (1962). 251. Munro, S.. and Raval, R., Surf Rev. Lett. 1, 645 (1994). 252. Delbecq, F., and Sautet, P., Cutul. Lett. 28, 89 (1994). 252a. Carter, R. N., Anton, A. B., and Apai, G., J. Amer. Chem. Soc. 114,4410 (1992). 253 Avery, N. R., and Sheppard, N., Proc. R. Sac. A405,27 ( I 986). 254. Chesters, M. A., Horn, A. B., Ilharco, L. M., Ransley, I. A., Sakakini, B. H.. and Vickerman, J. C., J. Electron Spectrosc. Relat. Phenom. 54/55, 677 (1990). 254a. Chesters, M. A., Horn, A. B., Ilharco, L. M., Ransley, 1. A,, Sakakini, B. H., and Vickerman, J. C., Surf Sci. 2511252, 291 (1991). 255. Fricke, A,, Graupner, H., Hammer, L., and Miiller, K., Surf Sci. 272, 182 (1992). 256. Morrow, B. A,, and Sheppard, N., Proc. R. Sac. A311,415 (1969). 257. Shahid, G., and Sheppard, N., J. Chem. Sac. Faraday Trans. 90, 507 and 512 (1994). 257a. Itoh, K.,Tsukada, M., Koyama. T., and Kobayashi, Y.,J. Phys. Chem. 90, 5286 (1986). 258. Avery, N. R., J. Catal. 24,92 (1972). 258a. Basu, P., and Yates, J. T., Jr., J. Phys. Chem. 93, 61 10 (1989). 259. Shirakawa, H., and Ikeda, S., Polym. J. 2, 231 (1971). 260. Eischens, R. P., and Mertens, F. P., J. Electrochem. Sac. 115, C81 (1968). 261. Selwood, P. W., J. Amer. Chem. Soc. 79, 3346 (1957). 262. Erkelens, J., and Liekens, Th. J., J. Catal. 27, 165 (1972). 263. Soma, Y., Bull. Chem. Soc. Jpn. 50,2119 (1977). 264. Basu, P., and Yates, J. T., Jr., J. Phys. Chem. 93, 2028 (1989). 265. Hammer, L., Dotsch, B., Brandenstein, F., Fricke, A,, and Miiller, K., J. Electron Spectrosc Relat. Phenom. 54/55, 687 (1990). 266. Pudney, P., PhD. thesis, University of East Anglia, 1989. 267. Shahid, G., and Sheppard, N., Canad. J. Chem. 69, 1812 (1991).

ADVANCES IN CATALYSIS, VOLUME 41

Catalytic Chemistry of Heteropoly Compounds TOSHIO OKUHARA Graduate School of Environmenral Earth Science Hokkaido University Sapporo 060, Japan

NORITAKA MIZUNO Institute of Industrial Science The University of Tokyo Roppongi, Minaro-ku, Tokyo 106. Japan AND

MAKOTO MISONO Deparrment of Applied Chemistry Graduate School of Engineering The University of Tokyo Bunkyo-ku. Tokyo 113. Japan

1. A.

Introduction

HETEROPOLYCOMPOUNDS AS CATALYSTS

The catalytic properties of heteropoly compounds have drawn wide attention in the preceding two decades owing to the versatility of these compounds as catalysts, which has been demonstrated both by successhl large-scale applications and by promising laboratory results. Heteropolyanions are polymeric oxoanions formed by condensation of more than two different mononuclear oxoanions, as shown in Eq. (1): 12WO:-

+ HPOi- + 23H'

-+PW120:i

+ 12H20

(1)

Heteropolyanions formed from one kind of polyanion are called isopolyanions, as shown in Eq. (2): 7Mo0:-

+ 8Hf

+

Mo,O:i

+ 4H20

(2)

Acidic elements such as Mo, W, V, Nb and Ta, which are present as oxoanions I13 Copyright 0 1996 by Academic Press. Inc. All rights of reproduction in any form reserved.

114

TOSHlO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

in aqueous solution, tend to polymerize by dehydration at low pH, forming polyanions and water (1-3). The term “heteropoly compound” is used in this review for the acid forms, e.g., H3PW12040.and their salts, e.g., Cs3PW1204~.Catalysts of which the main components are heteropoly or heteropoly-derived compounds are referred to here as “heteropoly catalysts,” and they are the subject of this review. Heteropolyanion-derived compounds are, for example, organic and metallo-organic complexes of polyanions (see Section 1.D for the terminology and nomenclature). Although there are many kinds of heteropolyanions (Section II), heteropolyanions having the Keggin structure are the most widely investigated as catalysts because of their stabilities and ease of synthesis. However, other heteropolyanions are also expected to be recognized as good catalysts. Heteropoly catalysts can be applied in various ways (4-1 0). They are used as acid as well as oxidation catalysts. They are used in various phases, as homogeneous liquids, in two-phase liquids (in phase-transfer catalysis), and in liquidsolid and in gas-solid combinations, etc. The liquid-solid and gas-solid combinations are represented by the classes of catalysis shown in Fig. 1 and described in the following sections. The advantages of heteropoly catalysts stem from the characteristics summarized in Table I. As excellent candidates for design at the atomic or molecular level, heteropoly catalysts have proven to be of value in fundamental studies as well as practical applications. But it is also true that much remains to be done. Efforts to establish methodologies for design of practical catalysts are still under way. The acid strength and acid site density can be controlled quite well both in solution and in the solid state, but the redox properties in the solid state are much less well understood because of the lack of sufficient thermal stability of mixedmetal (mixed-addenda) heteropolyanions. The acid strengths of some solid heteropolyacids have been suggested to reach the range of superacids, but they reactant 0

product -&

reactant product

0

4

\\ !/

f

product

(

.... .. .. .. .. . . . . . . . . . . . . .. .. .. .. .. .. .. .. .. .. .. .. .. . . . .. .. .. .. .. .. ....... .. .. .. .. .. . . . .. .. .. .. .. .. .. .. .. .

u

.............. .............

Solid (Surface catalysis)

reactant

Pseudoliquid (Bulk type I catalysis)

Solid (Bulk type I1 catalysis)

FIG. I . Three types of catalysis by heteropoly compounds.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

Advaniages

115

TABLE I Heieropoly Caialysis

OJ

1. Catalvst design at aromic/molecular levels based on the following: 1-1. Acidic and redox properties

These two important properties for catalysis can be controlled by choosing appropriate constituent elements (type of polyanion, addenda atom, heteroatom. countercation, etc.). 1-2. Multifunctionality Acid-redox, acid-base, multi-electron transfer, photosensitivity, etc. 1-3. Tertiary structure, bulk-type behavior, etc., for solid state These are well controlled by countercations. 2. Moleculariiy-meial oxide clusier 2- I . Molecular design of catalysts 2-2. Cluster models of mixed oxide catalysts and of relationships between solid and solution catalysts 2-3. Description of catalytic processes at atomicimolecular levels Spectroscopic study and stoichiometry are realistic Model compounds of reaction intermediates.

3 . Unique reaciionfield 3- I. Bulk-type catalysis "Pseudoliquid" and bulk type I1 behavior provide unique three-dimensional reaction environments for catalysis. 3-2. Pseudoliquid behavior This makes spectroscopic and stoichiometric studies feasible and realistic. 3-3. Phase-transfer catalysis 3-4. Shape selectivity. 4. Unique basicity of polyanion 4-1. Selective coordination and stabilization of reaction intermediates in solution and in pseudoliquid phase, and possibly also on the surface 4-2. Ligands and supports for metals and organometallics.

are still weaker acids than sulfated zirconia. Unique complexing or basic properties of polyanions have not been clarified sufficiently, although it appears that they play important roles in industrial liquid-phase processes. The efforts to describe catalytic processes at the molecular level have also made significant progress in the preceding decade, but the number of well-elucidated reactions remains very small. Early attempts to use heteropoly compounds as catalysts are summarized in reviews published in 1952 (IZ) and 1978 (I). The first industrial process using a heteropoly catalyst was started up in 1972 for the hydration of propylene in the liquid phase. The essential role of the Keggin structure in a solid heteropoly catalyst was explicitly shown in 1975 in a patent concerning catalytic oxidation of methacrolein. Systematic research in heterogeneous catalysis with these materials started in the mid-1970s and led to the recognition of quantitative relationships between the acid or redox properties and catalytic performance

1 16

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

(4-9). Pseudoliquid-phase catalysis (bulk type I catalysis) was reported in 1979, and bulk type I1 behavior in 1983. In the 1980s, several new large-scale industrial processes started in Japan based on applications of heteropoly catalysts that had been described before (5, 6, 12): namely, oxidation of methacrolein (1982), hydration of isobutylene (1984), hydration of n-butene ( 1985), and polymerization of tetrahydrofuran (1987). In addition, there are a few small- to medium-scale processes (9, 10). Thus the level of research activity in heteropoly catalysis is very high and growing rapidly. One of the authors of this chapter has previously reviewed heterogeneous catalysis by heteropoly compounds (4-6). Catalysis in solution has also been described (7-10). In this chapter, we critically survey the literature and attempt to describe the essence of the catalytic chemistry of heteropoly compounds in solution and in the solid state. We have attempted to highlight the advantages of heteropoly catalysts as described in Table I.

B.

CLASSES OF CATALYSIS BY

HETEROPOLY COMPOUNDS

As will be described in more detail in later sections, in acid and oxidation catalysis by solid heteropoly compounds, that is, gas-solid and liquid-solid systems, there are three different classes of catalysis: (1) surface catalysis, (2) bulk type 1 (pseudoliquid catalysis), and (3) bulk type I1 catalysis, as shown in Fig. 1. The latter two have been specifically demonstrated for heteropoly catalysts, and they could be found for other solid catalysts as well. Surface-type catalysis is ordinary heterogeneous catalysis, whereby the reactions take place on the two-dimensional surface (on the outer surface and pore walls) of solid catalysts. The reaction rate is proportional to the catalyst surface area. Bulk type I catalysis was found in acid catalysis with the acid forms and some salts at relatively low temperatures. The reactant molecules are absorbed between the polyanions (not in a polyanion) in the ionic crystal by replacing water of crystallization or expanding the lattice, and reaction occurs there. The polyanion structure itself is usually intact. The solid behaves like a solution and the reaction medium is three-dimensional. This is called “pseudoliquid” catalysis (Sections 1.A and VI). The reaction rate is proportional to the volume of the catalyst in the ideal case; the rate of an acid-catalyzed reaction is proportional to the total number of acidic groups in the solid bulk. Bulk type 11 catalysis was discovered later for some oxidation reactions at high temperatures. Although the principal reaction may proceed on the surface, the whole solid bulk takes part in redox catalysis owing to the rapid migration into the bulk of redox carriers such as protons and electrons (Sections VII and IX). The rate is proportional to the volume of catalyst in the ideal case.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

117

These three classes of catalysis are distinctly different from each other in the ideal cases. But the extent of the contribution of the inner bulk of the catalyst depends on the rate of the catalytic reaction relative to the rate of diffusion of reactant and product molecules in bulk type I catalysis and on the rate of reaction relative to the rate of diffusion of redox carriers for the bulk type I1 catalysis. c . CATALYST DESIGNBASEDON CRYSTALLINE MIXEDOXIDES

To develop efficient catalytic technology capable of solving contemporary problems related to energy and resource limitations, synthesis of materials, and environmental protection, novel concepts for catalyst design are needed. Catalyst design at the atomic level utilizing the techniques of advanced surface science is one of the possibilities; but this can be applied only for model catalysts, and the syntheses of industrial catalysts by this method are not yet realistic (6). Alternatively, we have attempted the molecular design of mixed-oxide catalysts by using crystalline mixed oxides whose bulk structures are known and whose potential for practical use is good. Heteropoly compounds, perovskites, and zeolites are the candidate catalysts. Since we believe that the relationships in Scheme 1 are useful for the design of catalysts (13), we place stress in this chapter on these relationships at atomic/ molecular levels of heteropoly compounds. In our opinion, sufficient care must be taken on the structure and stoichiometry in order to design catalysts taking advantage of the molecular nature of heteropoly compounds. D. TERMINOLOGY A N D NOMENCLATURE 1.

Generic Terms

Various generic names have been used for oxoacids and oxoanions. Because there are many of them, it is difficult to define the terms unambiguously and consistently. But the following statements may be helpful. For acid forms, polyacids = polyoxoacids, including heteropolyacids (e.g., H3PW12040) and isopolyacids (e.g., H2Mo6OI9); and for oxoanions, polyanions = polyoxoanions = polyoxometalates, including heteropolyanions (e.g., PWf20:O) and isopolyanions (e.g., Mo60:9). Performance of Catalyst

*

Chemiaal and Physical Properties

Composition

* Ekture

SCHEME I

Method of Synthesis

1 18

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

In addition, such generic terms as metal-oxygen cluster ion, metal-oxide molecule, etc., are used for polyanions (and polyacids). Since the traditional “heteropoly-” and “isopoly-” are unsatisfactory terms to express a variety of polyacids and polyanions, the terms “polyoxometalates” and “polyoxoacids” have been used recently (3). The terminology is still changing, thus reflecting the rapid expansion of the chemistry. Nonoxygen elements in the inner part of polyanions (usually P, Si, As, etc.) are called heteroatoms (in some cases, central atoms) and those in the peripheral part (usually Mo, W, V, Nb, etc.) are called addenda atoms or polyatoms (Section 1I.A). We use more or less conventional terminology here. 2. Nomenclature The rigorous and systematic nomenclature addressed by IUPAC ( 1 4 , in which all atoms and their topological connections are defined unambiguously, is too complicated here. Thus we use traditional names. But the semi-systematic nomenclature accepted by IUPAC (15) is mentioned briefly. For example, a heteropolyacid, H4SiMo12040,is called tetrahydrogen hexatriacontaoxo(tetraoxosilicato)dodecamolybdate(4-) [hydrogen nomenclature] or tetrahydrogen silicododecamolybdate [abbreviated semi-trivial name]. Or this is called 12-(or dodeca)molybdosilicicacid for the acid form and 12-(or dodeca)molybdosilicate for the anion [recommendations of IUPAC, 1971 ( I @ ] .

II. Structure, Synthesis, Stability, and Characterization A.

PRIMARY,

SECONDARY, AND TERTIARY STRUCTURES

Heteropolyanions and isopolyanions are polymeric oxoanions (polyoxometalates) (2, 3, 5, 6). The structure of a heteropolyanion or polyoxoanion molecule itself is called a “primary structure” (5, 6, 17). There are various kinds of polyoxoanion structure (Section 1I.A. 1). In solution, heteropolyanions are present in the unit of the primary structure, being coordinated with solvent molecules a n d or protonated. Most heteropolyanions tend to hydrolyze readily at high pH (Section 1I.C). Protonation and hydrolysis of the primary structure may be major structural concerns in solution catalysis. Heteropoly compounds in the solid state are ionic crystals (sometimes amorphous) consisting of large polyanions, cations, water of crystallization,and other molecules. This three-dimensional arrangement is called the “secondary structure.” For understanding catalysis by solid heteropoly compounds, it is important to distinguish between the primary structure and the secondary structure (5, 6, 17). Recently, it has been realized that, in addition

119

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

?d

P b

nn

."I& a@ .............._,

d a

..'& , .. ..............: ..' I.

I

cs+

PW,~O~,,~-

C

FIG. 2. Primary, secondary, and tertiary structures of heteropoly compounds. (a) Primary structure (Keggin structure, XM12040); (b) secondary structure (H3PW 12040. 6H20); (c) secondary structure (CS~PWI~O~O); (d) tertiary structure [Csz.sHosPWIz04O, cubic structure as in (c)].

to these structures, tertiary and higher-order structures influence the catalytic function (6). These structures are exemplified in Fig. 2 (3, 18). 1.

Primary Structure

a. Keggin Structure (1-3, 18, 19). Figures 3a and 3b illustrate the Keggin anions, which are the most popular heteropolyanions in catalysis. The ideal Kegging structure of the a type has G symmetry and consists of a central X 0 4 tetrahedron (X = heteroatom or central atom) surrounded by twelve M 0 6 octahedra (M = addenda atom). The twelve M06 octahedra comprise four groups of three edge-shared octahedra, the M3013 triplet (19), which have a common oxygen vertex connected to the central heteroatom. The oxygen atoms in this structure fall into four classes of symmetry-equivalent oxygens: X-0,-(M)3, M-Ob-M, connecting two M3013 units by comer sharing; M-0,-M, connecting two M3013 units by edge sharing; and Od-M, where M is the addenda atom and X the heteroatom. This Keggin structure is called an a isomer (18).

120

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

d

b

a

e

f

FIG. 3. Primary structures of heteropoly and isopolyanions. (a) Keggin structure, a-XM 120'& (the fourth M3013set and the X04 tetrahedron are not shown for clarity) (from Refs. 2 and 19); (b) Keggin structure, /?-XM120;; (the fourth M301) set and the XO4 tetrahedron are not shown for clarity) (from Refs. 24 and 25); (c) lacunary Keggin anion (the central XO4 tetrahedron is not shown) (from Ref. 2); (d) Dawson structure, X2M180gi (from Refs. 29 and 30); ( e ) Anderson structure, XM602/- (shaded tetrahedron indicates the heteroatom site) (from Refs. 18 and 33); (0 XMPO;Q (from Ref. 2); (g) isopolyanions, W,,O:; (from Ref. 2).

The known addenda and heteroatoms incorporated in heteropolyanions are summarized in Table I1 (20). The structures in this table of polyanions with Se(IV), Te(IV), Sb(III), Bi(III), Ti(IV), and Zr(IV) still need to be confirmed, since tetradendral coordination of these ions with oxide ions is seldom observed (2). In Table 111 (21, 22), the bond distances in various heteropoly corn ounds having the Keggin structure are listed. Bond lengths in PW120;for H3PW12040.6H20 are 1.71, 1.90, 1.91, and 2.44A for Od-W, 0,-W, Oh-W, and 0,-W bonds, respectively. The existence of isomers has been established for the Keggin anion. Figures 3a and 3b show the a-and /I-isomers. They can be separated by fractional crystallization (X = B, Si) or prepared separately (X = Si, Ge) (23). In /?-SiWlzO:,, one of the three edge-shared W3OI3 triplets of the a structure is rotated by 60°, thereby reducing the symmetry of the anion from to CJ, (24, 25). The other isomers involving the

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

121

TABLE I1 Known Addenda-(0) and Hetero-(0) Atoms Incorporated in Heteropolvacid.7

Pr Nb Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Pu Am Cm Bk Cf Es Fm Md No Lr

60” rotation of two, three, and all four W3013 groups are called the y, 6 , and c structure (26), respectively. Compounds containing fluoride ions in metatungstate have been synthesized (27): (A) ( W I Z F Z O ~ R H Z(B) ) H ~(W12F3037HW4, , (C) ( W I ~ ~ ~ R F Z Hand ) H (D) S, ( W I ~ O ~ ~ F Hwhere ~ ) H the ~ , central atoms are protons. b. Lacunqv Keggin Anion (2, 28). In solution, several species are present in equilibrium, the composition depending on pH. Figure 4 shows an example of an aqueous solution containing MOO:- and HPOt- in a molar ratio of 12: 1 (2, 28). The reactions to form polyoxoanions other than Keggin anions are shown by Eqs. (3)-(5). 17H’ + I1MoO:- + HPOSP M o I , O : ~+ 9H20 (3)

-.

17H+ + 9M00:8H‘

+ HP0:-

+ 5Mo0:- + 2HPO:

+

+

PMO&I(OH~):P2MosO;;

+ 9H20

+ 5H20

(4) (5)

In the case of PW120:0, the lacunary P W I I O : ~is formed at pH = 2. The to give PW90i, occurs at pH > 8 (2). These lacunary degradation of PWI or defect derivatives of the Keggin structure are illustrated in Fig. 3. c. Dawson Structure. The Dawson structure, M2XIgOgZ, is shown in Fig. 3d (29, 30). Two PW90:, units, “lacunary Keggin anions,” fuse to form a Dawson structure. Three isomers exist, depending on the number of rotated

122

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

Bond Lengihs in

Compounds H3PW12040.6H2O H3PM01204.13H20 H3PM012040.30H20 H&M012040.13H20 &SiW12040. 16H20

Mob

TABLE 111 and Won Group in Heteropolyanion (A) (21, 22)

M-Od

M-0,.

M-oh

MhO,

XhO,

1.71 1.66 1.68 1.67 1.68

I .90 I .96 1.91 1.94 1.91

1.91 I .97 I .92 I .96 I .96

2.44 2.43 2.44 2.35 2.38

I .53 1.53 1.54 1.62 1.63

M3013groups. H6As2MoI8O62 and S2M0180:2 have also been synthesized (2, 31). A complex containing F ions, H Z F ~ N ~ W I ~(32), O : is ~ isostructural with the Dawson species, P2W180gY.

-

PH

FIG. 4. Distribution diagram for species present in fresh solutions containing MOO:- and HP0:- in a molar ratio of 12 : I at different pH values. (From Ref. 28.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

123

d. Anderson Structure. The Anderson structure, XM60;4, comprises seven edge-shared octahedra (Fig. 3e) (18, 33). 2. Secondary Structure of Solid Heteropoly Compounds In heteropoly acids (acid form) in the solid state, protons play an essential role in the structure of the crystal, by linking the neighboring heteropolyanions. Protons of crystalline H3PW12040. 6 H 2 0 are present in hydrated species, H50:, each of which links four neighboring heteropolyanions by hydrogen bonding to the terminal W-Od oxygen atoms, and the polyanions are packed in a bcc structure (Fig. 2b) (21). Various heteropolyacid hydrates that differ in the number of waters of crystallization have been reported (34-36); H3PW12040.nH20 (n = 14, 21, 24, and 29). The loss of water brings about changes in the anion packing; n = 29 [cubic (diamond-like)], n = 21 (orthorhombic), n = 6 [cubic (bcc)]. Cs3PW12040,in which the Cs ions are at the sites of H50: ions of hexahydrate (Fig. 2b), has a dense secondary structure and is O~~ anhydrous (Fig. 2c) (34). The lattice constants of C S J P W ~ ~and H3PW12040.6H20 are 12.14 and 11.86 A, respectively (21, 26, 37). Secondary structures containing organic molecules are known. H4SiW12040 * 9DMSO [DMSO = (CH3)2SO] contains nine molecules of DMSO in a unit cell, where there are weak hydrogen bonds between methyl groups and oxygen atoms of the heteropolyanion, p~lyanion--(CH~)~SO--H+ --OS(CH3)2--polyanion (38). Eight independent DMSO molecules join in four pairs of cations [H(O=S(CH3)2)2]+ by strong hydrogen bonds, and one DMSO molecule is weakly bonded. Another example is PW12040. [(C5H5N)2H]3,which is obtained by the reaction of anhydrous H3PW12040 with pyridine (39). As shown in Fig. 5 , six pyridine molecules lie almost in a plane, and the pyridine molecules

0 FIG. 5. Structure of [(C5HSN)2H],[PW12040]. (From Ref. 3Y.)

124

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

are paired, forming N--H--N hydrogen bonding. Similar structures are also formed by the contact of pyridine vapor with heteropolyacids (5). Other examples are PW12040 (H+-quinolin-8-01)3 4C2H50H* 2H20 (40) and PM012040[H+-TMU213(TMU = 1,1,3,3-tetramethyl urea) (41). 3.

Tertiary Structure of Solid Heteropoly Compounds Tertiary structure is the structure of solid heteropoly compounds as assembled

(5, 6). The size of the primary and secondary particles, pore structure, distribu-

tion of protons and cations, etc. are the elements of the tertiary structure. a. Group A and B Salts. Countercations greatly influence the tertiary structure of a heteropoly compound. The salts of small ions such as Na+ [classified into group A salts (42)] behave similarly to the acid form in several respects. The group A salts are highly soluble in water and other polar organic solvents. The surface areas of group A salts are usually low. Polar molecules are readily absorbed in interstitial positions (between polyanions) of the secondary structure (Section VI). On the other hand, the salts of large cations such as NH: and Cs+ (classified as group B salts) are insoluble in water and exhibit low absorptivity for polar molecules. Low solubility is due to the low energy of solvation of large cations. The surface areas of group B salts are usually high due to the smaller sizes of the primary particles, giving favorable properties for heterogeneous catalysis (43-46a). The thermal stability of most group B salts is relatively high, which is also important in heterogeneous catalysis. b. Surface Area and Pore Structure. The surface area and pore structure are closely related. Figure 6 shows the surface areas as a function of the extent 200

0 : cs

0

1

2

3

x in MxH3.xPW12040

as a function of the extent of Na or FIG.6. Surface areas of Na or Cs acidic salt of H3PW12040 Cs substitution. (From Refs. 46 and 47.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

125

of Na or Cs substitution of H3PW12040 (46, 47). As the Na content increases, the surface area decreases monotonically (46b).The change of the surface area with increasing Cs content is remarkably different. The surface area increases significantly when the Cs content, x in Cs,H3-,PW12040, changes from x = 2 (1 m2 g - ' ) to x = 3 (156 mz g-'), although it decreases slightly from x = 0 (6 m2 ! - I ) to x = 2. The surface area increases significantly to more than 130m g-I when the Cs content exceeds 2.5. Cs2.5Ho.5PWIz040 (and also C S ~ P W ~ consists ~ O ~ ~of) very fine particles (8-10 nm in diameter) (Fig. 2d). Pore structure is an important property of solid catalysts. Gregg and Tayyab ( 4 3 ) reported that (NH4)3PW1 2 0has 4 ~a microporous structure (pore diameter <20 A) as estimated from the adsorption isotherms of N2, n-hexane, and carbon tetrachloride. Moffat et al. (44) reported that the salts of NH: and Csf possess pore structures in the microporous-mesoporous range as revealed by nitrogen adsorption. Figure 7 shows NZ adsorption isotherms for (NH4)3PW1z040and C S ~ P W ~ In ~O the~ case ~ . of C S ~ P W ~ ZaOhysteresis ~~, is evident, showing that this salt has mesopores (pore size >20A) (44c) as well as micropores. They proposed that these pores exist in the crystal structure. Mizuno and Misono (37) examined the tertiary structure of C S ~ P W , ~byOestimating ~~ the surface area with three different methods: particle-size distribution measured by TEM (assuming spherical particles), the pore-size distribution measured by N2 adsorption (assuming cylindrical pores), and the BET equation. The three values are in good agreement with each other, showing that this material is composed of fine primary particles observed by TEM, with the pores being intercrystalline, not

0

0.5

1.o

PPO FIG.7. Nitrogen adsorption4esorption isotherms (77 K) for Cs3PW1204,,and (NH4)3PW12040, (From Ref. 44c.)

126

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE IV Adsorption Data for C s , H j - , P W , ~ 0 4 0(x = 2.1. 2.2, and 2.5) (48)

Molecule c.s.'

Kinetic diameter

(A')

(A)

N2 (16.2) Benzene (30.5) Neopentane (37.2) 1,3,5-TMBd (41.1) I ,3,5-TIPBr (59.4)

3.6 5.9 6.2 1.5 8.5

Adsorption amount (Imol ti-7

137 (0.18) 21 (0.20) 99 (0.19) 0.6 (0.20) 9 x lo-' (0.20)

77 300 273 300 300

Cs2.1

Cs2.2

Cs2.5

Ratio"

487

861 124 179 I1 15

1648 232 390 237 236

0.52 0.53 0.46 0.05 0.06

10

5 -

-

"Cross section calculated from the molecular weight and density of liquid. hThe ratio of the partial pressure introduced (P) to the saturated vapor pressure (PO). 'Adsorption amount on Cs2.2 divided by that on Cs2.5. ,I 1,3,5-Trimethylbenzene. 'I ,3,5-Triisopropylbenzene.

intracrystalline. Considering the size and shape of the Keggin anion and the structure of C S ~ P (Fig. W ~2c),~there ~ ~are~no open pores in the crystal through which an N2 molecule (3.6 8, in diameter) can pass. W ~ ~ O ~ as ~ Cs2.2) is Recently, it was found that C S ~ . ~ H O . ~ P(abbreviated microporous and, according to adsorption experiments, has effective pores of about 7 A (48). Table IV is a comparison of the adsorption capacities of Cs2.2 (abbreviated as (32.5) for various molecules. Benzene and C S ~ , ~ H ~,2040 .~PW (kinetic diameter = 5.9 A) and neopentane (kinetic diameter = 6.2 A) are adsorbed on Cs2.2 and Cs2.5, and the relative adsorption capacity of Cs2.2 and Cs2.5 are similar to the corresponding ratio for N2 adsorption. On the other hand, both benzene and neopentane are little adsorbed on Cs2.1Ho.9PW12040 (Cs2. l), indicating that the effective pore size of Cs2.1 is less than 5.9 A. Of particular interest are the results observed with l13,5-trimethylbenzene (kinetic diameter = 7.5 8) and 1,3,5-triisopropylbenzene (kinetic diameter = 8.5 A). These two alkylbenzenes are adsorbed significantly on (32.5, but little on (32.2, so that the pore size of Cs2.2 is in the range of 6.2-7.5 A and that of Cs2.5 is larger than 8.5 8. This result demonstrates that the pore structure can be controlled by the substitution of H+ by Cs' (48). B. SYNTHESIS Heteropolyacids are prepared in solution by acidifying and heating in the appropriate pH range (I, 49-54). For example, 12-tungstophosphate is formed according to Eq. (1). Free acids are synthesized primarily by the following two methods: (1) by extraction with ether from acidified aqueous solutions and (2) by ion exchange from salts of heteropolyacids. Dawson-type heteropolyanions,

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

127

X2W180:2, are isolated as soluble ammonium or potassium salts or as free acids by extraction into ether (29, 30). Mixed addenda heteropolyanions with regiospecific substitution need careful preparation by use of lacunary heteropolyanions. If they are prepared from aqueous solutions of corresponding oxoanions, the products are usually mixtures of heteropolyanions having different compositions of addenda atoms. General procedures for the syntheses of various kinds of heteropolyacids are described in the literature (51-54).

C. STABILITY Particular attention should be paid to both the stability in solution and the thermal stability. The condesation-hydrolysis equilibria of heteropolyanions in aqueous media are shown in Fig. 8. Each heteropolyanion is stable only at pH values lower than the corresponding solid line (55). Some solid heteropolyacids are thermally stable and applicable in reactions with vapor-phase reactants conducted at high temperatures. The thermal stability is measured mainly by X-ray diffraction (XRD), thermal gravimetric analysis, and different thermal analysis (TG-DTA) experiments. According to Yamazoe et al. (56), the decomposition temperatures of H3PMo12040 and its salts depend on the kinds of cations: Ba2+, CoZf (673 K) < Cu2+, Ni2+ (683 K) < H', Cd2+ (693 K) < Ca2+, Mn2+ (700 K) < Mg2+ (710 K) < La3+, Ce3+ (730 K), where the

60 80 Cornposition/%

20 40

FIG.8. Stabilities of aqueous heteropolyacids (from Ref. 55): ( I ) PMoI20:{, (2) PW,,O:,, ( 3 ) GeMo120:o, (4) GeW120&-, (5) P2W180:2, ( 6 ) SiW120:i, (7) PMoIIO:9, (8) P2M0502;, (9) HzWi202Rr (10) P W I I O ; ~ .

128

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

decomposition temperature (in parentheses) is estimated by the exothermic peak in DTA. Herve et al. (57) investigated the thermal changes of structures by means of XRD and TG-DTA for Keggin-type heteropolyacids and proposed Scheme 2. Infrared spectroscopy of H4PMoI1VO40 showed the release of vanadium atoms to form and vanadium phosphate species (58). Exposure to water vapor induces the decomposition of the latter (indicated by the disappearance of a band at ca. 1037-1030 cm-I) (58). Results from TG and DTA show the presence of two types of water in heteropoly compounds, i.e., water of crystallization and “constitutional water molecules” (59). Loss of the former usually occurs at temperatures below 473 K. At temperatures exceeding 543 K for H3PMo12040or 623 K for H3PW12040,the constitutional water molecules (acidic protons bound to the oxygen of the polyanion) are lost. Data obtained by in situ XRD, 3 1P NMR, and thermoanalysis show that thermolysis of H3PMo12040 proceeds in two steps, as shown by Eq. (6) (60).

-

473 - 623 K

H3PMo12040-nH20

-

nH20

658 K

L

HxPM012038.5+x/2

-1.5H20

(x = 0.01)

H3PM012040

<

> 673 K

(PMo12038.5)~

Mo03(0.01P)

(6)

I>723K

+ (MbO2I2P2O7

The Moo3 phase appears at temperatures higher than 573 K. Thermal gravimetric analysis of H3PW12040and of C S ~ . ~ H O . ~ P showed W~~O~O that entire water molecules of crystallization are lost at temperatures as low as 573 K, and acidic groups are removed as water is formed from protons and lattice oxygens at temperatures exceeding 623 K. The numbers of protons lost, were 0.24, 0.31, and 0.32 after treatment at 623, x, in 673, and 773 K, respectively, whereas infrared spectra of Cs2.sHo.sPW1 2 0 4 0 remained unchanged at temperatures up to 773 K ( 6 1 ~ ) Similar . removal of protons of K~.5Ho.5PMo12040 begins by 500 K (61b).

D. CHARACTERIZATION OF HETEROPOLY COMPOUNDS 1 . Infrared Spectroscopy Infrared (IR) spectroscopy is a convenient and widely used method for the characterization of heteropolyanions. Keggin, Dawson, and lacunary heteropolyanions can be distinguished by their characteristic bands. a. Keggin Structure. In Table V (62-64), a partial list of the reported IR bands is given with their assignments. IR spectra of XWI2O;O, XMO,~O:~, and

129

CATALYTlC CHEMISTRY OF HETEROPOLY COMPOUNDS

curc

298 K

298 K

I

H4PMollV040- 13H20 triclinic T

H3PMo12040*13H20 triclinic T

I I H~PMo~~VO~O-~H~O

I I H3PM01204~7-8H20 333

- 353 K

333 - 353 K

unstable (cubic?)

unstable (cubic?)

I I

373

I I (0.5P205) 12 MOO3

723 K

undetected orthorhombic

-I353 K

I

H3PW120qty 6H20 cubic C 453 623 K

I

I

H3PMol 1vo40 tetragonal

t

333

-I

- I623 K

H3PMo12°40 tetragonal

H3PW12040 tetragonal

I

723 K

I

H3PW120q0.13H20 triclinic T

I I (0.5P205)

823 K

I

(0.5P205 + 0.5V205) t 12 Moo3 undetected orthorhombic

t 12 Wo3 undetected orthorhombic

SCHEME2

metatungstate are shown in Fig. 9 (62a).The vibrational mode of XO, is almost independent of the others for X = P, but it is mixed with other vibrational modes for X = Si, Ge, B, etc. (64).The X - 0 stretching bands (Si, 923 cm-'; Ge, 830 cm- I ; P, 1080 cm- I ) for XW,,O;, show higher frequencies than those TABLE V Infrared Absorption Bands of Heteropolyacids, em-' (62-64)

1080 982 893 812

926 980 878 779

I070 965 870 790

818 978 886 765

910 958 860 780

914 960 902 810

445 960 895 738

802 955 875 765

1091 962 914 780

v,,(X-0) v,,(W=O) v.,(W-0-W) v,,(W-O-W)

(X-0) v.,(Mo=O) v(Mo-0-Mo)

V",

130

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

FIG.9. Infrared spectra of XW12O;O and XMo120;U. (From Ref. 62a.) of XOl- anions, suggesting higher Ir-bond character of X - 0 bonds in the Keggin anion. Cation size influences the v(W-Od) frequencies for b-SiW120:O; the value of v(W-Od) decreases as the van der Waals radii of tetraalkylammonium cations increase (65). Due to the loss of hydrogen bonding between Od and water, the M-Od band (M = Mo, W) shifts to a higher frequency and the P-0 and M-Od peak intensities change (5). The W-Od ~ O ~into O doublets, suggesting a direct band for anhydrous C U ~ . ~ P W ~splits interaction between the polyanion and Cu2+ (66). In Table VI, IR bands of lacunary X M I 1 0 & are summarized (67).The P-0 stretching band for PWIIO:9 is split into 1085 and 1040 cm-’. It is believed

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

131

that the change of the symmetry from 5 (XMI2) to C,y(XMII) leads to the broadening of the band and sometimes causes bond splitting (67). The isomers, (Y- and P-PMo120:0 or SiW1203,, can also be distinguished by IR spectroscopy (68). The effects of solvent on the frequency have been reported (69). IR spectra of hydrated H3PW12040include a broad OH stretching band and two OH bending bands, at 1610 and 1720 cm- I . The latter two correspond to water and protonated water, respectively ( I ) . b. Dawson Structure (70). The IR spectrum of ( Y - P ~ W I ~resembles O~~ that of PW120:0. Three IR bands are observed in the PO4 stretching region, at 1090 (s), 1022 (w), and about 975 cm-I (sh) due to D3h symmetry of two PO4 groups. IR bands at 960, 912, and 780 cm- I are assigned to v(W-Oh), v(W-O,), and v(W-0b). respectively. 2. Raman Spectroscopy Vibrational frequencies in the Raman spectra of X M l l and XM12(X = Si, P; M = Mo, W) are summarized in Table VII (62, 65, 67, 71).The X - 0 vibration in Td symmetry of X04 is Raman-inactive. Among M - 0 bonds, M-Od is Raman-active. Raman spectra indicate that with an increase in pH, the structure of PMo120:i in aqueous solution changes, as shown by Eq. (7).

The states of hydrates of heteropolyacids are best distinguished by Raman spectroscopy, since Raman spectra give better resolved OH stretching bands than IR spectra (62, 72). H3PW12040.29H20has bands at 3570, 3525, 3490, 3450, 3205, and 3140 cm-', which correspond to 0-H distances of 2.94, 2.89, 2.86, 2.84, 2.72, and 2.65, respectively. The bands at 3205 and 3 140 cm- I are assigned to v(0H) of H(H2O);. The OH frequencies of H3PMo12040and H4SiW12040appear at 3570 and 3525 cm-', respectively. TABLE VII Raman Vibrational Frequencies of Keggin Anions (Aqueous Solutions), e m - ' (62. 65.67, 71)

pv,(M-OJ

shows polarized vibration.

132

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

3. NMR Spectroscopy a. Solid-state ' H NMR Spectroscopy. The broad-line NMR spectrum of H3PMo12040 29H20 shows that H30f and H20 become indistinguishable at temperatures greater than 298 K (73, 74). The NMR spectrum of anhydrous H3PMo12040 includes a narrow resonance (width <1 G) at 298 K due to equivalent protons (75). The H--H distance in anhydrous H3PW12040is estimated to be 4.0-4.5 A (76). Three different states of protons are detected by solid-state 'H MAS NMR at room temperature, as shown in Fig. 10 spectroscopy for H3PW12040.nH20 (77). A sharp 'Hresonance observed for n = 17 shows that the protons are in a uniform state and highly mobile. Much broader and weaker lines for less hydrated states indicate lower mobility of protons. The chemical shifts for n = 17 and n = 6 (7.3-7.5 ppm) correspond to clusters of hydrated water, as in Fig. 2a. The resonance at 9.2 ppm, for anhydrous H1PW12040was assigned to protons attached to the most basic bridging oxygen atoms, on the basis of IR results (78) and the basicity estimated by "0 NMR spectroscopy (see below).

1

1

1

1

1

14 10 6

1

1

1

1

1

2 -2

PPm FIG. 10. 'H MAS NMR spectra of H3PW12040'nH20: (a)n = 17, (b) n = 6, (c) n Ref. 77.)

=

1.5. (From

133

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE VIll Chemical Shifis of "P f i r Heteropoly Compounds Solution (ppm)

- 3.9 - 1.4 -2.9 to -3.4 - 1.7 to -3.4 - 14.9 - 10.4 - 12.7 - 14.2 - 14.2 - 13.6 - 13.4

+ 1.0

~~

~

Reference 79 79 79 79" 79" 79u 79

85 85 85 h

Solid (ppm)

- 3.9 - 2.9 - 4.0 - 4.4 - 4.5 - 4.5 -3.1 - 5.5 - 15.1 - 15.6 - 11.1 - 10.9 - 15.3

Reference 6lb 6lb 82 82 82 82 82 82

- - 14.9

77 77 77

47 47 ~~

O'Donnell, S. E., and Pope, M. T., J Chem SOC.Dalton, 2290 (1976). Kato, R., Kobayashi, A., and Sasaki, Y., Inorg. Chem. 21, 240 (1982).

b. "P NMR Spectroscopy. The 3'P chemical shift provides important information concerning the structure, composition, and electronic states of these materials. The chemical shifts of typical heteropoly compounds are summarized in Table VIII. The chemical shift in aqueous solutions is correlated with the P-0, bond strength [v(P-O,)] (70). Little solvent effect on chemical shift indicates that the central atom is effectively shielded. The decrease of the chemical shift in the series PWI2O:O > PMol20;0 > PVl2O:0 parallels the decrease in the IR frequency of PO4, with both reflecting P - 0 n-bonding character. Five resonances observed for H5PV2M010040 correspond to the possible five isomers, in which the locations of two vanadium atoms are different (80). P M O ~ W ~ Oin: ~aqueous solution gives 13 peaks expected from the statistical distribution of P W 1 2 - x M ~ x 0 4(x0 = 0-12) (81). Figure 11 shows the solid-state 31 P NMR spectrum of Cs3PMollW040 (82) prepared from mixed aqueous solutions of H3PMo12040, H3PW12040, and CsN03. Five resonances agree well with the spectrum observed from the reaction mixture in solution and are assigned to a statistical mixture of (x = 0-4).M3PMo12040 (M = H, Na, K, Cs, and NH4) give similar chemical shifts. The 31PNMR chemical shift is greatly dependent on n in H3PW12040*nH20 (47, 77), the values being - 15.1 to - 15.6ppm for n = 6 and - 11.1 to - 10.5 ppm for n = 0. This difference is explained as follows: In the former, protonated water, H(H20):, is connected with the heteropolyanion by

134

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

0 FIG. 1 I .

5 Chemical ShiWppm

10

"P MAS NMR spectrum of C S ~ P M O ~ ~ W (From O ~ ~Ref. . 82.)

hydrogen-bonding at terminal oxygens, and in the latter, protons are directly attached to oxygen atoms of the polyanion. c. I83 W, "Mo, and *'Si NMR Spectroscopy. 183W Chemical shifts are sensitive to the heteroatom (83), being - 130.4 ppm for BWl20;0, -111.3ppm for H2WI20:O, -103.8ppm for SiW120:0, and -98.8ppm for PWl20:O. SiWIIO!J has five pairs of structurally identical W atoms and one unique W atom. Accordingly, SiW, gives six clearly separated resonances having the intensity ratio of 2 : 1.6 : 1 : 2 : 2 : 2, which is close to the expected ratio of 2 :2 : 1 : 2 :2 : 2. Similarly, P2Wl&ir which has twelve equivalent W atoms around its belt and six equivalent W atoms capping its ends (26), has two resonances; the larger one at - 170.14 ppm and the smaller doublet at - 124.87 ppm. The 183WNMR spectrum of 6-electron reduced cr-SiW120:0 shows three narrow resonances with the intensity ratio 1 : 1 : 2 (Fig. 12) (84). Two (3W, 6W)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

135

have chemical shifts very close to that in the oxidized state and assigned to W(VI), whereas the third ( - 1500 ppm) is attributed to a W(IV) anion. This assignment is supported by the chemical shift of the W(IV) cation, w~o~(H~o):+. It is suggested that the protonated w6(v)w6(vI) polyanion undergoes intramolecular disproportionation to give W3(IV)W9(VI) species. The structures of positional isomers, a-1,2-XV2Wl~O:0, and a-l,2,3XV3W90kngf1)- (X = Si, P), are established from the 2-D connectivity pattern and 1-D spectra of 183W(85). The spectra of XV3W90:; (X = Si, P) are exclusively of the a-1,2,3-isomers. a- And p-SiWlzO:O can be also distinguished by 183WNMR spectroscopy; a-SiW120:0, gives one singlet (12 equivalent W atoms), and b-SiWI2O:[ three resonances in the ratio 1 : 2 : 1 (86). d. 170 NME Spectroscopy. 170 NMR spectroscopy gives information about the bonding nature of oxygen atoms. Figure 13 shows the assignments of the chemical shifts (87-89). There is a correlation between the downfield shift and the decreasing number of metal atoms to which the oxygen atom is bonded. The chemical shifts for the SiMoI20:0, SiW120:6, PMo120;;, and PW120:O are summarized in Table IX. 4. Electronic Spectra

Electronic absorption spectra give information about the electronic states of heteropolyanions (2). PWI20;; shows absorption at about 38,000 cm- I , due to

-1000 -I

FIG. 13. Assignments of " 0 NMR shifts due to various salts of mixed oxides. (From Refs. 87-8Y. )

136

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE IX 0 NMR Spectral Data for Heteropolyanions (87-89)

17

Chemical shift (ppm) Anions

M-Od

SiMo120!; S~W 12~:” SiMoWIIO:; PMo120:R Pwt2Ok-

M-0h-M.

92 8 76 I 929(Mo-O~) 726(W-Od) 936 769

M-Oc-M

M-O,-(M),

580, 555 427,405 504,469

41 21 27

583, 550 431,405

78 -

ligand (oxygen) to metal (W) charge transfer (LMCT) (90). In most cases of XW’ Wl (one-electron reduction heteropoly blues), three absorption bands are observed at 8000-10,000 cm-l (band A), 13,000-16,000 cm-’ (band B), and ca. 20,000 cm-l (band C) in addition to that at 38,000 cm-I. The bands B and A are the results of intervalence charge-transfer transitions (IVCT) between metal atoms. Two types of transition are possible for reduced Keggin anions, that is, within an edge-shared group of M 0 6 octahedra (“intra” transition) and those between metals atoms linked by corner sharing. The band A has tentatively been assigned to an “intra” and the band B to an “extra” transition ( 2 ) . The UV band positions of various heteropolyacids are summarized in Table x (9&93). +

TABLE X UV Absorption Bands of Heteropolyanions Compounds

Absorption“ (kK)

Half-wave potentialh (V)

38.0, 50.0 38. I 37.8 39.0 38.3 34.0, 39.0 3 1.O, 47.0 32.3, 47.2 20.0, 25.0 20.2

- 0.023 -0.187 - 0.349 - 0.520 -0.510 + 0.02“ - 0.55“

See Refs. 9&93. Pope, M. T., and Varga, G. M., Inorg. Chem. 5, 1249 (1966); Prados, R. A., and Pope, M. T., Inorg. Chem. IS, 2547 (1976).

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

137

5 . Others a. Extended X-Ray Absorption Fine Structure (EXAFS) Spectroscopy. EXAFS spectra of PMoI20:; and PW120:; as amine salts have been measured (94, 95). PW120:; gives distinct peaks due to W-Od ( 0 . 1 6 ~ )and W-0-W (0.196 nm), whereas Mo-0 peaks of PMo120:i are not discernible. For the latter, a large Debye-Waller factor for Mo-0 bonds and also multiple scattering effects are presumed.

b. Scanning Tunneling Microscopy (STM) and Transmission Electron Microscopy (TEM). H3PWI2O4~ deposited on freshly cleaved and highly oriented pyrolytic graphite in air has been measured by STM (96). A fairly regular periodic pattern is observed, suggesting that the individual heteropoly species were directly imaged. Individual anions of the Dawson-type cyclopentadienyl titanium (CpTi) heteropoly compound, K7(C5H5)TiP2W17061 , were observed by TEM (97).The size of the anion is estimated to be 1 .&IS nm. This is consistent with a size determined from X-ray crystallographic data indicating an ellipsoid of about 1.0 X 1.5 nm. c. Electron Spin Resonance (ESR) Spectroscopy. ESR spectra give information about mixed-valence structures of reduced heteropoly compounds. ESR data for Keggin-type molybdates reduced by one electron (1e- reduction) in solution are shown in Table XI (2, 92, 98, 99). A Mo5+ signal with hyperfine at temstructure due to Mo5+ ( I = 5/2) is observed for PMo5+Mo7:O:; peratures less than 40 K (99). At higher temperatures line broadening is observed, and the hyperfine structure disappears. At room temperature, no Mo5+ signal is observed. These results indicate the rapid hopping of an electron among TABLE XI ESR Parameters of Some Reduced Polymolybdates (from ReJ 2) Anion 1.916 1.917 1.938 1.913 1.935 I .93 1 1.914 I .935

1.930 1.924 1.949 1.939 1.948 1.944 1.93 I 1.951

138

TOSHIOOKUHARA, NORITAKAMIZUNO, AND MAKOTO MISONO

I2 equivalent Mo atoms in a Keggin anion at higher temperature, and the mixed valence behavior is classified as class I1 (99, 100). The extent of electron delocalization increases as the number of molybdenum atoms in a Keggin anion increases. The order of extent of electron hopping is estimated to be PM0120:; > GeMo120g> Mo60:, (99). A similar electron hopping is estimated by the very weak signal intensity of MoS+ for solid H3+xPM~1204~ reduced by H2 at 423 K or room temperature (101, 102). Upon 2e- reduction, about 60% and 100% of the electrons are paired for SiMo120:0 and PWIZO:U, respectively, and the latter shows diamagnetism (2). Solid H3PMo12040 reduced by H2 at a lower temperature shows a very weak ESR signal intensity of Mo", probably because most of the Mo5+ ions are not detectable due to the rapid hopping of electrons. Heat treatment, which eliminates oxide ions from the heteropoly anion, leads to development of the Mo5+ signal, indicating the localization of electrons (101, 102). Early reports of ESR are likely due to these species. Several different spectra of reduced H3PMo12040 species are observed in highly reduced samples. The extent of reduction of H3PM0120~during the oxidation of methacrolein has also been investigated by application of ESR spectroscopy for detection of MoS+ (103). The states of V in the mixed-valence Keggin anion and Cu countercation were also investigated. The results show that more reducible countercations or addenda atoms such as Cu2+ and V5+ are reduced first, and an electron is localized on them (104-106). d. X-Ray Photoelectron Spectroscopy (XPS)). XPS gives information about electron density of solid heteropoly compounds (107-115). It was reported that Mo5+ was fairly uniformly present throughout the bulk of H3PMo12040formed by H2 reduction, whereas the surface was preferentially reduced by cyclohexane (101). The oxidation states of countercations Pd, Zn, and Cu were investigated by XPS in relation to redox properties and oxidation catalysis. Cu2+ and Pd2+, having higher electron affinities than Mo6+,in 12-molybdophosphatesare easily reduced by H2 and act as electron reservoirs (107-109). The following observations have been made for bulk heteropoly compounds. 1. The binding energy of 0 1 s electrons decreases with increasing negative charge of the heteropolyanion: PW12O:O > SiW120:; 2 BW120:; > H2WI20:; > PWIIO& > SiWl1O;9 > H2WI20::- (113). 2. For isostructural heteropolyanions, the 01s binding energy is higher for tungstates than for molybdates (I 13). 3. In PMo120:0 and SiMo120:U (2e- reduced state), two Mo atoms are in the oxidation state of 5 and ten Mo atoms are in the + 6 state (113, 114). A similar result was obtained for 2e--reduced PW120:; (113, 114).

+

139

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

4. A good correlation exists between the acid strength and the difference between 01s and W4f binding energies of HjFWl2040 and its salts (115).

111. A.

Acidic Properties ACIDITYIN

SOLUTION

Typical heteropolyacids having the Keggin structure, such as t -,PW12040and H4SiW 1 ~ 0 4 0are , strong acids; protons are dissociated completely from the structures in aqueous solution (8, 116). The dissociation constants, pK,, of heteropolyacids depend on the solvent. These are summarized in Table XU, together with pK, values for mineral acids ( I 17-1 19). Heteropolyacids are much stronger acids than H2S04, HBr, HCl, HN03, and HC104. For example, in acetic acid, the acid strength of H3PW12040is greater than that of H2S04 by about 2 pK, units. In acetic acid, which is less polar than water, heteropolyacids behave as relatively weak 1-1 electrolytes. The effect of solvent is evident for mixed solvents (120); as the concentration of water in aqueous acetic acid changes from 100 vol% to 4 vol%, the Hammett acidity function, Ho, for H3PW12040(0.1 M) decreases from 0.01 to - 1.78, TABLE XI1 Acid Constants of Heteropolvacids in Nonaqueous Media at 298 K ( I 1 6 1 19) Acetone Acid

PKI

PK2

Ethanol pK3

PKI

pK2

pK3

(3.0)

4.1

-

-

Acetic acid PKI

~~

(1.6) (1.8) -

(2.0) (2.0) (2.1) (2.1) (2.1) . -

(3.0) (3.2)

3.98 4.37

(1.6) -

-

-

-

-

-

3.61 3.62 3.69 3.73 3.90

(5.3) (5.3) (5.5) (5.6) (5.9)

(2.0) (1.8) (1.9) (1.9)

3.96 3.41 3.77 3.74

(6.3) (5.1) (5.9) (5.8) -

-

-

-

-

-

-

4.77 4.74 4.78 4.91 4.70 -

4.68 4.78 4.25

140

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

indicating an increase in acid strength. In the case of concentrated aqueous solution of H3PW12040,the acidity hnction is less than that of H2SO4 by 1-1.5 units (120). Titration curves indicate that the three protons of H3PW1204"are equivalent in aqueous solution. Three protons of H3PW 1 2 0 4 0 dissociate independently in acetic acid, as measured by 13C NMR spectroscopy (121). The greater acid strength of heteropolyacids than that of mineral acids is explained as follows (122). Since in heteropolyanions the negative charge of similar value is spread over much larger anions than those formed from mineral acids, the electrostatic interaction between proton and anion is much less for heteropolyacids than for mineral acids. An additional important factor is possibly the dynamic delocalizability of the charge or electron. The change in the electronic charge caused by deprotonation may be spread over the entire polyanion unit. As for the acid strengths of heteropolyacids, the following order has been reported for the compound in acetone (Table XI) (116): H3PW12040> H4SiW12040 = H3PMo12040 > H4PMoIIVO40 > H4SiMo12040.The acid strength decreases when W is replaced by Mo or V and when the central P atom is replaced by Si. The effect of the central atom has been demonstrated for acetonitrile solutions of Keggin-type heteropolytungstates. As shown in Fig. 14, the acidity increases in general with a decrease in the negative charge of the heteropolyanion, or an increase in the valence of the central atom (the valence of the central atom increases in the order Co < B < Si, Ge < P) (63). This order is reasonable, since the sizes of the polyanions are nearly the same; thus the interaction between the proton and the polyanion would decrease as the negative charge of polyanion decreases. 4

3

r" 2

1 2 3 4 5 6 7 Negative charge of polyanion

FIG.14. Values of Hammett acidity function (Ho) of H,XWt204o as a function of the negative charge of the polyanion. (From Ref. 63.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

141

Pope et al. (123) measured the formation constant of the 1 : 1 complex of 1,l -dihydroxyl-2,2,2-trichloroethane(chloral hydrate) and polyanions in nitrobenzene by using NMR spectrometry [Eq. @)]: ,0--H,

C13CCH,/o-H + A-

C13CCH:

0-H

“O--

‘:A-

(8)

H’

The formation constants (in parentheses) of the complexes are as follows: PW120:o (1.30) < PMo120:0 (3.1 1) < SiMo1204, (24.7). These values represent the capability of the heteropolyanion to form hydrogen bonds. Thus the acid strength is in the order H3PW12040 > H3PMo12040 > H4SiMo12040.Izumi et al. (124) obtained the following order in acid strength by the same technique: H3PW12040 > H3PM012040 > H4SiW12040= H4GeW12040> H4SiMoI2O4,,> H4GeM012040. These orders are in genera! agreement with those obtained (118, 119) with indicator tests (Table XII). In addition to the acidity, the softness of the heteropolyanion is an important characteristic relevant in catalysis (124). The softness has been estimated by the equilibrium constant in aqueous solution of the following reaction [Eq. (9)] at 298 K. AgnX + nNa1

* nAgl + Na,X

(X

=

polyanion)

(9)

The order of softness was found to be the following: SiWl2O4, > GeWl20:; > PWl20:O > PMo120iU > SiMo120d, > SO:-. The softness greatly influences the catalytic behavior in concentrated aqueous solution or in organic solution, as described below.

B. ACIDITYI N

THE SOLID STATE

1. Acid Forms The strength and the number of acid centers as well as related properties of heteropolyacids can be controlled by the structure and composition of heteropolyanions, the extent of hydration, the type of support, the thermal pretreatment, etc. Solid heteropolyacids such as H3PW12040and H3PM012040 are pure Br~insted acids and are stronger than conventional solid acids such as Si02-A1203 (45, 125). According to an indicator test, H3PW12040 has a Hammett acidity function less than - 8.2 (126), and it has even been suggested to be a superacid (127, 128). A superacid is an acid with a strength greater than that of 100% H2SO4, i.e., a value of If0 < - 12 (129). Thermal desorption of basic molecules also reveals the acidic properties. Figure 15 compares the acid strengths of heteropolyacids and Si02-A1203 (125). Pyridine adsorbed on Si02-A1203 is

142

TOSHIO OKUHARA, NORITAKA MIZLTNO, A N D MAKOTO MISONO

.ss 4

%' 1

0

300

400 500 TemperatureK

600

FIG.15. Thermal desorption of pyridine from heteropolyacids (from Ref. 125). H3PW12 refers to WWi2040.

completely desorbed at 573 K. On the other hand, sorbed pyridine (see Section VI) in H3PW12040mostly remains at 573 K, indicating that H3PW12040is a very strong acid. The acid strength can also be demonstrated by temperature-programmed desorption (TPD) of NH3 (Fig. 16) (127). H3PW12040 gives a relatively sharp peak at about 800K, together with peaks for N2 and H20 at the same temperatures (not shown). N2 is formed by the reaction of NH3 with oxygen atoms in the heteropolyanions. NH3 adsorbed on Si02-A1203 and on HZSM-5 is mostly desorbed at temperatures less than 800 K (127, 130). Thus according to TPD results for NH3, the order of the acid strength is sulfated zirconia (SO;-/ZtQ) > H3PW12040 > HZSM-5 > SiO2-AI2O3. The temperature of NH3 desorption from (or decomposition of) ammonium salts of heteropolyacids is in the following order: (NH&PW1204o > (NH4)4SiW12040> (NH&PM012040 > (NH414SiMoI 2 0 4 0 (131). The strengths of heteropolytungstic acids have been determined more quantitatively by calorimetry of NH3 absorption (132, 133). Figure 17 shows the differential heats of NH3 absorption after treatment at 423 K. The initial heats of absorption of NH3 are as follows: 196 kJ mol-I for H3PW12040; 185 kJ mol-' for H4SiW12040; 164 kJ rno1-I for H6P2W210,1(H20)3; and 156 kl rno1-I for H6P~W18062. These values show that the order of acid strength is H3PW12040 > H4SiW12040> H6P2W21071(H20)3 > H6P2W18062, in afleement with the statements above. The data also indicate that the Keggin-type heteropolyacids are much stronger acids than Dawson-type heteropolyacids (132).

143

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS 21

373

'

'

573

'

'

I

a

'

773 973 TernperaturdK

1173

~ ~H3PW120m. O~U. FIG.16. TPD profiles of NH3 from various solid acids: (a) C S ~ . S H ~ . ~ P W(b) (c) SO:-/Zr02, (d) SiO2-AI2O3. (e) H-ZSM-5. Solid line: NH3 (m/e = 17); dotted line: N2 (mle = 28). (From Ref. 127.)

Heats of NH3 absorption further confirm that the heteropolyacids are stronger acids than zeolites or simple metal oxides: 150 kJ mol- for HZSM-5 and 140 kJ mol-' for y-A1203(133). Table XI11 shows the strengths measured by Hammett indicators with pK, values ranging from -5.6 to - 14.5. As described above, dried H3PW12040 possesses superacidity (127). The order of the acid strengths agrees with that

'

r

200

E al c .- 100 -5 L

C

f

E

n 0 0

200

400

600

800

loo0

1200

pmol of NH3 9-1

FIG. 17. Differential heat of NH3 absorption as a hnction o f the ammonia uptake for heteropolyacids. (From Ref. 132.)

144

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XI11 Acid Strengths Measured by Hammett Indicators (127) pK. of indicatorh

Catalyst" C~2.sHo.sPWi20a(573 K) HiPWi2040 (573 K) HZSM-5 (808 K) SO: 12102 (643 K) SiO2-AI2O3(723 K) ~

-5.6

-8.2

-11.35

-12.70

-13.16

+

+ + + + +

+ + + + +

+ + + + +

+ + +

+ + + 4-

-

-13.75

-14.52

-

-

-

-

-

-

+-

+

-

-

a The figures in parentheses are the pretreatment temperatures. Acidic color of the indicator was observed ( + ), or not ( - ). Indicators: - 5.6. benzalacetophenone; - 8.2, anthraquinone; - I I .35. p-nitrotoluene; - 12.70, p-nitrochlorobenzene; - 13.16, m-nitrochlorobenzene; - 13.75. 2,4-dinitrotoluene; - 14.52. 2.4-dinitrofluorobenzene.

estimated from NH3 TPD (127). In Fig. 18, various superacids in both solution and solid are summarized with reference to the Ho function (128, 129). H3PW12040is stronger than Nafion (Ho= - 12), but weaker than SO:-/ZrOz and AIC13-CuS04.

FIG. 18. Acid strengths of liquid and solid superacids. (From Ref. 128.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

145

Viswanathan et al. (115) reported small differences in binding energy between 01s and W4f: H3PW12040 (495.5 eV) > All.5PW12040(495.4 eV) > Na3PW12040(495.1 eV). This is the same order as the order of acid strength: > Na3PW12040. H ~ P W I Z O>~ A11.5PWl2040 O The locations of protons in solid heteropolyacids have been studied by NMR and IR spectroscopies. The protons of H3PW12040 with a high water content exist in the form of dioxonium ions H50: or H+(HzO), (21). Data obtained by 17 0 NMR spectroscopy confirm that the bridging oxygen atoms are protonated in CH3CN solution (88). In the solid state of dehydrated H3PW12040, there are two possible sites for protonation: the terminal W=O oxygen atoms or the bridging W-0-W oxygen atoms. Kozhevnikov et al. (134) assumed on the basis of an 170 NMR study that the terminal W = O oxygen atoms are the predominant protonation sites. The resonance of the terminal W=O oxygen atom shifts upfield 60 ppm, relative to that of the solution spectrum, whereas no such large shift was observed for the bridging W-0-W oxygens. On the basis of IR band broadening of the W-0,-W band for anhydrous H3PW12040 and D3PW12040,Lee et al. (78) concluded that protons are located on the bridging oxygens of PWI2O:,j-. In contrast, there was no difference observed for the W-Od band. As described above, the most basic oxygen is the bridging oxygen according to 170 NMR spectra of the solution. It was reported in a crystallographic study that reduced PMo12040 is protonated at bridging oxygen ( 135). Quantum-chemical calculations at the EHMO level (136) and bond-lengthhond-strength correlations (13 7) indicated that the bridging oxygen atoms carry the highest electron density (are the most basic).

2 . Salts Acidic properties of heteropoly compounds in the solid state are sensitive to countercations, constituent elements of polyanions, and tertiary structure. Partial hydrolysis and inhomogeneity of composition brought about during preparation are also important in governing the acidic properties. There are several possible types of origins of acidity (5): 1. Dissociation of coordinated water: for example, Ni(H20)b+ +Ni(H20),,,- ,(OH)+ + H i

2. Lewis acidity of metal ions. 3. Protons formed by the reduction of metal ions: for example, Ag'

+ tH2+Ago + Hf

4. Protons present in the acidic salts: for example, Cs,H3 -.rPW12040. 5. Partial hydrolysis during the preparation process: for example, P W 1 2 0 g-PW,,O:C

+ W0:- + 6H'

146

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

As for type ( I ) , Niiyama et al. (42) proposed that protons are generated by dissociation of water and that the equilibrium of the dissociation is a function of the electronegativity of the metal cations. Formation of Brensted acid sites in the aluminum salt of H3PW12040as a result of exposure to water vapor at 573 K was confirmed by IR spectra of sorbed pyridine (138). The presence of Lewis acidity [type (2)] or Brensted acidity [type ( l ) ] is revealed by the IR spectrum of pyridine sorbed on All,5PW12040 (138, 139). IR spectra of sorbed of NH3 show that C U ~ . ~ P W Ihas ~ OLewis ~ O acidity as well as Brensted acidity (140). Ghosh and Moffat (141) measured the acidities of several salts of H3PW12040 with Hammett indicators. The acid strength increases with an increase in the calculated charge on the peripheral oxygen atom of the polyanion: Zr > A1 > Zn > Mg > Ca > Na (141). Proton formation from H2 [type (3)] has been demonstrated by IR and NMR spectroscopies for Ag3PW12040(142). Ag3PW 1 2 0 4 0 treated with hydrogen at 573 K and then pyridine gives IR bands of pyridinium ion. Such bands are not observed for Ag3PW12040which had simply been evacuated at 573 K prior to sorption of pyridine. The former sample (in the absence of pyridine) was catalytically much more active than the latter. As shown in Fig. 19, the introduction of H2 to Ag3PW12040at 488 K leads to three resonances at 9.3, 6.4, and 4.1 ppm in the 'H NMR spectrum (143-145). The resonance at 4.1 ppm is due to H20 molecules formed by the hydrogen treatment. That at 6.4 ppm increased as the H2 pressure increased and disappeared as a result of the addition of deuterated pyridine (C5D5N) at 303 K. The peak at 9.3 ppm was nearly unchanged during the experiments. These results indicate that the acid strength of the proton-containing groups indicated by the resonance at 6.4 ppm is higher than that of the groups indicated by the resonance at 9.3 ppm. This difference in acid strengths is attributed to the differences in the interaction between proton and polyanion. Group B salts, incorporating acid groups, have high surface areas and high thermal stabilities. In TPD of NH3, gives a desorption peak in the temperature range 633-923 K, and the peak temperature is close to that for H3PWIZ040,indicating that the acid strength of Cs2.5Ho.5PW12040 is similar to that of H3PW12040(127); C S ~ . ~ H ~ has . ~a slightly P W ~broader ~ ~ ~distribution ~ of acid strengths than H3PW12040 (Fig. 16). In group A salts, the acid amount exceeds the number of protons expected from the formula for Na salts of H3PW12040. Partial hydrolysis has been proposed [type ( 5 ) ] (46a). Solid-state 31PNMR spectroscopy was used to demonstrate that all protons of C S ~ . S H O . ~ P Ware ~ ~distributed O~O randomly through the bulk of the material. Figure 20 is a 31P NMR spectrum for C S ~ . ~ H ~ . ~ P(47). W ~ ~Special O ~ O care was necessary in the measurement to protect the sample from moisture. Four resonances ( - 14.9, - 13.5, - 12.1, and - 10.9 ppm) appeared when

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

147

. 6.4

I

I

30

20

10

0

-10

Shift I ppm FIG. 19. ' H NMR spectra (room temperature) in reduced Ag3PW12040.(a) Ag3PW12040reduced with 40 kPa of H2 at 488 K and recorded in the presence o f H2; (b-g) recorded after evacuating H l at 77 (b), 303 (c), 333 (d), 488 (e), 523 (0 and 623 K (g). Reprinted with permission from Ref. 143. Copyright 1993 American Chemical Society.

Cs~.5Ho.sPW12040 was evacuated at 573 K. Cs3PW12O40, which has no proton, gives a resonance at - 15.3 ppm, which is close to the first resonance ( - 14.9 ppm). Anhydrous H3PW12040,which has three protons directly attached to the polyanion, has a resonance at - 10.9 ppm. Thus the chemical shift is determined by the number of the protons attached to the polyanion, and the four

148

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

FIG. 20. "P NMR spectrum (bottom) and one of the primary particles (top) of Csz.sHo.sPWlzOao. The dotted lines in the spectrum show the relative peak intensities expected for the statistical distribution of protons. (From Ref. 47.)

peaks correspond to the Keggin anions having 0, 1, 2, and 3 protons, respectively. The relative intensities of these peaks for C S ~ . ~ H ~ . ~are P in W good ~ ~ O ~ ~ agreement with those expected from a random distribution of protons (broken line, Fig. 20). The number of protons on the surface (shown in Fig. 21) was estimated by multiplying the formal concentration of protons on the surface by the quantity of polyanions present on the surface calculated from the surface area. As the Cs content increases, the number of surface protons at first decreases, but greatly increases when the Cs content exceeds 2 . The increase in the number of surface protons is mainly due to the sharp increase in the surface area (47). When x

x in C~,H3.~PW12040

FIG.21. The amount of surface protons (surface acidity) as a function of Cs content for C S , H I - , P W , ~ O ~(From ~ . Refs. 47, 48.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

149

increases from 2.5 to 3.0, since the formal concentrations of protons become low or zero, the number of surface protons decreases greatly. As described in Section V, the catalytic activity is in close correspondence to this surface acidity.

3 . Supported Heteropolyacids Supporting heteropolyacids on solids with high surface areas is a useful method for improving catalytic performance. It is necessary to pay attention to the changes in the acid strength, the structures of the aggregates, and the possibility of decomposition. In general, strong interactions of heteropolyacids with supports are observed at low loadings and the intrinsic properties of heteropolyacids prevail at high loadings. Basic solids such as A1203 and MgO tend to decompose heteropolyacids ( 146-1 49). Microcalorimetry of NH3 absorption (adsorption) reveals that when H3PW12040is supported on S O 2 , the acid strength decreases, as shown in Fig. 22 (150). The acid strength of H3PW12040diminishes in the sequence of supports Si02 > A1203 > activated charcoal. The acid strengths of heteropolyacids supported on Si02 (151) measured by NH3 TPD is in the following order: H3PW12040(865 K) > H4SiW12040(805 K) > H3PMo12040(736 K) > H4SiM~12040 (696 K), where the figures in parentheses are the temperatures of desorption. The interaction between H3PW12040 and the surface OH groups of Si02 has been detected by ' H and "P MAS NMR spectroscopies (152). The OH resonance of Si02 at 1.8 ppm (relative to tetramethylsilane) becomes smaller as the amount of H3PW12040on Si02 increases. At 20-50 wt% H3PW12040,a new type of proton appears at 5 ppm. At loadings >50 wt%, a resonance at 9.3 ppm

170

-

150

T

130

3 110 I

g

90

70

0

0.5

1.o

mmOl Of NH3 g 1

FIG.22. Differential heats ofabsorption of NH, on ( 1 ) H3PW12040and (2) 20 wt% H,PWl,O,$ SiO2. (From Ref. 150.)

150

TOSHIO OKUHARA, NORITAKA MIZUNO, A N D MAKOTO MlSONO

due to crystalline H3PW12040appears. At 20 wt%, 31P NMR also gives a broad resonance at - 15.8 ppm, which is different from the - 12.4 ppm resonance of crystalline H3PW12040.The interaction between H3PWI2O40and OH groups on SiOz is assumed to be as shown in Eq. (10). H3PW12040+ mOH-Si

+

H3-,,,PW12040-Si

+ mH2O

(10)

Lefebvre (153) reported for 1 3 4 7 % H3PW12040 supported on Si02 that two 31P resonances are due to the bulk H3PW12040 (15.1 ppm) and to SOH:H2PWI2O4&,( - 14.5 ppm). 31PSpin-lattice relaxation of H3PMo12040/Si02gives 155). For unsupported an estimated coverage of H ~ P M O on ~ ~ Si02 O ~ (154, ~ H3PMo12040.13H20, TI is 20 s; but TI drastically decreases as a result of dispersion on the support: 2.5 s (loading amount as Mo, 35 wt%) to 430 ms (loading amount as Mo, 2 wt%). Raman spectroscopy shows that the structure of P M o 1 2 0 ~is~preserved at high coverages on Si02 (156). When the loading decreases, the v(Mo=O) frequency decreases, which is probably due to weakening of the anion-anion interaction. On heating to temperatures >573 K, 23 wt% H3PMo12040/Si02 produced an unidentified species andor Moo3 (157). Upon exposure to water vapor, the unknown species was reconverted to the Keggin anion. When Moo3 was supported on Si02 and treated with H20, H4SiMoI2O40formed on the S i 0 2 surface (I58). Soled et al. (159) prepared a unique Cs acidic salt of H3PW1 2 0 4 0 , which is present in a ring 10 pm thick located about half-way into an S i 0 2 extrudate -1.5 mm in diameter.

-

IV. Acid-Catalyzed Reactions in the Liquid Phase

Heteropolyacids are more active catalysts for various reactions in solution than conventional inorganic and organic acids, and they are used as industrial catalysts for several liquid-phase reactions (6, 12). Important characteristics accounting for the high catalytic activities are the acid strength, softness of the heteropolyanion, catalyst concentration, and nature of the solvent (6, 7, 9, 116, 160, 161).

Figure 23 shows a correlation between the catalytic activity and the Hammett acidity function (Ho)of Keggin-type heteropolyacids in CH3CN. The catalytic activity increases with the acid strength (63). Softness of the polyanion is also important in bringing about unique activity. It has been suggested that reaction intermediates like oxonium ions or carbenium ions are stabilized on the surfaces of heteropolyanions (see below), probably by virtue of the softness (162). lzumi et al. (124) observed that H4SiW12040 was more active for the reaction of dibutyl ether with acetic anhydride than H3PW12040,although H4SiW12040 is a weaker acid than H3PW12040. This difference was explained by the stabilization of the cationic intermediate

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

7 -

151

1.0

0

E

4

I

c

zt

B m 3

0.5

0.5

1.5

2.5

-H0

FIG.23. Rate constants (per mole of protons) for decomposition of isobutyl propionate as a function of Hu values of HnXW12040 (acetonitrile solutions). (From Ref. 63.)

through the formation of the intermediate-polyanion complex due to the softness of the polyanion, which is in the order SiW120:0 > PWl2O;O ( 9 ) .The effect of the softness becomes significant for reactions in aqueous solutions, in which the influence of the difference in the acid strength is slight, since most heteropolyacids are completely dissociated. The concentration of the heteropolyacid is sometimes critical. Excellent performance was demonstrated for the hydration of butenes when heteropolyacids were used in a highly concentrated solution (163). As shown in Fig. 24, the rate increases remarkably with an increase in the concentration of

0 Concentrationof H3PM0~~0~drnol.drn-3 FIG. 24. Dependences of the initial rate of TBA formation and ratio of DIP/TBA on H~PMO~IO concentration ~O in the hydration of butenes. TBA, tert-butyl alcohol; DIB, diisobutylene. 357 K, isobutylene/l-butene = l / l (molar). (From Ref. 163.)

152

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XIV Acid-Catalyzed Reactions in the Liquid Phase Reaction

-

c C=C

/

\

+ H2O C

+ HzO

C=C-C-C

Catalyst

Remarks"

Ref.

c I I

C-C-C

-

H3PW12040

T = 313-353 K

162. 166

HiPW12040

T = 473 K, 200 atm

12

OH C-C-C-C

I

OH

+ 2HCHO

Ph-CH=CHz

K2KIC-CH2

\ /

-

P h p O OJ

+ KIOH

H~PW12040 T = 328 K H3PW12040: H2SO4 = 300: I (ratio of activity)

162

H,PWlz04o

T = 318 K H,PW1204o : HlS04 = 30: I

124

H3PW12040

T = 368 K 124 H,PW1204o : BF3 ' E t 2 0 = 50:l

H,PWl2040. nH20 (n = 0 6)

T = 333 K MW = 3000

H,PW

T = 298 K, Y

0

/

\

R,O

(2

+ R2Rl-C-CH2

RzRI-C-CHz

+ AcOH

-

OH

-

I

\

HO

OR,

AcO-(CH~)~-OAC

HOf(CH2)4-OkH (PTMG)

-

Cyclotrimerization of propionaldehyde

+ CHIOH 0

II

-

+o/

HbP~W1~062T = 315 K. S

- cR -A - -

C,H,OCH,

85%

176

z

100%

17'7'

II

HlPW12040

T

363-393 K

161

H3PW12040

T = 373 K H3PW 12040: HlS04 = 60:l

63

COR II

0

+ CH,COOH

A O C O C H ,

Ph -CH,OH

=

0

W O + 2ROH C II

0

12040

I64

f CnHa-CH,t

+ CH,COCl

C&0CH1(COCH,)

H4SiMo12040T = 298 K

I80

H4SiM012040 Reflux

180

continued

I53

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XIV-Continued Reaction

Catalyst

0 II

~~

"

Remarks"

H3PWI2O4,) T = 333 K. Y

=

72% (p,p)

Ref.

"

~

7' = temperature, Y = yield. S = selectivity " Asahi Chem Co , LTD , JP 1990-45439

H3PMo1204n.In a certain range of concentration of H3PW120jo,polymerization of THF proceeds efficiently, whereby the molecular weight of the product is also well controlled (164). In some cases, the effective acid strengths of heteropolyacids level off due to the leveling effect of reactant molecules (63). For example, the difference in activity between heteropolyacids and HzS04 is not significant for ester exchange and esterification because alcohols are good proton acceptors ( 6 3 ) . In organic reactants or solvents, high activities of heteropolyacids are often realized (Y). Typical reactions catalyzed by heteropolyacids are listed in Table XIV. A. 1.

REACTIONSIN AQUEOUS SOLUTION

Hydration and Dehydration

Tokuyama Soda commercialized a catalytic process for propylene hydration catalyzed by aqueous solutions of heteropolyacids such as H3PW I 2 0 4 0 ( 165). In Table XV, the activities of various acids are compared at a constant proton concentration. H3PW12040is two or three times more active than H2S04 or H3P04. The reason for the high activity is assumed to be the stabilization of intermediate propyl cations by coordination. Izumi et af. (162, 166) observed high activities of heteropolyacids for the hydration of isobutylene in dilute solution, as follows. The activation energy is 4 kcal mol-I lower for H3PW12040than for HN03. The reaction order with

154

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XV Effect of Anion on Rate of Propylene Hydration" (165)

Anion PMolzO:i pw120:;

so: Po:

-

Concentration of anion (IO~-'moldm-') 1 .oo I .oo I .50 1.50

Relative specific rate 3.4 I .7 I .o 1.o

"423 K, I4 atm, [H,O+]; 3.0 X lo-' mol dm-'

respect to heteropolyacid is about 1.5, whereas that for HN03 is 1. The proposed reaction mechanism is shown in Scheme 3. Path A corresponds to specific acid catalysis. Path B involves an intermediate n-complex formed from protonated isobutylene and a heteropolyanion. The overall rate is expressed by the sum of the rates of paths A and B [Eq. (1 1a)], where kl and kll are the rate constants, PR is the partial pressure of isobutylene, and [heteropolyanion] is the concentration of heteropolyanion, respectively. r = kl(Ps[H~O+])+ k~l(P~[H~O'])[heteropolyanion]

( 1 la)

The existence of a significant interaction between the heteropolyanion and the carbenium ion is supported by the presence of alkyl-heteropoly complexes. b o t h and Harlow (167) synthesized 0-alkylated compounds such as [(CZH5)30]3PWIZ037, where a CzHs cation is bonded to the polyanion. Farneth et al. (168) and Lee et al. (169) reported the formation of methyl groups and

Path A

SCHEME 3

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

155

ethyl groups directly attached to the heteropolyanion (methoxy and ethoxy), respectively. A commercial process for the separation of isobutylene from a mixture of isobutylene and n-butenes through direct hydration of isobutylene to give tertbutyl alcohol has been established by use of a concentrated solution of heteropolyacids (6, 163, 170). The reaction order in the heteropolyanion varies from 1 to 2 as the concentration of heteropolyanion increases from 0.05 to 10 mol dm-3. This increase corresponds to a change from the first to the second term in Eq. (Ila). At concentration of the heteropolyanion greater than 0.5 mol dm-3, path B in Scheme 3 becomes dominant. In addition, the solubility of isobutylene increases linearly with the concentration of H3PWI2O4,,whereas the solubility is little affected by the concentration of HN03 (169). At a 1.5 mol dm-3 concentration of heteropolyanion, the solubility of isobutylene is about 2.3 times higher than for mineral acids. Hence the high catalytic activities of heteropolyacids (about 10 times higher than that of HN03) are explained by the combination of the three effects of (a) the high solubility, (b) the coordination of isobutylene to the heteropolyanion, and (c) the high acid strength (162, 166). Isobutylene in a mixture of isobutylene and 1 -butene is very selectively hydrated in the concentrated solution; diisobutylene formation is negligible, and sec-butyl alcohol concentrations are < 100 ppm. An apparently different explanation for the catalysis has been advanced by Kozhevnikov et al. (171, f 72). The rate constant, k (min-I) for H3PW12040and that for H4SiW12040at room temperature show linear relationships with the Hammett acidity function (H,) as shown by Eq. ( I Ib). log k

= - 1.04H0 -

3.46

(1 lb)

The data of H2SO4, HCI, H N 0 3 , and HC104 also fit this equation. On this basis, they suggested that the hydration of isobutylene in the presence of heteropolyacids and inorganic acids proceeds via a common mechanism, in which the rate-limiting step is the conversion of the n-complex into a carbenium ion (Scheme 3). The complexation effect as described above is possibly included in the value of Ho according to this explanation. Heteropolyacids are much more active than H2SO4 and HC104 for hydration of phenylacetylene [Eq. (12)] ( I 73). Also in this case, the rate of reaction in the presence of heteropolyacids shows an approximately second-order dependence on the catalyst concentration. This observation suggests that this reaction proceeds by a mechanism similar to that of Scheme 3: PhCECH

+ HzO

+

PhCOCHi

(12)

The higher activities of heteropolyacids are also found for the dehydration of lP-butanediol to give tetrahydrofuran (THF) (I 74). The activity order is H4SiW12040> H3PW1204,> H3PMoI2O4,. Rate data are summarized by

156

TOSHIO OKUHARA, NORITAKA MIZUNO, A N D MAKOTO MISONO

Eq. (13), and the reaction probably proceeds by a mechanism similar to that shown in Scheme 3. I' =

( k , + k2[SiW120:i])[ 1,4-butanediol][Ht]

(13)

2 . Prins Reaction

Heteropolyacids catalyze the Prins reaction of alkenes [Eq. (14)] more efficiently than H2SO4 and p-toluenesulfonic acid (PTS). For example, H3PW12040 is 10-50 times more active than H2SO4 or PTS (162). In this reaction, oxocarbocations may be stabilized through complexation with heteropolyanions: PhCHzCH2

+ 2HCHO

-

,CHz-CH, PhCH, O+H2

\

(14)

/O

B. REACTIONSIN ORGANIC SOLUTION 1.

Ether Cleavage Reactions

H3PW12040shows a higher activity for the alcoholysis of epoxides [Eq. (15)] than H2SO4, PTS, or HC104 (9, 124, 175). While rapid deactivation is observed with H2SO4, which is probably due to the formation of an alkyl sulfate, H3PW12040maintains its high catalytic activity.

Tetrahydrofuran is cleaved with acetic acid to give 1,Cdiacetoxy acetate selectively (124) (Table XIV). The activity order is H3PW12040> H4SiW12040> H3PMo12040%- BF3 * EtzO > PTS. The catalytic activities for the reaction of THF with acetic anhydride are in the order H4SiW12040= H3PW12040> H4GeW1204~> H4SiMo12040 > H3PMo12040 S H2S04 (124). The cleavage of dibutyl ether with acetic acid is also catalyzed by heteropolyacids (124). H4SiW12040 is the most active among the heteropolyacids, probably due to its higher softness. 2 . Polymerization of Tetrahydrofuran

Aoshima et al. (164) found that THF can be directly polymerized to give polyoxotetramethyleneglycol (PTMG) is a highly concentration solution of heteropolyacids [Eq. (16)].

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

157

H@/H3PW12040 (Mol ratio)

FIG.25. Influence of the molar ratio of H20/H,PW12040 on THF polymerization (333 K). ( A ) Solid-liquid phase, (B) two-liquid phase, (C) homogeneous liquid phase. (From Ref. 1 6 4 . )

PTMG

The water content greatly influences the activity and the molecular weight of PTMG, as shown in Fig. 25. At the ratio of H20/PW120:; = 10, the reaction mixture consists of two liquid phases; the upper phase is mainly THF and the lower phase is the complex of H3PW12040 and THF (in the catalyst phase). The THF polymer is formed in the catalyst phase and is transferred to the THF phase. This “phase-transfer polymerization” is illustrated in Fig. 26. IR spectra of the C-0-C stretching region of the catalyst phase are shown in Fig. 27. The peaks a, b, and c were assigned by Aoshima et al. (264) to uncoordinated THF, hydrogen-bonded THF, and protonated THF, respectively. The polymerization activity is associated with the amount of protonated THF.

3 . Condensation Reactions Table XVl is a summary of typical results observed for cyclotrimerization of propionaldehyde to give 2,4,6-triethyl- 1,3,5-trioxane ( I 76). In catalysis by H ~ P M O ~ the ~ Oreaction ~ ~ , mixture separates into two phases during the course of the batch reaction. The products are present in the upper layer and the catalyst in the lower layer, so that the catalyst solution can be used repeatedly without a catalyst isolation step. Selectivities exceeding 97% and turnovers exceeding 300 moles of product per mole of catalyst have been obtained. Polymerization of 1,3-trioxane, a cyclic trimer of formaldehyde, is catalyzed by H3PMoI2O40 (277). The polymerization is very fast at concentrations of mol dm-3. To obtain comparable rates using BF3 H3PMo12040as low as catalyst, a BF3 concentration of mol m-3 is required.

158

TOSHlO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

FIG.26. Reaction model of phase-transfer polymerization. (From Ref.

164.)

Condensation of acetone to give mesitylene is catalyzed by H3PW12040at room temperature ( I 78).

4. MTBE Synthesis Table XVlI is a comparison of the catalytic activities for liquid-phase MTBE synthesis from isobutylene and methanol (I 79). The catalyst structure and composition have a strong effect on the activity. The highest activity per proton was obtained with a Dawson-type heteropolyacid, H6P2W18062,although the acid strength of H6P2WI8O62is lower than that of the Keggin-type H3PW1204,) (Section 111). Water added to the mixture has little effect on the reaction rate at water concentrations less that 2 wt%, but at 5 wt% the rate is less by a factor of 2.5. At the same time the selectivity is less due to the formation of tert-butyl alcohol.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

159

v(WE0) A

1100

1070

1040

1010

980

950

Wave number/cm-’

FIG.27. Infrared spectra of C-0-C stretching vibrations of THF in catalyst phase and their deconvolution results. ( I ) H4SiW12O4dH2OiTHF= 1/7/20 (molar); (2) H3PW1204dH20/THF= 1/7/20 (molar). (From Ref. 164.)

TABLE XVI Liquid-Phase Cyclotrimerization of‘ PropionaldehyJL.“ ( I 76)

Catalyst HiPMo I 2 0 4 ” HiPMo I 2 0 4 / Hd’W12040 H4SiW ,?04” Si02-A1203 AICIj ZnC12 ZnCIg P-TsOH P-TSOH’ H3Po4’

Amount (mmol)

Time (h)

Conversion (%)

Selectivity” (mol%)

0.4 0.4 0.3 0.3

2 2 2 2 2 2 2 12 2 2 4

87. I 84.7 86.3 66.2 2.8 91.5 87.6 69.2 78.5 47.3 58.2

97.2 97.5 97.2 97.3 3. I 88.6 91.5 92.5 4n.9 97.6 97.3

~

7.5 7.3 -

5.3 -

8.7

TON‘ 350 340 490 370 -

19 20 15 13 I5 II

“Reacted at room temperature, I g of catalyst to 10 g of propionaldehyde. Selectrivity to 2,4,6-triethyl-I ,3,5-trioxane. “Turnover number (molar ratio of aldehyde reacted to trioxane to catalyst). For twelfth run. Second run. ’Fifth run. ‘Contained I5 wt% of water.

160

TOSHlO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

Activiry

TABLE XVlI Methyl tert-Bury1 Ether' ( I 79)

01Heteropolyacid in Svnthesis of

2.0 x 10' 2.0 x 10' 1.5 x 10' 0.86 x 10' 1.9 x 10' 0.72 x 103 0.6 x 10'

2.0 3.3 2.3 1.8 3.3 2.0 2.3 5.5 3.1 0.76 9.0

4.4 x lo3 1.6 x 10' 0.55 x 10' 0.45 x 10'

Reaction conditions: 315 K, 12 ml CHIOH, 1 g cat., P

= 100 kPa.

5 . Esterijication and Ester Decomposition The addition of carboxylic acids to olefins proceeds in solution in the presence of 10-4-10-2 mol dm-3 of HjPWl2040 at 2 9 3 4 1 3 K with a selectivity of 100% (7). H2SO4 is a less active catalyst than H3PW12040by a factor of 30-90. Esterification of p-nitrobenzoic acid with ethanol has been carried out by using H2SO4 catalyst in an industrial process (160). In the presence of H3PWl2040, this reaction takes place with a yield >99% (160). At the end of the reaction, the reaction solution separates into two phases. The upper layer contains ethyl-p-nitrobenzoate, toluene, and ethanol, and the lower layer consists of a solution of the heteropolyacid in ethanol. Consequently, the catalyst can be readily separated and reused. The activities of heteropolyacids for the decomposition of isobutyl propionate were found to be 60-1 00 times higher than those H2S04 and p-toluenesulfonic acid (63). The activity increases with increasing acid strength of heteropolyacids. 6. Alkylation and Dealkylation Nomiya et al. (180) demonstrated that alkylation and acylation [Eqs. (1 7) and (18)] proceed in the presence of H4SiM012040: nPhCH2OH ArH

+ CH3COCI

+

(-C~H~-CHZ-)"

(17)

-+

ArCOCH3

(18)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

161

In an acetic acid solution of H3PW 1 2 0 4 0 , dealkylation of 2,6-di-tert-butylphenol to give 2-tert-butylphenol takes place at 357 K. The catalytic activity of H3PW12040is more than 100 times greater than that of H2S04(160). 7. Others In an acetone solution, decomposition of isopropylbenzene hydroperoxide to give phenol and acetone is catalyzed by H3PMo12040(181). The heteropolyacid is 3 times more active than H2SO4. Synthesis of I , 1 -diacetate from aldehyde and acetic anhydride has been attempted by using H4SiW12040and the zeolite HZSM-5 (182). Both catalysts gave more than 98% yield of the diacetate from benzaldehyde. However, the reaction rate was far higher for H4SiW12040than for HZSM-5.

V.

Heterogeneous Acid-Catalyzed Reactions A. GENERAL CHARACTERISTICS

Among heteropolyacids, polytungstic acids are the most widely used catalysts owing to their high acid strengths, thermal stabilities, and low reducibilities. As summarized in Section 111, the acidic properties (acid strength, acid amount, type of acid, etc.) are controlled by (i) the structure and composition of the heteropolyanion, (ii) the formation of salts (or the countercations), (iii) the tertiary structure, and (iv) supporting onto carriers such as Si02 and active carbon. Factors (iit(iv) are specific to the solid state. The acid strength can be controlled mainly by (i), and the number of acid sites is greatly influenced by (ii) and (iii). In this section, these influences will be described. Besides the acidic properties, the absorption properties of solid heteropolyacids for polar molecules are often critical in determining the catalytic function in “pseudoliquid phase” behavior. This is a new concept in heterogeneous catalysis by inorganic materials and is described separately in Section V1. With this behavior, reactions catalyzed by solid heteropoly compounds can be classified into three types: surface type, bulk type I, and bulk type I1 (Sections VII and IX). Softness of the heteropolyanion is important for high catalytic activity, although the concept has not yet been sufficiently clarified. The influence of the heteroatom on the acid strength is shown, for example, in Fig. 28. Here the rates of alkylation of 1,3,5-trimethylbenzene and the decomposition of cyclohexyl acetate catalyzed by the acid forms of several 12-tungstates are plotted against the negative charge of polyanion for solid heteropolyacids (63, 183). The catalytic activities correlate well with the acid strength in solution (Fig. 14). This correlation indicates that the acid strength of

162

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

2

3 4 5 6 7 Nagative charge of polyanion

FIG. 28. Catalytic activities (per surface proton) as a function o f the negative charge of polyanion, XW 120;". (0)Alkylation of I ,3.5-trimethylbenzene with cyclohexene; ( 0 )decomposition of cyclohexyl acetate. (From Ref. 183.)

the acid form in the solid state reflects that in solution and decreases with increasing negative charge of the polyanion (Section 111). A correlation between the acid amount of the surface and the catalytic activity for the Cs salts of H3PW12040is shown in Fig. 29 (128). The number of surface

a 0 R

7

0 0.5

-

J

-

. ms

a"

0

20

40 60 200 240 Surface Acidity/pmol g-1

280

FIG.29. Surface acidity and catalytic activities for alkylation of 1.3.5-trimethylbenzene with cyclohexene (closed symbols) and decomposition of cyclohexyl acetate (open symbols). (From Ref. 128.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

163

acid sites was estimated by multiplying the formal concentration of protons on the surface by the specific surface area [e.g., (0.5) X (number of polyanion on the surface) X (number of protons per polyanion)] (6, 47). The catalytic activities correlate linearly with the number of surface acid sites. These results are reasonable because the acid strengths of acidic Cs salts are similar to that of the acid form (127).The high catalytic activity of C S ~ . ~ H O . ~ P W(Cs2.5) I ~ O ~that ~) was reported previously (46a, 48, 127) is thus inferred to be due to the high surface acidity (high surface area and presence of protons). As described in Section 11, the pore size of the acidic Cs salts can be controlled by the Cs content. C S ~ . ~ H O . ~ P(Cs2.2) W ~ ~ has O ~pores ~ having an effective size of 6.2-7.5 A, and the pore size of Cs2.5 is larger than 8.5 A. Figure 30 shows the catalytic activities of Cs2.1 (the pore size is less than 5.9 A) and (32.2 for each reaction relative to the activity of Cs2.5 (48).Cs2.5 catalyzed all the reactions with considerable activities (the reaction rates are shown in parentheses in Fig. 30). On the other hand, although Cs2.2 was found to be as active as Cs2.5 for dehydration of 2-hexanol (molecular size, 5.0 A) and decomposition of isopropyl acetate (molecular size, 5.0 A), it was much less active for the decomposition of cyclohexyl acetate (molecular size, 6.0 A) and alkylation of 1,3,5-trimethyIbenzene (molecular size, 7.5 A).Therefore, Cs2.2 is

1-

Relative Activity I Reaction

Dehydration :

.. .. .. .. .... . . . . . .. ..... .. . . . .. .. .. .. ....(r:4 . . . . :. .. .. ..... .)id): ... Y

Decomposition :

* ' . .""'L

.. .. .. .. . . . . . . . . . . . . . . .. .. ... .446): . . . . . . :. . :.:.: .:(w ...

COCH,

I

Akylation:

~ O2.2, ~ ~and 2.5) for various kinds of FIG. 30. Relative activities of C S ~ H ~ - ~ P (xW=~2.1, reactions in liquid-solid reaction systems. Catalytic activity was estimated from the initial rate of the reaction. The activity of Cs2.5 for each reaction is taken to be unity. The figures in parentheses are reaction rates in units o f mmol g- h - I. (From Ref. 48.)

'

164

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

active only for smaller molecules; the reactions are influenced by reactant or transition-state shape selectivity. As for Cs2.1, it is active for the dehydration of 2-hexanol but inactive for other reactions, notwithstanding its high surface area (55 m2 g-I). To our knowledge, this is the first example of shape-selective catalysis by a solid superacid (47, 48). A remarkable effect of the countercation is demonstrated in Fig. 3 1, where the rates of several reactions are plotted against the extent of neutralization by Na or Cs, that is, x, in M,H3-.PW12040 (M = Na or Cs) (46, 128). In the case of Na salts, the rates decrease more or less monotonically as the Na content increases. On the other hand, peculiar changes in activity are observed for the Cs salts; activity maxima occur at x = 0 and x = 2.5. The activity of Cs2.5 relative to that of H3PW12040changes depending on the reaction. As will be described in more detail in Section VI, the activity pattern is principally governed by the extent of the contribution of the bulk type I catalysis. As was described in Section 111, the acid strength usually decreases when the catalyst is supported on SiOz. Figure 32 shows the influence of the loading of a

1.o

0.8

0.6

.b >

‘ i ;

0

0.4

m

$ m

c

0.2

m u

O

.z m -

1.0

0,

d

0.8 0.6 0.4

0.2

0 0

1

x in M,Hs-,PW

3

2 12040

FIG. 31. Catalytic activities of acidic Na or Cs salts of H3PWI2O4,)as a function of Na or Cs content. (a) M = Na: (0) dehydration of 2-propanol, (A)decomposition of formic acid, (0) conversion of methanol, (W) conversion of dimethyl ether. (b) M = Cs: (0) dehydration of 2-propano1, (m) conversion of dimethyl ether, (A) alkylation of 1,3,5-trimethylbenzene with cyclohexene. (From Refs. 46 and 128.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

roo^ 2.3 ,6.9, I

165

Mo (wt%) 12.1 ,

2?3

26;2

Coverage, fraction of monolayer

30;7

1 .o

FIG.32. Dependence of selectivities in methanol oxidation on surface coverage and weight YOof Mo for H4SiMo12040. (a) (CH3)20,(b) HCHO, (c) (CH30)2CH2.(d) HC02CH3. (From Ref. IX4.)

H4SiM~12040 on the support for catalytic oxidation of methanol (184). At coverages larger than 0.25 monolayer, the selectivities remain constant; dimethy1 ether is the main product (about SO%), showing the acidic character of the catalyst. Below 0.25 monolayer, the acidic character is lowered, and dimethyl ether rapidly disappears. Typical examples of reactions catalyzed by heteropoly compounds in the solid state are summarized in Table XVIII. B. DEHYDRATION A N D HYDRATION In catalytic dehydration of alcohols, pseudoliquid phase behavior (bulk type I reaction) of heteropolyacids has been demonstrated (Section VI). The high catalytic activity is associated with this behavior and the strong acidity. Unique pressure dependences of the catalytic activity and selectivity are found for H3PW12040 due to the pseudoliquid phase (Fig. 40). H3PWIZ040was found to be much more active for the dehydration of 2-propanol than Si02-Al203 (by a factor of about 30 per weight and about 2000 per surface area) at 398 K (185).The activities of heteropolyacids decrease in the sequence PW12 (H3PWI2040)> SiW12 > PWlOV2 > PMo12 > PMoIOV2 = SiMol2, which is close to the acid-strength series in solution. Niiyama et al. ( 4 2 ) found that the catalytic activities of the water-soluble salts correlate with the electronegativity of the metal cations. Dissociation of coordinated water [type (1) in Section 1111 has been proposed to account for the acidity.

166

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XVllI Heterogeneous Acid-Catalyzed Reactions

>OH

-

=( + H 2 0

Cata Iyst

Reaction

-

A1203 = 30 I (ratio of activity/g) H4SiWI2O40/ Amberlyst I5

+OH

-

Cl& Hydrocarbons

+ CzHsOH

-CHACOOC~H~

O O C O C H ,

-,&+A

Ref.

+ H,O

CH30H(CH30CH3)

CHiCOOH

Remarks"

+ HzO

- 0+

-

+,

CHICOOH

etc.

T = 343 K, S = 98.6% (1 1 % conv.)

I86

H3PW12040 T = 573 K I 89 C~z.sHo.sPW1204o T = 563 K, S = 74% 195 (Cz-C4 olefins) H3PW1~040/Si02T = 423 K S = 91% (90% conv.)

I51

C~2.sHo.sPWi2040T = 373 K csz 5 : so: -/zro2 = 43 : I (activity/g)

I98

H3PW12O4dSi02 T = 308 K, sensitive to pretreatment

207

H3PW12040

209

T = 298 K

I YY

continued

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

167

TABLE XV I I I-Contin ued Reaction

~

C

CHiNH: NH,

O

C

Catalyst

Ref.

d

+I

H4SiWI2O4,JSiO2 T = 263 K S = 94.1% (30% conv.) T = 323 K HcP~WIHO~

-HCN

+ CHIOH

Remarks"

T = 113 K

226

( N H ~ ) I P W I Z O ~ T~ = 150 K

227

CSIsH,~ P W I Z O T~ =~ 413 K s = 97%

h

H1PWi20411

(CH3).NH,,,

21x 220

(94% conv.) ~

"

T = Temperature. Y

=

~~~

yield, S = selectivity. *Sumitorno Chem. Co., LTD., JP 1991-300150.

The Cs salts of H3PW12040show a unique activity pattern for dehydration of 2-propanol (Fig. 32) (46a). Although the change in the catalytic activity resembles that of the surface acidity (Fig. 21), the activity of H3PW12040 for this reaction is relatively high. The high activity is explained by the pseudoliquid phase behavior of H3PW12040and its affinity for the polar molecule, 2-propanol. H ~ P W , Z OH4SiW12040, ~~, and the supported heteropolyacids catalyze the hydration of isobutylene (186). The supports are listed as follows in the order of activity: Amberlyst 15 [porous sulfonated poly(styrene-divinylbenzene)] = activated carbon > SiOz > TiOz. The maximum activity of H4SiWI2O4dArnberlyst 15 was obtained at about 30 wt% H4SiW12040. It was found that polypyrrole and polyacetylene-supported H3PMo12040are much more active than the unsupported parent acid (187). The activity per unit mass of heteropolyacid of the supported catalyst is about 10 times higher for the production of ethylene and ether than that of the unsupported catalyst. The dehydration of ethanol catalyzed by a membrane comprising H3PW12040 and polysulfone was reported (188). The polysulfone is more permeable for

168

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

ethylene than for diethyl ether. Thus the selectivity for ethylene in the membrane reactor was higher than that in a fixed-bed reactor containing the same catalyst. c . CONVERSION

OF

METHANOL INTO HYDROCARBONS

Ono et al. (189) reported that heteropolyacids such as H3PW12040 and catalyze the conversion of methanol into hydrocarbons, although H4SiW12040 the activities are less than that of HZSM-5. In contrast to HZSM-5, the main products observed with heteropolyacids are aliphatic C hydrocarbons, the selectivities for aromatic hydrocarbons being small (Table XIX). Countercations influence the rate and selectivity of this reaction. The activity order, as for cations, was found to be Ag > Cu, H > Fe > A1 > Pd > La > Zn (1YO). The distributions of product hydrocarbons were found to be similar to those observed for H ~ P W , Z (Table O ~ ~ XIX), suggesting similar reaction mechanisms. Ag and Cu salts of H,PW12040 are much more active than the acid form catalyst. Protons generated by the reaction of Ag' with H2 are presumed to give the more active catalyst (191). Hayashi and Moffat (192) reported that the A1 salt and the NH4 salt were effective catalysts for the conversion of methanol to hydrocarbons. They claimed that the N& salts show high catalytic activity and selectivity for the formation of saturated hydrocarbons rather than olefins. The salts of organic TABLE XIX Product Distribution in Conversion of Methanol into Hydrocarbons (189) Catalyst HTP"

CuTP

AgTP

HTS"

CUTS

AgTS

Product distribution (%)* MeOH I .3 MeOMe 38.6 Hydrocarbons 60.1

I .2 37.3 61.5

0 2.0 98.0

3.9 57.5 38.6

4.7 35.5 50.8

0 20. I 79.9

Hydrocarbons distribution (%)h CH4 3.7 C2H4 11.3 C2H6 0.9 C3H6 8.3 C3Hs 16.1 c4 39.3 cs 12.5 c 6 7.9

5.2 9.5 0.8 8.5 13.4 39.2 13.7 9.1

9.0 9.0 5.2 3.8 34.0 26.1 7.2 5.7

I .6 10.3 0.5 8.3 9.8 41.8 15.8

7.3

3.2 10.3 1.2 8.3 2 I .9 36.4 13.3 5.4

11.9

11.2

0.5 8.7 14.5 35.1 15.1 7.6

TP and TS indicate dodecatungstophosphateand dodecatungstosilicate, respectively. 'Calculated on a carbon-number basis.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

169

bases are effective for the formation of olefins (193, 194). In the case of an acidic Cs salt, Cs2.sHo.sPW12040, the selectivity for C 2 4 4 olefins increased continuously as the Cs content increased, e.g., being 43% for H3PW1204o and 64% for S2.5Ho.5PW12040(195). Pseudoliquid phase behavior is important in this reaction. The kinetics observed with H3PW12040is quite different from that observed with HZSM-5. No induction period in the rate of hydrocarbon formation was observed in the former, in contrast to the latter, suggesting that the methanol conversion catalyzed by H3PW12040 is not autocatalytic (196). The results were explained by pseudoliquid phase behavior of H3PWI2040 (Section VI). The selectivity to lower olefins is improved by control of the pseudoliquid phase behavior. The olefidparaffin ratio in the product hydrocarbon depends inversely on the ability of the heteropoly compounds to absorb reactant dimethyl ether. The ratio is greatly increased as the contribution to catalysis of the bulk phase of the catalyst (pseudoliquid phase) decreases (Section VI) (195, 197). It was confirmed that the acid strength is not significant in influencing the selectivity of this reaction.

D. ESTERIFICATION A N D ESTER DECOMPOSITION Okuhara et af. (198, 199) found that C S ~ . ~ H ~ . S P(Cs2.5) W I ~ Ois~much ~ more active for the decomposition of cyclohexyl acetate in liquid-solid mixture than Nafion-H (sulfonated fluorocarbon resin), HY zeolite, HZSM-5, SiO2-AI2O3, and SO:-/Zr02. Figure 33 demonstrates the superiority of Cs2.5 for this reaction as well as the alkylation of 1,3,5-trirnethylbenzene.The activity of Cs2.5 is more than 200 times as high as that of Si02-A1203. The difference is much greater than that observed for gas-solid mixtures (46a, 15)s).Cs2.5 works as an insoluble catalyst in esterification of acetic acid with ethanol (200). The Activity I mmol I I g-l h-l

Catalyst

2TL7T-i

Cs2.5H0.5PW12040 H3PW12040 S042-/Zr02 Nafion-H Si02-Al203 HZSMQ H2S04

FIG.33. Catalytic activities of various acid catalysts for liquid-phase reactions. TMB: Alkylation of 1,3,5-trimethylbenzene with cyclohexene; CA: Decomposition of cyclohexyl acetate. (From Ref. 47.)

170

TOSHIO OKUHARA, NORITAKA MIZUNO. AND MAKOTO MISONO

order of catalytic activities for the esterification at 333 K (acetic acid/ethanol/ catalyst = 100/100/1 by weight) is as follows: Amberlyst 15 (1.6 X > Cs2.5 (0.52 X > HZSM-5 (0.03 X where the -I figures in parentheses are rate constants in units of min- g . In the hydrolysis of ethyl acetate (ethyl acetate/water/catalyst = 1080/28/0.1 by weight), the activities are in the order Amberlyst 15 > Cs2.5 > HZSM-5. Izumi et al. (201) found that activated carbon firmly entraps H3PWI2O40and the acid is resistant to extraction with hot water or hot methanol. The entrapped H3PW12040 showed a catalytic activity for the esterification of acetic acid with ethanol comparable to that of Nafion-H. With H3PW12040supported on carbon, a selectivity to ethyl acetate of 99.5% at 95% conversion was obtained. Heteropolyacids supported on Si02 or carbon are more active than Si02-A1203 (202), but A1203 is not suitable as a support, because A1203 readily decomposes heteropolyacids due to the high basicity of the surface. Cs2.5 in water is hardly separable by filtration because of its very fine particle size. Cs2.5/Si02 prepared by the hydrolysis of ethyl orthosilicate in the presence of colloidal Cs2.5 is efficient for the hydrolysis of ethyl acetate, although it is less active than Amberlyst 15 (200). The catalytic activity of H3PW 12040 for esterification of phthalic anhydride with 2-ethylhexanol is higher than those of conventional soluble acids such as p-toluenesulfonic acid (202).

E. ALKYLATION A N D DEALKYLATION Conventional soluble catalysts such as H2SO4 and AIC13 have been used in industry for alkylation reactions, but these are characterized by operational problems of corrosion, catalyst removal, waste formation, etc. Insoluble solid catalysts are desirable for these reactions. H3PW, 2 0 4 0 catalyzes the monoalkylation of p-xylene with isobutylene (203) [Eq. (19)]. The product tert-butyl-p-xylene (BPX) is an important precursor for liquid crystalline polyesters and pol yamides having low melting points and good solubilities. The introduction of the tert-butyl group in the position ortho to the methyl group of p-xylene takes place very slowly, with H2SO4 or AlC13 as the catalysts (whereas alkylation at the meta or para position of 0- or m-xylene occurs rapidly):

(BPX) The results of Table XX demonstrate that H3PW12040is an excellent catalyst for this reaction. The selectivities are in the order H3PW12040(75%) >

171

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XX Selectivity for Alkylation of p-Xylene with lsoburylene by Various Acid Cutulysts (203) Selectivity" (%) Conversion

Mass balance (%)

Catalyst

BPX~

Oligomers'

OPX'

PXM'

(TO)

Hd'W12040 H4SiW 1 2 0 4 0 HsBW 1 2 0 4 0 CSZ5HosPW12040

74.6 26.6 6.1 50.5 50.9 4.2 1.3

11.5 70.2 93.9 48.0 26.0 95.8 98.7 92.8 I00

6.3 0.4 0 0.8 3.4 0 0 0 0

7.6 2.8 0 0.7 19.7 0 0 0 0

90 91 97 89 86 92 75 86 10

so: /zro2 ~

Amberlyst-I 5 SiO2-AI2O3

1.2

CFjCOOH

0

96 I08 I I9 I10 86 109 101

68 81

" Mol%. 'f-Butyl-p-xylene. 'Sum of dimer, trimer, and tetramer o f isobutylene. "Octyl-p-xylene. 'Di-p-xylylmethane. The reaction was performed at 303 K for 30 min. p-Xylene, 0.28 mol; isobutylene, 0.37 mmol m i - - ' ;catalyst, 0.45 g.

H4SiWI2O40(26.6%) > HSBW1204,, (6.1%) (203). The results indicate that the acid strength is essential for the selectivity; that is, strong-acid sites are effective for the alkylation to BPX, but weak-acid sites preferentially catalyze the oligomerization of isobutylene. Although SO:-/Zr02 is a stronger acid than H ~ P W I ~ itO is~ less ~ , selective. SO:-/Zr02 also has a large number of weakacid sites which are probably active for the oligomerization. Alkylation of p-cresol by isobutylene is an important reaction in the synthesis of phenolic antioxidants. The activity of H3PW12040 for this reaction is greater by four orders of magnitude than that of H2SO4 (160). Alkylation of p-(tevtbuty1)phenol (TBP) with cyclohexene, 1-hexene, styrene, or benzyl chloride proceeds in the presence of H3PW12040 at 3373423 K (204). Alkylation with styrene gives 2,6-dialkyl TPB with a selectivity of 90% at 100% conversion. When the alkylation of TBP is completed, an excess of o-xylene is introduced into the reaction system, and 2,6-dialkylphenol is obtained through the trans alkylation without the need for separation of 2,6-dialkyl-4-tert-butylphenol(160) [Eq. (20)].

$+--

6

+

%But

(20)

R R Dealkylation of 2,6-di-tert-butylphenol takes place at 4 0 3 4 2 3 K in the presence of solid H3PWI2O4,,. H3PW12040is two orders of magnitude more active than aluminum sulfate for this reaction (205).

172

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

60

5 40 .Q

Y?

I

5 20 0 0

30

60

90

120

Timdmln FIG. 34. Time course of alkylation of 1,3,5-trimethylbenzene with cyclohexene at 343 K.

(0) C S ~ ~ H " ~ P W , ~ O ~ (0) ~ ( Csecond S ~ . ~ run ) , for Cs2.5, ( 0 )third run for Cs2.5. (A) SOi-/ZrOz, (V)Nafion-H, (M) HY zeolite. (From Ref. I Y Y . )

Figure 34 shows the time course of alkylation of 1,3,5-trimethylbenzene with cyclohexene (199). In the presence of Cs2.5, the reaction rate reached a steady value after approximately 1.5 h. When Cs2.5 was filtered and reused (in the second and third runs), the rates were nearly equal to the steady-state value of the first run, indicating that there was little catalyst deactivation and that the reaction did not take place in solution. The primary reason for the high activity of Cs2.5 is both its strength of acid sites and acid strength (Section 111). The specific activities of heteropoly catalysts (rates per acid group) are much greater than the activities of Si02-A1203, SO:-/Zr02, or zeolites. This result cannot be explained simply on the basis of the acidic properties. There are additional effects such acid-base bihnctional acceleration by the cooperation of proton (acid) and polyanion (base) groups (127). A proposed reaction model is illustrated in Scheme 4. Supported heteropolyacids are also used for alkylation reactions. Alkylation of benzene with I-dodecene was examined with H4SiWI2O40/SiO2as the catalyst (206). Si02 is a better support for the catalyst in this reaction than AI2O3 or Si02-Al203 ; H3PWI2O40/SiO2 catalyzes the reaction at room temperature (207). The pretreatment temperature of the catalyst has a significant effect on the activity. As shown in Fig. 35, the maximum conversion of 1-octene was obtained when the catalyst was treated at 423 K. Pretreatment at a lower temperature such as 373 K gives a reduced acid strength, probably because of the remaining water of crystallization. H3PW1204dMCM-41 exhibits a higher activity than H2SO4 in liquid phase alkylation of TBP with isobutylene or

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

-

173

Me Me-

Me

Me

SCHEME 4

styrene (208). Supported heteropolyacids are also active catalysts for alkylation of benzene with ethylene at 473 K in the presence of vapor-phase reactants (151). Recently, Soled et al. (159) reported that C2,5Ho.sPW1204~ supported on Si02 extrudate is effective for this reaction. The alkylation of isobutane with n-butenes to give C8 alkylate [Eq. (21)] is a widely used and increasingly important process in petroleum refining. The

40

20 0

274

374

474

574

674

Calcination ternperature/K FIG. 35. Alkylation ofbenzene with I-octene with H3PW12040(303 K). ( 0 )Supported on silica A (15 wt%); (0)supported on silica B (15 wt%), (A) unsupported. (From Ref. 207.)

I74

TOSHIO OKUHARA, NORITAKA MIZUNO. AND MAKOTO MISONO

commercial catalysts are HF or H2SO4; HF has the disadvantages of being highly toxic and corrosive, and H2SO4 processes produce large amounts of spent acid. Both liquid acids constitute environmental hazards because of potential spills. -

9

(21)

Therefore, new solid catalysts to replace these liquid acids are desirable. W ~ ~ O ~this O alkylation Okuhara et al. (209) reported that C S ~ . ~ H O . ~ P catalyzes reaction at room temperature and that it is much more active than SO:-/Zr02. Yield and selectivity data are summarized in Table XXI. The catalytic activities are in the following order: Cs2.5 > H3PW12040> SO:-/Zr02. The selectivity to C8 alkylates in the total products was 73.3% for (32.5. A patent from ldemitsu Co. (210) also describes the high activity of (32.5 for this reaction. Another patent from Mobil gives data for H3PW1204dMCM-41 (211). The alkylation of toluene with methanol is catalyzed by the NH4 salt of H3PW12040 (212). TABLE XXI Yields and Selectivities of Producis in Alylation of isobuiane with I-Buiene Catalyzed by Solid Acids ar Room Temperature (209)

Total yield (w%)“ Selectivity (wt%)* 224-TMP [RON: 223-TMP [RON: 234-TMP [RON: 233-TMP [RON: 23-DMH (C8 alkylates) C547‘‘ Dimersd C9-C 12‘

1001

1101

1031 1061

19.4

25.1

23.0

0.3 24.1 23.6 14.5 10.8 (73.3) 1.5 13.9 11.3

0.6 18.4 15.2 13.9 8. I (56.2) 0.8 8.5 34.0

1.6 28.0 13.9 10.9 7.2 (61.6) 0.9 9.2 26.6

“Yield (wt%) is defined by 100 X [the weight of products divided by the weight of I-butene charged]. ”224-TMP = 2.2,4-trimethylpentane, 23-DMH = 2,3-dimethylhexane, etc. Figures in parentheses are research octane number (RON). ‘Hydrocarbons containing 5-7 carbon atoms. dOctenes. Hydrocarbons constaining 9-12 carbon atoms. Catalyst, 1 .O g; I-butene, 0.94 g; isobutane, 9.4 g. All data were collected at 7 h.

175

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

F. ACYLATION Industrial acylation of alkylaromatics is generally carried out with acid chlorides as reactants or stoichiometric amounts of AlC13 (catalyst). Solid acid catalysts would be desirable. Izumi et al. (207) found that Si02-supported H3PW12040 and H4SiW12040 are good insoluble catalysts for acylation, with anhydrides or chlorides as acylating agents [Eq. (22)].

It was confirmed that the reaction did not proceed in the liquid phase. A rapid decline in catalytic activity was observed. The deactivation is probably caused by strong adsorption of the product benzophenone on the catalyst; the acylation was retarded when the reaction was started in the presence of benzophenone. The reaction also proceeded with H3PMoI2O40/SiO2as the catalyst, but the H3PM012040 decomposed during the reaction. It was presumed that the catalytically active species was not the heteropolyacid on the support, but was probably Mo chloride instead (213). C~2.5Ho.sPW12040 (Cs2.5) catalyzes the acylation of aromatic compounds with benzyl chloride, benzoyl chloride, benzoic anhydride, benzoic acid, or acetic acid (214). As shown in Table XXIl, H3PW12040 is usually less active than TABLE XXll Friedel-Crajis Acylation Catalyzed by C SsH,, ~ .TPW,,O;" (214)

Substrates

Product yeld' (%)

Acylating agent

Aromatic compound

CszsHn.sPWizO40

HJ'Wi2040

(PhC0)20 (PhC0)20 (PhC0)20 PhCOzH PhCOzH Ac~O AcOH n-C7HISCOCI

p-xylene anisole chlorobenzene p-xylene anisole anisole anisole mesitylene

57

3 69'

85 0

I I" 3 89

0 8" 4'

sd

16

I 5'

80

44'

Yield is based on acylating agent. Acylating agenuaromatic compoundcatalyst = 5/100/0.1 mmol; reflux 2 h. ' Catalyst was dissolved. "Acylating agenuaromatic compound catalyst = 5/100/0.05 mmol. The water liberated was continuously removed by means of DeanStark equipment. 'Acylating agentlaromatic compoundkatalyst = 5/100/0.10 mmol. 'Catalyst was partly dissolved.

176

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

Cs2.5 for the acylation. Anisole and p-xylene are acylated with benzoic anhydride and acetic anhydride in the presence of Cs2.5 without the dissolution of this catalyst. Carboxylic acids are much less reactive as acylating agents than the corresponding anhydrides because of the liberation of water. But when the water is removed, the acylation proceeds smoothly (214). Although the reaction of benzene with acetic acid is attractive in prospect, there is no report of heteropoly compounds as catalysts for this reaction. G . SKELETAL ISOMERIZATION

OF

ALKANES

The skeletal isomerization of straight-chain paraffins is important for the enhancement of the octane numbers of light petroleum fractions. The isomerization of n-butane to isobutane has attracted much attention because isobutane is a feedstock for alkylation with olefins and MTBE synthesis. It is widely believed that the low-temperature transformation of n-alkanes can be catalyzed only by superacidic sites, and this reaction has often been used to test for the presence of these sites. Nowinska et al. (215) reported the isomerization of n-hexane catalyzed by ~ by H3PW12040/Si02.H3PWI2O4dSiO2has an appreciable (NH&PW 1 2 0 4 and activity at 423 K, but the activity of H3PW12040 is low, and Si02-A1203 is inactive at this temperature. Cs2.5 catalyzes the isomerization of n-butane at 573 K, and the rates of isobutane formation and the selectivity were much higher than those of SOi-/ZrO2, as shown in Table X X I I I (216, 217). The initial activity of SO;-/ZrO2 is very high, but the conversion decreases considerably during the initial stage of reaction. In contrast, the deactivation is relatively small for Cs2.5. Figure 36 shows the effects of reaction temperature for catalysis by Cs2.5. Deactivation was observed at 473-573 K, but not at 423 K. At temperatures less than 473 K, S0:-/Zr02 is more active than (32.5. The TABLE XXlII Activiw and Selectivi@,for Skeletal lsomerization of n-Butane" (21 7)

Selectivity' (mol%) CataI y st C~z.sHo.sPWi2040 H3PWizOw

so: - lZr02

H-ZSM-5 H-Y"

C,

loMX Rateh

CI

2.0 0.4 0.4 2.9 0.03

1.1 1.1

3.1 0.7 15.8

C2 + C; 2.0 2.4 9.1 2.3 33.3

Cs

+ C;

8.5 11.7 23.0 74.5 18.1

Isobutane 83.1 80.9 60.8 14.1 11.1

C; 0.8 0 0 0.4 18.1

C,(+) 4.5 3.9 4.0 8.0 2.9

"573 K, butane 5%. 'Formation of isobutane, mol g - l s - ' . ' C , , CH4; C2 + C;, C2H4 + CzH6; + CT, C3H6+ C,Hg; C4=,CdHg; Cs( +), hydrocarbons containing more that 5 carbons. d673 K.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

0 0

--- ---A

1

A

#A

A

2

3

A

A

4

A

177

A

5

Tim& FIG. 36. Time course of n-butane isomerization catalyzed by Csz sHo 5PW12040 at various 523 K, (W) 473 K, (0) 423 K, (A) 373 K. (From Ref. 217.) temperatures: ( 0 )573 K, (0)

skeletal isomerization of alkanes catalyzed by metal-promoted heteropoly compounds is described in Section XI.

H. MTBE SYNTHESIS Methyl tert-butyl ether (MTBE) is a good, widely used octane improver of gasoline. lgarashi et al. (218) reported that H3PW12040, H3PMoI2O40, etc. and their SiO2-supported analogs have catalytic activities superior to those of mixed oxides, fluorinated oxides, and mounted minteral acids for the MTBE synthesis from isobutylene and methanol in gas-solid reactors. Among the heteropoly catalysts, 20 wt% H4SiMoI2040/Si02was most effective; the selectivity was 95% at 30% conversion of isobutylene at 363 K. Ono and Baba (219) used Ag3PW12040 supported on carbon as a catalyst for MTBE synthesis. The treatment of the Ag salt with H2 greatly enhanced the activity, whereas no effect was observed for the acid form and Al salt. Effects of acid strength and structure of heteropolyanion on catalytic activity have been examined (220). The activity order is H&W18062 %- H3PW12040> H4SiW12040= H4GeW12040> H5BW12040> H6CoWI2O40, whereby the selectivity for MTBE exceeded 95%. The results are much better than those observed with SOi-/Zr02, Si02-A1203, and HZSM-5 (220). It is probable that the pseudoliquid phase behavior of H6P2W18062 is responsible for its high performance. By supporting H6P2W18062or H3PW12040on S O * , the yield of MTBE increased greatly and became comparable to that of the resin Amberlyst 15, which is representative of today's industrial catalysts for MTBE synthesis.

178

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

The reaction of tert-butyl alcohol and methanol to form MTBE is also catalyzed by heteropoly compounds (221-223). A relationship was found between the amount of pyridine sorbed in or on heteropoly compounds and tertbutyl alcohol conversion (221). The dependence of the rate on methanol partial pressure resembles that for the absorption of methanol in the bulk, suggesting pseudoliquid phase behavior (223).

I. OTHERREACTIONS A wide variety of acid-catalyzed reactions besides those described above have been investigated with heteropoly compounds as catalysts. A1203-supported H3PW1 2 0 4 0 (probably decomposed) catalyzed propylene-ethylene codimerization at 573 K to form pentenes with a selectivity of 56% (butenes 17%, hexenes 27%) (224). Propylene oligomerization proceeded on various kinds of salts of H3PWI2O40 (225). The activities of the salts decrease in the order A1 % Co > Ni, NH4 > H, Cu > Fe, Ce > K. The A1 salt gave trimers with 90% conversion at 503 K. The selectivities to trimer are about 40% for Al, Ce, Co, and Cu, while that of the acid form is 25%. Dehydrogenation of monomethylamine to give hydrogen cyanide is catalyzed by H3PW12040 at 773 K (226). A1203and Si02-A1203 have no activity for the reaction under the same conditions. (NH4)3PW12040 is active for synthesis of methylamines from ammonia and methanol (227). The formation of trimethylamine is suppressed with this catalyst, which is explained by the strong adsorption of trimethylamine. Cracking of paraffins (228, 229), isomerization of olefins (230, 231), transformation of alkylbenzene (232), etc. have also been reported.

VI.

Pseudollquld Phase

In ordinary heterogeneous catalysis of gas-solid and liquid-solid reactions, the reactions take place on the two-dimensional surfaces of solid catalysts (both on the outer surface and on the surfaces of pore walls). In contrast, the reactions of polar molecules in the presence of heteropoly catalysts often proceed not only on the surface but also in the bulk phase. We call this “pseudoliquid phase” behavior. The pseudoliquid phase is a unique reaction medium consisting of the three-dimensional solid bulk, as was first proposed in 1979 ( 1 7, 233, 234). Because of the flexible and hydrophilic nature of the secondary structures of the acid forms and group A salts (Section 11), polar molecules like alcohols and amines are readily absorbed into the solid bulk by substituting for water molecules and/or by expanding the distance between polyanions. The number of absorbed molecules is 10-102 times greater than the amount of monolayer

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

179

TABLE XXIV Types of Acid Catalpis by Salts of Heteropolyacids

Salts"

Polar molecules (e.g., dehydration of ethanol)

Nonpolar molecules (e.g., isomerization of butenes)

A salts (e.g., Na) B salts (e.g., Cs)

Pseudoliquid (bulk type I) Suface type

Surface type Surface type

Besides the above types, there is bulk type I1 (Section IX).

adsorption estimated from N2 adsorption. Heteropoly compounds absorbing significant amounts of polar compounds behave in a sense like concentrated solutions. As classified in Table XXIV, the pseudoliquid behavior is found for acid catalysis of reactions of polar molecules by group A salts at relatively low temperatures. Another bulk-type catalysis (type 11) is described in Section IX. A.

ABSORPTION OF POLARMOLECULES

When pyridine vapor at 298 K is brought in contact with solid H3PW12040or H3PMo12040after dehydration at 403 K, several molecules of pyridine per polyanion (in the entire bulk) are sorbed (5, 125). After evacuation at 298 K, the number of pyridine molecules per polyanion becomes about 6 (twice the number of protons). Single-crystal X-ray and IR analyses revealed that H3PW12040.6C5H5N consists of [(C5H5NH)3* PW120401(39). Considering the surface area of H3PW12040( 5 m'g-I) and the cross section of a pyridine molecule (3 1 A '), the uptake of 6 pyridine molecules per polyanion corresponds to about 80 times that corresponding to a surface monolayer (235). Some heteropoly compounds swell by absorbing a great number of polar molecules. Deliquescence is observed upon excess absorption. Large uptakes of alcohols into H3PW12040have also been reported (125, 235-238). In Table XXV, the rates and amounts of absorption of various molecules are summarized (235).Polar molecules such as alcohol, ether, and amine are readily absorbed. In contrast, nonpolar molecules like hydrocarbons are sorbed only on the surface. The initial rates of absorption of molecules are plotted against the molecular size in Fig. 37. The initial rates of alcohol sorption greatly decrease as the molecular size increases from 20 A2 (methanol) to 35 A2 (1-butanol). The rates are higher for amines than for alcohols, regardless of the molecular size. This difference is due to the greater basicity of amines. Thus, it may be stated that the rate is primarily determined by the basicity (or polarity) and secondarily by the molecular size (235). Diffusion coefficients of molecules in the lattice of H3PW12040are ca. lo3 times less than those of molecules in the micropores of zeolites (235).Niiyama et al. reported that the effective diffusion coefficient is in the order of

180

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XXV Rate and Amount ofdbsorption of Molecule into H j P W 1 2 0 4 ~(235) Absorption amounts‘

Molecule

(PK,)

Size”

p”

P‘

Rate*

d

1’

( - 2.0) ( - 3.0) ( - 3.0) ( - 3.2) ( - 3.8) ( - 3.8)

20 25 30 31 35 30 37

1.71 1.73 1.73 1.67

30 (20) 30 (43) 20 (80) 30 (57 4 (50) 760 (15) 350 (59) 30 (52) 30 (76) 30 (9) 30 (5) 30 (28) 15 (62) 65 (43) 60 (0.13) 0.8 (0.1)

5.2 5.5 3.8 3.1 2.0

6.0 14.1 14.0 12.1 5.4 4.0 6.3 4.9 8.3 12.0 9.5 15.2 8.5 0.50 0.04 2.8

3.1 (26) 6.1 (64) 6.2 (78) 6.1 (76) 2.8 (41) 2.6 (33) 5.8 (89) 4.6 (86) 4.8 (66) 7.0 (89) 6.9 (86) 8.6 (125) 6.0 (78) 0.10 (1.4) 0.03 (0.2)

( - 3.6)

( - 4.3) ( - 2.9)

(10.7) (10.6) (10.8)

(5.2)

45

1.81

1.30 1.17 I .20 I .40

33 32 33 35 31 34 16

1.32 2.32 0 0

13

-

1.34 -

1.6

6.2 1.1 2.2 3.1 4. I 6.0 4.0 -

-

-

“Cross section (A’). hDipole moment, Debye. “Introduced at 301 K, Torr. The figures in parentheses are the relative pressure (%). *The initial absorption rate; number of molecules (anion. 10 min)- I. ‘Number of molecule. anion- I. ’Saturated amount. ‘Irreversible amount. The figures in parentheses are the numbers of surface layers (see text). ‘Feed gas: 0.1% NO + 5% 0 2 + 5% H20 + 89.9% He at 423 K (Yang, R. T., and Chen, N., Ind. Eng. Chem. Res. 33, 825 ( 1994)).

10-”-10-13 m s-I, much less than those in the gas phase but close to those in the liquid phase (239). The amounts of absorbed molecules in H3PW12040tend to be integral multiples of the number of protons (3, 6,9, etc.), suggesting that these molecules form stable secondary structures throughout the bulk. Transitions between the different absorption states take place as a result of pressure changes (235).These transitions are closely related to the catalytic behavior for the dehydration of alcohol, as described below.

B. EVIDENCE FOR

THE

PSEUDOLIQUID PHASE

There is circumstantial evidence in our early studies indicating the pseudoliquid behavior: (1) rapid absorption of a large quantity of polar compounds as described above; (2) expansion of the solid volume of the materials upon absorption; and (3) quite high catalytic activity of H3PWI2O4,,, H3PMo12040, etc., despite their low surface areas (234). Firm evidence that catalytic reactions

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

.,

10

20

30

40

181

50

Molecular Cross Section /A2 Initial rates of absorption of various compounds by H3PW12040rt ted to cross section FIG. of molecules. ( I ) Ethylene, (2) dichloroethane, (3) benzene, (4) toluene, ( 5 ) methanol, (6) ethanol, (7) I-propanol, (8) 2-propano1, (9) 1.4-dioxan. (10) I-butanol, ( I I ) I-propanamine, (12) 2-propanamine, (13) I-butanamine, (14) pyridine. (From Ref. 235.)

take place in the pseudoliquid phase has been obtained by a transient response analysis using isotopically labeled reactants (236, 240). Figure 38 illustrates a typical result observed for the dehydration of 2-propanol catalyzed by H3PW12040,The feed gas was instantaneously changed from 2-propanol-do to 2-propanol-dx after a steady state of the reaction had been attained. 2-Propanoldo at the outlet was slowly replaced by 2-propanol-dx (solid lines, and open and solid circles in Fig. 38), whereas the change was rapid in the absence of the do-4 100

4

80

. g

E

60 ..4-

v)

5 0

40

20 0

0

5

10 15 Time I min

20

30

FIG.38. Transient response in the gas-phase composition resulting from a change of feed from 2-propanol-do to -4in the dehydration catalyzed by H3PW1~0m at 353 K (2-propanol, 3.4%; flow rate, 100 cm3 min-'; 50 mg of H3PW12040).(From Ref. 236.)

182

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

catalyst (broken lines). The amount of 2-propanol-do held by the catalyst under the reaction conditions corresponds to the shaded area, and the amount of 2-propanol-d8 newly absorbed equals the dotted area. The agreement between these two areas led to the estimate of about 7 molecules per polyanion. This value corresponds to about 100 surface layers, which shows that 2-propanol was mostly in the bulk during the reaction. Furthermore, the rates of absorption and desorption were estimated from these data to be 50 times the reaction rate. It is noteworthy that the concentration of 2-propanol in the bulk (6 X mol ~ m - under ~ ) the reaction conditions is comparable to that of liquid mol cm-’). These results justify the term “pseudoliquid phase.” It was confirmed by the same methods that dehydration of ethanol also proceeds in the pseudoliquid phase of H3PW12040(240). Saito and Niiyama (241) investigated the transient behavior of ethanol dehydration catalyzed by Ba1.5PW12040. When the ethanol feed was stopped after a steady state had been attained, ethylene continued to form for a prolonged period, whereas ether, formation decreased rapidly. This transient behavior, as well as the kinetics under stationary conditions, was well simulated with a model based on the assumption that the ethylene and ether are formed by unimolecular and bimolecular reactions in the bulk, respectively.

c.

UNUSUAL

KINETICSIN

PSEUDOLIQUID PHASE

There is evidence that at least two different pseudoliquid phases may be present, even during catalytic reactions, and these may change reversibly with changes in the reactant partial pressures, as shown for dehydration of 2-propanol in Fig. 39 (242). A small change in reactant partial pressure led to an abrupt

B 6 -

4 2 0-

0.2

0.5

1

2

5

10

102 x Pressure of a - ~ r ~ ~ a t r n

FIG.39. Pressure dependence of the catalytic reaction rate and the amount of absorbed propanol in the dehydration of 2-propanol catalyzed by H3PW12040at 353 K. (From Ref. 242.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

183

change from a high- to a low-activity state. The transition partial pressure tends to be higher as the reaction temperature increases. The amount of 2-propanol absorbed, which was determined by the transient response method, changed along with the reaction rate. At 353 K, the numbers of molecules absorbed per anion were 3 and 8 for the high- and low-activity states, respectively. At 373 K, these two states vaned reversibly and rapidly upon the change in the pressure of 2-propanol; but at a lower temperature, the transition was slower upon the change of the partial pressure. The unusual pressure dependences of the rate and selectivity associated with the pseudoliquid that were observed for ethanol dehydration catalyzed by H3PW12040 are shown in Fig. 40 (169, 243). The rates of ether and ethylene

-1

1

0

2

WPlkPa)

FIG.40. Rates of formation of diethyl ether and ethylene from ethanol catalyzed by H3PW12040 as well as the amount of absorbed ethanol under the working conditions as a function of the partial pressure of ethanol at 403 K. (From Refs. I I Y , 243.)

184

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

formation increased at first with increasing ethanol partial pressure, but decreased markedly at higher ethanol partial pressures. The maximum rate was observed at a higher pressure for the ether formation than for the ethylene formation. The amounts of ethanol absorbed corresponded to 4-80 times the monolayer value (20-400 times the number of the surface polyanions), demonstrating that ethanol was absorbed in the bulk. These results are contrasted to the pressure dependences observed for ordinary solid acids; on A1203and SiO2-AI203 the formation of ethylene is usually zero-order in ethanol and the formation of ether is zero- to first-order in ethanol (244). Furthermore, it is emphasized that the activity of H3PW12040is lo2 times greater than that of Si02-AI203. Since ethylene is formed from one molecule of ethanol and ether from two molecules, it is understandable that ethylene is preferentially formed when the ratio of ethanol to protons in the pseudoliquid phase is low and ether is favored as this ratio increases. Equations (23)-(25) represent a possible mechanism that explains the essential trend in Fig. 40. C ~ H ~ O+HH + P C ~ H ~ O H-'c?H~ ; +H ~ O C~HSOH+ C~HSOH; P (C~HSOH); + (C2H&O

(n - 2)C2HSOH + (C2HSOH)2HCP C2HSOH),Ht

(23)

+ H20

(24)

(not reactive)

(25)

Simulation of the pressure dependence-assuming that the reactions of the first steps of Eqs. (23) and (24), and of Eq. (25) are in equilibrium-reproduced essential trends of the rates and the amounts of absorption.

ANALYSISOF PSEUDOLIQUID PHASE D. SPECTROSCOPIC Pseudoliquid phase behavior facilitates the spectroscopic investigation of the catalysts as the phenomena occur nearly uniformly in the bulk. The IR spectrum of absorbed diethyl ether and its changes during the thermal desorption are shown in Fig. 41 (78). A distinct peak at 1527 cm-' is characteristic of protonated dimer species { v(0-H) mode of [(C2H5)20..*H...0(C2H5)2]} (245). Moffat et al. also detected this peak by IR-photoacoustic spectroscopy (PAS) (238). It is reasonable that this band was not observed for diethyl ether adsorbed on A1203 or SiOz due to the absence of strong protonic acids on the surfaces of these solids, or for D3PWI2O4". 6(C2H&O due to the formation of [(C2H5)20...D-..0(C2H5)2] instead of [(C2H5)20***H..*O(C2H5)2]. A protonated monomer, [(C2H5)20H]3PW12040 (ethedproton = I), remained after evacuation at 328 K, and the dimer (indicated by the band at 1527 cm- I ) disappeared. Further evacuation at 423 K gave four peaks (2954, 2924, 2897, and 2870 cm- ') due to an ethoxide bonded to the polyanion.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

185

7

I

I

1700 1500 Wave nurnberlcm'

I

1300

FIG. 41. Changes in the IR spectra of diethyl ether absorbed in H3PWIZ040during stepwise heating in vacuum. (From Ref. 78.)

NMR spectroscopy also provides useful information. As shown in Fig. 42, H3PW12040.6C2H50Hgives three 'H NMR resonances at 9.5,4.2, and 1.6 ppm, which are assigned to OH, CH2, and CH3, respectively (169, 246). This is probably the first observation of a well-resolved solid-state 'H NMR spectrum for protonated organic compounds in this catalyst system. The relative intensities (OH/CH2/CH3 = 1.45 : 1.83 : 3.0), as well as the stoichiometry of ethanol to proton (2 : I), are consistent with the protonated dimer, (C2H50H)2H+.The high resolution is explained by the high mobility of the protonated ethanol and the homogeneity of the bulk phase. The chemical shift of the hydroxyl proton of H3PW12040* 6C2H50H (9.4 ppm) is close to those reported for protonated ethanol in superacids: 8.3 in HF-BF3,9.3 in FS03F-SbF5-SO2, and 9.9 ppm in HS03F (247) (as compared with a value of 1.0ppm for a dilute ethanol

186

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

1 6b*9d 31P

'H

l3C

CH2

-15

PPm

CH3

CH3

ppm

ppm

Reprinted I ~ ~ ~ ~ with . FIG. 42. "P, I3C, and 'H MAS NMR spectra of [ ( C ~ H S O H ) ~ H ] ~ P W permission from Ref. 169. Copyright 1992 American Chemical Society.

solution). Hence, the pseudoliquid phase of H3PW12040may be regarded as a superacidic medium. The chemical shifts of I3C (61.9 and 17.2ppm for CH2 and CH3, respectively) are different from those of pure ethanol (57.0 and 17.6ppm, respectively). Changes in the 13C NMR spectra upon heat treatment are shown in Fig. 43. Peaks at 65.0 and 16.8 ppm detected at 333 K indicate * 3C2H50H(169). Further heating gave a new set of peaks at 82.1 H3PW12040 (CH2) and about 14.3 ppm (CH3). The peak at 82.1 ppm is assigned to an ethoxide. The shift from 65.0 to 82.1 ppm is of similar magnitude to the transformation observed for transformation of adsorbed methanol to methoxide on K3-,H,PM01204~ (from 51 to 75 ppm) reported by Farneth et al. (168). These shifts are significant but smaller than that observed for alkyl cation formation. For example, sec-propyl and tert-butyl cations in a superacid solution show a downfield shift of about 260 ppm. Hence, the species at 82 ppm is more like ethoxide than ethyl cation (247, 248). Main reaction paths of the thermal desorption of ethanol are proposed in Scheme 5. The species observed directly by NMR spectrscopy are surrounded by broken lines. At temperatures less than 323 K, the dehydration did not proceed, and only reversible desorption took place. The protonated ethanol dimer is transformed into protonated ether at temperatures exceeding 323 K. Diethyl ether is formed only in the gas phase by replacement with ethanol. Protonated ethanol monomer probably gives ethylene via the ethoxide at temperatures exceeding 333 K (169). E. SURFACE AND BULK TYPEI REACTIONS As shown in Fig. 44a, the catalytic activity of NaxHj-,PW12040 for dehydration of 2-propanol, conversion of methanol, and decomposition of formic acid decreased monotonically with the Na content in the salts. The activities for these

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

I87

12.0 i

b

-

100

60

20

-20

P P FIG.43. Transformation of protonated ethanol dimer in HjPW120~by heat treatment. Solidstate "C CP/MAS NMR spectra were obtained by using high purity ''C ethanol: (a) Dimer, (b) 333 K, (c) 343 K, (d) 363 K, (e) 373 K, (f) 423 K . Reprinted with permission from Ref. 169. Copyright 1992 American Chemical Society.

reactions correlate well with the bulk acidities measured by the thermal desorption of pyridine (46b).All the reactants are polar molecules, so that the reactions proceed in the bulk. On the other hand, the activity pattern for butene isomerization is quite different (Fig. 44b). Butene is nonpolar and not absorbed, and it reacts only on the catalyst surface. The irregular variations probably reflect the surface acidity which changed depending on the Na content and pretreatment. Niiyama et al. (223) found that the reaction rate characterizing MTBE synthesis from methanol and tert-butyl alcohol catalyzed by H3PW12040 increases in proportion to the amount of methanol absorbed in the bulk of Hd'W12040.

188

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

The influence of bulk-type behavior also exists in supported heteropoly catalysts. Changes in the activity as a function of the loading of heteropolyacids depend on the reaction type (151). At low loadings, the rates of both the bulk and the surface reactions increase as the loading increases because the

Number of pyridine molecules anion-1

FIG. 44. Relationships between catalytic activity and bulk acidity. (a): (0) Dehydration o f 2-propanol, (A) decomposition o f formic acid, (0)conversion o f methanol. (b): (M) lsomerization of cis-2-butene after treatment at 423 K, ( 0 )isomerization of cis-2-butene after treatment at 573 K. (From Ref. 46h.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

189

:tivity/ mmol g-lh-l

FIG. 45. Catalytic activities of solid acids for reaction of liquids: Alkylation of 1,3.5-trimethylbenzene with cyclohexene (373 K);alkylation of phenol with I -dodecene: rearrangement of benzopinacol. (From Ref. 249.)

dispersion of the heteropolyacid becomes high. At high loadings, the dispersion becomes low, so that there is little increase in the rate of the surface reaction at increasingly high loadings. Pseudoliquid phase behavior is also evidenced by the liquid-solid reactions of polar compounds (249). An example is shown in Fig. 45. Heteropoly compounds are much more active than HzSO4 and other solid acids. The rate of alkylation of 1,3,5-trimethylbenzene increases in proportion to the surface acidity. Thus Cs2.5H0.5PW12040is much more active than H3PW12040. On the other hand, H3PW12040 is more active for the alkylation of phenol and the rearrangement of benzopinacol than C S ~ . ~ H ~1.2~0 4P0W . Since the latter two reactants are polar, the pseudoliquid phase is formed for H3PW12040 in the latter reactions. The activity of H3PW12040relative to that of Cs2.5Ho.5PW12040 decreases for these reactions in the order alkylation of trimethylbenzene > alkylation of phenol > pinacol rearrangement. The results indicate that the pseudoliquid phase is most important for the first reaction in this series and least important for the last. When the rearrangement of pinacol in I ,2-dichloroethane was examined in detail at 323 K, the amount of pinacol held by the catalysts was about 20 times the amount corresponding to a surface monolayer for H3PW12040rwhereas this ratio was less than 1 for the other catalysts. Niiyama et al. (250) attempted to separate isobutylene from a mixture of isobutylene and I-butene by using a H3PW12040-porousglass hybrid membrane, in which H3PWI2040 was loaded into the pores of the glass. A mixture of butenes and water vapor was brought in contact with one side of the membrane. Since the hydration of isobutylene takes place preferentially, tert-butyl alcohol formed on one side of the membrane is absorbed in the H3PW12040 and diffused through the bulk to the opposite side. After the tert-butyl alcohol reached the surface of H3PW12040 at the opposite side, it decomposed to give isobutylene and water.

190

TOSHIO OKUHARA. NORITAKA MIZUNO, AND MAKOTO MISONO

F. CONTROL OF ABSORPTION PROPERTIES AND CATALYTIC REACTIONS Absorption is influenced significantly by the cations in the salts (235). Typical examples for absorption of ethanol are shown in Fig. 46. It is clear that the absorption of H3PW12040is greatly suppressed when H + is replaced by Cs', while the absorptivity vanes little when H + is replaced by Na'. In contrast, there is little difference between the Na and Cs salts when they are characterized by the acid amounts measured by pyridine absorption. The pressure dependences as well as the amounts of ethanol held by CsxH3-.PW12040 (CsX) during the catalytic dehydration of ethanol were measured experimentally (240). The pressure dependences for Cs2 and Csl resemble those of the low- to intermediate-pressure region observed for ethanol in H3PW12040,suggesting that the pseudoliquid phase was also present in the acidic Cs salts. Slightly higher pressures are required for the latter to give the same trend because of the lower absorption capability of the Cs salts. The selectivity is affected strongly by the presence of the pseudoliquid phase. An example is given in Fig. 47. The oiefin-to-paraffin ratio in the product of the conversion of dimethyl ether increased markedly as the absorptivity of the heteropoly compounds decreased (195, 197). The ratio observed for catalysis by Na,H3-,PW12040 was similar to those observed for catalysis by H ~ P W I ~ O ~ ~ . The acidity governs the reaction rate but plays a minor role in determining the selectivity. The change in the olefidparaffin ratio can be explained by the formation of the pseudoliquid phase as follows. When olefins are produced in the bulk phase, they have a relatively high probability of undergoing hydrogena

x in MxH3-xPW12040(M = Na,Cs)

x in MxH3-xPW12040(M = Na,Cs)

FIG.46. Total amount (a) and absorption capacities (b) of Na and Cs salts of H3PW12040: (0) Na, (0) Cs. The acid amounts were measured by pyridine absorption (adsorption). (From Ref. 235.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

191

- 4

-3 - 2

10

20

30

40

Uptake of DMVnumbers of surface layers

FIG.47. Olefin-to-paraffin ratio in the product hydrocarbons from dirnethyl ether conversion as a function of absorptivity of the heteropoly compounds, expressed in term of the absorption of dirnethyl ether in surface layer units: (1) PW12 (H3PW12040), (2) NaH2PWI2, (3) Na2HPWI2, (4) C&PWiz, (5) C S ~ S H O S P W I( Z 6 ). CszssHo I s P W I ~(7) . ( N H ~ H Z P W I(8) ~. (NW~HPWIZ. (9) (NH4)2sHo5PW12, (10) 1,4-diazine, ( I I ) 1,3-diazine, (12) 1,4-bis(aminornethyl)benzene, (13) triazine. (From Ref. 197.)

transfer reaction to form paraffins and coke before they desorb. On the other hand, when the reaction occurs near or on the surface, the olefins formed may desorb rapidly into the gas phase without undergoing significant hydrogentransfer reactions. Similar selectivity patterns associated with the pseudoliquid phase were observed in the dehydration of ethanol to give ethylene and diethyl ether.

VII. A.

Redox Properties

REDOXCHEMISTRY IN SOLUTION

A general feature of heteropolymolybdates and heteropolytungstates is their high reducibility. Electrochemical investigations of Keggin-type heteropolyanions in aqueous or nonaqueous solutions have revealed sequences of reversible one- or two-electron ( l e - or 2eC) reduction steps [Fig. 48 (3)] which yield deeply colored mixed-valence species (“heteropoly blues”). Electronic spectra of the reduced heteropolyanions show intensified d-d bands in this visible region and intervalence charge-transfer (IVCT) bands in the near-1R region. Depending on the solvent, the acidity of the solution, and the charge of the polyanion, the reductions involve either single-electron or multi-electron steps, often accompanied by protonation. In protic solvents, the Keggin anions exhibit

192

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

EN FIG. 48. Cyclic voltammogram of a-PMol2OG in 0.1 M HCI (50% H20/EtOH). (From Ref. 3.)

electrochemical reductions by two electrons. The reduction potentials depend on pH as a result of protonation [e.g., Eq. (26)] (251): PMo120&

-+

H2PMo120:;

-t

H4PMo120&

(26)

Acidification of nonaqueous solutions of an unprotonated 1e --reduced species causes disproportionation to a 2eC-reduced anion and an oxidized anion (251). At higher pH values, polarograms show a series of le- reductions (252). Reduction of Keggin anions beyond 2e- reduction leads to modest changes in electronic and molecular structure, although the reductions remain reversible. For example, in the case of molybdates, a 4e--reduced P-Keggin structure is stabilized, with the bridging oxygen atoms being partially protonated [Eq. (27)] (3, 253, 254): a-PMo120+ ~ ~4e-

+ 3H’

-+/Y-H3PMol20;;

(27)

Electronic and NMR spectra of these complexes, together with X-ray structural determinations, indicate that the valence is completely averaged over at least six Mo atoms (251, 253, 254). ESR and I7O NMR spectra of le--reduced SiW120:0 demonstrate that the unpaired electron is weakly trapped on a W atom at low temperatures but undergoes rapid hopping (intramolecular electron transfer) at room temperature (Section 11). Anions generated by 2e- (and 4e-) reduction are ESR-silent, but 17 0 and Ig3WNMR spectra show that the additional electrons are fully delocalized (on the NMR timescale) at room temperature and generate “ring currents” analogous to those produced by the n-electrons of benzene. In contrast, in the case of le--reduced PMoWIIO:i, the electron is localized on a more reducible Mo atom at room temperature (251). The reduction potentials of heteropolyanions containing Mo and V are high, and they are easily reduced. Oxidizing ability decreases in general in the order V- > Mo- > W-containing heteropolyanions (Fig. 49) (8, 91). As for heteroatoms, the reduction potential (or oxidizing ability) decreases linearly with a

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

193

Gi

x

. '0

Y

-3

-4

Anion

-5

-6

-7

charge

FIG. 49. Dependence of reduction potentials on anion charge: ( 0 ) XMWllO&; (0) XMMolIO[;R;SCE = saturated calomel electrode. (From Ref. 91.)

decrease in their valence or an increase in the negative charge of the heteropolyanion (Fig. 49) (91). For polyanions with mixed-addenda atoms, the reduction potentials have been reported to be PMoloV20:G > PMollVO:o > PMo120:,; and PMo6W60& > PMo120i0 (2). XWI IMn039(X = Si, Ge) forms an oxygen adduct in nonpolar solvents. This was claimed to be the first example of an inorganic oxygen camer (255). Recently, it was reported that XWI ICrO39 reacted with PhIO, H 2 0 2 , or NaOCl to form XW11(CrO)O39 [Eq. (28)]: PWIICr3+(H?O)O:<+ DO

+

PWII(Cr5+O&+ D (DO

=

+ HzO

(28)

PhIO, HzOz, or NaOCI)

The 0x0 species further reacted with olefins to form the corresponding epoxides, ketones, and alcohols (256). In the reoxidation of heteropolyanions, reduced forms of polyoxomolybdates and polyoxovanadates are stable and hardly reoxidized by molecular oxygen, whereas polyoxotungstates undergo facile reoxidation (257).

B. REDUCTION-OXIDATION I N THE SOLIDSTATE 1. Reduction Mechanism and Oxidizing Abiliry Misono et al. (17, 258-262) have shown that the reduction of H3PM12040 (M = Mo, W) and its alkali-metal salts by H2 consists of the following three steps. [A similar mechanism has also been proposed by Eguchi et al. (101).]

194

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO Ri

xH2(gas) + P M I 2 0 ~ ( s u r f a c eF? ) 2xH+(bulk) + PM120r2”’-(bulk) 1 R, I1

2 r H ‘(bulk)

+ P M I ~ O $ + ~(bulk) ’-

R1

-+

I1

PM120:&.(bulk)

+ xH20(bulk)

(29) (30)

111

xH20(bulk)

+

(31)

xH20(gas)

[RI and R2 are the rates of the forward and reverse reactions of Eq. (29), respectively, and R3 is the rate of Eq. (30).] The first step is H2 2H’ + 2e-. Here, the number of electrons introduced per anion may be variable. In the second step, the protons formed in the first step react with the oxygen of the polyanion to form water. Accordingly, in TPR experiments with H2, the H2 uptake takes place but no H20 is observed at low temperatures (I 11),while H2O appears in the gas phase at higher temperatures (11 111) (259, 261). Further reduction produces several irreversibly reduced species. IR spectra indicate that reactive oxygen in the reduction of H3PM01204~is the bridging oxygen in the Keggin anion (101, 261, 263). According to a quantumchemical calculation by the Xa method, the LUMO is a mixture of 4d orbitals of Mo (50%) and 2p orbitals of the bridging oxygen atoms (Ob 21% and 0, 27%), while the HOMO mostly (94%) consists of 2p orbitals of bridging oxygen. The LUMO is antibonding with respect to Mo-Oh and Mo-0, bonds. This result indicates that the interaction with a reductant preferentially takes place at the bridging oxygen atom and loosens the Mo-01, and Mo-0, bonds (264). IR spectra change only slightly upon the reduction to 11, since the Keggin structure is maintained. The spectra change greatly upon the reduction to 111. [Some authors reported that even the first reduction step (I 11) significantly changed the IR spectrum (265). At present this discrepancy is still controversial.] H3PMo12040reduced by H2 at a lower temperature shows a very weak ESR signal of Mo”. It is probable that as long as the Keggin structure is maintained (I -+II), most of the Mo” ions are not detectable due to the rapid hopping of electrons. A heat treatment converts I1 into 111, and the Mo” signal grows significantly, indicating the localization of electrons. Early reports of ESR spectra of reduced H3PMo12040 probably should be attributed to these species. ) been determined by the quantitative The rates of these reactions ( R I - R ~have analysis of the reduction of H3PM12040 (M = Mo, W) by a mixture of H2 and D z . With H3PW12040, the isotopic equilibration of H2 and D2 in the gas phase, as well as the isotopic exchange between the entire solid and the gas phase, is very rapid, so that, to our surprise, the content of H in the gas phase increased rapidly (Fig. 50). The detailed kinetics analysis shows that the reactions of Eqs. (29) and (30) are very rapid and that of Eq. (31) is the slow step, the equilibrium strongly favoring the reactions on the left-hand side of Eq. (29) (Fig. 51, left). -+

-+

-+

-+

195

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

0. I00

41

. ae

Y

50

8 E

0 Time I min

10

20 30 40 Time I min

120

0

in Fig. 50a show the FIG. 50. Isotopic exchange of Hz-D2 at 573 K: A, 0,0 , and -.amounts of H2. HD, and D2, and decrease of the sum,respectively. Sample: H3PW12040. 19Hz0 ( 1 . 1 g). (From Ref. 260.)

The result obtained for H3PM012040 was essentially the same, except that the equilibrium of Eq. (29) favors the right-hand side (Fig. 51, right) (262). The reduction rates of H3PM12040(M = Mo, W) by Hz depend only slightly on the specific surface areas, as shown, for example, in Fig. 52 (260). This dependence is quantitatively explained on the basis of the assumption that R3 is

FIG.51. Models of H2-D2 reactions for H3PMo1204"(right) and H3PW1204"(left). Values are given in units of 1 0 ~ 6 m o l m i n -g1 - ' . Outer circle indicates the surface of H ~ P M O and ~ ~ O ~ ~ H3PW12040; p and b, respectively, indicate bulk phase and hypothetical portion in which the isotopic concentration is uniform. (From Ref. 262.)

196

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

proportional to the number of polyanions in the bulk (p), in which protons as well as electrons migrate rapidly in the bulk (Fig. 52). Thus, it can be concluded that the rates of reduction of H3PM12040by H2 reflect the oxidizing ability of the catalyst bulk (bulk type I1 behavior) (260). The reduction is slower for H3PW12040 than for H3PMo12040 not because the dissociation of H2 is slower, but because the reactions of either Eq. (30) or Eq. (31) are slower (Fig. 51). In contrast, the rates of reduction of H3PMo12040and its alkali salts by CO under dry conditions are proportional to the specific surface areas, as is observed for ordinary heterogeneous catalysis (Fig. 52). In this case, because of the slow diffision of oxide ion, the reduction mainly proceeds only near the surface (266). The slow diffusion of oxide ion is deduced from the following result, as well as the observation of a much slower isotopic exchange for 16 0 2 - 1 8 0 2 - H ~ P M ~ 1 2 0 4 0than for H2'80-H3PMo12040 (267). When the reduction by CO (on the surface) is stopped halfway by evacuation, then resumed after a delay, the rate becomes several times greater than the rate just before the interruption in contrast to the small change observed for the reduction by H2 (bulk type 11) in a similar experiment (266). The rate of reoxidation by O2 is similar to the rate of reduction by CO (both are surface-type reactions). In the case of reduction by CO, the oxide ions difise very slowly, so the surface is reduced to a much greater extent than the bulk: During the interruption, the r

.-c

E e

5

.-0 C

m

e 0

'E3 'D

?!

r 0

X

Surface area / m2g-' FIG.52. Dependence of rates of reduction of H J P W ~ by ~ OH2~ and ~ of N ~ ~ H P M O by ~CO ~ on O ~ ~ O ~ broken ~. line shows the calculated data the specific surface area. (0)Reduction of H J P W ~ ~ The (see text). H J P W , ~19H20, ~ ~ ~ . 1.0 g; reaction temperature, 573 K. (A) Reduction of Na2HPMo1204,,. Reaction temperature, 623 K. (From Refs. 260, 266.)

197

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

overreduced surface is reoxidized by the oxide ions that difhse from the bulk; therefore, the reduction rate greatly increases after the interruption. Hence, to a first approximation, the rate of reduction of these heteropoly compounds by CO expresses the oxidizing ability of the surface, whereas, as described above, the rate of reduction by H2 reflects the oxidizing ability of the catalyst bulk. If the former rate is divided by the surface area and the latter normalized to the catalyst mass, both oxidizing abilities decrease monotonically with the extent of neutralization with alkali (Figs. 53a and 53c). Although it is not shown in Fig. 53, Cs2.sHo.sPMo1204~, a class B salt that has a high surface area, is reduced exceptionally rapidly. The contrast between the two types is also found for organic reactions. The reduction of H3PMoI2O4,-,by dehydrogenation of isobutyric acid or cyclohexene belongs to the bulk type I1 classification, and the reduction by oxygenation of methacrolein or acetaldehyde is a surface reaction. For example, when HjPMol2040 is reduced by repeated pulses of isobutyric acid and methacrolein

1

0

2

1

3 0

0

1

2 X

2

3

X

X

30

1

2 X

30

1

2

3

X

FIG. 53. Stoichiometric reduction of H3 - r C ~ , P M ~ ~ ~by0 4(a) 0 H2, (b) isobutync acid, (c) CO, and (d) methacrolein, and reoxidation by 0 2 [Degree o f reoxidation after 10 min at 623 K. Sample was prereduced by l e - at 623 K (from Refs. 258, 266)], (e) at 623 K.

198

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

Bulk reduction

Surface reduction Scmm 6

and the conversions are plotted against the average extent of reduction of the bulk solid, the conversion of isobutyric acid to methacrolein decreases only gradually, whereas the conversion of methacrolein to methacrylic acid diminishes sharply (268). This contrast is explained as follows. In the oxidative dehydrogenation, protons and electrons formed on the surface rapidly diffuse into the inner bulk, so that whole bulk solid is reduced nearly uniformly. On the other hand, in the case of methacrolein, the surface oxygen removed by the reaction is only slowly replenished by oxide ion diffusion, so that the reduction takes place mainly near the surface (Scheme 6). 2. Reoxidation As for the reoxidation of reduced heteropoly compounds in the solid state, few reliable studies have been reported. It was reported that the reoxidizability increases with an increase in standard electrode potentials of countercations (108). In the case of reoxidation by 0 2 of le--reduced C ~ , H ~ - ~ P M o 1 2 0 4the 0, rates divided by the surface area show a monotonic variation (Fig. 53e) as in Figs. 53c and d, indicating a surface reaction. A similar variation was observed for the Na and K salts. The presence of water vapor sometimes accelerates the migration of oxide ion, probably in the form of OH- or H 2 0 , and makes surface-type reactions more like bulk type I1 reactions (266). As for the rate and reversibility of redox cycles, the following have been observed. [I, 11, I11 refer to Eqs. (29x31) (258b).]

I (fully oxidized heteropolyanion) 11: very rapid and reversible; I ++ 111: rapid near the surface, but slow in the bulk and reversible; I III’(excess reduction species): slow and mostly irreversible. ++

There are two reversible redox cycles, I I1 and I t* 111. Upon reoxidation, 0, water is evolved in I --* 11, but not in 111+ I. In the case of H 3 P M ~ 1 2 0 4 the redox cycle is nearly reversible when the average extent of reduction is less than 3e-/anion at 573 K. The second redox cycle (I --* 111) tends to dominate at high temperatures and for extensive reductions. ++

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

I '

I

0.5

'

I

"

,

,

'

,

'

I

I

I

'

'

,

,

0.0 EIV

'

,

'

" I

I

,

199

I

-0.5

FIG.54. Cyclic voltammogram of H4SiWI~040recorded at conventional size glassy carbon disk electrode. (a) Single-crystal sample; (b) 4 mM H4SiW12040 in 0.5 M HzS04. (From Ref. 269.)

Recently, Kulesza et al. (269) reported cyclic voltammetry with a single crystal of H4SiW12040.31H20.In the potential range of -0.1 to - 0.65 V, three reversible redox transitions similar to that in solution (Fig. 54b) (269) are observed with the ratios of numbers of electrons being 1 : 1 : 2 (Fig. 54a). The reduction probably corresponds to Eq. (32): SiWIZO:" + z e - + z H ' -tH2SiW120:U

( z = 1,2,4)

(32)

The cyclic voltammogram of Fig. 54a is not fully symmetrical. The distortion probably originates from the catalytic discharge of protons and evolution of hydrogen in the solid phase. These results suggest the possibility that by using cyclic voltammetry with a single crystal, the reduction potential of solid heteropoly compounds can be measured and that the effects of constituent elements described below can be made clearer.

c.

EFFECTSOF CONSTITUENT ELEMENTS ON REDOXPROPERTIES IN THE SOLID STATE

Methods for estimating the oxidizing ability of heteropoly compounds in the solid state involve measurements of the reduction rate at a constant temperature as described above, temperature-programmed reduction ( 1 7, 258, 259, 270-273), and ESR and XP spectra indicating changes by reduction (107, 108, 274, 275). As described previously (254), the orders of oxidizing ability obtained experimentally differ significantly, depending on the method adopted and the kind of the reductant as well as the inhomogeneity, nonstoichiometry, and decomposition of the samples. Inconsistencies in the literature may also be explained by lack of recognition of the occurrence of both surface and bulk reactions.

200

TOSHIO OKUHARA, NORITAKA MIZUNO, A N D MAKOTO MISONO

Various parameters such as the heat of oxide formation, ionic potential, electronegativity, and the extent of cation exchange have been proposed to control the redox properties. However, a full understanding is lacking. Nonetheless, the following trends are evident:

1. When MIFW mixed-addenda heteropolyacids are reduced by H2, the rate of reduction decreases in the order PMo6W,O:; > PMo120i0 > PMoIOV20:0 > PWI~O:;, in parallel with the reduction potentials in solution (except for PMoloV2O:; ) (276). 2. For a given polyanion, the effects of metals are divided into two groups: (a) Transition metals play roles in the redox processes; e.g., they activate reducing agents and molecular oxygen and possibly provide reservoirs of electrons (108, 272, 273, 276, 278). (b) Alkali and alkaline-earth metals are not reduced. Oxidizing abilities measured by the rate of reduction decrease upon the formation of alkali salts (Fig. 5 3 ) (258, 272, 273, 277). The reason for the decrease of the oxidizing ability with alkali content is not hlly understood, although suggestions have been made concerning the electronegativity of the cation and the role of protons in the reduction process. The mixed-addenda atoms affect the redox properties; mixed-addenda heteropoly compounds are used as industrial oxidation catalysts. For example, the rate of reduction by H2 is slower and less reversible for solid P M o 1 2 - , V , 0 ~ ~ " ) than for solid PMoI20:;, although the former are stronger oxidants than the latter in solution (279, 280). The effects of substituting V for Mo on the catalytic activity are controversial (279, 281-284). Differences in redox processes between solutions and solids, the thermal or chemical stability of the heteropoly compounds, and the effects of countercations in solids have been suggested to account for the discrepancies. (M = Mo, W) is It has been demonstrated that V 5 + in H3+xPMLZ-xVx040 eliminated from the polyanion framework upon thermal treatment or during catalytic oxidation, and the V 0 2 + salt of H3PMI2O40is formed (284). It has been reported (I03) that H3PMo12040 is re-formed from thermally decomposed H3PMoI2O4,,under the conditions of methacrolein oxidation.

VIII.

Liquid-Phase Oxidation Reactions A.

OXIDATION WITH

DIOXYGEN

Typical examples of liquid-phase oxidation with molecular oxygen catalyzed by heteropoly compounds are listed in Table XXVI. Introduction of VSf or

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

20 1

other transition-metal ions such as C 0 2 + or Mn2+ as addenda atoms usually enhances the catalytic activity, reflecting changes in the reduction potential. Recently, paraffin oxidation in the liquid phase, which does not proceed with the V' and mono-Co2+-substituted heteropolyanions, has become possible, catalyzed by heteropolyanions having tri-transition-metal (Fe2Ni or Fe3) sites. The use of heteropoly compounds in combination with transition metals as in Wacker-type reactions is described in Section XI. PV2MoloO:i has usually been used in the acid form. H5PV2MolOO40catalyzes aerobic oxidative cleavage of cycloalkanes, 1-phenylalkanes, and ketones. For example, the oxidation of 2,4-dimethyl cyclopentanone and 2-methylcyclohexanone gives 5-0x0-3-methylhexanoic acid and 6-oxoheptanoic acid, respectively, in yields higher than 90% (285, 286). Bromination of arenes with HBr (289, oxidative dehydrogenation of cyclohexadiene (288, 289) and a-terpinene (290), oxidation of 2,4-dimethylphenol (291) and sulfides (292) are other examples. The mechanism of aerobic oxidative dehydrogenation of a-terpinene to give p-cymene catalyzed by PV2MoloOii has been investigated (290). On the basis of kinetics along with the use of UV-visible, ESR, 3'P-NMR, and IR spectroscopies, a reaction scheme was proposed, as shown in Fig. 55. In this scheme, a stable a-tetrapinene'-H6PV s+ V4 + M010O40 complex (b) is formed via an electron transfer complex (a), which is a reduced heteropolyanion attached to an oxidized cation radical of a-terpinene. A doubly protonated reduced heteropolyanion and p-cymene are generated via the intermediate. On the basis of the kinetics (zero-order reaction in a-terpinene, second-order reaction in PV2MoIoO:~,and first-order reaction in 02)and the IR and NMR data, the ratelimiting step is proposed to be the reoxidation of the catalyst. In the reoxidation heteropolyanions are reoxidized in a 4estep, two reduced PV:!'MoloO:i redox reaction via a p-peroxo intermediate (c). Co2+-Substitution at the addenda atoms gives catalysts for the epoxidation of olefins in the presence of aldehyde (293). PWII-Co is the most active among the mono-transition-metal-substituted polyanions; the order of activity is PWII-Co 9 -Mn 2 -Fe 2 -Cu > -Ni. Here, PWII(M"+)O\~-")-(M = Co2+, Cu2+, Fe3+,Ni2+, Mn2+) is denoted by PWII-M. The same order was observed for the oxidation of isobutyraldehyde, suggesting that the oxidation of aldehyde to give peracid is an important step in the reaction. It has been reported that substitution of V5+ for Mo6+ in PMo120:i gives a good catalyst for epoxidation and the Baeyer-Villiger reaction (294). Styrene and 1-decene are selectively epoxidized, as shown in Table XXVII (293). The rates observed for PWIl-Co are greater than those observed for Ni(dmp), and Fe(dmph, and the selectivities are comparable or higher for the former (295). It is also remarkable that PWII-Co polyanion exhibits a steric effect comparable to that of a moderately hindered TTMPP ligand in the +-

202

TOSHIO OKUHARA. NORITAKA MIZUNO. AND MAKOTO MISONO

TABLE XXVI Liquid-Phase Oxidation Reactions with Molecular Oxygen Catalyzed by Heteropoly Compounds

-

Reaction

~~

PhCOCHzPh + 0

R2S

Catalyst

Temp. (K)

Ref.

+ PhCHO

HSPMOIOVZO~O

333

285

6

H~PMoYV~O~O

363

287

HSPMOIOVZO~O

293

289

HSPMOIOVZO~O

343

288. 289

HSPMOIOVZO~I

298

290

HsPMoioVz040

333

291

PW 1coo:; PMo,VhOI,

303

293,294

- A +A

H7PWyFezNi037

423

297

-

(TBA)4HJ’WyFezNi037 TBA, ( ~ - C ~ H Y ) ~ N

355

299

PdC12 + Na,.H3 + .-,PMoIz (X = 2-3)

393

See Section IX

2

PhCOOH

-

RzSO or RzSOZ

6

+ HBr + 1 / 2 0 ,

-

+ H20

Br

$

($+02-

&+oz-& 0

)=( + 0, + aldehyde 0

n+ 0 2 O + O 2

HzC=CHz

+0 2

-% OH

6 . 6 CH3CHO

-x V , O ~

continued

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

203

TABLE XXVl-Continued Reaction

Catalyst

Temp. (K)

Ref.

N ~ , P M O ~ V ~ O ~ ~293

0

SO2

+ 112 0 2 + H2O

-

HzS04

Hd'M07Vs040

298

"J. E. Lyons et al., Stud. SurJ Sci. Catal. 67, 99 (1991). h R . Neumnann er a/.. Dioxygen Activation and Homogeneous Catalysis Oxidation L. 1. Simanded, Ed. Elsevier, Amsterdam, 12 1 ( 1991). ' B. S. Jumakaeva et a/., J. Mol. Catal. 35, 303 ( 1986).

epoxidation of (R)-(+ )-limonene, 4-vinyl- 1-cyclohexene, and I -methyl- 1,4cyclohexadiene (296). Activation of paraffins by heteropolyanions has been attempted by using tritransition-metal-substituted heteropolyanions. Propane, ethane, and methane are oxidized to the corresponding alcohols and ketones in the presence of tritransition-metal-ion (Fe2Ni)-substituted PW I 20:;, although the composition and structure of the catalyst are still to be examined (297). Recently, Mizuno et al. (298) characterized its structure, and confirmed that it catalyzed the oxygenation of adamantane, ethylbenzene, and cyclohexane with molecular oxygen alone (299). For example, adamantane was catalytically oxidized mainly to 1-adamantanol with small amounts of 2-adamantanol and 2-adamantanone. The total number of turnovers was 25, calculated on the basis of the bulk polyanion, or 3750, calculated per surface polyanion. This number of turnovers is the highest for the dioxygen oxidation of adamantane on p-0x0 di- or tri-iron and Ru complexes with or without any reductants (300, 301). Fe=O, Fe-OH, or Fe-OOH species are assumed to play an important role in the reaction. B. OXIDATION WITH OXIDANTS OTHERTHANO2

Hydrogen peroxide and alkyl hydroperoxides are important oxidants in organic synthesis, but they usually need to be activated by catalysts such as tungsten, molybdenum, and titanium oxides. Heteropoly compounds are good catalysts for oxygenation of olefins or paraffins and oxidative cleavage of vic-diols.

204

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

ICatalyst Reduction)

batalyst Reoxidation

h

FIG. 55. Mechanistic scheme for oxidative dehydrogenation of a-terpinene catalyzed by H S P M O ~ O V (From ~ O ~ ~Ref. . 290.)

1. Hydrogen Peroxide

Typical examples are collected in Table XXVIII. Mixed addenda heteropolyanions show unique catalytic properties. For example, in the oxidation of cyclopentene to glutaraldehyde with hydrogen peroxide, the catalytic activity of

205

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XXVII Expoxidation of Olefins Catalyzed by PW,,-Co at 303 K" (from ReJ 293)

Alkene

Conversion" PA)

Selectivity"(%)

Cyclohexened Styrene' 1 -Decene/

78

90

100

91

73

80

" Olefin, 250 pmol; isobutyraldehyde, 1000 pmol; solvent, 1.2-dichloroethane; 1 h. Based on the starting olefin. Based on the converted olefin. "PWII-Co,1.2 pmol; 1.3 h. "PWII-Co,4.5 pmol; I h. 'PWII-Co, 1.2 pmol; 4 h.

"

H3PW6M06040 was observed to be the highest among H3PW12-xM~r040 species, as shown in Fig. 56 (302-304). A theoretical explanation of this activity pattern or synergistic effect has been attempted by accounting for excess electronic energy of nonbonding levels (305). However, there is a possibility that the active peroxo catalysts are formed by the degradation of the Keggin structure (304). Phase-transfer catalysis has been developed by the combination of Keggintype heteropolyanions and quaternary countercations such as tetrahexylammonium or cetylpyridinium ion. The oxidations of alcohols (306), ally1 alcohols ( 3 0 3 , olefins (308), alkynes (309),j3-unsaturated acids (310), vic-diols (311), phenol (312),and amines (313) are the examples. Ishii et al. (306, 307, 310, 311, 313) and Venture110 et al. (308) have developed various oxidation reactions using organoammonium salts of PM ,O:, (M = Mo, W) for phase-transfer catalysts. A reaction mechanism of the epoxidation of olefin is proposed in Scheme 7. In epoxidation of long-chain olefins, epoxides are produced in the organic phase by peroxo species which are only slightly soluble in the organic phase. Since the salts dissolve only slightly in the organic phase, they do not catalyze undesirable ring opening of epoxides. At a conversion of olefin of 82-98%, the selectivity to epoxide is more than 98%. A simple molybdenum compound, H2MoO4, shows no catalytic activity under these reaction conditions. It is probable that neither metal oxides nor peroxomolybdates are soluble in the organic phase (288). Recently, it was claimed that the active catalysts are not the starting Keggin-type heteropolyanions, but rather PO4[WO(O2)2]: and/or [W203(02),( H20)2l2 -, which are peroxo compounds formed in aqueous solution (308, 314-316). P04[WO(02)& was two orders of magnitude more active than [W203(02)4(H20)2]2- for olefin epoxidation. It is suggested that the catalytic cycle is mainly PO4[WO(O,)2]:[P,W,Os(O2),],- [r is 4 or 3 (316)], although the composition of the species formed by the reaction depends on the ratio [H2O2]/[H3PWI2O40](316, 31 7). ~

-

206

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

TABLE XXVlll Liquid-Phase Oxidation Reactions with Hydrogen Peroxide Catalyzed by Heteropoly Compounds Reaction

Catalyst

Temp. (K)

298

Ref.

307a

OH

--

F OH

0

continued

207

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XXVlII-Coniinued Catalyst

React ion

0-0. 0"-

Temp. (K)

Ref.

303

312

r.t.

313

27 I

319

reflux

(1

D C O C H 3 353

h

0 -

-d 0

" D. Attanasio el al., J. Mol. Caial. 51, LI ( I 989). S. Sakaguchi er al., J. Org. Chem. 59, 568 I (1 994).

75

,

I

0

3 6 9 12 x in H Q P M O ~ ~ - ~ W ~ O U

FIG.56. Effects of addenda atoms on the catalytic oxidation of cyclopentene by H202 at 303 K. ( 5 ) H3PMol2040 + H3PW12040.Reaction time, 3 h. (From Ref. 304.)

208

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO H3pw12040

Aqueous phase

..................................................

P04[W0(02)21~ _._____. (w203(02)4(H20>2)2-

SCHEME 7

Cetylpyridinium salts of H3PM12040 (M = W, Mo) also catalyze the oxidation of secondary alcohols or alkynes to give the corresponding ketones, but they are not active for primary alcohols (308b, 309b). Schwegler (308c)reported that lacunary heteropoly- 1 1-tungstates are better catalysts than tetrahexylammonium 12-tungstophosphate for the oxidation of cyclohexene in biphasic systems. As described above, catalytic oxidation with hydrogen peroxide is usually limited to polyoxometalates containing only metals in the d o state. Recently, however, a tetrairon-substituted heteropolyanion, Fe4(PW9034):o-, and dimanganese-substituted heteropolyanion, [ W Z I M ~ ~ ( Z ~ W ~ O (sandwich~~)~]'~type compound with a WZnMn2 ring between two B-ZnW9034 units), have been reported to catalyze selectively the epoxidation of olefins (328). Table XXIX summarizes the oxidation of various reactants with hydrogen peroxide and [WZnMn2(ZnW9034)2]12-. At 275 K, the selectivities were more than 99% in all cases. High yields of more substituted olefins indicate that the reactivity of

209

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XXIX Oxidation of Various Reactants with 30% H202 Catalyzed by pZnMn2(Zn W90,4)J’2- at 275 K“ (f?om ReJ 318) ~~

Substrate Cyclohexene

I-Octene 2-Methyl-I -heptene trans-2-Octene Cyclohexanol

Products

Turnovers

Cyclohexene oxide Cyclohexen-2-01 Cyclohexen-2-one I-Octene oxide 2-Methyl-I-heptene oxide trans-2-Octene oxide tran+2-Octen-4-one Cyclohexanone

1450

0 5

25 190 340 0 510

“Catalyst, 0.2 p o l ; reactant, I mmol; H202. 2 mmol; 5 pmol, methyltricaprylammonium chloride.

the reactant is a function of the nucleophilicity of the carbon-carbon double bond.

2. tert-BuWl Hydroperoxide and Others Mixed addenda or transition-metal-ion-substituted heteropolyanions containing Co, Mn, and Ru are catalysts for oxidation reactions with tert-butyl hydroperoxide and other oxidants. Typical examples are listed in Table XXX. TABLE XXX Liquid-Phase 0-ridarion Reurtions with rert-Burylhydroperoxide o r Other Oxidants Catalvzed bv Heteropoly Compounds Reaction

0

+

fert-BuOOH or PhIO

0 +PhIO

-

00”

0

0

Catalyst

+0

0 PWI~COO:~

PWlIMO:9 PW17MO:;(M = Co, Mn)

D. Mansuy et al., J. Amer. Chem. SOC.113, 7222 (1991).

Temp. ( K ) Ref.

298

320, 32 I

297

321“

2 10

TOSHIO OKUHARA. NORITAKA MIZUNO. A N D MAKOTO MISONO

Strongly acidic H3PM12040(M = Mo, W) species catalyze oxidation of thioether into sulfoxide and sulfone with 98-99% and 1-3% selectivities, respectively. The V5+ substitution increases the selectivity to sulfone up to >99% (319). Cobalt- or manganese-substituted PWl20:O and SiWl1039Ru(OH2)S catalyze the oxidation of paraffins such as cyclohexane and adamantane (320, 321) as well as the epoxidation of cyclohexene with tert-butyl hydroperoxide, iodosylbenzene potassium persulfate, and sodium periodate (321, 322). The reactivity depends on the transition metals. In the case of epoxidation of cyclohexene with iodosylbenzene, the order of catalytic activity of PWII(M)O:; is M = Co > Mn > Cu > Fe, Cr. A Ni-containing sandwich complex, Ni3(a-PW9039)2,is a better catalyst than PWll(M)O:; (M = Co, Mn) for the formation of N-alkylacetamide from adamantane and isobutane with tert-butylhydroperoxide as the oxidant (323). As for the mechanism of oxygenation of paraffins with oxygen donors (DO) such as iodosylbenzene and potassium persulfate, Eqs. (33) and (34) have been proposed, whereby oxgenation of metal centers by oxygen donor is followed by 0x0 transfer from the transition metal to CH bonds of paraffins: ~

IX.

LM+DO

+

RO+D

LM=O t RH

+

ROH

+ LM

(L = heteropolyanion)

(33) (34)

Oxidation Catalyzed by Solid Heteropoly Compounds

Solid heteropoly compounds are suitable oxidation catalysts for various reactions such as dehydrogenation of 0- and N-containing compounds (aldehydes, carboxylic acids, ketones, nitriles, and alcohols) as well as oxidation of aldehydes. Heteropoly catalysts are inferior to Mo-Bi oxide-based catalysts for the allylic oxidation of olefins, but they are much better than these for oxidation of methacrolein (5). Mo-V mixed-oxide catalysts used commercially for the oxidation of acrolein are not good catalysts for methacrolein oxidation. The presence of an a-methyl group in methacrolein makes the oxidation difficult (12). The oxidation of lower paraffins such as propane, butanes, and pentanes has been attempted (324).Typical oxidation reactions are listed in Table XXXI and described in more detail in the following sections. Keggin-type heteropoly compounds having Mo and V as addenda atoms are usually used for such oxidations. The catalysts reported in patents often contain several elements other than Mo, V, and P. An excess amount of P is added to stabilize the structure, and the presence of additional transition elements like Cu improves redox reversibility. Supported heteropoly catalysts are also important for industrial applications and have been characterized (69, 325, 326). To understand oxidation catalysis by solid heteropoly compounds, the contrast between surface and bulk type I1 catalysis, and acid-redox bifunctionality

21 1

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS TABLE XXXl Heterogeneous Oxidation Reactions Catalyzed by Heteropoly Compounds

Reaction

Catalyst

Temp. (K) 533

= + 02

-

CH3COOH

+ O2 N

-

C

I

12

573

333

573

34P

Pd + H4SiW1204dSi02

423

h

PbFeBiPMo120,

673

C=C-CHO

- no -

563

J

+ 0 2

0

0

CH4 + NzO HCHO, CH3OH CHxCHO CzH6 + N 2 0 ( 0 2 ) -C2H4, ---+

- - no A -A

Ref.

+ O2

CH2=CHCOOl

c

H3PMo12040/Si02 H3PMo1204dSi02

843 540

I

Hd’M01z040 ( +As)

613

352, 353

633

324, 350

623

354-356

583

106

473-563

345

443

110

+ 0 2

0

0

C

I

+ o2

+o,-

CH3OH

+0 2

-OH

+ 0,

H3PM012040

C=C-COOH

0a 0

-

0

HCHO, (CH3)2O, etc.

CH3CH0,(C2H5)20

H P M o 12040 H~PMoIzO~O + polysulfone)

(

OM. Akimoto et al., J. Cural. 89, 196 (1984). ’T. Suzuki ef al., US Patent, 5405996 (1995). ‘T. Ohara, Shokubai (Catalysf) 19, 157 (1977). ’M. Ai, J. Catal. 85, 324 (1984). ‘S. S. Hong ef a/., Appl. Catal. A 109, I17 (1994). ’S. Kasztelam et al., J. Cats/. 116, 82 (1989).

of heteropoly catalysts must be properly taken into account, along with relationships between the oxidizing properties of catalysts and their catalytic activities (5, 6, 258, 266, 327, 328). A. 1.

CONCEPT OF SURFACE AND BULK TYPEI1 CATALYSIS AND REDOX (MARS-VANKLEVELEN)MECHANISM

Bulk Type II Catalysis

As described above, heterogeneous catalytic reactions on heteropoly compounds are classified into three different types, surface, bulk type I (pseudoliquid phase), and bulk type I1 (Fig. 1). The surface reactions are typical of

212

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

ordinary heterogeneous catalysis, and therefore the surface acidity and oxidizing ability are important. Bulk type I reactions proceed within the catalyst. In the bulk type I1 reduction, the rapid migration of redox carriers enables the whole catalyst to be reduced (260). The differences between the noncatalytic surface and bulk type I1 reduction are reflected in catalytic oxidations proceeding via redox mechanisms, as described below (258, 260-262, 329). If a catalytic oxidation proceeds by the cyclic reduction and reoxidation of the catalyst (a redox mechanism), the rate of catalytic oxidation and the rates of reduction and reoxidation of the catalyst must coincide if they are measured in the stationary state of catalytic oxidation. Measurements were made for the oxidation of H2 catalyzed by H ~ P M Ohaving , ~ ~ different ~ ~ specific surface areas (Fig. 57). While the degree of reduction in the stationary state differs among the catalysts, the rates of reduction of the catalysts by H2, the rates of reoxidation, and the rates of catalytic oxidation agree quite well. The agreement indicates that the catalytic oxidation of H2 proceeds by a redox mechanism. A similar redox mechanism was confirmed for H3PW12040 and alkali salts of H3PMol2040. In addition, the rate of reduction depends only slightly on the surface area (Fig. 5 2 ) , although the rate of reoxidation is proportional to the surface area. If these rates are compared for a particular degree of reduction at the steady state of the catalytic oxidation, the differences in rates from one catalyst to another

d

0

.CI

Y

I

;; f

cw

-a

41

0

,

,

0.1 0.2 Degree of reduction 1 e- anion1

FIG. 57. Rates of catalytic oxidation of H2. reduction by H2. and reoxidation by O2 for H3PMo12040catalysts having different specific surface areas: Dependence on the degree of reduction of each catalyst at the stationary state. The numbers next to the symbols indicate the surface areas (after reaction) in m' g I . ( 0 )Catalytic oxidation; (-) reduction; (0)reoxidation. Reaction temperature, 573 K. (From Ref. 262.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

213

are small. Consistent with this pattern, the rate of the catalytic reaction is only weakly dependent on the surface area, as shown in Fig. 58. On the other hand, the rate of catalytic oxidation of CO is proportional to the specific surface area, as shown in Fig. 58. This dependence indicates ordinary heterogeneous catalysis. The linear dependence in Fig. 58 can also be explained on the basis of the redox mechanism, as both the rate of CO conversion and the rate of O2 conversion are proportional to surface area (Fig. 53). A weak support effect, as shown in Fig. 59, is another indication of bulk type 11 behavior (327). With an increase in the loading of the heteropoly compound on the support, the rate of bulk type I1 catalysis increases to high loading levels, whereas the rate of surface catalysis shows saturation at relatively low loadings because of the decrease in the dispersion of the heteropoly compound on the support (327). Good correlations are observed between the oxidizing abilities of catalysts and the catalytic activities for oxidation, provided that the bulk and surface catalysis are properly accounted for. Examples are shown in Figs. 60a and b. A linear correlation is observed between the rates of catalytic oxidation of

Surface area / m 2 gl

FIG. 58. Rates of catalytic oxidation of H2 and o f CO in the presence of catalysts having different surface areas. (a) H3PMo12040 at 573 K, (b) Na2HPMo12040 at 623 K. ( 0 )H 2 4 2 ; (A)C G 0 2 (From Ref. 262.)

214

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

100

0

30 Amount of H,PMo,,O, 10

20

40

50

loaded / wt 8

FIG.59. Catalytic oxidative dehydrogenation of cyclohexene (0, surface catalysis) and oxidation of acetaldehyde (0.bulk-type 11); the catalyst was H~PMOIZON supported on Si02. Masses catalyst: 0.2 g for cyclohexene and 0.1 g for acetaldehyde. (From Ref. 327.)

acetaldehyde (surface reaction) and the rate of reduction of catalysts by CO (indicating surface oxidizing ability). A similarly good relationship for oxidative dehydrogenation of cyclohexene (bulk type I1 reaction) and the rate of reduction of catalysts by H2 (bulk oxidizing ability) has also been found. However, the

~1

tb

0

1

Na2-2,3,4

2

FIG. Correlations between catalytic activity and oxidizing ability for (a) oxidation a acetaldehyde (surface reaction) and (b) oxidative dehydrogenation- of cyclohexene (bulk-type I1 reaction). (From Ref. 327.) r(aldehyde) and r(hexene) show the rates of catalytic oxidation of acetaldehyde and oxidative dehydrogenation of cyclohexene, respectively. (From Ref. 337.) r(C0) is the rate of reduction of catalysts by CO; r(H2) is the rate of reduction of catalysts by H2. M, denotes M,Hj-xPMo1204~.Na2-I, 2, 3, and 4 are Na2HPMoI2O4,,of different lots, of which the surface areas are 2.8, 2.2, 1.7, and 1.2 m2 g-', respectively.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

215

correlation is poor if the rate of the surface reaction is plotted against the bulk oxidizing ability. The following are typical reactions that have already been found to be described by either of the two types of catalysis by H3PMoI2O4"and its alkali salts: Surface type: Oxidation of CO, acetaldehyde, and methacrolein. Bulk type II Oxidation of H2, oxidative dehydrogenation of cyclohexene, isobutyric acid.

Since the classification is essentially based on rates of catalytic reactions relative to rates of difhsion of redox carriers, there are oxidation reactions that are intermediate between the two limiting cases. We note that neither the molecular size nor the polarity of reactant molecules is the principal characteristic determining the type of catalysis. Although oxide ions migrate rapidly in the bulk, bulk type 11 catalysis is not observed for oxidation catalyzed by Bi-Mo oxides. In this case the rate-limiting step is a surface reaction. It is noteworthy that in the two industrial processes to produce methacrylic acid, both involving catalysis by H ~ P M O and ~ ~ its O alkali ~ ~ salts, one involves bulk type I1 catalysis and the other, surface type catalysis, as described in the following section.

I

CH,=C-CHO

B.

SurfacewF

OXIDATION OF ALDEHYDES

Methacrylic acid has been used for the synthesis of poly(methy1 methacrylate). It has been synthesized industrially via a reaction of acetone with hydrogen cyanide (12, 17, 330, 331). However, the process produces ammonium bisulfate and uses the toxic hydrogen cyanide. Recently, an alternative, a twostep oxidation of isobutylene, has been developed. The first step is the oxidation of isobutylene to methacrolein, and the second is the oxidation of methacrolein to methacrylic acid: CH3 I CH~=C-CHJ

CH3 I + CHz=C-CHO

+

CH3 I CH2=C-COOH

(36)

1. Mechanism and Roles of Acidity and Oxidizing Ability In the second step of Eq. (36). methacrolein is oxidized by heteropoly catalysts, of which the active component is essentially H3PMo12040 or its

216

TOSHIO OKUHARA, NORITAKA MIZUNO, A N D MAKOTO MISONO

equivalent. Therefore, the oxidation mechanism was investigated with H3PMo12040and its salts. It was concluded that the reaction proceeds by a redox mechanism according to the reaction scheme showin in Eq. (37): RCHO F & RCH(OMo)(OM)

RCOOMo

-

RCOOH (M = H or Mo)

(37)

Equation (37) was derived on basis of the following experimental facts: (i) A fair correlation between the rate of catalytic oxidation and the oxidizing ability of the catalyst measured by the reduction with CO was observed (Fig. 61) (17, 332). This result shows that the rate-limiting step is part of the second reaction, that is, the oxidative dehydrogenation of the intermediate. The first reaction requires acidic sites, as nonacidic catalysts were inactive. But the rate-determining step is inferred not to be part of the first reaction because there was no parallel between the acidity and the rate (332). (ii) Catalyst oxygen is involved in the reaction, since the reaction continued and the selectivity remained essentially the same for a prolonged period after the supply of oxygen was stopped, with the catalytic reaction proceeding in the stationary state (17).

Na3 Cs3 C s l

N a l Cs2.86 Na2 \ / H

. C

0

0

1

2

3

lo6 x Rate of reduction by CO / mol rnin-lg-l FIG.6 I. Correlation between the conversion of methacrolein and the rate of reduction of catalyst by CO. M, denotes MxH3--xPMo12040.(From Ref. 332.)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

217

oxid

CO,

acid SCHEME 8

(iii) Rapid and direct exchanges of isotopic oxygen between either two of the species, methacrolein, water, and M,H3- rPMolOOOO(M = Na, Cs; x = 0-3.15), were confirmed by elaborate pulse-mass spectrometry experiments ( I 7, 332). In addition, the rate of oxygen exchange between methacrolein and the polyanion increased with the number of Brsnsted acid sites. This result indicates that the first reaction in Eq. (37) is catalyzed by Brsnsted acids. (iv) The presence of water vapor had a strong influence on the rate and selectivity. This effect was reversible and instantaneous (103). The roles of acidity and oxidizing ability were investigated in detail for the oxidation of acetaldehyde catalyzed by various salts of H3PMoI2O4(); the data were interpreted on the basis of the reaction scheme deduced (Scheme 8 ) (333). The rate of each reaction in the scheme was estimated from the rates of oxidations of acetaldehyde and acetic acid, and compared with the acidities and oxidizing abilities of the catalyst surfaces. The comparisons indicate that the oxidizing ability influences mainly the reactions of CH3CHO -+ CH3COOH and CHJCHO CO,, and the acidity accelerates CH3COOH CH30H + CO and CHJOH + CHjCOOH CH3COOCH3. Ai (334a) investigated the effect of countercation and additives for the oxidation of methacrolein. It was found that the catalytic activity of C S ~ H P M O ~is ~enhanced O ~ ~ by the addition of oxoanions such as BO:- and PO:-. The effect was explained on the basis of acidic and basic properties of the heteropoly compounds. Ai el al. (334h) also investigated the oxidation of crotonaldehyde to furan catalyzed by H3PMo12040 and its salts. The rate increased markedly with an increase in the steam concentration but was almost independent of the partial pressures of oxygen and crotonaldehyde. The reaction catalyzed by the Cs salt was faster than that catalyzed by the parent acid. This result was explained as follows. The addition of a basic species such as Cs ion reduces the acidity of the catalyst and the affinity for furan. which is basic. The weaker interaction facilitates the furan desorption, which is assumed to be ratedetermining. -+

+

+

2. Catalysts In industry, heteropoly catalysts of H3-,Cs,PMo12- ,,V,,04(,(2 < x < 3 ; 0 < v < 2) are used to oxidize methacrolein into methacrylic acid with 60-70%

2 18

TOSHIO OKUHARA. NORITAKA MIZUNO, A N D MAKOTO MISONO

yields (12, 335). Formation of Cs salts markedly increases the surface area and thermal stability of the catalysts, but the stoichiometric Cs salt is not catalytically active, probably because of the absence of acidity. Thus acidic salts that are nearly stoichiometric are preferred. The acidic salts are often mixtures or solid solutions of the acid form and salts. It is assumed in some cases that the acidic Cs or K salts are the acid form epitaxially formed as thin films on the . 337). surface of Cs or K salts ( 4 6 ~336, It has been claimed (335) that preparation of an acid form catalyst by the thermal decomposition of pyridinium salts results in a cubic crystal structure and increases the surface area and pore volume. For example, the surface area of H4PMoIIV040increases from 1.O to 5.3 m2g- by the creation of macropores having radii of 103-104 8. As a result of macropore formation, higher yields are obtained (Fig. 62). The formation of acetic acid, CO and CO2 at high conversion is suppressed by treatment of the catalyst with pyridine. The application of this method to acidic Cs salts further improves the activity and selectivity.

'

c.

DEHYDROGENATION OF ISOBUTYRIC ACID

This reaction is another possible route for the production of methacrylic acid, since isobutyric acid can be obtained by an 0x0 process from propene and CO. Heteropoly compounds and iron phosphates are so far the most efficient catalysts for the reaction. The favorable role of the presence of an a-methyl group is remarkable for oxidative dehydrogenation, as the heteropoly compounds are not good catalysts for the dehydrogenation of propionic acid (338, 339).

t 0 . 0 ' ' 0

' 20

'

' ' 40

"

60

I

'

80

100

Conversion of methacrolein / % and treated (0) FIG.62. Oxidation of methacrolein catalyzed by H4PMoIIV040.untreated (0) with pyridine. Catalyst, 10 cm3; reaction temperature, 553 K; SV, 1000 h- I. Reactant composition: methacrolein 2%. oxygen 6%0, water 20%. (From Ref. 335.)

219

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

The effects of countercations as well as heteroatoms have been investigated by Akimoto et al. (340). The catalytic activity increased in the order H < Li < Na < Rb < Cs at 573 K, and the reverse order was found at 523 K. The authors concluded that the reaction proceeds by a redox cycle involving heteropoly catalysts and that the reduction of the catalyst is rate-determining at 573 K, whereas the reoxidation is rate-determining at 523 K. It was supposed that oxygen atoms bonded to Mo become more reactive as the electronegativity of the countercation decreases. The authors hrther suggested that a metal ion having a high oxidation potential, such as Pd or Ag, acts as an electron reservoir and accelerates the reaction. It was reported that Cu2+ is also an effective cation (341). Recently, Lee et al. (342) have compared the catalytic performance with that of C S , H ~ - . ~ P M O ~ ~ ~ , .( xV, ,y, = O ~W) ~ and found that Cs2.75H1.2sPMoIIVO40 was effective for the oxidative dehydrogenation of isobutyric acid. For example, the selectivity to methacrylic acid was 78% at 97% conversion and 623 K. Herve et al. (284) demonstrated that the elimination of VS+ from H4PMoI IVO40 took place during the reaction. Notwithstanding the extensive investigations, a consistent explanation is lacking; further investigations are necessary. The heteropoly catalyst is deactivated during a prolonged reaction period by the loss of Mo to form volatile Mo-containing species via the interaction of isobutyric acid andor methacrylic acid with the catalyst (343).The deactivation was suppressed by presaturating the feed by flow over a bed of Moo3 (343). In the dehydrogenation of isobutyric acid, the by-products in addition to CO and C 0 2 are propylene and acetone. Two reaction mechanisms were proposed (340, 341) and the latter is shown in Scheme 9 (340). The formation of methacrylic acid and acetone involves a common intermediate: The El elimination of a proton from I yields the methacrylic acid while a nucleophilic SN1 attack of oxide ion produces COZ and acetone (344). On the other hand,

CH3 \CH-COOH CH3

'

CH3

I

CH2=C -COOH

[

-

CH3 \ cH3, CH COOH]

H'

C3H6'

co

ads

<[

a

3

cH3,

\

C'-COOH

I SCHEME 9

]

@-

CH3COCH3 t

ads

CO or COz

220

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

propylene forms by an acid-catalyzed reaction. Therefore, the acid-basic and redox properties of the catalysts are both important. A similar competitive action of acidity and oxidizing ability was demonstrated for the reaction of methanol and ethanol (345, 110) and the blending of H3PMo12040with polysulfone greatly increased the oxidation performance.

D. OXIDATION OF PARAFFINS Oxyfunctionalization of lower paraffins such as methane, ethane, propane, and butanes has recently attracted much attention (5, 330, 331, 347-350). Oxidation of n-butane to maleic anhydride is an industrial example (346, 351). The oxidation of propane and isobutane with heteropoly catalysts was first reported in 1979 (352). Ai (324a) and Centi et al. (324b, 324c) reported that heteropoly compounds catalyze the oxidation of lower paraffins, especially propane, isobutane, and pentane (324). The oxidation of propane into acrylic acid in the presence of heteropoly catalysts prepared from H3PMo12040 and antimony pentachloride gave rather low conversion and selectivity [lo and 19%, respectively (2% yield)] (352). Recently, a yield of ca. 9% was obtained with H5PV2Molo040(353). The addition of Cr ion also enhanced the catalytic performance (354). Heteropoly catalysts have significant activities for the oxidation of isobutane into methacrolein and methacrylic acid. The yield increased up to 6% by vanadium substitution or salt formation, as follows. With C S ~ , ~ N ~ O .+xPVxMol O ~ H O . 2~- x~ 0 4 ~ , the highest conversion and selectivity were observed at x = 1 (355). Increases in the reaction temperature to 613 K led to increased yields, up to 9.0%. A similar increase in the yield resulted from the substitution of As for P as a heteroatom or from the addition of various transition metals (106, 356). When the oxidation of n-pentane was camed out in the presence of H3+xPVxMo12-x040 (x = 0-3), the only product observed in addition to oxides of carbon was maleic anhydride (357). The activity and selectivity increases when one Mo was replaced by a V atom.

E. HETEROGENEOUS LIQUIDPHASEOXIDATION REACTIONS Na5PV2M01f1040 supported on active carbon is active for oxidative dehydrogenation of benzylic alcohols and amines without overoxidation of benzaldehyde and benzylamine in the liquid phase (357). The suppression of the overoxidation may be due to the lower oxidizing ability of Na5PV2Molo04fl relative to its acid form.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

22 1

Soeda et al. (358) used Pd-supported heteropoly compounds for a heterogeneous Wacker-type reaction and found that P ~ / C S ~ . ~ H ~ . ~ P W was ~MO~O~~ active for oxidation of cyclohexene to produce cyclohexanone and cyclohexenone. The active sites are assumed to be Pd2+ and Pdo for the two products, respectively. Homogeneous Wacker-type reactions are described in Section XI.

X.

Fine Chemicals Synthesis

The use of heteropolyacids as catalysts for fine organic synthetic processes is developing. Syntheses of antioxidants, medicinal preparations, vitamins, biologically active substances, etc., have been reported and some are already applied in practice (10, 160). Alkylation of p-cresol with isobutylene to give tert-butyl-4-methyl phenol [Eq. (38)] is the first step in the synthesis of agidol-2, an antioxidant for polymeric materials (160). OH

1

2

Me

Me

Me Me agidol-2

The activity of H3PW12040is greater by four orders of magnitude than that of sulfuric acid. The use of heteropolyacids instead of H2S04 not only eliminated the formation of waste water containing toxic cresol sulfate, but also reduced the loss of p-cresol during neutralization and washing. H3PWI2O4,,and H4SiW12040are active for transalkylation of 2,6-di-tert-butyl 4-R-phenol (R = H, alkyl, aryl, etc.) [Eq. (39)] (359, 360). OH

OH

R

R

These heteropolyacids are superior to Nafion and give 4-R-phenol with 92-98% yields at 373-413 K by use of toluene or p-xylene for tert-butyl group acceptors. It is claimed that these catalysts can be readily separated from the reaction mixture and reused. Heteropolyacids are also active for the conversion of sugar derivatives (360). HnXM12040(X = P, Si, M = W, Mo) are much more active than conventional catalysts such as p-toluenesulfonic acid and ZnClz for nucleophilic substitution

222

TOSHIO OKUHARA, NORITAKA MIZUNO. AND MAKOTO MISONO

of acetylated aldohexose to give the corresponding glycosides [Eq. (40)], which are used as biodegradable surfactants. AcO AcO

OAc

+ ROH(RSH)

a

-

AcO AcO

AcO

AcO

(40)

OR (SR)

The reaction is performed in a homogeneous liquid phase by use of less than 2 mol% of heteropolyacids with respect to acetylated aldohexose, and 70-90%0 yields and 60-98% mol% /3-anomer selectivity are obtained. The reaction of L-sorbose with acetone, a step in the synthesis of L-ascorbic acid (vitamin C), takes place in acetone solution [Eq. (41)] (361).

+ 2Me2CO

-

JMe H

2

7

CH20H P -I-2H20 (41)

0

The yield of diacetone-L-sorbose in the presence of 0.1435% of H3PW12040 or H4SiW12040is 85% under reflux. Heteropolyacids catalyze the condensation reaction in the synthesis of vitamin E (a-tocopherol acetate) as shown in Eq. (42) (362). Me

Me

OH

Me

+

*-

-

-H20

Me

HO

Ac~O

Me

HPA

(42)

a-tocopherol Me L

vitamin E

With 1 % of H3PW12040relative to 2,3,5-trimethylhydroquinone, the yield reaches about 92%, and vitamin E in the isolated product is 95%. H3PW12040/ S O z is less active than bulk H3PW12040, probably because of the decrease in the acid strength that results from supporting the catalyst. In the case of ZnC12 or H2S04, a large amount of catalyst is required.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

223

H3PW12040 and H4SiWI2O40are also active for the condensation of isophytol which is a key step in the synand I -acetoxy-4-hydroxy-2-methylnaphthalene, thesis of vitamin K I (2-methyl-3-phytyl-1 ,Cnaphthoquinone) [Eq. (43)]; heteropolyacids are approximately 50 times more active than ZnC12 (362).

fi+ \

p *

\

OH

-OH-

t

-H20

(431

0

02

H vitamin K1

The esterification of steroids is a step in the synthesis of modified hormones. HzPW12040and 25% H3PWI2O40/SiO2 (1-9%) in CH3CN (at 3 13-353 K) give a quantitative yield (363).The activities of the heteropolyacids are close to that of HC104 and much greater than that of 5-sulfosalicylic acid.

XI.

Hybrid Catalysts

An attractive research target is the design of heteropoly catalysts complexed with organometallics. The use of heteropoly catalysts in combination with noble metals is also promising. Research in these directions has attracted much attention recently. A.

MONO-TRANSITION-METAL-ION-SUBSTITUTED HETEROPOLYANIONS AS INORGANIC SYNZYMES

A recent development concerns the use of polyanions of the type [XMI10j9M'(OHz)]"-. In this type, the M' atom easily becomes coordinatively unsaturated by dehydration (255). The resulting dehydrated anion, [XM I 1039M']n-, can be considered an inorganic metalloporphyrin analog (322, 364, 365). Oxidation catalysis by these polyanions is described in Sections VIIl and IX. Here, the catalytic performance and stability are compared with that of metalloporphyrin. In the majority of the homogeneous oxidations of hydrocarbons by oxometalbased catalysts (including metalloporphyrins), there is appreciable decomposition of catalyst ligands by oxidation, and hence appreciable loss in activity after a few turnovers. A similar degradation of organic ligands, often hydrophobic long-chain carboxylates, is also observed in industrial processes of hydrocarbon

224

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

autoxidation. These disadvantages of organic ligands, can be overcome by the use of heteropolyanions. The rates, selectivities, and stabilities of transition-metal-substituted polyoxometalates in olefin epoxidation are compared with those of metalloporphyrins, Schiff base complexes, and triflate salts, as follows (320b): Activity: PWI ICO(II)O& > PWIIMn(II)O:9 2 Fe(III)(TDCPP)Cl > Fe(III)(TPP)Cl > M(OTQ2. Selectiviry to epoxide: PWIICO(II)O:~= PWIIMn(II)O:~> M(OTQ2 > Fe( III)(TDCPP)Cl > Fe( III)(TPP)Cl. Stability: TMSP > Fe(III)(TDCPP)Cl > Fe(III)(TPP)Cl > M(OTQ2 (Abbreviations: TDCPP, tetrakis-2,6-dichlorophenylporphyrin;TPP, tetraphenylporphyrin; OTf, triflate ion).

TMSP is also active for paraffin hydroxylation. Figure 63 is a summary indicating the number of turnovers of a variety of 0x0 transfer catalysts before catalyst decomposition (320b). When tert-butyl hydroperoxide is used as an

M(TDCPP)

-

FIG. 63. Stability of transition-metal-substituted polyoxometalate for 0x0 transfer to hydrocarbons. Values on the ordinate indicate numbers of turnovers for paraffin oxidation. shows the ranges of the numbers of turnovers. (From Ref. 3206.)

225

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

oxygen donor, the turnover of TMSP is higher than those of metalloporphyrins, Schiff base complexes, triflate or nitrate salts, or other soluble transition-metal . Br)'complexes. Recently, Mansuy et al. (366) reported that PzW17061(Mn3+ was oxidation-resistant and the most active for the epoxidation of cyclooctene with PhIO among P2W17061(Mn+ *Br)(I1-"- (Mn3+,Fe3+, Co2+,Ni2+, C u 2 + ) . Br)8- also catalyzes the oxygenation of cyclocatalysts. P2W17061(Mn3f hexane, adamantane, and heptane; the hydroxylation of naphthalene; etc. P2W17061(Mn3+ * Br)' shows similar regioselectivities to those of Mnporphyrin. B. METALION-

OR

METALSALT-HETEROPOLYANION COMPLEXES

1. Simple Combinations

Oxidation reactions catalyzed by heteropoly compounds alone are described in Section VIII. Oxidation reactions catalyzed by transition-metal ions in combination with heteropolyanions are shown in Table XXXII (288, 292, 368-375). TABLE XXXIl Reactions Cutulyzrd bv Combinations of Hetrropolyanions with Metal ions Reaction RCH=CH2 GHt, + 0

2

-

PhNOz

+0 2

-

RCHOCH,

-(Ct,H5)2

+ 3CO + CHIOH PhNHC02CHI + 2C02

Catalyst

Ref.

PMolz-,,V,O$+"'~ + PdS04

292,368

P M O ~ ~ - , , V . O $ ~ "+ ' - Pd(OCOCH3)2

292

+ PdC12 H~PV~MOIUO~"

369

CH,OH (or CH30CHI) + CO CHJCO~CH~

370

--+

-OH

Li4SiMoI2OW+ RhCI(PPh,)I

+ Hz --OH

371

3 72

0 + H 2

-0

3 73

0

CH4

+0 2

-

CH,OH

+ CHiCl

3 74

Pt salt

+ NaRHPMo6Vh040

3 75

226

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

FIG.64. Redox cycle for the oxidation of ethylene to acetaldehyde. V5+ (oxidized heteropolyanion) represents vanadium in the oxidized heteropolyanion. (From Ref. 368.)

The Wacker process is employed industrially for converting ethylene into acetaldehyde. The reaction is usually carried out with a PdCl2-CuCl2-HC1 catalyst in an aqueous solution at 370-403 K. This process has several drawbacks, such as formation of chlorine-containing by-products and extensive corrosion of the reaction vessel. The development of chloride-free oxidants to replace CuCl2 for oxidation of Pdo has long been desired. Heteropolyanions can be used as reoxidizing reagents for Pdo in place of CuCl2, as shown in Fig. 64. The idea was first reported by Matveev et al. (292) for the oxidation of ethylene (x = 1-4). 1-Octene is also with chloride-free PdS04 and H3+xPVxMo12-x040 converted into 2-octanone with a selectivity of 95% at 333-353 K in the presence of PdS04 and H9PMo6V6040 (367). lzumi et al. (367) found that the rate-determining step is the reoxidation of Pdo to Pd2+ for the oxidation of 1-butene and that PMo6W60& is an effective heteropolyanion. Recently, researchers at Catalytica proposed a new technology for ethylene oxidation (368). Typical compositions are aqueous ca. 0.1 mM Pd2+, 5-25 mM (preferably x = 2-3). The C1-, and ca. 0.30 M Na,H~3+x-y~PVxMo12-x040 Pd" and chloride concentrations are only 1/100 those in the oridinary Wacker system. The solutions at pH 0-1 result in high reaction rates and stability of Pd2+, as shown in Fig. 65. The stability of Pd2+ is further improved by the presence of chloride ion in a concentration of about 0.01 M. In this system, the phosphomolybdate serves two functions in the Pdo reoxidation: (1) It solubilizes high concentrations of V5+ in aqueous solution and (2) it accelerates the reoxidation of V4+ by dioxygen. Kinetics (the reaction is first-order in Pd2+ and in ethylene concentrations and zero-order in Vs+ concentration) shows that the oxidation of ethylene to produce acetaldehyde is rate-determining. lzumi et al. (369) found that Keggin-type heteropolyanions containing Mo or V show a promoting effect in the reductive carbonylation of nitrobenzene by PdC12 to form methyl N-phenylcarbamate in the presence of methanol (Eq. 44): PhN02

+ 3CO + CH3OH

+

PhNHCOOCH3

+ 2C02

(44)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

227

y in Na,H5.YPMo10V2040

PH FIG. 65. Dependence of turnover frequency of palladium on pH. Ethylene, l50psi; Na,.HS -,.PMoloV2040, 0.30 M ; Pd(OAc)z, 0.10 mM; reaction temperature, 393 K. (From Ref. 3/18.)

Figure 66 shows the catalytic activities of H3 +,PMo~~-,V.rO~o-modified PdClz and the highest reduction potentials of the heteropolyanions. The more reducible heteropolyacids show greater promoting effects, and the highest activity was 100 I

0F -0.0

1.0

2.0

3.0

4.0

12.0

in ~3+xPM012-xVx040 FIG.66. Catalytic activities of H, +,PMo12-,V,Odo-modified PdClz and the maximum reduction potential of heteropolyacids: PhNOz, 0.01 mol; CO, 41 atm; reaction time, 3 h; reaction temperature. 423 K; 12-vanadophosphoric acid, Li2H5PV12038. SCE = saturated calomel electrode. (From Ref. 369.)

228

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO TABLE XXXlll Vapor-Phuse Curbonylation oJMethanol or Dimeihvl Ether to Methyl Acetate Catulvred hy Metal-Exchanged HjPW,2Od0 Supported on SO, (498 K) from ReJ 370) ~

~

Product yield (%) CataI yst lrPW 12040 RhPWl204n HPdPW12040 HMnPW I2040 HCoPWI204o HNiPW 12040 FePW 1 2 0 4 0

MeOH

DME

MeC(0)OMe

8 17 0 0 5 7 7

52 49 92 96 92 90 92

40 34 8 4 3 3 I

obtained for H4PMoIIV040.It is proposed that the rate enhancement is brought about by the coordination of Pd2+ on partially reduced heteropolyanion, which is regarded as a macroligand. Rh- or Ir-exchanged heteropolyacids supported on Si02 catalyze vapor-phase carbonylation of methanol or dimethyl ether to give methyl acetate at 498 K and 1 atm (370). As shown in Table XXXIII, with RhPW12O4dSiO2,the yield of methyl acetate is 44%. At this temperature, the yield of methyl acetate dropped rapidly to < 1% during 6 h of reaction time accompanied by increases in the yields of dimethyl ether, methanol, and hydrocarbons. catalyst modified by Li4SiW12040shows sterically controlled A shape selectivity in homogeneous hydrogenation of olefins ( 371). The presence of SiWl2O:O retards the reduction of sterically more crowded 1,2- or 1,l-disubstituted ethylene, although the overall catalytic activity for hydrogenation is lower. Probably SiW120:0 exists in close proximity to the coordination sphere of the Rh complex and hinders the access of bulky olefins. 2. Heteropolyanion-Supported Metals

Heteropolyanion-supported metals are not simple combinations as described above, but instead are complexes of heteropolyanions with organometallic comor pounds. Siedle et al. (372) have shown that [L2Rh(CO)(CH3CN)]4[SiW12040], catalyzes olefin the 16-electron rhodium complex in [L3Rh(CO)]4[SiWI 2040], hydroformylations when suspended in toluene (L = PPh3) as a solid catalyst. However, EXAFS data show that the Rh atom was not bound directly to the polyoxoanion. This result is reasonable because the classical heteropolyanions such as SiW120:0 [(Si04)4-(W12036)o]have low surface charge densities or basicities (in fact, they are C104--like) and cannot coordinate with cations such as (CSMes)Rh2+or CpTi3 + . Therefore, the organometallic cations need to be

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

229

supported on (or bound to) reduced heteropolyanions or more negative polyoxoanions such as SiW9M30:i and P2Wl5M30:; (M = V 5 + ,Nb"). The latter has three full units of negative surface charge that enable tight and covalent bonding of transition metals. Finke et al. (376) synthesized highly negative P2WI 5Nb30:, having a high negative surface charge density. They demonstrated that there is direct bonding in the P2W15Nb30:2 -supported Ir catalyst, [ ( ~ I - C , H ~ ) ~ N ] ~1,5-COD)lr N ~ ~ [ ( * P2W15Nb3062]( 3 7 9 , and that this is active for both hydrogenation (373) and oxygenation (374) of cyclohexene. The turnover frequency was found to be 2.9 h- I at 3 1 1 K in CH2C12,which is 100-fold greater than that observed with the parent compound, [( 1 ,5-COD)IrC1I2. C. METAL-HETEROPOLY BIFUNCTIONAL CATALYSTS Pd1,5PW12040supported on Si02 catalyzes the skeletal isomerization of C5 and C6 paraffins (378-380). The presence of H2 is necessary to maintain the high activity in the stationary state. This requirement indicates the bifunctional character of the catalysis, with protons being generated by the reduction of Pd2+, as shown in Eq. (45). Pd2+ + Hz -+ Pd" + 2H'

(45)

This is the same mechanism as that for the generation of H + with Ag3PW12040 (Section 111). However, Pd1,5PW12040 is active for esterification and MTBE synthesis even in the absence of H2 (378).Therefore, it is concluded that this catalyst is not as simple as Ag3PW12040. The catalytic activity of Pd.rH3-.rPW12040/Si02for hexane isomerization is plotted as a function of x in Fig. 67. The addition of a 100

80

20

0

0.5

1.o

' 0 1.5

x in P d x H 3 - 2 ~ ~ ~ 1 2 ~ 4 0 FIG. 67. Effect of Pd content on activities for isomerization of hexane catalyzed by Pd,H3-~,PW12OW. (From Ref. 378.)

230

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

small amount of Pd (x = 0.05) greatly enhances the activity, but further increases in x improve the activity only slightly. The combination of Pd and the heteropolyacid is effective for the isomerization of paraffins, as heteropolyacids are very effective components in the bifunctional catalysts. The combination of Pt or Pd with CS~.~HO.SPWIZO~O (Cs2.5) is also very effective for the isomerization of n-butane to isobutane (381).The reaction rate and selectivity for conversion to isobutane are summarized in Table XXXIV (381,382).The activity in the presence of H2 changed little with time. Pt- and Pd-02.5 show very high selectivities (94-96%) relative to those of Pt/SO:-/ Zr02 (47%) and Pt/HZSM-5 (34%),whereas the activities of Pt- and Pd-Cs2.5 for the formation of isobutane are comparable to those of Pt/HZSM-5 and Pt/SO:-/ZrOz. Pt-Cs2.5 catalyzes the reaction even at 473 K and 0.05 atm of H2 *

The important roles of H2 and Pt are demonstrated in Fig. 68, in which the conversions for Pt-CsZS(A) and Cs2.5 are compared. It is clear that Pt and H2 enhance the activity and selectivity. In the absence of H2, the initial activity of Pt-Cs2.5 is high, but the stationary-state conversion (at 5 h) is very low. H2 suppresses the deactivation by the hydrogenation of coke or coke precursors, resulting in a high steady-state activity. The high selectivity to isobutane observed with Pt-Cs2.5 is brought about by the unique roles of protons, which greatly suppress hydrogenolysis (382). When TABLE XXXIV Activity and Selectivit), for Skeletal lsomerization of n-Butane in the Presence of H2 and MetalPromoted Catalysts at 573 K (3R1, 382) Selectivity' (mol%) Conversion

lon X Rateh

i-C4

CI

C2

C3

C4=

C5

In the presence of 0.5 atm of Hz Pt-Cs2.5 25' 34" Pd-Cs2 .5 Pt-SO:-/Zr02 65" Pt-H-ZSM-5 51'

7.9 10.9 10.4 12.0

93.9 95.6 47.3 34.0

1.4 0.5 6.0 18.8

2.4 0.5 11.6 24.9

1.8 2.0 30.3 21.4

0 0 0 0

0.5 I .4 4.8 0.9

In the presence of 0.05 atm of H2 Pt-Cs2.5 20.5" Pd-(32.5 12.9" Pt-SO: - /zr02 4.8" Pt-H-ZSM-5 70.4'

6.2 3.4 1.2 8.8

88.3 78.4 72.6 16.4

0.3 0.3 1.0 3.9

0.5 0.4 2.3 6.5

5.8 11.3 16.6 66.8

0.7 1.8 0 0

4.4 7.8 7.5 6.4

Catalyst'

P O )

" Csz 5 indicates Csz 5Ho 5PW 1 2 0 4 ~ . The rate for isobutane formation: mol g- I s I . i-C4, C I , C2. Cp, C4=, and C5 indicate isobutane, methane, ethane, propane, butenes, and pentanes. ' M / F = 41 g h mol-I. 'MIF = 18 g h mol-I, where M is the catalyst mass and F is the total flow rate. ~

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

2

. E

100

%c

s

I

I

l

.-L?

23 1

Pt-Cs2.5(A)

I

, f -(Pt-Cs2.5 in N2)

20 0

0

1

2

3

4

5

Time/h

(0. A) Fir;. 68. Time course of n-butane isomerization catalyzed by 1 % Pt-CszSHOcPWIZ040 and CszSH05PWIz040 (0,A ) at 573 K. +Butane: Hz : Nz = 0.05 : 0.5 : 0.45 ( M ) . I % Pt-Csz s(A) was prepared from H2PtCI6. (From Ref. 382.)

a small amount of H3PW12040 was impregnated onto Pt-Cs3PW12040 (which exhibited high activity for hydrogenolysis), the isomerization of n-butane proceeded very selectively, as the hydrogenolysis activity of Pt-Cs3PW 12040 was almost suppressed. Scheme 10 is tentatively proposed (382). In the absence of protons, the intermediates (I) react with hydrogen to cause hydrogenolysis, whereas in the presence of protons, the intermediates (1) interact with protons on Pt or at the interface between Pt and Cs2.5 to form carbenium ions (11) which are the intermediates for the skeletal isomerization. D. INTERCALATED POLYANIONS An isopoly cation, [A11304(OH)24(OH2)Iz]7+, has the Keggin structure and can be intercalated into the layer structure of montmorillonite clay,

-

C1, C2, C3 (without H+)

butenes

H+ (1)

SCHEME 10

i-C4 (with H+)

232

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MlSONO

[Si4(A15~3Mgl12)(0H)2010]1/3 - . Na;?. By calcination, microporous solid catalysts with pore sizes of about 8 A have been prepared (383).This catalyst is active for cumene cracking (384), conversion of methanol (385),and alkylation (386); and it can be used as a support for metal catalysts (387, 388). Ru supported on A1203 intercalating montmorillonite is a catalyst for the production of c6412 hydrocarbons in CO hydrogenation (388). Heteropoly anions with Keggin structures are effective reagents for the pillaring of layered double hydroxide (389, 390). [Zn2Al(OH)6]N03 . 2H20 (abbreviated as Zn2Al) undergoes facile and complete intercalation by an ionexchange reaction with H2WI20:0, SiWl or SiV3W90&. On the other hand, no reaction is observed for Keggin anions such as PW120$ and SiW 120:,, probably because they have smaller negative charges. XRD patterns for Zn2Al-SiW I 1 0 3 9 and Zn2Al-SiV3W9040 include several 001 harmonics corresponding to a basal spacing at 14.5 A. If the thickness of the double hydroxide layer is taken to be 4.7 A, the gallery height is 9.8 A, in accordance with the expected size of the Keggin anions. IR spectra of Zn2A1-SiV3W9040 confirm that the Keggin structure of SiV3W90:; is retained. The surface areas of Zn2AI-SiWI1039 and Zn2AlSiV3W9040were found to be 98 and 113 m2 g-I, respectively, whereas that of Zn2Al was only 26 m2 g - I . Besides heteropolyanions, isopolyanions can be intercalated into the layer of various clays. Zn2Al-V10028has been synthesized from [Zn2Al(OH)6]Cl. 2H20 and [NH&,(VIOO~S)] * 6H2O (391). It has been pointed out that these pillared intercalates are intrinsically difficult to synthesize in highly crystalline form because the layered hosts are basic, whereas most heteropolyacids are acidic and tend to decompose. Narita et al. (392) tried direct synthesis of a heteropolyanion-pillared layered double hydroxide by a coprecipitation reaction of Zn2+ and AI3+ ions in the presence of a moderately acidic lacunary Keggin anion, a-SiW I 10;;. XRD of the product showed a basal spacing of 14.6 A, which corres onds to a gallery height of 9.9 A. The surface area was found to be 97 m2 g - q which is three times that of the layered double hydroxide. Drezdzon (393) reported a new route to the synthesis of pillared clay intercalates based on the ion-exchange reaction of isopolyanions with a clay that had been intercalated by a large organic anion. By this method, Mgl2Al6(0H)36(Mo7O24). xH20 and Mg12A16(OH)36(V10028)* xH2O were synthesized; they were claimed to be stable at temperatures up to 773 K. Analogous reaction of Mg3Al-adipate with the lacunary Keggin species SiWl 10;; yielded the corresponding pillared product as a single crystalline phase (basal plane spacing = 14.8 A), having a surface area of 155 m2 g-I (394). The intercalated compounds of hydrotalcite [Mg2&110(OH)66.~] with Mo70;; or W 120:y- catalyze the shape-selective epoxidation of olefins; epoxidation of 2-hexene was favored over theat of cyclohexene (395).

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

XII.

233

Photocatalysis and Electrocatalysis

Various polyoxometalates can be reduced electrochemically and reversibly by several electrons at modest potentials (Section VILA), and these properties are exploited in photocatalysis and electrocatalysis. In both cases, redox properties of heteropolyanions (Fig. 49) and the organic reactants (Table XXXV) are the principal properties that control the catalytic performance. The selection of the electrode is also important in electrocatalysis. Photocatalysis by hereopolyanions has been reported extensively, but there are only a few reports of electrocatalysis by these compounds.

TABLE XXXV Ranges qf Redox Potential of Organic Compounds Ranges of electric potential, V (vs. SCE”) Reduction Azo compounds Aldehydes Activated esters Activated alkenes Diazo compounds Sulfones Nitro compounds Halogenated compounds Aromatics Hydroxyl amines Oxidation Azo compounds Amides Alcohols Carboxylates Ketones Diazo compounds Sulfides Phenols Aromatic amines Aromatic hydrocarbons

+0.05 -1.5 - 1.0 -1.5-

- -0.1 - -2.0

- -2.0 -3.5 -0.3 - - 1.0 - 1.5 - -2.5 -0.5 - - 1.5 -0.3 - -3.0 - 1.5 - -3.0 -0.5 - - 1.5 + 1.5 - +2.0

+ 1.5 - +2.5

+IS - +2.0 +2.0

- +2.5

+ 1.0 - +2.0 + 1.0 - +2.0

- + 1.5

f0.5 +0.5 +1.0-

- + 1.0 +2.5 + 1.0 - +2.5

Saturated calomel electrode. Source: D. K. Kyriacou, Basics of EIectroorganic Synthesis, John Wiley & Sons, New York (1981).

234

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

A. PHOTOCATALYSIS

Reports of the photosensitivity of polyoxometalates appeared as early as 19 I6 (396). Early systematic work on photocatalysis was done by Yamase et al. ( 3 9 7 4 0 2 ) .For example, methanol is photooxidized to give formaldehyde in the presence of Mo70:, according to Eq. (46). Mo70;;

+ CH3OH 2 MoI40&- + HCHO + 20H

~

(46) to form

This reaction involves electron transfer from water to Mo70:4 M o , O ~ ~ O Hand ~ - OH radicals. Papaconstantinou et al. (403, 404) applied heteropolyacids to the photooxidation of alcohols and indicated the potential importance of such chemistry [Eq. (WI. RlRzCHOH

+ 4 0 2 2 RlRzCO + HzO

(47)

1. Activity of Isopoly and Heteropoly Catalysts

It is known that molecules in photoexcited states are stronger oxidants and reductants than those in the ground states. The potentials of the excited states of polyoxometalates can be estimated by adding the energy of 0,O-transitions to the ground-state reduction potentials (257, 405). For example, the approximate 0,O-transition energies in acetonitrile for WIOO:2, PWI2O:U, PMo120:,, VIoO:R, and VI3O:4 are 2.52, 2.75, 2.29, 2.01, and 1.97 V [vs. saturated calomel electrode (SCE)], respectively, and the 0,O-transition energies are roughly one half the ground-state potentials (257). The ground-state redox potentials of these five polyoxoanions are - 1.21 , -0.61, 0.10, -0.48, and 0.39 V (vs. SCE), respectively. From these data, the potentials of the excited states of Q.1W10032, Q3PW12040, Q ~ P M o I ~ O : ~ , Q3H3VIOO:R,and Q3V13034(Q is tetra-n-butylammonium) are estimated to be approximately 1.55, 2.39, 2.43, 1.77, and 2.60 V (vs. SCE), respectively. However, the emission from the excited states of polyoxometalates is usually so weak that it is difficult to determine the 0,O-transition energies exactly (257). Polyoxometalates are classified into three groups in terms of reactivity (257, 405): 1 . Polyoxometalates such as Nb60fP and Ta60yG, which are hardly photoreduced. 2. Polyoxometalates of Mo and V, the oxidized forms of which are appreciably photoreduced but the reduced forms are too stable to be reoxidized. 3. Polyoxometalates of W, which undergo both photoreduction and facile reoxidation.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

235

Therefore, tungstates are efficient oxidizing photocatalysts under UV irradiation and near-visible light. PW120:0 and protonated WlOO:, are reported to promote radical oxidation. On the other hand, unprotonated W 100:Z promotes the radical-radical reaction. 2. Reactions Catalyzed by Heteropoly Compounds Table XXXVI is a list of some catalytic photochemical redox transformation of organic reactants by (Q or H)3PW12040. In the presence of UV light, Q3PW 1 2 0 4 0 reacts with paraffins, arenes, alcohols, alkyl halides, ketones, nitriles, thioethers, and water. Under either anaerobic or aerobic conditions, decarboxylation, dehydrogenation, dimerization, polymerization, oxidation, and acylation takes place. Alcohols such as methanol, 2-propanol, and benzhydrol are cleanly oxidized to the corresponding carbonyl compounds upon photoexcitation with Na3PW12040in water or with (n-Pr4N)3PWI2O40in CH3CN (406).The quantum yields appear to be governed by the oxidation potential of the alcohol, the availability of a-hydrogens, and the tightness of complexation with the photocatalyst: The reactivity order is primary alcohol > secondary alcohol %= tertiary alcohol. Noteworthy features of photoreactions of paraffins are the following (257): 1. The product distributions are affected in some cases by the presence of a hydrogen evolution catalyst such as Pt(0). 2. The secondary reactions of initial products are sometimes significant and informative. 3. Unactivated C-H bonds can be selectively replaced with C-C bonds in some cases. 4. The addition of acids increases the quantum yield. The heteropolytungstates, which in general have formal redox potentials of grounds states that are less negative than - 1.0 V vs. Ag/AgN03(CH3CN) [such as a - P W 1 2 0 ~ 0(-0.63 to -0.67 V) and a-P~W18022(-0.78 V)], photodehydrogenate paraffins in high selectivity to give the most substituted olefins under anaerobic conditions. For example, 2,3-dimethylbutane was dehydrogenated to give 2,3-dimethyl-2-butene with >80% selectivity (407).The major influence on the regioselectivity is the stability of alkyl cations generated from the intermediate radicals. Photoreactions of p-xylene are interesting from the standpoint of C-H activation (408).Figure 69 shows the time course of the anaerobic photoreaction of p-xylene with (TBA)3PW12040 in the presence or absence of PtO2. In the presence of an equimolar amount of Pt02, ca. 4% of the p-xylene was selectively converted to 1,2-di-p-tolylethane after 30 h of irradiation. No trimers or higher oligomers were detected. It is suggested that the reaction proceeds

236

TOSHIO OKUHARA, NORITAKA MIZUNO. AND MAKOTO MISONO TABLE XXXVl

Photocutulytic Transformution of Orgunic Reactants in the Presence of Heteropolv Compounds

Reaction H2

Catalyst

evolution

02

2-PrOH

RR’CHOH

-+

Ref. 403, 405. 410 404

MeCN RR’C=O

406 407

408

P*WIsOg+n ’ PW 120rn’ -

CH4 -CO, +H

COz. HzO

-Gz

+ H,

$N -.‘HCOCH,

m ar n 0

h

h

+HZ

O-O+O-O CHiOH --t HCHO C2HsOH

409

--

d

CHiCHO

COz + 4MeOH

CH4 + 4HCHO + 2H20

CHO

CO,H

1

P

continued

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

237

TABLE XXXVI-Continued

"M.Gr;itz ef a / . ,J. Phys. Chem. 93, 4128 (1989). hR. F. Renneke et a/.,J. Am. Chem. SOC. 108, 3528 (1986). ' R. F. Renneke et a/.,Angew. Chem. In/. Ed. Eng. 27, 1526 (1988). dT. Yamase e t a l . , J. Chem. SOC. Dalton Trans.. 1669 (1986). 'E. S. Ganolina et a / . , Russ. J. lnorg. Chem. 29, 51 (1984). 'T. Yamase et a/., Inorg. Chim. Acta 172. 131 (1990). 'R. C. Chambers etal., J. Am. Chem. SOC.112,8427 (1990). *D. Sattari etal.,J. Chem. SOC. Chem. Commun. 1990,634. 'R. C. Chambers et a/., Inorg. Chem. 20, 2776 (1991).

according to Eqs. (48)-(50), and the rate-determining step is associated with Eq. (50). h 11 --t

pw12o:; PWIzO:i'

+ 2MePhMe

PW120!,tt"'-

+ 2H'

Pwl*o:o*

(48)

PWI20!,i+')--t

PW120:(I

+ MePh(CH2)jPhMe + 2 H t

+ Hz

(49) (50)

When the photoreaction was carried out under aerobic conditions, different products, such as p-tolualdehyde, p-toluic acid, and 4-methylbenzyl alcohol, were obtained in total yields of about 25%. It is probable in this case that benzyl

0

10 20 Time I h

FIG. 69. Anaerobic catalytic photoreaction of p-xylene. I ,2-di-p-tolylethane. (From Ref. 408.)

30

(A)p-Xylene; (0 and 0) ( + PtOz),

238

TOSHIO OKUHARA, NORITAKA MIZUNO, A N D MAKOTO MISONO

a

t2H'

sepeirate subsequent step

Hz0 m3cNsolvent; Ar; 298 K; A d 6 0 m

-=!E

b

cls- 1

trans- 1

CH3CN solvent; Ar; 298 K; 1 9 6 0 nm

SCHEME 1I

radicals react with dioxygen to form ArCH202, giving rise to a classical autooxidation mechanism. C Y - P ~ Wand ~ ~aO - P~W~1 2 0 4have ~ selectivities different from that of WloO:, for the transformation of unactivated C-H bonds in ketone (409). The irradiation of the heteropolytungstates selectively transforms cis-2-decalone into octalones with the nonthennodynamic isomer, A9*'0-2-octalone,in comparable or greater quantity than the conventional thermodynamic isomer, A1'9-2-octalone (Scheme 1 la). In contrast, WlOO;; transforms cis-2-decalone into trans2-decalone (Scheme 1 lb) (409). 3. Reaction Mechanism

The reaction mechanism shown in Eqs. (51)-(57) was proposed for photooxidation of isopropyl alcohol to acetone under irradiation of PWl2O;< : PwI?o:o PW120&*

hv +

+ (CH,),CHOH 3

(51)

Pw,*o:;' PW12Og

+ (CH&C'OH + H t

(52)

PW,20:R

+ (CH,),C'OH -, P W l 2 O Z + (CH,),CO + H'

(53)

PWI2O.&

+ (CH&C'OH -,PWlzO: + (CH&CO + H t

(54)

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

+ PWl20L 2PWI2O:; + +02 + 2H’ PW120& + 40, + 2H’ pw120:;

-+

2pw120:;

+

2PWl2O:; PWI2O:;

239 (55)

+ H20

(56)

+ H20

(57) Photoexcitation involves oxygen-to-metal charge transfer in heteropolyanions [(Eq. (5 l)]. The photoexcited heteropolyanions react with organic species, accompanied by formation of H’ and/or electron transfer, resulting in the reduction of heteropolyanions and the oxidation of organic reactants [Eq. (52)]. The reduced heteropolyanions are reoxidized by dioxygen according to Eqs. (56) and (57). The mechanism is supported by the fact that in the absence of 0 2 the system leads to the generation of H2 by reduction of protons. This process is accelerated by the addition of Pt(0) as a catalyst for hydrogen evolution (410). +

B. ELECTROCATALYSIS Electrochemical reduction of protons to form H2 in an acidic aqueous solution can be catalyzed by SiW12040in the presence of glassy carbon, Pt, Si, or Ti02 as an electrode (411-414). Reduction of 0 2 is also catalyzed by SiW12040immobilized in polyaniline on a glassy carbon electrode (415). The following mechanism is proposed:

siwl20:; : siwI2o:; 5 siwl20i; 4SiWI2O:; + 0 2 + 4H+ +4SiW120z + 2H20 2SiW120G + 0 2 + 4H+ +2SiW120:; + 2H20

(58) (59)

(60) The idea is supported by the fact that 2e--reduced SiW12010 catalyzes the reduction of O2 (410). Iron-substituted heteropolytungstates such as XWI IFe3f0;9 (P, As, Si, Ge) are the catalysts for the electroreduction of nitrite to ammonia (Table XXXVII) TABLE XXXVll Electrocatalytic Reduction of Nitrite to Ammonia in the Presence of Iron-Substituted Heteropolytungstates, XW,,Fe” 0;; (41 6, 41 7) Xh Si Ge P AS

Reduction to NH: 21 29 31 36

Efficiencyd 35 22 49

35

Supporting electrolyte, 0.1 M CH3COONa-CH3COOH; pH 4. Stirred mercury pool kept at - 0.9 V. hHeteroatom in the catalysts. ‘Percentage of initial NO; converted to NH: as determined by ion chromatography. dCoulombic efficiency for generation of ammonia.

240

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

(416, 417). The pH dependence of the rate of formation of a nitrosyl complex

shows that nitrous acid is the reactive intermediate in the reaction when the pH is in the range of 2-8. The catalysts are not deactivated during repeat cycles between their oxidized and reduced states. The catalyzed reduction appears to depend on the ability of the multiply reduced heteropolyanions to deliver electrons to the NO group bound to the iron center.

XIII. Conclusions Heteropoly compounds are already important industrial catalysts, and more applications are anticipated. Much remains to be done. Future desirable goals of research (5, 6) are all related to the advantageous properties of heteropoly compounds listed in Table I: molecular design of catalysts by control of acid and redox properties; understanding of catalytic processes at the molecular level; use of cluster models of mixed oxide catalysts; application of novel polyanions (complexes with organometallics, synzymes, etc.); and photo- and electrocatalysis. Worthy targets of research are suggested to be the following: design of solid acids stronger than H3PW12040 and acids having moderate but uniform acid strengths; bifunctional acid-base or acid-redox catalysts; exploitation of the unique properties of assembly of heteropoly compounds (e.g., stereo- and shapeselective reactions in pseudoliquid or controlled pores). Worthy noncatalytic applications may include the use of heteropoly compounds as anti-retroviral active substances (418, 419) and as electronic materials, such as photoresists (420, 421), electrochromics (422, 423), and solid electrolytes (424, 425). REFERENCES I. Tsigdinos, G. A., Top. Curr. Chem. 76, 1 ( 1 978). 2. Pope, M. T., “Heteropoly and lsopoly Oxometalates.” Springer-Verlag. Berlin, 1983. 3. Pope, M. T., and Miiller, A., Angew. Chem. Inr. Ed. Engl. 30, 34 (1991); “Polyoxometalates:

4. 5. 6.

7. 8. 9.

10.

From Platonic Solids to Anti-Retroviral Activity.” (M. T. Pope and A. Miiller, Eds.). Kluwer, Dordrecht, 1994. Misono, M., in “Proc. Climax 4th Intern, Conf. Chem. Uses of Molybdenum” (H. F. Barry and P. C. H. Mitchell, Eds.), p. 289. 1982. Misono, M., Cafal.Rev.-Sci. Eng. 29,269 (1987); 30,339 (1988). Misono, M., in “Proc. 10th Int. Congr. Catal., Budapest, 1992,” p. 69. Elsevier, Amsterdam, and Akademiai Kiado, Budapest, 1993. Ono, Y . , in “Perspectives in Catalysis” (J. M. Thomas and K. 1. Zamaraev, Eds.), p. 341. Blackwell, London, 1992. Kozhevnikov, 1. V., and Matveev, K. I., Appl. Cafal.5,135 (1983). Izumi, Y.,Urabe, K., and Onaka, A., “Zeolite, Clay, and Heteropolyacids in Organic Reactions.” Kodansha, Tokyo-VCH, Weinheim, 1992. Kozhevnikov, I. V., in “Acid-Base Catalysis II” (H. Hattori, M. Misono, and Y.Ono, Eds.) (Stud. Surf. Sci. Catal., Vol. go), p. 21. Kodansha, Tokyo-Elsevier, Amsterdam, 1994; Caful. Rev. Sci. Eng. 37.3 1 1 (1 995).

CATALYTIC CHEMISTRY O F HETEROPOLY COMPOUNDS

24 1

11. Killefer, D. H., and Linz, A., “Molybdenum Compounds,” p. 87. Interscience, New York, 1952. 12. Misono, M., and Nojiri. N., Appl. Cural. 64, 1 (1990); Nojiri, N., and Misono, M., ihid. 93, 103 ( 1 993). 13. Misono, M., in “Future Opportunities in Catalytic and Engineering Technology” (M. Misono, Y. Moro-oka. S. Kimura. Eds.), p. 13. Elsevier, Amsterdam, 1990. 14. Jeanin. Y., and Fournier, M., Pure Appl. Chem. 59, 1529 (1987). 15. Leigh, G . J., Ed., “Nomenclature of Inorganic Chemistry, Recommendations 1990.” Blackwell. Oxford, 1990. 16. IUPAC, “Nomenclature of Inorganic Chemistry (I970 Rules), 2nd Ed.” Butterworths, London, 1971. 17. Misono, M., Sakata, K.. Yoneda, Y., and Lee, W. Y., in “Proc. 7th Int. Congr. Catal., Tokyo, 1980,” p. 1047. Kodansha, Tokyo, Elsevier, Amsterdam, 198 I . 18. (a) Jeannin, Y., Herve, G., and Proust, A,, Inorg. Chim. Actu 189, 319 (1992); (b) Weakly. T. J. R., Struct. Bonding 18, 131 (1974). 19. Keggin, J. F., Nature (London) 131, 908 (1933); Proc. R. Soc. London, Ser. A 144, 75 ( 1934). 20. Misono, M., and Hashimoto, M., Shokuhui (Cutulvst) 34, 152 (1992). 21. Brown, G . M., Noe-Spirlet, M. R., Busing, W. R., and Levy, H. A,, Acru Ctystullogr., Ser B 33, I038 ( 1977). 22. Spitsyn. V. I., Kazansky, L. P., and Torchenkova. E. A,, Sov. Sci. Rev. B.. Chem. Rev. 3, I I I (1981 ). 23. Souchay, P., Bull. Soc. Chim., France 18,365 (195 I ) . 24. Yamamura, K., and Sasaki, Y., J. Chem. Soc., Chem. Commun. 1973,648. 25. Matsurnoto, K. Y., Kobayashi. A., and Sasaki, Y., Bull. Chem. Soc. Jpn. 39,3 146 (1975). 26. Baker, L. C. W.. and Figgis, J. S., J. Am. Chem. Soc. 92,3794 (1970). 27. Chauveau, F., Doppelt. P., and Lefebvre, J., J. Chem. Res (S) 1978, 130; 1981, 155. 28. Pettersson, L., Anderson, I., and Ohman, L. 0.. Inorg. Chem. 25,4726 ( I 986). 29. Dawson, B., Actu Ciystullogr. 6, 113 (1953). 30. D’Amour, H., Acta Crystullogr. B32,729 (1976). 31. Hori, T., Tamada, 0..and Himeno, S., J. Chem. Soc. Dalton Truns. 1989, 1491. 32. Chauveau, F., Doppelt, P., and Lefebvre, J.. lnorg. Chem. 19,2803 (1980). 33. Anderson, J. S . , Nurure 140,850 (1937). 34. “Crystal Structure of Inorganic Compounds.” LB Neue Series, IIIi7f. PiesiWeiss. p. 200.

Springer, Berlin.

35. Spirlet, M. R.. and Busing, W. R.. Actu Cwstullogr. B34,907 (1978). 36. Bradley, A. J.. and Illingsworth, J. W.. Proc. R. Soc. (London) A157, I13 (1936). 37. Mizuno, N., and Misono, M., Chem. Lett. 1987,967. 38. Oniani, E. S., Sergienko, V. S., Chuaev. V. F., and Mistryukov, A. E., Russ. J. Inorg. Chem. 36, I157 (1991). 3Y. Hashimoto, M., and Misono, M., Actu Cwstullogr. C50,231 (1994). 40. Attanasio. D., Bonamico, M., Fares, V.. lmperatori, P., and Suber, L., J. Chem. Soc.. Dulton Truns. I 190,322 I . 41. Prosser-McCartha, C. M.. Kadkhodayan. M., Williamson, M. M.. Bouchard. D. A,. and Hill, C. L., J. Chem. Soc.. Chem. Commun. 1986, 1747. 42. Niiyama. H.. Saito, Y., and Echigoya, E., in “Proc. 7th Int. Congr. Catal., 1980.” p. 1416. Kodansha, Tokyo, Elsevier, Amsterdam, 198 1. 43. (a) Gregg, S. J., and Stock, R.. Truns. Furudu.v Soc. 53, 1355 (1957); (b) Gregg, S. J.. and Tayyab. M. M., J. Chem. Soc. Furuduy 174,348 ( I 978). 44. ( a ) Taylor. D. B.. McMonagle, J. B., and Moffat, J. B., J. Colloid Inter-uce Sci. 108, 2789 (1985); (b) MacMonagle, J. B., and Moffat, J. B., J. Colloid Interfuce Sci. 101, 479 (1984);

242

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

(c) Moffat, J. B., J. Mu/. Catal. 52, 169 (1989); (d) Moffat, J. B., and MacMonagle, J. B., and Taylor, D., Solid Sfafefonics 26, 101 (1988). 45. Furuta, M., Sakata, K., Misono. M., and Yoneda, Y., Chem. Len. 1979,31. 46. (a) Tatematsu, S., Hibi, H., Okuhara, T., and Misono, M., Chem. Left 1984, 865; (b) Okuhara. T., Kasai, A., Hayakawa, N., Yoneda, Y., and Misono, M., J. Catal. 83, 121 (1983). 47. Okuhara, T., Nishimura, T., Watanabe, H., Na, K., and Misono, M., in “Acid-Base Catalysis 11.” p. 419. Kodansha, Tokyo-Elsevier, Amsterdam, 1994. 48. Okuhara. T., Nishimura, T., and Misono, M., Chem. Leff.1995, 155. 49. Tsigdinos, G. A.. fnd. Eng. Chem. Prod. Res. Dev. 13,267 (1974). 50. Gruttner, B., and Jander, G., in “Handbook of Preparative Inorganic Chemistry,” 2nd Ed. ( G . Brauer, Ed.), Vol. 2, p. 1716. Academic Press, New York, 1965. 5 f . Tsigdinos, G . A,, in “Heteropoly Compounds. Methodicum Chimicum” (K. Niedenzu and H. Zimmer, Eds.), Vol. 8, Chap. 32. Academic Press, New York, 1976. 52. Fuchs, J., and Flindt, E. P., 2. Nafurfursch Teil B 34,412 (1979). 53. Bailar, J. C., fnurg. Synfh. 1, 132 (1939). 54. (a) Hu,H., J. B i d . Chem. 43, 189 (1920); (b) Hall, R. D., J. Am. Chem. SUC.24,780 (1907). 55. (a) Souchay, P., “Ions Moineraux Condenses.” Masson, Paris, 1969; (b) Izumi, Y., and Otake, M., “Kagaku Sosetsu 34,” p. I 18. Gakkai Shuppan Senta, 1982. 56. Eguchi, K., Yamazoe, N., and Seiyama, T., Nippon Kagaku Kaishi 1981,336. 57. Fournier, M., F-Jantou, C., Rabia, C., Herve, G., and Launay, S., J. Maf. Chem. 2,971 (1992). 58. Rocchiccioli-Deltcheff, C., and Fournier. M., J. Chem. Soc., Faraday Trans. 87,3913 (1991). 59. Rocchiccioli-Deltcheff, C., Inorg. Chem. 22,207 (1983). 60. Bondareva, V. M., Sandrushkevich, T. V., Maksimovskaya, R. I., Plyasova, L. M., Ziborov. A. V., Litvak, G. S., and Detusheva, L. G., Kinef. Cural. 35, 1 14 (1994). 61. (a) Na, K., Okuhara, T., and Misono. M., J. Chem. Suc. Faraday Trans. 91, 375 (1995); (b) Black, J. B., Senvicka, E., Clayden. N. J., Goodenough, J. B., and Scott, J. D., Polyhedron 5, 141 (1986). 62. (a) Rocchiccioli-Deltcheff, C., Thouvent, R., and Franck, R., Specfruchim. Acfu 32A 587 (1976); (b) Brown, D. H., Specrruchim. Acfa 19,583 (1963). 63. (a) Hu, C., Hashimoto. M., Okuhara, T., and Misono, M., J. Cafal. 143, 437 (1993); (b) Okuhara, T., Hu,C., Hashimoto, M., and Misono, M., Bull. Chem. Suc. Jpn. 67, I 186 (1994). 64. Boeschen, F., Brus, B., and Krebs, B., Acfa Crysfallugr. 830.48 (1 974). 65. Rocchiciolli, C.. Fournier, M., Franck, R., and Thouvenot, R., fnurg. Chem. 22,207 (1983). 66. Okuhara, T., Hashimoto, T., Hibi, T., and Misono. M.. J. Caral. 93,224 ( I 985). 67. Rocchiccioli-Deltcheff, C., and Thouvent, R., J. Chem. Res. (S) 1977,46; (M) 1977, 546. 68. Rocchiccioli-Deltcheff, C., and Thouvent, R., fnorg. Chem. 17, 1 I15 (1978). 69. Fournier, M., Thouvenot, R., and Rocchiccioli-Deltcheff, C., J. Chem. SUC.Faradav 87, 349 ( I 99 1 ). 70. Rocchiccioli, C., and Touvenot, R., Specfrusc.Leu. 12(2), 127 (1979). 71. Fukumoto, T., Murata, K., and Ikeda, S., Anal. Chem. 56,929 (1984). 72. Mioc, U., Colomban, P., and Novak, A,, J. Mu/. Sfrucr. 218, 123 (1990). 73. Wada, T., Cumpf.Rend. B263.5 I ( 1966). 74. Churaev, V. F., and Spitsyn, V. Y., Dukl. Akad. Nauk. SSSR 154,908 (1964). 75. Spitsyn, V. I.. Chuvaev, V. F., and Bakhchisaritseva, S. A., Dukl. Akad. Nauk SSSR 204, 1403 (1972). 76. Spitsyn, I., Chuvaev. V. F., and Bakhchisaraitseva, S. A,, Dukl. Akad. Nauk SSSR 165, I126 ( 1965). 77. Kanda, Y.. Lee, K. Y., Nakata, S., Asaoka, S.. and Misono, M., Chem. Leff.1988, 139. 78. Lee, K. Y . , Mizuno, N., Okuhara, T.. and Misono, M., Bull. Chem. SUC. Jpn. 62, 1731 (1989).

CATALYTIC CHEMISTRY O F HETEROPOLY COMPOUNDS

243

79. Massart, R., Contant, R., Fruchart, J., Ciabrini, J.. and Fournier, M., Inorg. Chem. 16, 2916 (1977). 80. (a) Pope, M. T., O’Donnell, S. E., and Prados, R. A,, J. Chem. Soc.. Chem. Commun. 1975,22; (b) Pope, M. T., and Scully, T. F., Inorg. Chem. 14,953 (1975). 81. (a) Lee, K. Y., Itoh, K., Hashimoto, M., Mizuno, N., and Misono, M., Stud. SurJ Sci. Catal. 82, 583 (1994); (b) Schwegler, M. A., Peters, J. A,, and van Bekkum, H., J. Mol. Catal. 63, 343 (1 990). 82. Black, J. B., and Clayden. N. J., J. Chem. Soc., Dalton Trans. 1984,2765. 83. Acerete, R., Hammer, C. F., and Baker, L. C. W., J. Am. Chem. Soc. 101,267 (1979). 84. Piepgrass, K., and Pope, M. T.,J.Am. Chem. Soc. 109, 1586 (1987). 85. Domaile, P. J., J. Am. Chem. Soc. 106, 1677 (1984). 86. Lefevre, J., Chauveau, F., and Doppelt. P.,J. Am. Chem. SOC.103,4581 (1981). 87. Filowitz, M., Klemperer, W. G., Messerle, L., and Shum, W., J. Am. Chem. Soc. 98, 2345 (1976). 88. Filowitz, M., Ho, R. K. C., Klemperer, W. G., and Shum, W., 1norg.Chem. 18,93 (1979). 89. English, A. D., Jesson, J. D.. Klemperer, W. G., Mamouneas, T.. Messerle, L., Shum, W., and Tramontano, A,. J. Am. Chem. SOC.97,4785 ( 1 975). 90. Varga, G . M., Papaconstantiou, E., and Pope, M. T.,Inorg. Chem. 9,667 (1970). Y I . Alternau, S. L.. Pope, M. T., Prados, R. A,, and So. H., Inorg. Chem.2,4 I7 ( 1975). 92. So, H., and Pope, M. T., Inorg. Chem. 11, 1441 (1972). 93. loannidis, A., and Papapconstatinou, E.. Inorg. Chem. 24,439 (1985). 94. Miyanaga, T.,Fujikawa, T., Matsubayashi, N., Fujimoto, T., Yokoi, K., Watanabe, I.. and Ikeda, S., Bull. Chern. Sac. Jpn. 62, I791 ( 1 989). 95. Miyanaga, T., Fujikawa, T., Matsubayashi, N.. Fujimoto, T., Yokoi, K., Watanabe, I., and Ikeda, S., Chem. Lett. 1988,487. 96. Keita, B.. and Ndjo, L., Sut$ Sci. Lett. 254, L443 (1991). 97. Keana, J. F., Ogan. M. D., Lu, Y., Beer, M., and Varkey, J., J. Am. Chem. Soc. 108, 7957 ( 1986). 98. Pope, M. T., Prog. Inorg. Chem. 39, 181 (1991). Y9. Sanchez, C., Livage, J., Launay, 3. P., Fournier, M., and Jeanin, Y., J. Am. Chem. SOC. 104, 3 I94 (1982). 100. Prados, R. A., and Pope, M. T., Inorg. Chem. 15,2547 (1976). l o / . Eguchi, K., Toyozawa, Y., Yamazoe, N., and Seiyama, T., J. Catal. 83,32 (1983). 102. Mizuno, N., Katamura, K., Yoneda, Y., and Misono, M.,J. Cufal.83,384 (1983). 103. Konishi, Y., Sakata, K., Misono, M., and Yoneda, Y., J. Catal. 77, 169 (1982). 104. Otake, M., Komiyama, Y., and Otaki, T., J. fhys. Chem.77,2896 (1973). 105. Akimoto, M., Ikeda, H., Okabe, A., and Echigoya, E., J. C a d 89, 196 (1984). 106. Centi, G., Nieto, J. L., and lapalucci, C., Appl. Catal. 46, 197 (1989). 107. Akimoto, M., Shima, K., and Echigoya, E., J. Chem. SOC.,Faraday Trans. 179,2467 (1983). 108. Akimoto, M., Shirna, K.. Ikeda, H., and Echigoya, E., J. Cafal.86, 173 (1984). 109. Aboukak, A., Ghoussoub, D., B-Crusson, E., Rigole, M., and Guelton, M.,Appl. Catal. A I 1 I, I09 (1994). 110. Song, I. K., Lee, J. K., and Lee, W. Y., Appl. Catal. A 119,107 (1994). 111. Black, J. B., Clayden, N. J., Gai, P. L., Scott, J. D., Serwicka, E. M., and Goodenough, J. B., J. Catal. 106, I(1987). 112. Serwicka, E. M., Black, J. B., and Goodenough, J. B., J. Calal. 106,23 (1987). 113. Kazansky, L. P., in “Proc. Climax 3rd Intern. Counf. Chemistry Uses Molybdenum” (H. F. Bany and P. C. H. Mitchell, Eds.), p. 70, 1979. 114. Dorokhova, E. N., and Kazansky, L. P., Dokl. Akad. NaukSSSR 229,622 (1976). 115. Viswanathan, B., Omana, M. J., and Varadarajan, T. K., Catal. Lett. 3,2 I7 ( I 989).

244

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

116. Kozhevnikov, 1. V., Russ. Chem. Rev. 56,811 (1987). 117. Polotebnova, N. A., Kozlenko, A. A., and Furtune, L. A,, Russ. Inorg. J. Chem. 21, 1511 ( 1976). 118. Kozhevnikov, 1. V., Kulikov, S. M., and Matveev, K. I., IN. AM. NuukSSSR, Ser. Khim. 1980, 2213. 119. Kulikov, S. M., and Kozhevnikov, I. V., Izv. Akad. NuukSSSR. Ser. Khim. 1981,498. 120. Kozhevnikov, 1. V., Khankhasaeva, S. Ts., and Kulikov, S. M., Kinet. Kuful. 29,76 (1987). 121. Farcasiu, D., and Li, J. Q.,J. Cuful.152, 198 (1995). 122. Kozhevnihov, 1. V., and Matveev, K. I., Usp. Khim. 51,1985 (1982). 123. Baraza, L., and Pope, M. T., J. Phys. Chem. 79,92 (1975). 124. Izumi, Y., Matsuo, K., and Urabe, K., J. Mol. Curd. 18,299 (1983). 125. Misono, M., Mizuno, N., Katamura, K., Kasai, A., Konishi, Y.,Sakata, K., Okuhara, T., and Yoneda, Y., Bull. Chem. SOC.Jpn. 55,400 (1982). 126. Otake, M., and Onoda, T., Shokubui (Curulysr) 17, 13 (1975). 127. Okuhara, T., Nishimura, T., Watanabe, H., and Misono, M.,J. Mol. Curul. 74,247 (1992). 128. Misono, M.. and Okuhara, T.. CHEMTECH 23,23 (1993). 129. Gillespie, R. J., Acc. Chem. Res. I, 202 (1968); Olah, G. A., Prakash, G. K. S., and Sommer, J., “Superacids.” Wiley, New York, 1985. 130. Hidalgo, C. V., Itoh, H., Hattori, T., Niwa, M., and Murakami, Y.,J. Card 85,362 (1984). 131. Jozefowicz, L. C., Karge, H. G., Vasilyeva, E., and Moffat, J., Microporous Mar. 1, 313 ( 1993). 132. Lefebvre, F., Cai, F. X.L., and Auroux, A.,J. Muter. Chem. 4. 125 (1994). 133. Auroux, A., and Vedrine, J. C., Stud. Surf Sci. Card. 20,3 I 1 (1 985). 134. Kozhevnikov, I. V., Sinnenma, A., Jansen, R. J. J., and Bekkum. H., J. Chem. Soc.. Chem. Commun. 1994,92; Card Leu. 27, 187 (1994). 135. Barrows, J. N., Jameson, G. B.,and Pope, M. T., J. Am. Chem. SOC.107,1771 (1985). 136. Moffat, J. B., er ul.. J. Mol. Curd. 26,285 (1984). 137. Kazanskii, L. P., Fedotov, M. A., and Spitsyn, V. I., Dokl. Akad. Nuuk SSSR 233, 152 (1977). 138. Baba, T., Watanabe, H., and Ono, Y., J. Phys. Chem. 87,2406 (1983). 139. Highfield, J. G., and Moffat. J. B.,J. Card. 88, 177 (1984). 140. Saito, Y., Abe, Y..Niiyama, H., and Echigoya, E., Nippon Kuguku Kuishi 1986,25. 141. Ghosh, A. K., and Moffat, J. B.,J. Card. 101,238 (1986). 142. Baba, T., and Ono, Y., J. Phys. Chem. 87,3406 (1983). 143. Baba, T., Nomura, M., Ono, T., and Ohno, Y., J. Phys. Chem. 97,12,888 ( 1993). 144. Baba, T., Nomura, M., Ono, Y., and Kansaki, Y., J. Chem. Soc., Furuduy Trans. 88,71 (1992).

145. Baba, T., Nomura. M., Ohno, Y., Hiroyama, Y., and Ono, Y., in “New Aspects of Spillover Effect in Catalysis” (T. h i . Ed.), p. 81. Elsevier. Amsterdam, 1993. 146. Nowinska, K., Fiedorow, R., and Adamiec, J., J. Chem. Soc.. Furuduy Trans. 87,749 (1991). 147. Cheng, W-C.. and Luthra, N. P.,J. Curul. 109, 163 (1988). 148. Rao, K. M., Gobetto, R., Lannibello, A,, and Zecchina, A., J. Curd. 119,512 (1989). 149. van Veer, J. A. R., Hendriks, P. A. J. M.,Andrea, R. R., Romers, E. J. M., and Wilson, A. E., J. Phys. Chem. 94, 183 I (1990). 150. Kapustin, G. I., Brueva, T. R., Klyachko, A. L., Timofeeva, M. N., Kulikov, S. M., and Kozhevnikov, 1. V., Kinef. Kurd. 31,896 (1990). 151. Izumi. Y., Hasebe, R., and Urabe, K., J. Cuful.84,402 (1983). 152. Mastihkin, V. M., Kulikov, S. M., Nosov, A. V., Kozhevnikov, I. K., Mudrakovsky, 1. L., and Timofeeva, M. V., J. Mol. Curd. 60,65 (1990). 153. Lefebvre, F.. J. Chem. SOC.,Chem. Commun. 1992,756. 154. Thouvenot, R., Rocchiccioli-Delcheff, C., and Fournier, M., J. Chem. SOC.,Chem. Commun. 1991, 1252.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOWDS

245

155. Thouvenot, R., Fournier, M., and Rocchiccioli-Delcheff, C., J. Chem. Soc.. Faraday Trans. 87,

2829 (1991). Rocchicioli-Deltcheff, C., Amirouche, M., and Fournier, M., J. Catal. 138,445 (1992). Payen, E., Kasztehan, S., and Moffat, J. B., J. Chem. Soc., Faraday Trans. 88,2263 (1992). Ogata, A.. Kazusaka, A., Yamazaki, A., and Enyo, M., Chem. Lett. 1989, 15. Soled, S. L., Miseo, S., McVicker, G. B., Baumgartner, J. E., Gates, W. E., Gutierrez, A,, and Paes, J., Preprints 209th ACS National Meeting, 1995,Petr. Div., 16. 160. Kozhevnikov, I. V., Russ. Chem. Rev. 62,473 (1993). 161. Jansen, R. J. J., van Veldhuizen, H. M., Schwegler, M. A., and van Bekkum, H.. Reel. Trav. Chim. Pays-Bas 113, I I5 (1 994). 162. Urabe, K, Fukita, K., and Izumi, Y., Shokubai (Catalyst) 22,223 (1983). 163. Aoshirna, A,, Yarnamatsu, and Yamaguchi, T., Nippon Kagaku Kaishi 1987,984. 164. Aoshirna, A,, Tonomura, S., Yamamatsu, S., Polym. Adv. Tech. 2, 127 (1990). 165. Onoue, Y., Mizutani, Y., Akiyama, S., and Izumi, Y., CHEMTECH 1978,432. 166. Izumi, Y., Matsuo, K., and Urabe, K., in “Proc. Climax 4th Intern. Conf. on the Chemistry and Use of Mo,” p. 302.Climax Molybdenum Co., Ann Arbor (1 982). 167. Knoth, W. H., and Harlow, R. L., Inorg. Chem. 23,4765 (1984). 168. Farneth, W.E., Stanley, R. H., Domaille, P. J., and Farlee, R. D., J. Am. Chem. Soc. 109,4018 (1987). 169. Lee, K. Y., Arai, T., Nakata, S., Asaoka, S., Okuhara, T., and Misono, M., J. Am. Chem. Soc. 114,2836 ( 1992). 170. Aoshima, A,, Yamamatsu, S., and Yamaguchi, T., Nippon Kagaku Kaishi 1987,976. 171. Kozhevnikov, I. V., Khankgasaeva, S. T.. and Kulikov, S. M., Kinet. Kural. 29.66 (1988). 172. Kozhevnikov, 1. V., Khankhasaeva, S. T., and Kulikov, S. M., Kinet. Catal. 30,39 (1987). 173. Matsuo, K., Urabe, K.. and Izumi, Y., Chem. Lett. 1981,1315. 174. Baba, T., and Ono, Y., J. Mol. Catal. 37,3 17 ( 1 986). 175. Izumi, Y., and Hayashi, K., Chem. Lett. 1980,787. 176. Sato, S.,Sakurai. C., Furuta, H., Sodesawa, T., and Nozaki, F.. J. Chem. Soc., Chem. Commun. 1991, 1327. 177. Bednarek, M., Brzezinska, K., Stasinski, J., Kubisa, P., and Penczek, S., Makromol. Chem. 190, 929 ( 1989). 178. Khankhasaeva, S.T., Kulikov, S. M., and Kozhevnikov. 1. V., Kinet. Katal. 1990, 188. 179. Maksimov, G. M., and Kozhevnikov, 1. V., React. Kinet. Katal. Lett. 39.3 17 (1987). 180. Nomiya, K., Ueno, T., and Miwa, M., Bull. Chem. Soc. Jpn. 53,827(1980). 181. Kozhevnikov, I. V., and Kulikov, S. M., Kinet. Kafal. 22,747 (1980). 182. Joshi, M. V., and Narasimhan, C. S . , J . Catal. 120,281 (1989). 183. Hu, C., Nishimura, T., Okuhara, T., and Misono, M., Sekiyu Gakkaishi 36,386 (1993). 184. Tatibouet, J., Che, M., Amirouche, M.. Fournier, M., and Rocchiccioli-Delcheff, C., J. Chem. Soc., Chem. Commun. 1988, 1260. 185. Hayakawa, N., Okuhara, T., Misono, M., and Yoneda, Y., Nippon Kagaku Kaishi 1982,356. 186. Baba, T., and Ono, Y., Appl. Catal. 22,321 (1986). 187. Pomiczek, J., Kulszewicz-Bajer, I., Zagorska, M., Hasik, M., Bielanski, A., and Pron, A,, Stud. Surf: Sci. Catal. 75,2593 (1993). 188. Song, 1. K., Lee, W. Y., and Kim, J. J., Catal. Lett. 9,339 (1991). 189. Ono, Y.,Baba, T.. Sakai, J. and Keii, T., J. Chem. Soc., Chem. Commun. 1982,400. 190. Baba, T.,Sakai, J., and Ono, Y . , Bull. Chem. Soc. Jpn. 55,2657 (1982). 191. Ono, Y.,Taguchi, M., Gerile, Suuki, S., and Baba, T., in “Catalysis by Acids and Bases” (B. lmelik et al., Eds.), p. 167.Elsevier, Amsterdam, 1985;Ono, Y., Baba, T., Kanae, K., and Seo, S. G., Nippon Kagaku Kaishi 1988,985. 192. Hayashi, H., and Moffat, J. B.,J. Catal. 83, 192 (1983);81,66 (1983). 156. 157. 158. 159.

246

TOSHIO OKUHARA. NORITAKA MIZUNO, AND MAKOTO MISONO

193. Kasai, A., Okuhara, T., Misono, M., and Yoneda, Y., Chem. Left 1981,449. 194. Hibi, T., Okuhara, T., Misono, M., and Yoneda, Y., Chem. Left 1982, 1275. 195. Hibi, T., Takahashi, K., Okuhara, T., Misono, M., and Yoneda, Y., Appl. Cuful.24,69 (1986). IY6. Baba, T., Sakai, J., Watanabe, H., and Ono, Y., Bull. Chem. Soc. Jpn. 55,2555 (1982). 197. Okuhara, T., Hibi, T., Takahashi. K., Tatematsu, S., and Misono. M., J. Chem. Soc., Chem. Commun. 1984,697. 198. Okuhara, T., Nishimura, T., Takahashi, K., and Misono, M., Chem. Left. 1990, 1201. 199. Nishimura, T., Okuhara, T., and Misono, M., Appl. Cufal.73, L7 (1991). 200. Izumi, Y., Ono. M., Ogawa, M., and Urabe, K., Chem. Left 1993,825. 201. Izumi, Y . ,and Urabe, K., Chem. Lea 1981,663. 202. Schwegler. M. A., Bekkum, H., and Munck, N., Appl. Cuful.74, 191 (1991). 203. Soeda, H., Okuhara, T., and Misono, M., Chem. Left. 1994,909. 204. Kozhevnikov, 1. V., Tsyganok, A. I., Timofeeva, M. N., Kulikov, S. M., and Sidel’nikov, V. N., Kinef.Kaful. 35,430 (1992). 205. Kulikov, S . M., Kozhevnikov, 1. V., Fomina, M. N., and Krysin, A. P., Kinet. Kurd. 27, 651 (1986). 206. Sebulsky. R. T., and Henke, A. M., Ind. Eng.Chem. Process Des. Dev. 10,272 (197 I). 207. Izumi, Y., Natsume, N., Tamamine, H., Tamaoki, I., and Urabe, K., Bull. Chem. Soc. Jpn. 62, 2 I59 (1989). 208. Kozhevnikov, 1. V., Sinnema, A., Jansen, R. J., Pamin, K., and Van Bekkum, H., Cufal.Left. 30.24 I ( 1995). 209. Okuhara, T., Yamashita, M., Na, K., and Misono, M., Chem. Lett. 1994, 1451. 210. Ohgoshi, S., Kanai, J., and Sugimoto, M., EP 561284, 1993. 211. Kresge, C. T., Marler, D. 0..Rav, G. S., and Rose, B. H. (Mobil Oil Corp.) US 5324881, 1994. 212. Nishi, H., Nowinska, K.. and Moffat, J. B., J. Caful. 116,480 (1989). 213. Yamaguchi, T., Mitoh, A., andTanabe, K., Chem. Left. 1982, 1229. 214. Izumi, Y., Ogawa, M., Nohara, W., and Urabe, K., Chem. Leff. 1992, 1987. 215. Nowinska, K., Fiedorow. R., and Adamiec, J., J. Chem. Soc., Furuduy Trans. 87,749 (1991). 216. Na, K., Okuhara, T., and Misono, M., Chem. Leu. 1993,1141. 21 7. Na, K., Okuhara. T., and Misono, M., J. Chem. Soc., Furaday Trans. 91,367 (1995). 21 8. Igarashi, A., Matsuda, T., and Ogino, Y .,J. Japan Pefrol. Inst. 22,33 1 (1979). 219. Ono, Y., and Baba. T., in “Proc. 8th Intern. Congr. Catal,” V-405. Verlag Chemie, Berlin, 1984. 220. Shikata, S., Okuhara, T., and Misono, M., Seikiyu Gakkuishi 37,632 (1994);J. Mol. Cutal. 100, 49 ( 1995). 221. Kim, J., Kim, J., Seo, G., Oark, N., and Niiyama, H., Appl. Cufal.37,45 (1988). 222. Sugiyama, K., Kato, K., Miura, H., and Matsuda, T., J. Jpn. Pefrol. Insf. 26,243 (1983). 223. Saito, Y . ,Niiyama, H., and Echigoya, E., Nippon Kuguku Kuishi 1984,391. 224. Yamamura, T., and Nakatomi, S., J. Catal. 37, 142 (1975). 225. Vaughan, S., O’Connor, C. T., and Fletcher, J. C. Q., J. Cuful. 147,441 (1994). 226. Niiyama, H., Tamai, S., Saito, Y., and Echigoya, E., Chem. Leff. 1983, 1679. 227. Nasikin, M., Nakamura, R., and Niiyama, H., Chem. Leff. 1993,209. 228. Nayak, V., and Moffat, J. B., Appl. Cutal. 47,97 (1989). 229. Oulmekki. A.. and Lefebvre, F., React. Kinef. Kurul. Left.48, 601 (1992). 230. Matsuda, T., Sato, M., Kanno, M., Miura, H., and Sugiyama, K., J. Chem. Soc. Faraduy Trans. I77,3107 (1981). 231. Nayak, V . , and Moffat, J. B., Appl. Cutal. 36, 127 (1988). 232. Nishi, H.. and Moffat., J. B., J. Mol. Caful. 51, 193 (1989). 233. (a) Sakata, K., Furuta, M., Misono, M., and Yoneda, Y., ACS/CSJ Chemical Congr., Honolulu, April, 1979; (b) Misono, M., 1st Japan-France Catal. Seminar, July 1979.

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

247

234. Okuhara, T., and Misono, M., in “Dynamic Process on Solid Surface” (K. Tamaru, Ed.), p. 259. Plenum, New York, 1993. 235. Okuhara, T., Tatematsu, S., Lee, K. Y., and Misono, M., BUN. Chem. SOC. Jpn. 62,717 (1989). 236. Okuhara, T., Hashimoto, T., Misono, M., Yoneda, Y., Niiyama, H., Saito, Y., and Echigoya, E., Chem. Lett. 1983,573. 237. Highfield, J . G., and Moffat, J. B.,J. Caful. 95, 108 (1985). 238. Highfield, J . G., and Moffat, J. B., J. Cutul. 98,245 (1986). 239. Uchida, T . , Sugiura, N., Nakamura, R., and Niiyama, H., J. Chem. Eng. Japan 24,417 (1991). 240. Okuhara, T . , Arai, T., Ichiki, T., Lee, K. Y., and Misono, M., J. Mol. Cutal. 55,293 (1989). 241. (a) Saito, Y., and Niiyama, H., J. Cufal. 106, 329 (1987); (b) Saito, Y., Cook, P. N., Niiyama, H., and Echigoya, E., J. Cutal. 95,49 (1985). 242. Takahashi. K., Okuhara, T., and Misono, M., Chem. Lett. 1985,841. 243. Misono, M., Okuhara, T., Ichiki, T., Arai, T., and Kanda, Y., J. Am. Chem. SOC.109, 5535 ( I 987). 244. Knozinger, H., Angew Chem., In!. Ed. Engl. 7,791 (1968). 245. Klages, F., Gordon, J. E., and Jung, H. A,, Chem. Ber. 98,3748 ( I 985). 246. Lee, K. Y., Kanda, Y., Mizuno, N.. Okuhara, T., Misono, M., Nakata, S., and Asaoka, S., Chem. Lett. 1988, 1175. 247. (a) Olah, G . A,, and Namanworth, E., J. Am. Chem. SOC.89,3576 (1967); (b) Birchall. T., and Gillespi, R. J., Can. J. Chem. 43, 1045 (1965). 248. Maclean, C., and Mackor, E. L., J. Chem. Phys. 34,2207 (1961). 249. Nishimura, T., Okuhara, T., and Misono, M., Chem. Left. 1991, 1695. 250. Niiyama, H., Akratiwa, S., and Saito, Y., in “Proc. Regional Symposium on Petrochemical Engineering,” ST-I, Bangkok, Thailand, 1987. 251. Barrows J . N., and Pope, M. T., Adv. Chem. Ser. 226,403 (1990). 252. Massart, R., Ann. Chim. 3,507 ( I 968). 253. Miiller, A,, Krickemeyer, E., Penk, M., Witteneben, V., and Doring, J., Angew. Chem. In/. Ed. Engl. 29,88 ( 1990). 254. Barrows, J . N., Jameson, G. B., and Pope, M. T.,J. Am. Chem. SOC.107, 1771 (1985). 255. Katsolis P. E., and Pope, M. T., J. Am. Chem. SOC.106,2737 (1984). 256. Khenkin A. M., and Hill, C. L.,J. Am. Chem. SOC.115,8178 (1993). 257. Hill, C. L., and McCartha, C. M. P., in “Photosensitization and Photocatalysis Using Inorganic and Organometallic Compounds” (K. Kalyanasundaram and M. Gratzel, Eds.), p. 307. Kluwer, Dordrecht, The Netherlands, 1993. 258. (a) Komaya, T . , and Misono, M., Chem. Lett. 1983, 1177; (b) Misono, M., Mizuno, N., and Komaya, T., in “Proc. 8th Int. Congr. Catal., 1984,” Vol. 5, p. 487. Verlag Chemie, Weinheim. 1984. 259. Katamura, K., Nakamura, T., Sakata, K., Misono, M., and Yoneda, Y., Chem. Lett. 1981,89. 260. Mizuno. N . , and Misono, M.,J. Phys. Chem. 93,3334 (1989). 261. Mizuno, N., Katamura, K., Misono, M., and Yoneda, Y., J. Catal. 83,384 (1983). 262. Mizuno, N., Watanabe, T., and Misono, M., J. PhyJ.. Chem. 94,890 (1990). 263. Tsuneki, H., Niiyama, H., Echigoya, E., Chem. Lett 1978,645. 264. Taketa, H.. Katsuki, S.. Eguchi, K.. Seiyama, T., and Yamazoe, N., J. Phys. Chem. 90, 2959 ( 1986). 265. Akimoto, M., and Echigoya, E., Chem. Letf. 1981, 1759. 266. Mizuno, N., Watanabe, T., and Misono, M., J. Phys. Chem. 89,80 (1985). 267. Sakata, K., Ph. D. Dissertation, The University of Tokyo, 198 I . 268. Komaya, T., Masters Thesis, The University of Tokyo, 1983. 269. Kulesza, P. J . , Faulkner, L. R., Chen, J., and Klemperer, W. G., J. Am. Chem. Soc. 113, 379 (1991); Kulesza, P. J., and Faulkner, L. R., J. Am. Chem. SOC.115, 11879 (1993).

248

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MlSONO

270. Yoshida, S., Niiyama, H., and Echigoya, E., J. Phys. Chem. 86,3150 (1982); Nippon Kugaku Kuishi 1981, 1703; Yoshida, S., Miyata, Y., Niiyama, H., Echigoya, E., ibid. 1982, 1873. 271. Hodnett, B. K., and Moffat, I. B., J. Curd. 91,93 (1985). 272. Ai, M., Appl. Cutul. 4,245 (1982). 273. Kim, H. C., Moon, S. H., and Lee, W. Y., Chem. Lerr. 1991,447. 274. Kazanskii, L. P., Fedotov, M. A,, Potapova, I. V., and Spitsyn, A. V. I., DOH.Akud. NuukSSSR 244,372 (1979). 275. Massart, R.. Fournier, M., and Souchay, P., C. R. Akad. Sci. C267, 1805 (1968). 276. Misono, M.. Igarashi, H., Katamura, K., Okuhara, T., and Mizuno. N., Stud. Surf: Sci. Cural. 77, 105 (1993). 277. Niiyama, H., Tsuneki, H., and Echigoya, E., Nippon Kaguku Kuishi. 1979,996. 278. Kim, H. C.. Moon, S. H., and Lee, K. Y., Chem. Lea 1991,447. 279. Matsumoto, H., Lee, K. Y., and Misono, M., 61st National Meeting of The Chemical Society of Japan. Tokyo, 1991. 280. Zuemecki, S. B., in “Proc. 9th Int. Congr. Catal. 1988,” Vol. 4, p. 1798. 281. Senvicka, E. M . , Bruckman, K., Haber, J., Paukshtis, E. A., and Yurchenko, E. N., Appl. Card 73,153 (1991). 282. Bruckman, K., Haber, J., and Serwicka, E. M., Furuduy Discuss. Chem. SOC.87, 173 (1988). 283. Taouk, T., Ghoussoub, D., Bennani, A., Crusson, E., Rigole, M., Aboukais, A., Decressain, R., Fournier, M., and Guelton, A,, J. Chim. Phys. 89,435 (1992). 284. Cadot, E., Marchal, C., Foumier, M., Teze, A., and Herve, G., in “Polyoxometalates” (M. T. Pope and A. Miiller, Eds.), p. 135. Kluwer, Dordrecht, The Netherlands, 1994. 285. Ali, B. E., Bregeault, J.-M., Mercier, J., Martin, J., Martin, C., and Convert, 0.. J . Chem. Soc.. Chem. Commun. 1989,825. 286. Atlamsani, A,, Bregeault, J.-M., and Ziyad, M., J. Org. Chem. 58,5663 (1993). 287. Gorodetskaya, T. A., Kozhevnikov, 1. V., and Matveev, K. I., Kiner. Kurd 23,992 ( I 982). 288. Mizuno, N., and Misono, M., J. Mol. Curd. 86,3 19 (1994). 289. Neumann, R., and Lisel, M., J. Org. Chem. 54,4607 (1989). 290. Neumann, R., and Levin, M., J.Am. Chem. SOC.114,7278 (1992). 291. Lissel, M., Jansen, H., in de Wal, and R. Neumann, TetrahedronLett. 33, 1795 (1992). 292. Matveev, K. I., and Kozhevnikov, I. V., Kiner. Kurd. 21,1189 (1980) 293. Mizuno, N., Hirose, T., Tateishi, M., and Iwamoto, M., Chem. Leu. 1993, 1839; Mizuno, N., Hirose, T., and Iwamoto, M., Stud. SurJ Sci. Carol. 82,593 (1994). 294. Hamarnoto, M., Nakayama, K., Nishiyama, Y., and Ishii, Y., J. Org. Chem. 58,6421 (1993). 295. Yamada, T., Takai, T.,Rhode, O., and Mukaiyama, T.,BuN., Chem. SOC.Jpn. 64,2109 (1991); Takai, T., Hata, E., Yamada, T., and Mukaiyarna. T., ibid. 64, 2513 (1991); Mukaiyama, T., Yorozu K., Takai, T. and Yamada, T., Chem. Leu. 1993,439. 296. Mizuno, N., Tateishi, M., Hirose, T.,and Iwamoto, M., Chem. Lerr. 1993, 1985. 297. Lyons, J. E., Ellis, P. E., and Durante. V. A., in “Structure-Activity and Selectivity Relationships in Heterogeneous Catalysis” (R. K. Grasselli and A. W. Sleight, Eds.), p. 99. Elsevier, Amsterdam, 1991. 298. Mizuno, N., Tateishi, M., Hirose, T., and Iwamoto, M., J. Mol. Card 88, L125 (1994). 299. Mizuno, N., Tateishi, M., Hirose, T., and Iwamoto, M., Chem. Lett. 1993,2137. 300. (a) Kitajima, N., Fukui, H., and Moro-oka, Y., J. Chem. Soc., Chem. Commun. 1988, 485; Kitajima, N., Itoh, M., Fukui H., and Moro-oka, Y . ,ibid., 1991, 102; (b) Murahashi, S., Oda, Y.,and Naota, T, J. Am. Chem. SOC.114,7913 (1992); (c) Barton, D. H.R., Gastiger, M. J., and Mothemell, W. B. J. Chem. Soc., Chem. Commun. 1983,731; (d) Onophenko, A., and Shultz, J., J. Org. Chem. 38,999 (1973); 38; 3729 (1973). 301. Lau, T.-C., and Mak, C.-K., J. Chem. SOC., Chem. Commun. 1993, 766; Khan, M. M. T., Chattejee, D., Kumar S, S., Rao, A. P., and Khan, N. H., J. Mol. Cutul. 75, L49 (1992).

CATALYTIC CHEMISTRY OF HETEROPOLY COMPOUNDS

249

302. Furukawa, H., Nakamura, T., Inagaki, H., Nishikawa, E., Imai, C., and Misono, M., Chem. Lett. 1988,877.

303. Mizuno, N., Yokota, S., Miyazaki, I., and Misono, M., Nippon Kuguku Kuishi 1991, 1066. 304. Lee, K. Y., Itoh, K., Hashimoto, M., Mizuno, N., and Misono, M., Stud. Surf: Sci. Cutul. 82,583 ( I 994). 305. Burden, J. K., and Nguyen, C. K., J. Am. Chem. SOC.112,5366 (1990). 306. Ishii, Y., Yamawaki, K., Yoshida, T., Ura, T., and Ogawa, M., J. Org. Chem. 52, 1868 (1987); Yamawaki, K., Yoshida, T., Nishihara, H., Ishii, Y., and Ogawa, M., Synth. Commun. 16, 53 (1986); Daumas, M., Vo-Qaung, Y.,and Vo-Quang, L., Synthesis 1989,64. 307. (a) Motoba, Y., Inoue, H., Akagi, J., Okabayashi, T, Ishii, Y., and Ogawa, M.. Synth, Commun. 14,865 (1984); (b) Ishii, Y., and Ogawa, M., J. Synth. Org. Chem. 47,889 (1989). 308. (a) Venturello, C., Alneri, E., and Ricci, M., J. Org. Chem. 48, 3831 (1983); Venturello, C., D’Aloiso, R., Bart, J. C. J., and Ricci, M., J. Mol. Cutul. 32, 107 (1985); Venturello, C., and D’Aloisi, R., J. Org. Chem. 53, 1553 (1988); (b) Ishii, Y., Yamawaki, K, Ura, T., Yamada, H., Yoshida, T., and Ogawa, M., J. Org. Chem. 53,3587,5549 (1988); (c) Schwegler, M., Floor, M., and van Bekkum, H., Telu-uhedronLett. 29,823 (1988). 309. (a) Ishii, Y., and Sakata, Y., J. Org. Chem. 55, 5545 (1990); (b) Balistreri, F. P., Failla, S., Spina, E., and Tomaselli, G. A,, J. Org. Chem. 54,947 (1989). 310. Oguchi, T., Sakata, Y., Takeuchi, N., Kaneda, K., Ishii, Y., and Ogawa, M., Chem. Lett. 1989, 2053; Sakata, Y., and Ishii, Y.,J. Org. Chem. 56,6233 (1991). 311. Sakata, Y., and Ishii, Y.,J. Org. Chem. 56,6233 (1991). 312. Shimizu, M., Orita, H., Hayakawa, T., Watanabe, Y., and Takehira, K., Tetruhedron Lett. 30, 471 (1989).

313. Sakae, S., Sakata, Y., Nishiyama, Y., and Ishii, Y., Chem. Lett. 1992,289. 314. Csanyi, L. J., and Jaky, K.,J. Mol. Curd 61,75 (1991). 315. Aubry, C., Chottard, G.,Platzer, N., Bregeault, J.-M., Thouvenot, R., Chauveau, F., Huet, C., and Ledon, H., Inorg. Chem. 30,4409 ( 1990). 31 6. Duncan, D. C., Chambers, R. C., Hecht, E., and Hill, C. L., J. Am. Chem. SOC,117,68 I (1 995). 317. Itoh, K., Hashimoto, M.. and Misono, M., Proc. 70th National Meeting of The Catalysis Society of Japan, 4F424 (1992). 318. Neumann, R., and Gara, M., J. Am. Chem. SOC.116,5509 (1994). 319. Gall, R. D., Fajaj, M., and Hill, C. L., Inorg. Chem. 33,5015 (1994). 320. (a) Faraj, M., and Hill, C. L., J. Chem. Soc.. Chem. Commun. 1987, 1487; (b) Hill, C. L., in “Activation and Functionalization of Alkanes” (C. L. Hill, Ed.), p. 243. Wiley, New York, 1989.

321. Neumann, R., and A.-Gnim, C., J. Chem. SOC.,Chem. Commun. 1989, 1324. 322. Hill, C. L., and Brown, R. B., Jr., J. Am. Chem. SOC. 108, 536 (1986); Neumann, R., and A.-Gnim, C.,J. Am. Chem. SOC.112,6025 (1990). 323. Faraj, M., Lin, C.-H., and Hill, C. L., New J. Chem. 12,745 (1988). 324. (a) Ai, M., in “Proc. 8th Intern. Congr. Catal., ” Berlin, 1984. Verlag Chemie, Weinheim, 1985, Vol. 5, p. 475; (b) Centi, G.,Nieto, J. P., Briickman, C. K., and Serwicka, E. M., Appl. Cutul. 46, 197 (1989); (c) Centi, G.,Lena, V., Trifiro, F., Ghoussoub, D., Aissi, C. F., Guelton, M., and Bonnelle, J. P., J. Chem. Soc.. Furuduy Trans. 86,2775 (1990). 325. Tatibouet, J.-M., Che, M., Amirouche, M., Fournier, M., and Rocchiccioli-Deltcheff, C., J. Chem. Soc.. Chem. Commun. 1988, 1260. 326. Rocchiccioli-Deltcheff, C., Amirouche, M., Herve, G., Fournier, M., Che, M., and Tatibouet, J.-M.,J. Catd. 126, 591 (1990). 327. Mizuno, N., Watanabe, T., Mori, H., and Misono, M.,J. Cutul. 123, 157 (1990). 328. Misono, M., Mizuno, N., Mori, H., Lee, K. Y., Jiao, J., and Okuhara, T., Stud. Surf: Sci. Cutul. 67,87(1991).

250

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO

329. Mizuno, N., and Misono, M., Chem. Lett. 1984,669. 330. Artz, D., Catal. Today 18, 173 (1993). 331. Sheldon, R. A., “Dioxygen Activation and Homogeneous Catalytic Oxidation,” p. 573. Elsevier, Amsterdam, I99 1. 332. Mizuno, N.. Watanabe, T., and Misono, M., Bull. Chem. SOC.Jpn. 64,243 (1991). 333. Mori, H., Mizuno, N., and Misono, M., J. Catal. 131, 133 (1990). 334. (a) Ai, M., J. Catul. 71,88 (1981);Ai, M., in “Proc. Climax 4th Int. Conf. Chemistry and Usage of Molybdenum” (H. F. Barry and P. C. H. Mitchell, Eds.), p. 306. Climax Molybdenum Co.. Ann Arbor, 1982; (b) Ai, M., Tsai, T., and Ozaki, A., Bull. Chem. SOC.Jpn. 53,2647 (1980). 335. Ueshima, M., Tsuneki, H., and Shimizu, N., Hyoumen 24, 582 (1986); 130 Shimizu, N., fetrotech 6,778 (1983). 336. Black, J. B., Claydon, N. J.. Gai, P. L., Scott, J. D., Senvicke, E. M., and Goodenough, J. B., J. Catal. 106, 1 (1987). 337. Tatematsu, S.. Masters Thesis, The University of Tokyo, 1984. 338. Otake M., and Onoda, T., Shokubai (Catalyst) 18,169 (1976). 339. Otake, M., and Onoda, T., in “Proc. 7th Int. Congr. Catal., Tokyo, 1980,” p. 780. KodanshaElsevier. Tokyo and Amsterdam, 1981. 340. Akimoto, M., Tsuchida, Y., Sato, K., and Echigoya, E., J. Catal. 72,83 (1981). 341. Abouka’is, D., Ghoussoub, A,, B.-Crusson, E., Rigole, M., and Guelton, M., Appl. Catal. A 11I, 109 ( 1994). 342. Lee. K. W., Oishi, S., and Misono, M., 67th National Meeting of The Chemical Society of Japan, Tokyo, 2M 1 10, 1994. 343. Watzenberger, O., Emig, G., and Lynch, D. T., J. Catal. 124,247 (1990). 344. McGarvey, G . B., and Moffat, J. B., J. Catal. 132, 100 (1991). 345. Briickman, K., Tatibouet, J.-M., Che, M., Serwicka, E., and Haber, J., J. Catal. 139, 455 (1993). 346. Centi, G., Catal. Lett. 22, 53 (1993). 347. Krylov, 0. V., Cafal. Today 18, 209 (1993); Periana. R. A., Taube, D. J., Evitt, E. R., Loffler, D. G., Wentreck, P. R., Voss, G.. and Masuda, T.. Nafure 259,340 (1993); Hickman, D. A,, and Schmidt, L. D., Nature 259,343 (1 993). 348. Hong, S . S., and Moffat, J. B., Appl. Catal. A 109, 1 17 (1994). 349. Kim, Y.-C., Ueda W., and Moro-oka, Y., J. Chem. Sac., Chem. Commun. 1989,652. 350. Centi, G., Ed., “Vanadyl Pyrophosphate Catalysts,” Cafal. Today 16 (1993). 351. The Chemical Engineer, 13 Sept., 3 (1990). 352. Kringer, H., and Kirch, L. S., European Patent 0010902 (1979). 353. Centi, G., and Trifiro, F., Catal. Sci. Technol. 1,225 (1991). 354. Briickman, K., and Harber, J., “Proc. 10th Intern. congr. Catal., Budapest, 1992,” p. 742. Elsevier, Budapest. 1993. 355. Mizuno, N., Tateishi T., and Iwamoto, M., J. Chem. SOC., Chem. Cornmun. 1994, 141 I ; Appl. Catal. A 118, L I ( 1994). 356. JP Patent (Jpn. Kokai Tokkyo Koho); H.Imai, M. Nakatsukasa, and A. Aoshima, 62/132832, H. Imai, T. Yamaguchi, and M. Sugiyama, 631145249, S. Yamamatsu, and Y. Yamaguchi, 2142033 357. Neumann, R., and Levin, M., J. Org. Chem. 56,5707 (1991). 358. Soeda, H., Okuhara, T., and Misono, M., Nippon Kagaku Kaishi 1993,917. 359. Kozhevnikov, 1. V., Kulikov, S. M. M., Timeofeeva, N., Krysin, A. P., and Titova, T. F., React. Kinet. Caral. Lett. 45,257 (1991). 360. Kinomura, K., Kitazawa, S., Takata, Y., and Sakakibara, T., USP. 4874852; Kinomura, K., and Sakakibara, T., 56th National Meeting of the Chemical Society of Japan, 1988, Abstr. No. I IXB40-41.

CATALYTIC CHEMISTRY O F HETEROPOLY COMPOUNDS

25 1

3 6 l . Kozhevnikov, 1. V., Vasilieva, I . B., and Zarutskii, V. V., React. Kine/. Catul. Le//. 47, 83

( I 992). 362. Kozhevnikov, 1. V.. Kulikov, S. M., Chukaeva, N. G., Kirasanov, A. T., Letunova, A. B., and

Blinova, V. I., React. Kine/. Caral. Lett. 47,59 (1992). 363. Timofeeva, M. N., Grosheva, V. S . , Demchenko, B. I., and Kozhevnikov, 1. V., React. Kiner.

Cutul. Lett. 45.2 15 ( 199 I ). 364. Katsoulis, D. E., and Pope, M. T.,J. Chem. SOC.Chem. Commun. 1986, 1186. 365. Hill, C. L., and P., McCartha, C. M., Coord. Chem. Rev. 143,407 (1995). 366. Mansuy, D., Bartoli, J.-F., Battioni, P., Lyon, D. K., and Finke, R. G., J. Am. Chem. SOC. 113,

7222 (1991). 367. Urabe, K., Kimura, F., and Izumi, Y., “Proc. 7th Int. Congr. Catal., Tokyo, 1980,” p. 1418.

Kodansha, Tokyo, Elsevier, Amsterdam, I98 I . 368. Grate, J. H., Hamm, D. R., and Mahajan, A,, Preprint distributed at 14th Conference on 369. 370. 37l. 372. 373. 374. 375. 376. 377. 378.

379. 380. 381. 382. 383. 384. 385. 386. 387. 388. 389. 390. 391. 392. 393. 394. 3Y5. 396. 397. 398.

Catalysis of Organic Reactions, Albuquerque, New Mexico, April 27-29, 1992. Izumi, Y., Satoh, Y., Kondoh, H., and Urabe, K.. J. Mol. Ca/al. 72,37 (1992). Wegmann, R. W., J. Chem. Soc.. Chem. Commun. 1994,947. Urabe, K., Tanaka, Y., and Izumi, Y., Chem. Left. 1985, 1595. Siedle, A. R., Markell, C. G.. Lyon, P. A,, Hodgson, K. 0..and Roe, A. L., Inorg. Chem. 26, 2 I9 ( 1987). Lyon D. K., and Finke, R. G., Inorg. Chem. 29, 1787 (1990). Mizuno. N., Lyon, D. K., and Finke, R. G., J. Card. 128,84 ( 1991 ). .Geletti, Y.V., and Shilov. A. E., Kinet. Katal. 24,486 (1983). Finke, R. G., Lyon, D. K., Nomiya, K., and Weakley. T. R. J., Acta Cn.s/allogr. C46, 1592 ( 1990). Finke, R. G.. Lyon, D. K., Nomiya, K., Sur, S., and Mizuno, N., Inorg. Chem. 29, 1784 (1990). (a) Ono, Y., Taguchi. M., Gerile, Suzuki, S., and Baba, T., in “Catalysis by Acids and Bases” (B. lmelik ct a / . ,Ed.), p. 167. Elsevier, Amsterdam, 1985; (b) Ono, Y., Baba, T., Kanae, K., and Seo, S . G., Nippon Kagaku Kaishi 1988,985. Baba. T.. and Ono, Y., Appl. Cutal. 55,301 (1989). Suzuki, S.. Kogai, K., and Ono, Y., Chem. Le//.1984,699. Na, K., Okuhara, T., and Misono, M.,J. Chem. Soc.. Chem. Commun. 1993, 1422. Na, K., Okuhara, T., and Misono, M., in “Catalytic Science and Technology II” (Y. Izumi. H. Arai and M. Iwamoto, Eds.) p. 245. Kodansha, Tokyo, Elsevier, Amsterdam, 1995. Brindley. G. W., and Semples, R. E., Clav Miner 12,229 (1977). Shabtai. J.. Lazar, R., and Oblad, A. G., in “Proc. 7th Int. Congr. Catal., Tokyo. 1980” (T. Seiyama and K. Tanabe, Eds.), p. 828. Kodansha, Tokyo, Elsevier, Amsterdam, 1981. Kikuchi, E., Nakano, M.. Fukuda, I., and Morita, Y.,J. Jpn. fefrol.Ins/. 27, 153 (1984). Urabe, K., Sakurai, H., and Izumi, Y.,J. Chem. Soc.. Chem. Commun. 1986,1074. Bartley, G. J. J., and Burch, R., Appl. Cu/a/.28,209 (1986). Mori, H., Okuhara, T., and Misono, M., Chem. Lert. 1987,2305. Kwon, T., and Pinnavaia. T. J., Chem. Mat. 1,381 (1989). Narita, E., Kaviratna, P., and Pinnavaia, T. J.. Chem. Left. 1991, 805. Kwon, T.. Tsigdinos, G. A,, and Pinnavaia, T. J., J. Am. Chem. SOC. 110,3653 (1988). Narita. E., Kaviratna, P. D., and Pinnavaia, T. J., J. Chem. SOC.Chem. Cummun. 1993,60. Drezdon, M. A., Inorg. Chem. 27,4628 ( 1988). Dimotakis, E. D., and Pinnavaia, T. J., lnorg. Chem. 29,2393 (1994). Tatsumi, T.. Yamamoto, K., Tajima, H., and Tominaga, H., Chem. Left. 1992,815. Rindl. M., S. AfricunJ. Sci. 11,362 (1916). Yamase, T., Chrm. Lett. 1973,615. Yamase, T., Hayashi, H., and Ikawa, T., Chem. Left. 1974, 1055.

252 399. 400. 401. 402. 403. 404. 405.

406. 407. 408.

409. 410. 411. 412. 413. 414. 415.

416. 41 7.

TOSHIO OKUHARA, NORITAKA MIZUNO, AND MAKOTO MISONO Yamase, T., and Ikawa, T., Bull. Chem. SOC.Jpn. 50,746 (1977). Yamase, T., J. Chem. Soc.. Dalton Trans. 1978,283. Yamase, T., J. Chem. Soc.. Dalton Trans. 1987,1982. Yamase T., and Kurozumi, T., J. Chem. SOC.,Dalton Trans. 1983,2205. Papaconstantinou, E., Dimotikali, D., and Politou, A., Inorg. Chim. Acta 46, 155 (1980). Papaconstantinou, E., J. Chem. Soc.. Chem. Commun. 1982,12; Chem. SOC.Rev. 18,l (1989). Hill, C. L., Synlett. 127 (1995). Fox, M., Cardona, A. R., and Gaillard, E., J. Am. Chem. SOC.109,6347 (1987). Renneke, R., Pasquali, F. M., and Hill, C. L., J. Am. Chem. SOC.112,6585 (1990). Attanasio, D., and Suber, L., Inorg. Chem. 28,3779 (1989). C.-Walke, L. A., and Hill, C. L., J. Am. Chem. SOC.114,938 (1992). Danvent, J. R., J. Chem. SOC.,Chem. Commun. 1982, 798; Akid, R., and Danvent, J. R., J. Chem. Soc., Dalton Trans.. 1985,395. Rouillier, M. C.. These de 3eme cycle, Nancy, France, 1968. Keita, B., and Nadjo, L.,J. Electroanal. Chem. 191,441 (1985). Keita, B., and Nadjo, L., J. Electroanal. Chem. 199,229 (1986). Maevoy A. J., and Gratzel, M., J. Electronal. Chem. 209,. 391 (1986). Bidan, G., Genies, E. M., and Lapkowski, M., J. Chem. Soc.. Chem. Commun. 1988,533. Toth, J. E., and Anson, F. C., J. Am. Chem. Soc. 111,2444 (1989). Toth, J. E., Melton, J. D., Cabelli, D. B., Bielski, H. J., and Anson, F. C., Inorg. Chem. 29, 1952

( 1990). 418. Fukuma, M., Deto, Y., and Yamase, T., Antiviral Res. 16,327 (1991). 4 1 9 Hill, C. L., Weeks, M. S., and Schinazi, R. F., J. Med. Chem. 33,2767 (1990).

420. Kudo, T., Ishikawa, A,, Okamoto, H., Miyauchi, K., Murai, F., Mochiji, K., and Umezaki, H., J. Electrochem. Soc. 134,2607 (1987). 421. Yoshimura, T., Ishikawa, A., Okamoto, H., Miyazaki, H. Sawada, A., Tanimoto, T., and Okazaki, S., Microelecfron. Eng. 14, 149 (1991). 422. Tell, B., and Wagner, W., Appl. Phys. Lett. 33,837 (1978). 423. Yamase, T., and Naruke, H., Coord. Chem. Rev. 111,83 (1991). 424. Nakamura, O., Kodama, T., Ogino, I., and Miyake, Y., Chem. Lett. 1979, 17. 425. Kreuer, K. D.. Hampele, M., Dolde, K., and Rabenau, A.. Solid State lonics 28-30, 589 (1988).

ADVANCES IN CATALYSIS, VOLUME 41

Microporous Crystalline Titanium Silicates BRUNO NOTARI Notari Tecnologie. SnC 22 Via Triulziana. 20097 San Donato Milanese, Italy

New plants are successhlly operating with clean and efficient process technologies. This is the most significant development made possible by the discovery, in 1983, of titanium-containing crystalline silicas and their unique catalytic properties, especially in selective oxidation reactions with H202 as the oxidant. These important results have stimulated many research workers in universities and industrial research laboratories in the world to investigate the particular state of aggregation and coordination that Ti” assumes when forced into framework positions of hydrophobic crystalline silicas. Researchers are also engaged in the search for other compounds containing titanium and silicon oxides with Ti” in the same coordination and environment, on the assumption that similar catalytic properties would be obtained. Relevant discoveries have been made, and additional valuable information has been obtained on this new class of materials and on their catalytic performance in many different reactions. The purpose of this chapter is to review the subject after more than ten years of intense research activity, to summarize the well documented facts, and to present the hypotheses that have been advanced to explain them, providing guidelines for the correct evaluation of the large amount of data available in the literature. As such, this chapter should prove of value to research chemists and managers interested in a concise survey of this new important development in oxidation chemistry. 1.

Introduction

Zeolites have had a major impact on catalytic science and technology. The activities of zeolite-containing catalysts for numerous acid-catalyzed reactions are many orders of magnitude higher than that of the previously used SO2A1203;and the introduction of zeolites in the catalytic cracking process in the 1960s resulted in great benefits in product yields, reduced pollution, and energy 253 Copyright 0 1996 by Academic F’rcss. Inc. All rights of repduction in any form reserved.

254

BRUNO NOTARI

savings. In the 1970s, the discovery of ZSM-5 and other pentasil zeolites, with their intermediate pore sizes and high thermal stabilities in the protonic form, led to the discovery of new reactions made possible by the combination of shape selectivity and acidity. These discoveries were rapidly followed by the development of new industrial processes. The success of zeolites in numerous industrial catalytic processes stimulated the search for catalysts containing elements different from A13+ in a silica lattice. These elements could, in principle, impart catalytic properties to the solid for many reactions other than those catalyzed by acids while offering the advantages of atomic dispersion, which is difficult to achieve with traditional catalyst preparation methods. The idea of introducing foreign atoms into solids to change their catalytic properties is at the very heart of heterogeneous catalysis; many industrial catalysts are obtained by modifying the composition of a solid to change important physical, chemical, and catalytic properties. However, in many cases these changes involve only the surface or the outer layers of a stable solid, whereas the introduction of foreign atoms in the crystal lattice of a silica involves a change in the bulk structure. Attempts to produce zeolitic structures containing Ti" were made as early as 1967, but evidence of the crystalline nature of the solids and of the effectiveness of the substitution were not given (Young et al., 1967). The isomorphous substitution of Si" by Ti" was claimed by Taramasso, Perego, and Notari in 1983 for a new material with the composition xTi02( 1 - x)Si02 (0.0 5 x 5 0.04 M ) . This has the crystalline structure of silicalite-1 (or MFI) with Ti" in framework positions; it was named titanium silicalite-1 or TS-1 (Taramasso et al., 1983). The occurrence of isomorphous substitution was deduced from the regular increase in unit-cell parameters with the degree of substitution and from the good agreement between the observed and calculated values of the Si-0 and Ti-0 distances. The same type of evidence had already been obtained by the same authors in the synthesis of crystalline microporous boron silicates, where the smaller B-0 distance relative to Si- 0 causes a decrease in unit-cell parameters (Taramasso et al., 1980). The catalytic properties of TS-1 appeared to be unique. In oxidation reactions, with H202 as the oxidant, many organic compounds could be oxidized selectively and efficiently. Partial oxidation products could be obtained in high yields, and almost all the oxygen available from H202 was used to produce the desired compounds, with only slight decomposition to give H2O and 0 2 (Neri et al., 1985, 1986; Romano et al., 1990). For quite a few years, however, these new findings met with skepticism. Many objections were raised-on the one hand because isomorphous substitution of Si" by Ti" was considered unlikely, and on the other, because difficulties arose in duplicating the properties of the new material in other laboratories. The objections concerning the structure were based on the observation that, for

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

255

any substitution to take place, it is necessary for the guest ion to assume the same coordination of the ion being replaced, in this case the tetrahedral Si’v-O coordination. But the widespread belief was that “substitution of framework silicon by titanium does not occur in aluminosilicates owing to the preference of titanium to be octahedrally bound rather than tetrahedrally bound” (Hartman, 1969).In fact, the minerals zorite and vinogradovite, consisting of Na, Ti, and Si oxides, have Ti” in octahedral coordination. Applying the Pauling rule of coordination based on the ratio of the ionic radii pl of the cation and the anion, the value for Ti and 0 (p = 0.5 15) falls well outside the range (0.225-0.414) for which tetrahedral coordination is expected; and therefore, substitution of Si” by Ti” seemed unlikely (Ione et al., 1985). This led to reinforcement of the conclusion that “little replacement would be expected of silicon with titanium or zirconium” since only elements comparable in size and chemical properties could give rise to thermodynamically stable solids (Barrer, 1984; Tielen et al., 1986). With specific reference to the data presented for TS- 1, it was pointed out that “no significant change in the X-ray diffraction pattern of silicalite was observed and the minor changes observed may simply be due to the occluded titanium dioxide” (Lok et al., 1985). More detailed information on TS-I was presented (Perego et al., 1986), but because of the difficulties inherent in the synthesis of the new material, the skepticism continued. In 1987, the successful startup of a new process was announced for the production of 10,000 tonslyear of catechol and hydroquinone by the selective oxidation of phenol with HzOz catalyzed by TS-1 at the Enichem plant in Ravenna, Italy (Notari, 1988). Soon thereafter, it was disclosed that another new process for the production of cyclohexanone oxime from cyclohexanone, H 2 0 2 , and NH, with TS- 1 as the catalyst was being developed (Roffia et al., 1990).The fact that a material with unusual catalytic properties had been obtained was then finally recognized, and the interest in titanium-containing catalysts spread rapidly in the scientific community, especially in industrial research laboratories. In the meantime, the synthesis method was studied and described in more detail; and when all the necessary precautions were taken, TS-1 was reproduced in other laboratories, as were the highly selective catalytic reactions. The subsequent work confirmed that Ti” can assume the tetrahedral coordination necessary for isomorphous substitution of Si’” and added valuable information about the structure, properties and catalytic performance of the material. New reactions catalyzed by TS- 1 have been discovered, and new synthetic methods Abbreviations: /I, ratio between ionic radii of cation and anion; TBHP. terr-butyl hydroperoxide; EBHP, ethylbenzene hydroperoxide; PO, propylene oxide; TBOT, tetrabutyl orthotitanate; TEOT, tetraethyl orthotitanate; TIOT, tetraisopropyl orthotitanate; TEOS, tetraethyl orthosilicate; THF, tetrahydrofuran; TPA-OH, tetrapropylammonium hydroxide; TMOS, tetramethyl orthosilicate; TBA-OH, tetrabutylammonium hydroxide; TBP-OH, tetrabutylphophonium hydroxide; TEA-OH, tetraethylammonium hydroxide; TMA-OH, tetramethylammonium hydroxide.

256

BRUNO NOTARI

for producing TS- 1 have been developed. Furthermore, new titanium-containing microporous crystalline materials have been discovered: titanium silicalite-2 (TS-2), with the MEL structure (Bellussi et a1.,1989; Reddy, J. S. et al., 1990); Ti-ZSM-48, with the ZSM-48 structure (Serrano et al., 1992); Ti-HMS, with the hexagonal mesoporous silica structure; and Ti-MCM-41, with the MCM-41 silica structure (Tanev et al., 1994; Corma et al., 1994). Simultaneous incorporation of Ti" and other elements has been obtained in silicalite-1 and silicalite-2 with Ti-Al, Ti-Ga, Ti-Fe (Bellussi et al., 1988a, 1988b, 1988~;Reddy, J. S . et al., 1994a), Ti-Ge, Ti-Sn, and Ti-V combinations (Kornatowski et al., 1993); in zeolite /?with Ti and A1 (Camblor et al., 1992); and in silicoaluminophosphate SAPO-5 (Tuel et al., 1994c) aluminophosphates ALP04-5 and ALP04-11 (Ulagappan et al., 1995). Titanium containing materials have been investigated for various reactions, but selective oxidations with H202 as the oxidant have attracted the most interest. For these reactions, the formation of surface titanium peroxo compounds with H202 and the subsequent transfer of the peroxidic oxygen to the organic reactants have been proposed to explain the mechanism by which titanium participates in the catalytic cycle (Notari, 1988). Many industrial processes use air or oxygen as oxygen donors, and whenever possible, this is the most economical way of carrying out oxidations. But sometimes this proves to be impractical, in which case other oxygen donors must be used, with H202 often being preferred. H202 has the advantage of giving environmentally benign water as its by-product. Furthermore, H202 ranks first in terms of available oxygen in the list of oxygen donors (Shirmann et a]., 1980; Sheldon, 1980). The industrial use of H202has been limited by the low selectivity that could be obtained using available catalysts. Undesirable reactions readily set in, the most common being the decomposition of H202 into O2 and H20, which is catalyzed by many different materials, even in trace amounts. For this reason, other oxidants have been preferred in industrial applications, particularly tertbutyl hydroperoxide (TBHP) and ethylbenzene hydroperoxide (EBHP), which are used for the production of large-tonnage chemicals such as propylene oxide. The possibility of improving the economics of these and many other oxidation processes with catalysts selective enough to render the use of H202 costeffective is thus clearly justified. In oxidation processes one of the conditions for obtaining high yields is the fast removal of the initially formed oxidation products in order to minimize their hrther oxidation into less valuable products. In heterogeneous catalytic oxidations, this goal is met by the use of low-surface-area, large-pore catalysts that allow rapid diffision of products out of the particles. It is thus surprising that titanium-based catalysts, despite their high surface areas and microporous structures, exhibit selectivities as high as 80% and sometimes even 95%.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

257

This selectivity advantage, combined with the advantages of regenerability and the ease of separation from products that characterize any solid catalyst, justified the increased attention paid to these materials. The discovery and application of titanium silicalite are regarded as milestones in zeolite catalysis (Holderich, 1989). The unusual properties of titanium silicalites have been attributed to the presence of Ti" in framework positions of the Si02 lattice. It is important to realize that there is a limit to the extent of substitution;the exact value is still under discussion, but is certainly not more than a few percent. Very likely, the structure of crystalline silica is not stable at higher degrees of substitution. This suggestion is consistent with theoretical predictions of lack of substitution; these predictions referred mainly to high degrees of substitution, which have not been observed. If Ti'" is in excess of the amount that can fit in framework positions, extraframework oxide phases are formed; their high degrees of dispersion make their detection and correct identification very difficult, even with the most sophisticated techniques. They are indicated to be TiO, nanophases. With larger amounts of Ti'" or after severe treatments, Ti02 anatase phases are formed, and these are more easily detected (Millini et al., 1992). When extra-framework Ti'" is present, the H202efficiencies of the catalysts are drastically decreased. This behavior has been interpreted by assuming that TS-1 has a very low activity for the H202decomposition reaction; this consequently kinetically favors the oxidation of the organic reactants, whereas the TiO, or Ti02 anatase phases, as well as acids or bases that might be present, catalyze the H 2 0 2decomposition and other secondary reactions, lowering yields and efficiency of H 2 0 2 utilization (Notari, 1988; Huybrechts et al., 1992). Another important point is that, when prepared from pure raw materials, titanium silicates do not have an appreciable acidic character, as demonstrated by the high yields that can be obtained even in applications with acid-sensitive products like propylene oxide. In contrast, mixed oxides of titanium and silicon have been described as being strongly acidic (Tanabe et al., 1981). The reasons for the difference are not clear and deserve firther attention. The goal of this review is to summarize the literature and to provide guidelines for interpreting the rapidly expanding literature of titanium silicates and their role in oxidation catalysis.

II. Mixed Oxides of Ti and Si A.

CATALYTIC ACTIVITY

The catalytic activity of titanium compounds for many different reactions has been recognized for a long time. A Ti02-SO2 catalyst consisting of 2% Ti02 dispersed on high-surface-area Si02 is an active and selective catalyst for the

258

BRUNO NOTARI

epoxidation of olefins with hydroperoxides (Wulff et al., 1971 ). A process for the production of propylene oxide (PO) from propylene and EBHP or TBHP based on this catalyst has been developed by Shell Oil, and a plant for the production of 130,000 tons of PO per year using this technology has been operating for years in Moerdijk, The Netherlands. The Ti02-Si02 catalyst can operate only with hydroperoxides, being rapidly deactivated by aqueous H202. Other titanium compounds such as titanium diacetylacetonate or tetrabutyl orthotitanate (TBOT) were found to be active catalysts for the epoxidation of olefins with extremely high selectivity (98%), but only with hydroperoxides (Sheldon et al.., 1973). Additional applications have recently been demonstrated for Ti02-Si02 catalysts obtained as large pore aerogels. Their catalytic activity increases with Ti02 content up to 20% Ti02 as a consequence of the high degree of dispersion. Using cumene hydroperoxide as the oxidant, large molecules such as cyclododecene and norbornene have been oxidized to the corresponding epoxides with 97% selectivity (Hutter et ale.,1995). As will be made clear in Section 111, it is possible that finely dispersed Ti02 is present in titanium silicates. The observed catalytic activity would then be attributed to a combination of the two materials. It is therefore useful to review the chemistry of high-surface-area Ti02-Si02 and its catalytic behavior. B. STRUCTURE From the difference in electronegativity between 0 (3.5) and Ti (1.32), it is inferred that the Ti-0 bond has a marked ionic character, but the designation of these ions as Ti4+ and 02-seems exaggerated; the notation Ti" will therefore be used, except in the case of reductiotdoxidation reactions where Ti4+ and Ti3+ better clarify the transformation taking place. The composition of the mixed oxides can span the full range from Ti02 to Si02 and has a major influence on the properties of the materials. The procedures used to prepare the mixed oxides are also important in influencing structure and catalytic properties. The procedures can be summarized as follows: 1. Separate precipitation of titanium- and silicon-containing precursors followed by mixing of the precipitated products. 2. Coprecipitation of titanium and silicon hydroxides from their precursors, possibly by sol-gel techniques. 3. Solvent removal from the gels. 4. Deposition of a titanium precursor onto preformed Si02 followed by transformation of the precursor into Ti02. Frequently used precursors are TiC14; titanium alkoxides Ti(OR)4 such as tetraethyl orthotitanate (TEOT) and tetraisopropyl orthotitanate (TIOT); sodium

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

259

silicate; and the alkoxides Si(OR)4, mainly tetraethyl orthosilicate (TEOS), in water or water-alcohol solvents. The use of inorganic materials, particularly silicates, can introduce impurities that significantly change the properties of the final materials. For this reason, the use of organic derivatives, which can be more easily purified by fractional distillation, has been preferred. The pH at which precipitation is carried out has a major influence on the structure of the products obtained. For example, the amorphous silica obtained by hydrolysis of TEOS under acidic conditions has a surface area of 900 m2/g and is weakly crosslinked, with an asymmetric Si-0 stretching vibration at 1030 cm-', whereas the hydrolysis of TEOS obtained under neutral or basic conditions produces a silica having a surface area of 400 m2/g which is strongly crosslinked, with the asymmetric Si-0 stretching vibration at 1100 cm(Schraml-Marth et al., 1992; Miller et al., 1994; Liu et al., 1994). When prepared by precipitation, the mixed oxides have structures that depend largely on the relative concentration of the two precursors and on the conditions under which hydrolysis and mixing are carried out. In the separate precipitation of the two precursors prior to mixing, there is a tendency to form large clusters of the individual components. This condensation is not reversible; therefore, after calcination at high temperature, an intimate mixture of anatase embedded in a matrix of amorphous Si02 gel is obtained. Coprecipitation of precursors mixed before precipitation gives materials with a higher degree of interaction between the two metal oxides, although the titanium-containing precursors in general hydrolyze faster than the silicon-containing precursors, with the possibility that a separate precipitation may take place. Solvent removal from the gels has a strong influence on the structure of the materials. Simple evaporation of solvent gives the xerogel, whereas extraction with supercritical C02 gives the aerogel with much larger pore volume and pore diameter (Miller et al., 1994; Hutter et al., 1995). Obviously, the properties of the materials depend strongly on their compositions. Early studies have shown that when NH3 is used to obtain coprecipitated Ti02-SiO2, the material is X-ray-amorphous; no crystalline anatase phase could be detected up to TiOz concentrations of 15% (Tanabe et al., 1981; Anpo et al., 1986). It was later found that when the sol-gel method of coprecipitation is used with titanium and silicon alkoxides as precursors, the concentration of TiOz at which no crystalline phase can be observed is 67%, indicating a remarkable effect of the Si02 component in stabilizing small Ti02 particles. The high degree of dispersion persists even after calcination at 873 K; crystallization takes place at temperatures higher than 1000 K. In the Ti02-rich region, microdomains of crystalline anatase are formed (Schraml-Marth et al., 1992; Doolin et al., 1994; Toba et al., 1994; Miller et al., 1994). A method was developed to increase the formation of Si-0-Ti bonds; it involves the partial hydrolysis of TEOS followed by the addition of the

'

260

BRUNO NOTARl

appropriate amount of TIOT and the addition of H2O to complete the hydrolysis. A simplified representation of the many reactions taking place is the following:

* (C2H50)2Si(OH)2+ 2C2H50H (C2Hs0)2Si(OH)2 + Ti(OR)4 * (C2HSO)SiOTi(OR), + ROH I Si(OC2Hs)4+ 2H20

(1) (2)

OH

In this way the titanium-containing precursor is forced to react with the partly hydrolyzed silicon-containing precursor, and Si- 0-Ti bonds are formed. These bonds persist in the dry and calcined mixed oxides (Schraml-Marth et al., 1992). A significant improvement has been obtained by performing the dehydration of the gel with supercritical C 0 2 at low temperature. An aerogel is obtained with a high Si-0-Ti connectivity, with the result that many Ti" are surrounded by 0- Si -0- Si-0 structural elements in all directions, thus becoming isolated from other Ti". These isolated Ti" are considered the most active species in olefin epoxidation with hydroperoxides (Hutter et al., 1995a, b, c, d), similarly to what has been proposed for Titanium Silicalite-1 with H202. The UV-visible diffuse reflectance spectra of these mixed oxides show that the characteristic absorption edge of anatase at 30,500 cm-' (or 330 nm) shifts progressively as the TiOl content is decreased, reaching a value of 40,000 cm(or 250 nm) as shown in Fig. 1. A high energy shift of the absorption edge is indicative of decreasing dimensions of particles, and therefore in the low-Ti02 region, Ti02 must be present as very small particles dispersed in the Si02 carrier matrix. When the preparation involves deposition of a titanium-containing precursor onto a preformed Si02, there is no doubt that an overlayer of Ti02 on Si02 is obtained. The presence of H 2 0 adsorbed on the Si02 can cause the hydrolysis of

'

100 75 ' : \

50. 25. 0-

600

500

400

300

Wavelength, nm FIG.1. UV-Visible diffuse reflectance spectra of TiO2-SiO2 at 298 K. (Reprinted with permission from Anpo ef nl. Copyright 1986 American Chemical Society.)

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

26 1

the titanium alkoxide, with the resultant formation of bulk anatase. But when the adsorbed H 2 0 is removed, the alkoxide reacts only with the SiOH groups present at the surface of the Si02, producing a Ti02 coating with an extremely high dispersion (Muiioz-Paez et al., 1991). The unusual properties of Ti02 highly dispersed on Si02 have been investigated with materials containing 7% Ti02 on preformed Si02. Upon treatment in H2 at 723 K, Ti4+ is reduced to Ti3+. The formation of Ti3+ was demonstrated (Fig. 2) by the low-temperature adsorption of CO (Femandez el al., 1988). The results show that when titanium oxide is highly dispersed on S O 2 , the reducibility of Ti4+ is significantly enhanced. The dispersed phase that diluted Ti02 forms on S O 2 has been investigated with materials prepared by the reaction of Ti-(O-t-Bu), in tetrahydrofiran (THF) with Cabosil fumed silica; samples contained 0, 1, 2, or 3 wt% Ti02. Raman spectra (Fig. 3) indicate that at the 1% loading, Ti02 cannot be detected, whereas at the 2% loading, Ti02 phases identified as TiO, surface nanophases

I

3

x2

m d

Y L 469

461

453 B.E. /eV

FIG. 2. Ti 2p photoelectron spectra of Ti02-Si02: (a) calcined; (b) reduced in H2 at 773K; (c) difference spectrum (b - a). (From Femandez et al., 1988.)

262

BRUNO NOTARI

TiO$3i02 ( Cab-0-Sil )-973 K Treatment

.a s -E

-

surface titania species

v)

1200

1000

800

600

400

200

Raman shift / cm-1 FIG. 3. Raman spectra of Cabosil Si02 incorporating up to 3 w t % TiOl. Weak features suspected to be due to dispersed Ti02 are visible in the spectra of samples containing 2 and 3 wt% TiOz. Crystalline Ti02 is first evident in the sample containing 3 wt% Ti02. (From Srinivasan el d., 1991.)

were observed, but with difficulty. At the 3% loading, both surface nanophases and crystalline anatase phases were detected (Srinivasan et al., 1991). The mixed oxides obtained by hydrolysis at neutral pH of TIOT and TEOS over a wide range of compositions have also been investigated. The X-ray absorption near-edge spectra (XANES) of the mixed oxides having a composition corresponding to Ti : Si = 1 : 8 (Fig. 4) indicate that Ti" is present not in an octahedral coordination, as in anatase, but instead in a lower coordination, 5- or a mixture of 4-,5-, and 6-coordination. The structural arameters derived from XANES for the Ti : Si = 1 : 8 material indicate that Ti' $is present with a Ti-0 distance of 1.82 8, shorter than the Ti-0 distance in Ti02, which is 1.94 A. The value of 1.82 8 is typical of Ti'" residing in a tetrahedral environment. The domain size of the Ti02 particles could not be determined with precision, the only information being that these domains must be significantly smaller than 10 A. Most of the Ti'" in these domains must form Ti-0-Si linkages rather than Ti-0-Ti linkages. It has therefore been concluded that in the Si02-rich region and with dry samples, the majority of Ti" ions occupy sites of tetrahedral symmetry, directly substituting for Si" in the mixed oxides. The Ti" coordination increases when the material is exposed to HzO and reversibly returns to the previous state under dehydrating

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

4960

4980

5000

5020

263

5040

Energy I eW FIG.4. Ti-K edge XANES for Ti02 and Ti02-Si02. (Reprinted with permission from Liu ef (I/. Copyright I994a American Chemical Society.)

conditions. The effect of hydration is also demonstrated by a shift in the UV-visible absorption edge from 32,260 cm- I for the dehydrated material to 29,400 cm- for the hydrated one (Liu et al., 1994). The picture that can be derived from these observations is that in the mixed oxides at low Ti02 concentration and when the preparation procedure is adequate to produce an efficient interaction between the two metal oxides, Ti02 is not present as anatase, but rather in the form of very small domains, in which the normal octahedral coordination of crystalline TiOz has changed to tetrahedral. Upon exposure to water or other absorbable compounds, Ti" increases its coordination to 5 or 6, but the original structure can be restored by simple evacuation. There is a significant difference between the labile octahedral coordination of highly dispersed Ti" that easily reverts to tetrahedral and the stable octahedral coordination characteristic of anatase or rutile crystals. Furthermore, Ti4 + in such a state of high dispersion can be more easily reduced to Ti3+ than is Ti4+ in anatase. On the contrary, in the Ti02-rich region, octahedral Ti'" is present in the form of small crystalline anatase particles stabilized by a thin layer of surface SiOz.

264

BRUNO NOTARI

C. ACIDITY Acidic properties have been attributed to the mixed oxides, and the possibility that they are responsible for the catalytic activity of the materials must be taken into consideration. In the binary oxidcs that exhibit acidic properties, the acidity originates from the substitution of ions of the oxide lattice by a different ion. The best known examples are Si02-A1203 and the related zeolites. These solids are made up of Si04tetrahedra which are electrically neutral because the positive charge of Si" is balanced by the negative charges of the oxygens. When a Si4+ is replaced by an A?+, cations must be added to establish electrical neutrality. If the cation is H+,the solid exhibits the typical reactions of an acid (Thomas, 1949). Tanabe (1981) has proposed that this substitution of a metal ion into the structure of the host oxide follows rules that apply to many binary oxides including Ti02-Si02. The rules for such substitutions are that the cations maintain the coordination they have in their stable oxides, while the anions should have the coordination they have in the majority component. The application of these rules to Ti02-Si02 leads to the prediction that Ti" has octahedral coordination, as in anatase or rutile, and Si" has tetrahedral coordination, as in SiO2. In the Si02-rich region, the coordination of 0 should be 2, as in S O 2 .The following model structure represents the coordination of Ti", Sir", and 0" in the SiO2-rich region:

The Si04 group is neutral, as in Si02,but the TiOd group has an excess of two negative charges. According to Tanabe, two protons are required to reestablish electrical neutrality, with the consequence that the solid develops strong Brransted acidity (Tanabe, 1981). A different result is obtained in the TiO2-rich region, where the oxygens would have coordination of 3, as in anatase, and no Brransted acidity would be present; rather, Lewis acidity would be present. A different model has been proposed by Seiyama (1978), on the basis of which new acidic sites would form at the boundary of the two metal oxides. In TiOz-SiOz, 0 bridging between Ti" and Si" ions would have a negative ' H the solid charge, which must be neutralized by cations; when the cations are , would have Brransted acidity. According to the model of Seiyama, Brransted acidity develops regardless of composition, but being a boundary effect, it should depend on the degree of dispersion of the two metal oxides.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES r

in

9 E

5

I

4 -

I

1

1

40

60

80

2 65

T423 K

3 -

X

% -c

0

20

100

mtania Content / mol % FIG. 5. Rates of I-butene isomerization catalyzed by Ti02-Si02. (From Liu and Davis, 1994b.)

Many studies of the acidic properties that develop upon interaction of Ti02 with Si02 have been reported, but in many of these the purity of the raw materials has not been sufficiently controlled. When pure materials are used, it is found that the rate of 1-butene isomerization is close to zero in the Si02-rich region, with a maximum at a composition of 80% Ti02 (Fig. 5). Only linear butenes are present in the reaction products, whereas isobutylene and other products would be expected for strongly acidic catalysts. Furthermore, the activity for the dehydration of 2-propanol is low in the SiO2-rich region and greater in the TiO2-rich region. The measurement of acidic properties by the interaction of the solids with NH3 has been attempted. The results obtained from temperature-programmed desorption (TPD) indicate that NH3 is more strongly retained by pure Ti02 than by any of the mixed oxides. It is therefore doubtful that NH3 absorption can be used to measure acidity in Ti02-containing materials (Toba et al., 1994; Liu et af., 1994). In contrast, the adsorption of pyridine is an appropriate measure of acidity, as pyridine adsorption on pure Ti02 is zero and that on the Ti02-rich region at Ti/(Ti + Si) = 0.9 is a maximum, in agreement with the results obtained by the measurement of the rate of I-butene isomerization (Doolin et af.,1994). These results cannot be explained with Tanabe's hypothesis. Rather, in the SO2-rich region, where the maximum acidity should develop according to Tanabe, all acid-related properties have their minimum value, whereas acidrelated properties are present in the Ti02-rich region. The charge imbalance calculated by Tanabe for Ti'" in SO2-rich mixed oxides is based on the incorrect assumption that Ti'" is atomically dispersed in the Si02 matrix and octahedrally coordinated. When it is considered that the actual coordination is tetrahedral, no charge imbalance is created and hence no protonic acidity is present. Furthermore, only in a very few instances is atomic dispersion obtained. In most binary metal oxides, nanophase domains are formed with more complex forms of aggregation.

266

BRUNO NOTARI

It should also be considered that Ti" can catalyze reactions by virtue of the peculiar coordination properties it has with oxygen ligands. The best known is with six oxygens in octahedral coordination, as in anatase and rutile, but also four, five, and up to eight oxygens can be coordinated. Titanium nitrate, Ti(NO&, has eight oxygens arranged in pseudo-tetrahedral coordination. Once the oxygen containing groups are coordinated, they can interact, exchanging positions and giving rise to new products. A typical example is transesterification, in which esters, in the presence of Ti" compounds, exchange alcoholic residues and in which acidity plays no role (Seebach et al., 1982). These non-acid-related catalytic properties obviously increase with the degree of dispersion of Ti02 (Blandy et a]., 1991). Great care must therefore be used in selecting reactions aimed at detecting and measuring the presence and nature of acidity. The reactions should be insensitive to the coordination-related catalytic properties of Ti". The moderate acidity of TiO2-SiO2 at high Ti02 concentrations can be better explained with the model proposed by Seiyama, according to which the moderate acidity is the result of the interaction of Ti" at the boundaries of small Ti02 particles with Si02. In this heterolinkage, Ti" maintains its octahedral coordination since it belongs to an anatase-like crystal, while Silv maintains its tetrahedral coordination. In these Ti -0- Si heterolinkages, Ti" shares more negative charges from oxygen atoms of the SiOz than from oxygen atoms of the bulk, each of which shares its charge with three Ti" ions.The charge imbalance can be compensated by protons, thus giving rise to the moderate acidity observed in the Ti02-rich region. In summary, the acid properties of Ti02-Si02 are weak and limited to the Ti02-rich region. In particular, no acidity is present in the Si02-rich region. It is also useful to consider the acidity of crystalline titanium silicates, in which it is possible to recognize both octahedral and tetrahedral coordination of Ti". The minerals zorite, Na6[Ti$i 12]035(0H)4.1 1 H 2 0 , and vinogradovite, Na8[Ti8Si1 6 1 0 5 2 , contain Ti" in octahedral coordination. The Ti" ions, however, are not surrounded in all directions by Si04 tetrahedra, but rather are linked together, sharing edges. Consequently, the number of extraframework cations required to produce an electrically neutral solid does not correspond to that of the simple model structure, i.e., two monovalent cations for each Ti". Synthetic products have been obtained by hydrothermal treatment of strongly alkaline titanium silicate gels which have the same unit-cell parameters and chemical composition of the minerals. Treatment of these materials with HN03 has removed a large fraction of the cations, leaving the unit cell unchanged, and presumably producing the protonic form of the materials. Since Ti" is octahedrally coordinated in these materials, acidity would be expected. Unfortunately, the acidic properties of the protonic forms have not been measured (Chapman et al., 1990).

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

267

Octahedral coordination of Ti" is also present in the titanium silicates ETS-4 and ETS-10. The structure of these materials is reported to be similar to that of zorite, and they can be described as microporous crystals with uniform pores similar in dimensions to classical small- and large-pore zeolites. In ETS-4 and ETS-10, there are two monovalent cations or one divalent cation for each Ti" ion (Kuznicki, 1989, 1990; Kuznicki et al., 1991a, 1991b, 1991c, 1993; Deeba et al., 1994). A recent report of the synthesis of ETS-10 with tetramethylammonium chloride indicates a ratio of monovalent cations to Ti" of 1.6 (Valtchev et al., 1994). The acidic properties of these materials have not been reported. A material modified by the addition of A13+ has been obtained, ETAS10, which, after exchange with NH4 salts, exhibits acidic properties; but these are due to the presence of A13+ and not to the Ti" (Deeba et al., 1994).

111.

Titanium Silicates

The first discovered member of the group of crystalline microporous materials made of oxides of titanium and silicon is titanium silicalite-1 (TS-1). TS-1 has attracted much interest for its unique catalytic properties; it is also of interest by virtue of the proposal that Ti" assumes tetrahedral coordination in substituting for Si" in framework positions of crystalline silica, as stated above. To clarify this point, many detailed studies of the TS-I structure have been carried out. An outcome of the work was the discovery of new crystalline microporous titanium silicates. A.

TITANIUM SILICALITEI

TS-1 has been obtained by the hydrothermal crystallization of a gel obtained from TEOS and TEOT in the presence of tetrapropylammonium hydroxyde (TPA-OH). The structure of TS-1 has been demonstrated by X-ray diffraction (XRD), energy dispersive X-ray (EDX), microprobe analysis, and 29Simagicangle spinning (MAS) NMR spectroscopy. Furthermore, an absorption band in the IR spectrum at 960 cm-', present in TS-I and absent from that of silicalite, was initially considered a fingerprint for the characterization of TS-1. However, later work (discussed below) has shown that this band is also present in many other silica compounds, and therefore its relation to framework TiIV is not straightforward. The X-ray patterns of silicalite-1 and TS-1 (Figs. 6 and 7) demonstrate a change from the monoclinic structure of silicalite- 1 to orthorhombic when Ti'" is introduced into the framework (Taramasso et al., 1983). A more quantitative evaluation has been obtained from the changes in unitcell parameters derived from X-ray data. The relationship between the degree of

268

BRUNO NOTARI

Ec -9E

28 angle, degrees FIG.6. X-Ray diffraction pattern of silicalite-I. (From Taramasso el a/., 1983.)

28 angle, degrees FIG.7. X-Ray diffraction pattern of TS-I.(From Taramasso et al., 1983.)

269

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

+

substitution x = Ti/(Ti Si) and the unit-cell parameters, represented by the unit-cell volume (UCV), was calculated starting from the reasonable assumption that when Si04 tetrahedra are replaced by Ti04 tetrahedra in the SiOz framework, the volume of each tetrahedron is roportional to the cube of its tetra3 , where dTi hedral bond distance. For titanium VT~= dTi and for silicon Vsi = dsi and dsi are the tetrahedral Ti-0 and Si-0 bond distances. Their ratio provides a relationship between the two volumes:

P

The UCV of a mixed framework with a degree of substitution x is made by the contributions of Si04 and Ti04 tetrahedra proportional to their relative amounts in the crystal: From this, Eq. ( 5 ) is obtained:

The experimental values obtained for UCV of TS-I having different degrees of substitution of Si" by Ti'" are given in Fig. 8, together with the values calculated using Eq. (5). Calculated and experimental values are in good agreement and consistent with the isomorphous substitution of Si'" by Ti". The concentration of titanium measured along the axis of the crystals was found to be constant. A simple

5390

1

>

5340'

a01

Q02

I.

0,03

Atomic ratio Ti(Ti+Si) FIG.8. TS-I unit cell volume expansion as a hnction of titanium content. (From Taramasso er al., 1983.)

270

BRUNO NOTARI

model has been proposed, according to which TS-1 can be represented as a silicalite in which a few Ti'" species have taken the place of Si" (Taramasso et af., 1983; Perego et af., 1986; Notari, 1988). Measurement of the unit-cell parameters of TS-1 has been carried out recently by Rietveld analysis, which makes use of the whole profile of the diffraction pattern, thus providing very accurate measurements (Millini et af., 1992). The results (Fig. 9) show the same trend reported previously with a higher degree of accuracy and demonstrate, with remarkable precision, the relationship between the unit-cell parameters and the atomic fraction of Ti" in the framework. The agreement of experimental values with those calculated by Eq. ( 5 ) is excellent, confirming beyond any reasonable doubt the hypothesis of isomorphous substitution . Another property that distinguishes TS-1 from silicalite is the 29SiMAS NMR spectrum. The multiplet characteristic of silicalite-1 broadens in TS- 1 into a major signal at - 113 ppm relative to TMS, while a shoulder appears at - 116 ppm, the intensity of which increases with increasing titanium content (Perego et al., 1986). The state of dispersion of Ti" in TS-1 has been confirmed by X-ray photoemission spectroscopy (Carati et al., 1990). The Ti 2p peak, which in anatase is at 458.4 eV, is shifted to 460.2 eV in TS-1 (Fig. 10). Such a high binding-energy value for a Ti 2p electron is ascribed to a relaxation effect observed when photoemission occurs from a highly dispersed system or an isolated atom.

'.-",

::::m ;:::

4 20.12 X

19.90 19.95

a 20.10

20.08 0.000 0.010 0.020 0.005 0.015 0.025

19.80 0.000 0.010 0.020 0.005 0.015 0.025 Atomic ratio Ti/(Ti+Si)

Atomic ratio Ti/(Ti+Si) m

3

13.42

13.36 Y 0.000 0.010 0.020 0.005 0.015 0.025 Atomic ratio Ti/rI+si)

.r C

3

-

5320 0.000 0.010 0.020 0.005 0.015 0.025 Atomic ratio Ti/(Ti+Si)

FIG.9. Unit cell parameters of TS-I by Rietveld analysis. A, B, and C: axis dimensions; D: volume. (From Millini e/ al., 1992.)

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

27 1

Binding energy I eV FIG.10. XPS Ti 2p signal for (a) TS-1; (b) anatase. (From Carati et al., 1990.)

The electronic transitions of silicalite and TS-1 in the UV-visible spectrum have provided significant information about the structure of TS-1. The diffuse reflectance spectra of the two materials (Fig. 1 1 ) show a strong transition at 48,000 cm-' that is present in the spectrum of TS-I and absent from that of silicalite. This transition must be associated with a charge-transfer process localized on Ti'". The frequency of this transition is modified by the presence of H 2 0 (Fig. 12). As the H2O partial pressure increases, the peak at 48,000 cm-I is progressively eroded with formation of a lower-frequency absorption, which reaches a new stable maximum value at 42,000 cm- '. These frequencies come very close to those that can be calculated by the Jsrgensen equation for Ti'" tetrahedrally and octahedrally coordinated to oxygen, respectively. Furthermore,

10

20

30 cm-1.103

40

50

A

FIG. 1 I. UV-Visible reflectance spectra (Kubelka-Munk function vs. wavenumber) of (a) silicalite-1; (b) TS-I. (From Boccuti ef al., 1989.)

272

BRUNO NOTARI ~

~

l

'

l

'

l

~

l

F(RC0)

-

I

-

"

30 cm1.103 (-)

1

,

6

1

1

'

B 50

FIG. 12. Changes in the TS-I reflectance spectra upon exposure to H 2 0 . (......) 4TOW. saturated vapor. (From Boccuti er al , 1989.)

NH3 can be adsorbed on Ti", giving rise to a band in the UV-visible spectrum at 38,000 cm- I , ascribed to a charge-transfer transition. All these phenomena are reversible on evacuation (Boccuti et al., 1989). The increase in coordination around Ti" could take place by the simple addition of H2O to TiIvor as a result of hydrolysis of Ti-0-Si bonds and formation of TiOH groups, followed by the addition of H2O to the TiOH groups. Ti02 exhibits little tendency to form surface hydroxyl groups, but could be forced to do so by the unusual condition in which titanium is present in titanium silicates. The steric hindrance of the crystalline lattice prevents Ti" from acquiring the same regular octahedral coordination that it assumes in many of its compounds. A shorthand notation of H20 on Ti" will be used to indicate the increased coordination:

Indication for the formation of TiOH has been obtained from the observations of the IR and NMR spectra of TS-1 treated with three different isotopically labeled waters D20, H2170,and H2I80. The SiOH groups on silica have an absorption band at 985 cm-I which is D-exchange-sensitive. The 960 cm-' band of TS-1 does not change frequency after treatment with D20. It is, instead,

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

273

sensitive to H2170 and H2"O exchange, shifting from 960 cm-l to 952 and 937 cm- I , respectively. The frequency decrease is consistent with the expected shifts based on the change in the reduced mass of the harmonic oscillator Ti-0-Si when I6O is exchanged with I7O or I8O. The original frequency is restored upon treatment with H2I60. The I7O exchange is accompanied by the appearance, in the 170 MAS NMR spectrum of TS- 1, of a peak at 6360 ppm. This peak is absent in silicalite. Notwithstanding the difficulties of assignment of the 960 cm-' band, the observed 0 exchange could indicate that the coordination of H20 takes place with partial hydrolysis of the Ti-0-Si bond and 17 formation of TiOH. When the reverse reaction takes place, 0 becomes part of the solid (Bellussi et al., 1992).

The IR band at 3450-3540 cm-' has been considered characteristic of TiOH groups (Boccuti et al., 1989). However, this assignment has been questioned on the basis of the invariance of this band with the titanium content up to titaniumfree silicalite; thus it has been ascribed instead to internal SiOH groups, which are greatly influenced by the interaction with the solid and have a high degree of hydrogen bonding (Huybrechts et al., 1992). The UV-visible absorption at 48,000 cm-', which is observed only in anhydrous TS- 1, departs radically from the absorption frequencies observed in the various forms of Ti02 and Ti02-SiOz. Ti02 anatase has a characteristic absorption at 30,500 cm- I , whereas for the smallest Ti02 nanophase particles, detected in 3% Ti02-97% Si02 material (Anpo et af.,1986) and in 7% Ti02- 93% SiOz material (Fernandez et al., 1988), the absorption is in the range 30,500-40,OOO cm- I . Since the UV-visible absorption frequency is a strong function of Ti02 particle size for diameters smaller than 10 A, the frequency of 48,000 cm-' indicates that the Ti" ions in TS-1 are more highly dispersed than in any other material, providing further indications for atomic dispersion. UV-Visible spectroscopy has proved to be valuable for the identification of the different forms in which Ti" is present in these materials. If the synthesis conditions are not carefully controlled, nanophases of Ti02 can form; in this case, they can be detected by the absorption between 30,500 and 40,000 cmand distinguished from framework Ti". It has therefore become customary to control the purity of TS-1 by UV-visible spectroscopy, since the catalytic properties of TS-I are profoundly changed by the presence of even small amounts of TiO2. However, as previously discussed, the state of hydration of the materials produces a significant change in the absorption frequency, and therefore meaningful information is obtained only if measurements are carried out in a controlled atmosphere. This technique has also proved useful for detecting the

'

274

BRUNO NOTARI

transformations that TS- 1 undergoes when used in catalytic reactions. Investigations of the deactivation of TS-1 have demonstrated that under severe conditions of temperature and chemical environment, framework Ti'" is removed from the lattice and gives rise to materials having an absorption at 40,000 cm-', thus indicating the formation of Ti-0-Ti bonds, presumably nanophase TiO., (Petrini et al., 1991). An independent demonstration that Ti'" in TS-1 resides in framework tetrahedral positions derives from the analysis of the electron paramagnetic resonance (EPR) signals that are obtained upon the reduction of TS-1 with CO and subsequent reaction with O2.These signals are compared with those produced by Ti3+-exchangedZSM-5. The dry TS-1 sample was subjected to a dry carbon monoxide atmosphere, and the EPR signal was recorded while the temperature was increased; the signal reached a maximum value in 10 Torr of CO at 673 K. The analysis of the EPR spectrum shows that it is consistent only with Ti3+, demonstrating that the reduction of Ti4+ to Ti3+ takes place. After removal of excess CO at 373 K, the addition of O2 to the reduced sample produced a new signal at the expense of the signal ascribed to Ti3+ (Fig. 13). The signal was interpreted as evidence of the superoxide ion generated upon a charge-transfer reaction between Ti3+ ions and 02: Ti"

+ O2

+

Ti4+ + 0;

(9)

For comparison purposes, Ti3+-exchanged ZSM-5 was analyzed by the same technique. Addition of 0 2 under conditions similar to those used for TS-1 produced a signal attributed to the superoxide ion; however, the principal values differed significantly from those obtained in the case of TS-1 (Fig. 14). The change in g,, from 2.031 to 2.0185 is due to a significant change in the coordination sphere around the Ti center to which the superoxide is binding. The g-tensor ordering of Ti3+ in TS-1 may be interpreted as due to a tetrahedral coordination of the Ti3+ ions with tetragonal distortion, thus providing direct 2.03 1

10

a

FIG.13. 02-in TS-I generated following addition of 02 to prereduced TS-1. (From Tuel er a/., 1990.)

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

275

f 2.0006 FIG.14. 0; generated following addition of O2 to Ti'+-exchanged MFI. (From Tuel

er al.,

1990.)

and convincing evidence for the framework siting of the precursor tetravalent titanium (Tuel et al.,1990). The reduction of Ti4+ to Ti3+ in TS-1 has also been observed with cyclic voltammetry using zeolite-modified carbon paste electrodes. With silicalite, neither anodic nor cathodic processes can be observed. However, TS-1 is electrochemically active, with a reduction process at + 0.56 V versus a saturated calomel electrode (SCE) and an oxidation process at + 0.65 versus SCE. These observations must be attributed to the redox system Ti4+/Ti3+.The electrochemical process involves the Ti cations of the inner part of the zeolite crystals, provided that a suitable electrolyte cation can difhse inside the channels to compensate for the electrical imbalance caused by the redox process in the solid: (SiO),-Ti4+

+ e - + M+

(SiOk--Ti3+ M+

(10)

TiOz, TiOz on silicalite, and Ti-exchanged ZSM-5 show no electrochemical activity under the same conditions, suggesting that the nature and coordination state of the electroactive sites are characteristic of Ti" in TS-1. A carbon electrode modified with TEOT undergoes reduction and oxidation processes at 0.62 and 0.71 V versus SCE, values only 60 mV different from those of TS-1. This indicates a close similarity between Ti'" in TEOT and Tilv in TS- 1 ; in both compounds the Ti" ions are tetrahedrally coordinated by 0 and isolated from other Ti" ions (de Castro-Martins et al., 1993 ). The isomorphous substitution of Si" with Ti" has also been studied with X-ray absorption spectroscopy at the Ti Kedge (XANES and EXAFS). The first results reported indicated that a large proportion of Ti" in TS-1 resided in an

276

BRUNO NOTARl

octahedrally coordinated environment and changed to lower coordination upon dehydration (Behrens et al., 1990, 1991). However, it has been objected that the lattice parameters reported correspond to those of pure silicalite-1 with no titanium content, and therefore the observations do not represent the real state of Ti" in TS-1 (Pei et al., 1993). An investigation of the effect of hydration on TS-1 has shown that the intensity of the first peak in the XANES spectrum changes position and intensity upon dehydration of the material ( Lopez et al., 1993; Bordiga et al., 1994). This behavior closely parallels that of Ti" in Ti02-Si02 in the Si02-rich re ion which, despite its chemical composition, is substantially different from Ti in anatase (Liu et al., 1994). XANES analysis does not allow a distinction between Ti" in TS-1 and Ti" in TiO, nano hases. It is necessary to ascertain independently the framework position of Tipv by X-ray crystallography and demonstrate the absence of TiO, phases by UV-visible spectroscopy. Only for pure materials do the XANES measurements provide true information representing Ti" in TS-1. The issue was clarified when dry, pure TS-1, having the correct crystallographic parameters, was used. The analysis of the XANES and EXAFS data in comparison with theoretical models gives the best fit for the structural parameters: NTi-o = 4.1(5), RTi-0 = 1.80A, Aa2 = 0.0026(10) A2, which are, respectively, the number of oxygen atoms surrounding a Ti atom, the Ti-0 distance, and the Debye-Waller (DW) factor. These values correspond to those of Ti'" in tetrahedral coordination. The analysis of the Ti-0 distances of structurally characterized 4-coordinated Ti" compounds along with several representative 5- and 6-coordinated compounds shows (Table I) that the average Ti-0 distances in 4-coordinated compounds lie only in a narrow window between 1.76 and 1.81 A. Furthermore, for a given compound, the length differences, AR, between the individual bonds are small (<0.07 A). The Ti-0 distances for 5-coordinated compounds are greater than 1.90 A, and the AR about 0.27 8. For 6-coordinated compounds, the Ti-0 distance is larger than for 4-coordinated compounds, being in the range 1.85-1.96 A. Considering that the Ti-0 distance obtained by fitting the first-shell data for TS-1 is 1.80A and that the variation in Ti-0 distance is not greater than 0.05 A, it has been concluded that the results are consistent only with 4-coordinated Ti, the other possibilities being ruled out by mismatches in Ti-0 distances or the AR values (Pei et al., 1993; Lopez et al., 1993; Bordiga et al., 1994). Many attempts have been made to interpret the origin of the IR band at 960 cm-' and to relate it to the catalytic properties of TS-1. This band is absent from the silicalite spectrum, and its intensity is linearly correlated with the amount of Ti" incorporated in framework positions. It has been proposed that it could be due to the presence in the solid of a titanyl group (Ti=O) because

R

277

MICROPOROUS CRYSTALLINE TITANIUM SILICATES TABLE I Individual T i - 0 Distances (R) and Diferences in T i - 0 Distances (AR)for Selected Titanium-Containing Compounds“ Ti compound

Ti-0 distance (A)

(A)

AR (A)

1.957 I .964 1.952 1.846

0.046 0.03 I 0.000 0.221

R

6-Coordinated Ti02 (anatase) Ti02 (rutile) SrTiO] MgTiO3

1.934 X 1.949 X 1.952 X 1.735 X

5-Coordinated BasTiSi2On (fresnoite)

1.698, 1.970 X 4

1.916

0.270

4-Coordinated Ba2Ti04 CsAITi04 C4nH6804Ti C40H5204Ti C~~HIWO8Ti2. 6C7H8 Titanium-silicalite- 1

1.766, 1.812, 1.817, 1.836 1.751, 1.757, 1.733 X 2 1.780 X 2, 1.781 X 2 1.778 X 4 1.782, 1.797, 1.813, 1.845 1.80

I .808 1.764 1.781 1.778 1.809 I .80

0.070 0.022 0.00 I 0.000 0.063

4, 1.980 X 2 4, 1.980 X 2 6 3, 1956 X 3

c0.05

Pei et al. ( 1 993).

compounds containing this group have an absorption band at 975 cm-I ascribed to the stretching of the Ti=O bond (Notari, 1988). However, the UV-visible spectra of TS-1 do not show the absorption at 25,000-35,000 cm-I, typical of Ti=O bonds. A different assignment has been proposed: “A local impurity mode of a Si04 structure bonded to a Tilv: 03Si-O-Ti, essentially the Si-0 stretching of the polarized Si-O’-...Ti” (Boccuti et al., 1989). The fact that the IR spectra of SiOz-Ti02 glasses have a similar band has been considered to be evidence in favor to this assignment. It has already been mentioned that the 960 cm-l band is sensitive to exchange with H2”O and H21s0, and since the 0 exchange does not occur in silicalite, Ti” must somehow be involved in the exchange process. The changes undergone by the 960 cm- I band upon interaction of TS- 1 with H202 are not yet clear. One report indicated that the 960 cm- band disappears following addition of H202 and reappears after heating to 333 K, the temperature at which titanium peroxo compounds decompose (Huybrechts et al., 1992). Recent results do not confirm these observations: The 960 cm-’ band does not disappear following addition of H202, but rather, disappears under the action of strong bases, which also cause deactivation of the catalyst. This deactivation is very likely due to the formation of stable alkali-titanium peroxo complexes when H202 is added. Treatment with H2S04 restores the 960 cm-’ band and the catalytic activity (Clerici et al., 1993; Khouw et al., 1995).

’

278

BRUNO NOTARI

The 960 cm-' band has been considered to be a critical, although not necessarily sufficient, condition for the catalytic activity of titanium silicates in H202 oxidation of hydrocarbons (Huybrechts et al., 1992). Later studies have shown that a band in the same frequency region, 960 cm-I, is present in many other silicas with or without titanium. It is present in Ti02-Si02 (Schraml-Marth et al., 1992) and in Ti-ZSM-48 (Khouw et al., 1992; Dartt et al., 1993), which are both catalytically inactive for H202 oxidation. Is also present in zeolite-beta with only aluminum (Dartt et al., 1993), in vanadiumsilicalite (Hari Prasad Rao et al., 1992) and in chromium-silicalite (Chapus et al., 1994). The fact that the 960 cm-' band is also found in many different silicas without titanium means that titanium is not involved in the group causing the absorption at this frequency. A recent proposal has been made that this band is better assigned to the stretching vibration of the SiO- groups, which can have different countercations, H', Na', and Nk', where R is alkyl. This assignment is in agreement with the fact that the IR bands of calcined and uncalcined samples are different from each other, and also with the shifting of the frequency in alkali-containing materials (Camblor et al., 1993a). The presence of titanium, and also of the other metals that cause the appearance of this band, could therefore cause an increase in the number of SiO- groups which give rise to this band. This band therefore merely indicates the presence of perturbing or defect groups which, in some but not all cases, could be related to the catalytically active groups. Several factors have contributed to the difficulties experienced in obtaining consistent results in different laboratories, especially with EXAFS and XANES: the difficulty of producing pure samples of TS-I; the difficulty of detecting TiO, nanophases when present in minute amounts; and the change in coordination of Ti'" in both TS-1 and TiO2-SiOz in the Si02-rich region, from tetrahedral in the dry state to octahedral upon adsorption of molecules from the gas phase. But once these problems have been solved, the combined information obtained with the different techniques and investigations-in particular the regular changes in UCV with titanium content, the UV-visible absorption data, the properties of the superoxide ion on Ti3+ obtained by the reduction of Ti4+ of TS-1, the reductiodoxidation behavior in cyclic voltammetry, the binding energy values, and the recent EXAFS and XANES results-all converge to indicate that, in pure TS-1, Ti" is indeed present in framework positions of silicalite-1 crystals and that it occupies these positions at random, thus confirming the model originally proposed, which is the simplest possible model. In TS- 1, Ti'" has the highest degree of dispersion observed in Ti02-containing materials, and its unique chemical and catalytic properties are very likely due to a unique state of dispersion, i.e., atomic dispersion. The structure of TS-1 can then be represented as shown in Fig. 15.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

1. Maximum Amount of Ti'

279

in Framework

The maximum amount of Ti" that can be accommodated in framework positions in TS-1 has been evaluated by numerous authors, and the subject is still actively debated. The problem is mainly of scientific, rather than practical interest. Although a relationship between titanium content and catalytic activity certainly exists, a direct proportionality is limited to very low titanium concentrations because other factors, notably diffusion, limit higher catalytic activity. An upper limit to the value x = Ti/(Ti + Si) of 0.04 on a molar basis was indicated in the original patent, although a good fit of experimental data with calculated values was limited to x = 0.025. Attempts made to synthesize TS-I with titanium content exceeding 0.025 with the synthesis methods known at the time caused the formation of a TiOz phase, and the UCV did not follow the correlation with x described by Eq. (5). It was therefore assumed that the limit for Ti" in framework positions should be found in the range O.W.04 (Taramasso et al., 1983). It was recently reported that by using a modified synthesis method, much higher amounts of Ti'" could be inserted into lattice positions, up to x = 0.10 (Thangaraj et al., 1 99 1; Thangaraj et al., 1992; Mirajkar et al., 1992). However,

FIG. 15. Crystalline structure of TS-I obtained by isomorphous substitution of Si" by Ti" in framework positions of silicalite-l .

280

BRUNO NOTARI “-.YV

-

5420

s

m

5410

0

5400t ;

B

0

0.02

5390

0.04

0.06

0.08

0.1

t

0.12

0.14

Atomic ratio Til(Ti + Si) FIG. 16. UCV changes vs. X. (0)Data of Taramasso era/. (1983); ( + ) data of Thangaraj el al. (1991a). (From Notari, 1993a.)

the experimental values reported for the UCV of these highly substituted materials do not follow the linear correlation of Eq. (5) that was found to hold for low TiIv-substituted materials (Fig. 16) (Notari, 1993). To further clarify matters, the modified synthesis method was repeated and the products analyzed by XRD with Rietveld analysis. In no case could framework Ti” in excess of x = 0.025 be found. Instead, Ti02 phase was present, and its amount was shown to quantitatively account for the remaining titanium (Millini et al., 1992). The characterization of the high-Ti-content TS- 1 synthesized by the modified method was also attempted by using cyclic voltammetry, which distinguishes between framework Ti” and extra-framework Ti02. The results (Fig. 17) indicate that while the maximum current intensity increases linearly up to 2% Ti content, no further increase is observed beyond this value. The same maximum value is obtained when TS-1 is synthesized by the conventional method. From this result, it is concluded that the limit for Ti” in framework position in silicalite-1 corresponds to about x = 0.025 and that the excess titanium introduced with the modified method is not in framework positions (de CastroMartins et al., 1994). A difference in the relationship between UCV and degree of substitution x has recently been reported in connection with the use of tetramethyl orthosilicate (TMOS) as the silicon source instead of TEOS. The Ti content can be increased up to a value of 0.05 without evidence for extra-framework titanium species in the UV-visible spectra. However the UCV increases do not follow the relationship of Eq. (5) and the maximum value obtained, 5390 A3 at x = 0.05, is the

MICROPOROUS CRYSTALLINE TITANIUM SILICATES 1 20

f-

-2

TS-1(5)

..

15*'

28 1

~s-1(7)

TS-1(4) a%\ -]

TS-1(3)

'1

\ P

TS-2(2)

'TS-2(1)

10 . . T S - ~ ( ~ ) / J \ ~ ~ - ~ ( ~ ) 5..

7

/\TS-l(l)

0

L

same as the one obtained with the conventional method of synthesis at x = 0.025 (Tuel et al., 1994). If the well established criterion of the regular

increase in UCV with substitution is taken as evidence of, and a necessary condition for, isomorphous substitution, the conclusion is that the value of x = 0.025 constitutes the upper limit for isomorphous substitution in this new material, as well. It must, however, be explained in which form the excess titanium is present, since according to the authors nanophase TiO, particles were not detected by UV-visible spectroscopy. The high reactivity of TMOS has been indicated to promote homogeneous component mixing in the case of zirconia-silica aerogels (Miller et al., 1994). It is therefore possible that TMOS produces domains of TiO, nanophase so small that their absorption frequency in the UV-visible spectrum is shifted and overlaps that of framework Ti". This would make their detection more difficult. The catalytic tests performed with materials obtained from TMOS at x = 0.05 are limited to the hydroxylation of phenol; and when the results are compared with those for TS-1 with x = 0.025, they indicate neither increased activity (as would be expected for a higher concentration of framework Ti") nor decreased selectivity (as would be expected for extraframework TiO, nanophases). Further investigation to clarify this point would be desirable, but the difficulties should not be underestimated: All the experimental techniques previously described provide, at best, rather crude results that hardly allow one to measure nanophase TiO, particles smaller than 10 A and distinguish them from atomically dispersed Ti".

282

BRUNO NOTARI

2. TS-1 Modified with A13+, Ga3+,or Fe3+ TS-I containing a trivalent element (A13+,Ga3+,or Fe3+)in lattice positions has been synthesized by the same procedure used for pure TS- 1, except for the addition of suitable precursors containing the second element. The catalytic properties of the materials are modified with respect to those of pure TS-I. In the epoxidation of propylene, 1-butene, and ally1 alcohol with H202,TS-I gives the epoxides as the major reaction products. With H-[AI,Ti]-MFI, H-[Ga,Ti]-MFI, and H-[Fe,Ti]-MFI, glycols and glycol ethers are the major reaction products. Most probably, the epoxides initially formed undergo acid-catalyzed reactions with water or with the solvent (Section V.C.3.b). These catalysts are also active for the oligomerization of 1-octene to dimers and trimers (Bellussi et al., 1988a, 1988b, 1988~).In some case A13+ was accidentally introduced as an impurity of the Si02 raw material. Consequently, in the oxidation of methanol, the dehydration product dimethyl ether was obtained (Deo et al., 1993; Section V.C.1); in the epoxidation of 1-octene in acetone, the condensation product of 1,2 epoxyoctane with acetone, 2,2-dimethyl-4-hexyl3-dioxolane, was obtained (Huybrechts et al., 1991b, 1992). Activity in acidcatalyzed transformations of hydrocarbons was assigned to TS- 1, but was very likely due instead to the A13+ impurity present (Chen et al., 1988). B.

TITANIUM SILICALITE-2

A microporous crystalline titanium silicate having the MEL or silicalite-2 structure (TS-2) has been obtained by performing the synthesis with alkoxides of titanium and silicon and the organic base tetrabutylammonium hydroxide (TBA-OH). Other organic bases, such as 1,8-diaminooctane or trimethylbenzylammonium hydroxide, which can be used for the synthesis of the aluminum analog ZSM-11, do not produce TS-2. Framework insertion of Ti has been established with the same methods used for TS-1, including the regular expansion of UCV with increasing amounts of Ti (Bellussi et al., 1989; Reddy, J. S., et al., 1990). TS-2 has also been synthesized from alkoxides of titanium and silicon with the base tetrabutylphosphonium hydroxide (TBP-OH). When this base is used, it has been reported that the maximum amount of framework titanium that can be incorporated without formation of extra-framework phases is 1.1 Tihnit cell, which corresponds to x = 0.01 15. Beyond this value, extraframework Ti02 phases are detected (Tuel et al., 1993~).Claims of high incorporation of titanium in framework positions have been made when the base TBA-OH was used, up to x = 0.06 (Reddy, J. S., et al., 1991d, 1992b). However, the UCV values reported do not justify these claims. By applying Eq. ( 5 ) to the reported experimental data, a maximum value of x = 0.032 is obtained, slightly higher than the values reported for TS-I. As in the case of

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

283

TS- 1, the possibility exists that extra-framework TiO, nanophases were present in the materials. The IR band at 960 cm-' present in TS-1 is also present in TS-2, and the same considerations apply. When the coordination of Ti" in TS-2 was studied by XANES and EXAFS spectroscopy, the results obtained suggested that Ti" is not located in a regular tetrahedral site, thus raising doubts about the proposal of isomorphous substitution of Si" with Ti". Different hypotheses were advanced to describe the environment of Tiiv, including dimeric species mixed with monomeric ones, framework defects, and edge-sharing linkages (Trong-On et af.,1992a, 1992b; Jacobs, 1992; Bittar et a f . , 1993; Bonneviot et af., 1993). These hypotheses have been rejected, as in the case of TS-I. Rather, the results have been ascribed to the limited accuracy of the XANES data, which were not up to the level that allows statistically meaningful fits involving multiple atomic shells and 17 fitting parameters (Pei et af., 1993). From the XPS values characterizing the samples, it is also clear that some of them contained extra-framework Ti02 ,which interfered with the correct identification of the coordination of Ti'" in TS-2. With the present state of knowledge, the most plausible description is that, as in TS-1, Ti'" in TS-2 occupies framework positions at random and that the maximum degree of titanium incorporation corresponds to a value x = 0.03-slightly higher than the maximum value found for TS- 1. There is, as is well known, a close similarity between the crystalline and porous structures of silicalite- 1 and silicalite-2. The same similarity therefore exists between TS-1 and TS-2, and it appears logical that they should have very similar catalytic properties. TS-2 has been evaluated as a catalyst for many different reactions, such as Beckmann rearrangement of cyclohexanone oxime with vapor-phase reactants; H202 oxidation of phenol, anisole, benzene, toluene, n-hexane, and cyclohexane; and ammoximation of cyclohexanone. As described in detail in Section V.C.3, differences that had been claimed between the catalytic properties of TS-1 and those of TS-2 have not been substantiated. Later investigations have shown that, when all the relevant parameters are identical, the catalytic activities of TS- 1 and TS-2 are also identical. The small differences in the crystalline structure between the two materials have no influence on their catalytic properties (Tuel et af.,1993a). 1. TS-2 Modified with Al'+ and Fe3' TS-2 substituted with A13+ and with Fe3+ have also been obtained. The incorporation of Ti, Al, and Fe in the MEL framework was demonstrated by the expansion of the UCV, the occurrence of the 960 cm-' IR band, and the ionexchange capacity; and, for Fe3+,by Mossbauer-effect spectroscopy and magnetic susceptibility.

284

BRUNO NOTARl

H-[Ti,Al]-MEL and H-[Ti,Fe]-MEL are active catalysts for both oxidation reactions and typical acid-catalyzed reactions such as rn-xylene isomerization. In this reaction, H-[Ti,AI]-MEL deactivates at a lower rate than the H-[All-MEL analog (Reddy, J. S., et al., 1994a). C. Ti-ZSM-48

The titanium-containing analog of the zeolite ZSM-48 has been synthesized from soluble peroxo titanates and Si02 by using diaminooctane as the structure directing agent. The incorporation of Ti" into framework positions of the ZSM48 crystal lattice was demonstrated by the regular increase of the UCV with titanium content. The experimental values agree with the values calculated by Eq. ( 5 ) up to x = 0.02. For higher titanium contents, no further increase in UCV was observed (Serrano et al., 1992). Attempts have been made to synthesize Ti-ZSM-48 with even higher titanium contents, but it was reported that no more than 2% titanium could be incorporated into framework positions. UV-Visible spectroscopic analysis of materials prepared with more titanium precursor indicates the presence of extra-framework Ti02, which in some cases is present also in the materials with low titanium contents. From these observations it is concluded that for Ti-ZSM-48 a limit exists in the amount of Ti'" that can substitute for Si" in framework positions (Reddy, K. M. et al., 1994b). The synthesis of Ti-ZSM-48 has also been obtained by using the base In this case hexamethonium hydroxide [(CH3)3NCH2(CH2)4CH2N(CH3)3(OH)2]. as well, the maximum amount of titanium in framework positions has been found to be x = 0.02 (Reddy, K. M. et al., 1994a). The 960 cm- band is present in the IR spectrum of Ti-ZSM-48. As has been mentioned above, this material has been found to be catalytically inactive for the oxidation of phenol, and therefore no correlation can be established between the 960cm-' band and catalytic activity. The absence of catalytic activity of Ti-ZSM-48 has been ascribed to the unidirectional, noninterpenetrating channel system which is prone to plugging by high-molecular-weight by-products. However, Ti-ZSM-48 has been reported to be an active catalyst for the oxidation of aniline (Section V.C.3.e).

'

D. Ti-BETA ZEOLITE The synthesis of titanium-containing beta zeolite has been carried out by direct hydrothermal synthesis and by secondary synthesis. In the direct synthesis with aluminum salts and titanium and silicon alkoxides, it is necessary to operate in the absence of alkalies, since these cause the formation of an amorphous precipitate containing Ti02 and S O 2 . Evidence for the substitution

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

285

of Si" by Ti" in framework positions was obtained by (1) the change in the interplanar d-spacing corresponding to the most intense XRD peak of beta zeolite (28 = 22.4') as a function of the titanium content of the materials; (2) the absorption at 48,000 cm-' in the UV-visible spectrum, which, by analogy with that of TS- 1, has been ascribed to highly dispersed Ti" in tetrahedral coordination; and (3) catalytic activity tests (Camblor et al., 1992, 1993). The secondary synthesis has been camed out by reaction of beta zeolite with TiC14 (Ferrini et al., 1992) or with ammonium titanyl oxalate (Reddy, J. S . et al., 1995). This second method appears to be more reliable, inserting Ti" in framework positions only, while with TiC14, Ti02 phases are also formed. The pore diameter of zeolite beta is 7 A, larger than those of silicalite-1 and silicalite-2 (5.5 A). Titanium incorporated into zeolite beta reacts with molecules whose dimensions are too big to diffuse in the pores and be oxidized by TS- 1 or TS-2. The drawback is that zeolite beta must contain A13+ to crystallize, and this imparts strong protonic acidity to the solid, with the consequence that secondary acid-catalyzed reactions also take place. However, the acidic properties can be neutralized in several ways and highly selective oxidations can be carried out on Ti-beta (Section V.C.3.b). The maximum content of titanium in Ti-beta zeolite appears to be higher than in the other materials. A value of x = 0.038 has been reported without formation of extra-framework Ti02. From the characterization of Ti-beta zeolite by XANES and EXAFS, it has been concluded that Ti" in the calcined material is tetrahedrally coordinated, isolated from other Ti" ions, and surrounded by OSi groups. In the presence of H20, Ti" increases its coordination and very likely undergoes hydrolysis of the Ti-0-Si bonds forming TiOH and SiOH groups (Blasco et al., 1993). These properties, except for the different pore diameter and the presence of aluminum, are very similar to the corresponding properties described above for TS- 1, indicating that the simple model developed for TS- 1 also applies, with the necessary modifications, to Ti-beta zeolite.

E. Ti-MCM-4 1, Ti-HMS New materials consisting of amorphous silica with regular pore structure and therefore called mesoporous molecular sieves have recently been described (Kresge et al., 1992b). The incorporation of titanium has been attempted by performing the synthesis in the presence of Ti" compounds, such as titanium alkoxides (Corma et al., 1994c; Tanev et al., 1994; Blasco et al,, 1995; Zhang et al., 1966; Liu et al., 1996). The interpretation of the results obtained is still under debate. The changes in UCV of the Ti-containing materials with respect to the pure silica ones are not

286

BRUNO NOTARl

those expected for the case of isomorphous substitution. Large variations are observed which cannot be attributed to the small difference in Ti-0 and Si-0 bond distances. Moreover, upon calcination these materials undergo significant changes, since the large population of SiOH groups, typical of amorphous silicas, upon heating condense by elimination of water and cause a contraction of the structure. The changes observed upon incorporation of Ti" in mesoporous silicas can be better explained by the effect that the presence of foreign elements exerts on the framework thickness and on the condensation process that takes place on heating. The same phenomenon is observed with other transition metal derivatives of mesoporous silicas such as V-MCM-48 and Cr-MCM-48 (Chen et al., 1993; Zhang et al., 1996). The XANES and EXAFS data indicate that Ti" in these mesoporous materials is in tetrahedral coordination and highly dispersed (Thomas et al., 1994; Maschmeyer et al., 1995; Blasco et al., 1995; Liu et al., 1996). Also the UV visible spectra, with absorption edges at 48,000 cm- and 43,500 cm- ', indicate the presence of isolated Ti'' in tetrahedral and octahedral coordination. A moderate increase in Q4-Si content along with a broadening of the Q4-Si NMR peak indicates that a part of the Ti'" can be embedded in the amorphous silica walls (Zhang et al., 1996). The picture that emerges from these data is that of a mesoporous silica in which Ti" is present as isolated Ti" ions, partly at the surface of the walls, partly embedded in the walls with four 0-Si surrounding it. The catalytic properties of these materials provide interesting information. They have been tested with both H202and hydroperoxides. There is agreement on their activity when hydroperoxides are used as oxidants, but disagreement on their activity with H202. The activity with hydroperoxides is not surprising, since it is common to that of well-dispersed TiO2-Si02 (Wulff et al., 1971; Sheldon et al., 1973; Hutter et al., 1995a, b, c). Also, an identical turnover number has been reported in the oxidation of 1-hexene with TBHP when using Ti-MCM-41 and Ti02 on silica gel, indicating that the Ti" species can be identical for the two materials (Blasco et al., 1995). For H 2 0 2 as the oxidant, one report indicates that Ti-MCM-41 is active for the epoxidation of 1-hexene, although its activity is lower than that of TS-1 and Ti-beta (Blasco et al., 1995). Another report, however, fails to detect any significant activity for the same reaction, in spite of the fact that the catalyst was prepared following exactly the same procedure (Liu et al., 1996). On the other hand, catalytic activity of Ti-MCM-48 in the oxidation of methyl methacrylate and styrene with H202has been reported. Because of the large amounts of oxidant used, mainly products of double bond cleavage were obtained, methyl piruvate and benzaldehyde (Zhang et al., 1996). The activity of Ti-HMS and Ti-MCM-41 for the oxidation of 2,6-di-tert-butylphenol with H z 0 2 to give the corresponding quinone has also been reported.

'

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

287

In summary, the properties of these mesoporous amorphous materials do not duplicate those of the crystalline silicas, but present a useful addition to the array of titanium catalysts useful for oxidation reactions.

F. Ti-MORDENITE The synthesis of Ti-mordenite has been conducted by reaction of TiC14 with dealuminated mordenite or by hydrothermal synthesis (Section 1V.F). The evidence for the incorporation of titanium is limited. The UV-visible spectra show that, in addition to the transition at 48,000 cm- I , assigned to isolated Ti" in tetrahedral coordination, there is also an absorption at 35,000 cm-I, indicating extra-framework T i 0 2 . The catalytic properties in oxidation reactions with H 2 0 z are significantly different from those of Ti02 deposited on mordenite, but they are limited to the hydroxylation of benzene and the oxidation of n-hexane (Kim and Cho, 1993).

G. Ti-Y, Ti-L, Ti-W, A N D Ti-Q The synthesis of these titanium-substituted zeolites has been described to occur by a secondary synthesis process involving the reaction of [NH4I2TiF6 with the preformed corresponding zeolite (Section 1V.G). The chemical and physicochemical properties described are not sufficient to establish the presence of Ti'" ions in framework positions. The titanium concentrations reported are much higher than the maximum values observed in titanium silicates for which isomorphous substitution has been demonstrated. The possible presence of Ti02 particles has not been investigated. No indication of the properties of these materials as catalysts in reactions typical of titanium silicates has been provided. It is therefore very doubtful that real isomorphous substitution has been obtained (Skeels et al., 1989; Skeels, 1993).

H. ETS-4. ETS-10 The high-temperature synthesis from strongly alkaline suspensions of salts of Ti" and Si" produces crystalline microporous materials in which Ti'" is present in octahedral coordination. These materials do not exhibit the catalytic properties typical of the other titanium silicates in which Ti" is in tetrahedral coordination (Kuznicki, 1989, 1990; Kuznicki et al., 1991a, 199 1 b, 1991c, 1993; Deeba et al., 1994). The acidic properties of these materials have been discussed (Section 1I.B).

288

BRUNO NOTAN

IV. Synthesis The synthesis of crystalline microporous titanium silicate requires conditions that are specific to each product, such as the chemical nature of the structure directing agent, reactant concentrations, temperatures, and times of crystallization. For brevity, only a short outline of the basic rules that apply to the synthesis of all titanium silicates will be given, with reference to a few selected papers in which clear, detailed descriptions of the syntheses can be found. Alkalies. Whereas alkalies do not interfere with the synthesis of most zeolites, and in many cases are even required, they do interfere with the synthesis of titanium silicates (Taramasso et al., 1983; Notari, 1988, 1991; Bellussi et al., 1991a). Attempts to perform the synthesis in the presence of alkalies either failed (El-Hage et al., 1989, 1990) or produced materials with inferior catalytic properties (van der Pol, 1992a). In the presence of alkalies, the formation of an amorphous precipitate containing Ti" is observed. It is possible that this material redissolves only with difficulty and makes the incorporation of Ti'" into the zeolitic structure more difficult. A recent reinvestigation has confirmed the negative effect of alkalies in the synthesis of TS-1. Samples of TS- 1 prepared in the presence of Li, Na, and K in the synthesis gel produce materials that have considerable amounts of extra-framework TiO, , poor catalytic properties, and give a high rate of H202 decomposition. It was, however, observed that when the amount of Na in the synthesis gel is very low (Si/Na = loo), the TS-1 obtained does contain some extra-framework TiO,, but can be reactivated by treatment with H2S04to give a catalytic activity comparable to that of TS-I synthesized in the absence of Na and having equal Ti content (Khouw et al., 1995). Processes for the production of alkali-free tetraalkylammonium bases have been developed and are used in industry (Buonomo et al., 1989). Alkali-free bases are now commercially available. Mixingprocedure. This turns out to be a delicate operation because, even in the absence of alkalies, the formation of a precipitate containing Ti" can occur and prevent successhl synthesis. Cooling of reagents to 273 K, efficient stirring, and low rates of mixing have been recommended to prevent this phenomenon and its consequences (Kraushaar, 1989; van der Pol et al., 1992a). Calcination. After crystallization, the solid contains a substantial amount of organic base which must be removed to give the catalytically active material. The removal can be achieved only by decomposition, and calcination is the most common process used. But calcination at temperatures insufficient to bum all the organic or in-static air generates less active catalysts (Martens et al., 1993). On the other hand, calcination without temperature control can cause the sudden combustion of the organic material, and, at the high temperatures that are reached, the Ti" separates from the crystalline structure to form Ti02.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

289

Calcination in a controlled atmosphere ( N 2 + 2% 0 2 ) at 823 K until complete removal of the organic material has produced satisfactory results. The normal analysis of residual carbon, XRD, and UV-visible spectra indicate whether the operation has been successful. A. TS-1 Here, essential information for the synthesis of TS-I will be reviewed briefly.

1. Classical Mixed Metal Alkoxides This procedure consists of the preparation of a solution of mixed alkoxides of titanium and silicon such as TEOS and TEOT followed by hydrolysis with alkali free solution of TPA-OH, distillation of the alcohol, and crystallization of the resulting gel at 448 K (Taramasso et al., 1983). As was already indicated, the temperature at which the reagents are mixed and the rate of hydrolysis are critical. In subsequent work, the hypothesis was formulated that during the mixing of the metal alkoxides a reaction takes place with the production of mixed oligomers. In these mixed oligomers, with the titanium being highly dilute, Ti-0- Si bonds largely prevail over Ti-0-Ti bonds. Thus, isolated Ti" species are obtained, which then become part of the crystal lattice of the zeolite, provided that no precipitation takes place during the hydrolysis with TPA-OH (Kraushaar, 1989). A procedure reported for the synthesis of TS-1 indicates the precautions to be taken for a successful synthesis (van der Pol et al., 1992a). The synthesis of TS-I has also been carried out, under conditions similar to those used with TPA-OH, with other bases such as hexapropyl-l,6-hexanediammonium hydroxide, (CH&NCH2(CH2)4CH2N(CH&(OH)2 , which can be considered the dimer of TPA-OH (Tuel et al., 1994b). Mixtures of bases have also been used successfully: TPA-OH and tetraethylammonium hydroxide (TEA-OH) or TBA-OH and TEA-OH have produced TS-I (Tuel et al., 1993c, 1993d). The use of TBOT as a titanium-containing precursor has been proposed to simplify the procedure and incorporate more Ti" in TS-I by reducing the rate of hydrolysis of the precursor (Thangaraj et al., 1991a, 1992b); but, as discussed in Section 1II.A.1, the results have been questioned. A comprehensive investigation on the use of alkoxides of Ti" and of Si" obtained from different alcohols has demonstrated that there were no significant changes associated with the nature of the alcohol, provided that the reaction conditions are carefully controlled. The only reported exception is that of TMOS, as discussed in Section III.A.1 (Tuel et al., 1994a). The synthesis of TS-1 containing A13+, Fe3+, or Ga3+ is performed by the addition of suitable precursors of these metals, such as the nitrates dissolved in

290

BRUNO NOTARI

ethyl alcohol, and the addition of the resulting solution to the TEOS before the titanium alkoxide. From this point on, the procedure is the same as that for pure TS-1. 2. Colloidal SiOz Another method described for the synthesis of TS-1 involves the use of colloidal Si02 and tetrapropylammonium peroxo titanate (Taramasso et al., 1983). Indeed, TS- 1 can be produced by this method, but subsequent experience has shown that it has the serious drawback of impurities contained in colloidal S O 2 , particularly A13+. These impurities are incorporated into the crystalline product and modify the catalytic properties, as discussed in Section V. It is certainly possible to obtain pure colloidal Si02 by hydrolysis of TEOS, but in this case the method does not offer advantages over the use of metal alkoxides. 3.

Wetness Impregnation

This method consists of the use of a Si02-TiOz coprecipitated dry gel which is impregnated with an aqueous solution of TPA-OH and crystallized under autogenous pressure. At a high concentration of the base, dissolution of the oxides occurs, followed by crystallization in the presence of TPA-OH. The method should offer the advantage of requiring a relatively small amount of TPA-OH (Padovan et al., 1989, 1991). TS-1 has indeed been obtained, but the impurities present in the starting material create the same problems discussed for the colloidal Si02 method. Again, if a pure gel has to be obtained from hydrolysis of metal alkoxides (Uguina et al., 1994), the advantages are limited. 4. Secondary Synthesis The incorporation of Ti" in the crystal lattice of silicalite has been attempted by the reaction of TiC14 with dealuminated ZSM-5 (Kraushaar et al., 1988) or deborated borosilicalite (Carati et al., 1990). The same reaction has been used in the attempt to incorporate titanium in the crystal lattice of zeolite beta, mordenite or zeolite Y. In many cases catalytic properties have resulted, but the way in which the incorporation takes place has been questioned. Because of its molecular dimensions, TiC14 cannot enter or leave the pore system of ZSM-5. It has been shown that 89% of the OH groups present in the preformed zeolite as SiOH remain unreacted after treatment at 573 K with TiCI4. The incorporation of titanium must therefore be limited to the outer part of the crystals or proceed through a severe chemical attack with removal of silicon and formation of a secondary pore system (De Ruiter et al., 1993). Deposits of Ti02 on the outer part of the crystal treated with TiC14 have indeed been observed (Carati et al., 1990), as has abnormal behavior in the oxidation of phenol (Section V.C.3.c).

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

29 I

5. Fluoride Method The synthesis of TS-1 in the presence of fluoride ions has been described (Popa et al., 1988; Guth et al., 1989; Shilun et al., 1989). Subsequent investigation has shown that extra-framework Ti" is formed when fluorides are used (Lopez ef al., 1993). B. TS-2 The synthesis of TS-2 has been camed out with the same procedures and precautions described for the synthesis of TS-I, but with the use of TBA-OH (Bellussi et al., 1989; Reddy, J. S. et al., 1990). The synthesis has also been performed with the use of the phosphonium base TBP-OH (Section 1II.B).

C. Ti-ZSM-48 The incorporation of Ti" into a zeolite having the structure of ZSM-48 has resulted from the use of fumed silica (Cabosil), TBOT, H 2 0 2 , and diaminooctane. The titanium alkoxide is transformed into a titanium peroxo compound and added to a suspension of SiOz in diaminooctane. Crystallization for 10 days at 448 K in rotated autoclaves produced Ti-ZSM-48 (Serrano et al., 1992). The synthesis of Ti-ZSM-48 has also been obtained by using the base hexamethonium hydroxide [(CH3)3NCH2(CH2)4CH2N(CH3)3(OH)2] (Section 1II.C). D. Ti-BETA This material has been obtained by direct hydrothermal synthesis and by secondary synthesis. In the hydrothermal synthesis, amorphous Si02 (Aerosil 200, Degussa), sodium aluminate, NaOH and KOH, TEOT, and TEA-OH are used. The hydrolysis of the titanium alkoxide with TEA-OH was followed by the addition of Si02 and of a solution containing A13+ salts (Camblor, 1993a, 1993b). The presence of A13+ is necessary for the crystallization of zeolite beta. It was later reported that the presence of alkalies promotes the formation of poorly crystalline precipitates containing the oxides of titanium and of the alkali, and that better results were obtained when the synthesis was performed in the absence of alkalies (Blasco et al., 1993). This demonstrates that the same alkali interference observed in the preparation of products containing only the oxides of titanium and silicon operates as well in these ternary compositions of titanium, silicon, and aluminum oxides. The incorporation of Ti" in framework position is slightly higher for Ti-beta than for TS-I (Section 1II.D). Ti-Beta has also been obtained by secondary synthesis, by reaction of beta zeolite with TiCI4 (Ferrini et al., 1992), or with ammonium titanyl oxalate

292

BRUNO NOTAIU

(Reddy, J. S. et al., 1995). The presence of framework cations such as A13+ that can be extracted by oxalate species or the presence of defects is considered requisite for the subsequent incorporation of titanium. This method could prove useful for obtaining other titanium-modified titanium silicates difficult to synthesize by other procedures (Reddy, J. S. et al., 1995). E. Ti-MCM-41, Ti-MCM-48, AND Ti-HMS These materials have been obtained by a modification of the procedure followed for the preparation of the pure silica analogs (Kresge et al., 1992a, 1992b). TEOS and TIOT dissolved in an ethanol-isopropanol mixture were added to an aqueous solution of dodecylamine and HCl. Aging of the resulting gel for 18 h afforded the crystalline material with HMS structure (Tanev et al., 1994). For MCM-41, amorphous SiOz (Aerosil 200, Degussa) is used as the Si02 source with tetramethylammonium hydroxide (TMA-OH), TEOT, and hexadecylammonium hydroxide and bromide as structure directing agents. The gel obtained was held at 413 K for 28 h to yield Ti-MCM-41 (Corma et al., 1994). Before use, all samples were carefully calcined to remove the organic materials.

F. Ti-MORDENITE Ti-Mordenite has been obtained by reaction of TiC14 with dealuminated mordenite and by hydrothermal synthesis. The reaction of TiC14 with dealuminated mordenite was carried out at 723 K following the method of Kraushaar and van Hooff (1988, 1989). In the hydrothermal synthesis, Si02 powder, NaOH, A1 isopropoxide, and TBOT react at 448 K for 3 days (Kim et al., 1993). G. Ti-Y, Ti-L, Ti-W, AND Ti-R The synthesis of these titanium-containing materials by treatment of the corresponding preformed zeolites with (NH4)2TiF6 at 348-368 K has been claimed (Skeels et al., 1989; Skeels, 1993). No evidence for Ti'" incorporation has been provided. H. ETS-4 AND ETS-10 Titanium-containing materials have been synthesized from strongly alkaline suspensions obtained by mixing TiC14 with alkali silicate and sodium aluminate solutions. The gel obtained was homogenized and mixed in an autoclave at

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

293

473 K for 24 h. The crystalline material was recovered after suspension in 5% NaOH solution by filtration (Kuznicki, 1989, 1990; Kumicki et al., 1991a, 1991b, 1991c, 1993). The synthesis of ETS-10 has also been carried out by the use of tetramethylammonium salts, with a significant acceleration of crystal growth (Valtchev et al., 1994). These titanium silicates contain Ti'" in octahedral coordination and do not have the catalytic properties of the titanium silicates discussed previously (see Sections I11 and VI).

V.

Catalytic Reactions

Titanium silicates catalyze a large number of different reactions. The literature of many of the reactions is characterized by discrepancies in the results reported by different groups. Very likely, these discrepancies are consequences of inconsistencies in catalyst preparation and structure, and in reaction conditions. lssues related to catalyst structure are briefly reviewed below, followed by a review of specific reactions. Framework and extra-framework titanium. In the previous sections it has been shown that titanium silicates are characterized by the isomorphous substitution of Si with Ti in crystalline silicas. Extra-framework Ti02 phases can be present, and indeed a major problem is to prevent their formation, both during syntheses and during use under severe operating conditions. The change in catalytic properties resulting from the presence of extra-framework Ti02 depends on the catalytic reaction. In the case of alkane and alkene oxidation, the differences are limited, whereas in phenol oxidation a remarkable dependence of selectivity on Ti02 content is observed. Ti02 is a very efficient catalyst for H202 decomposition, and because it is often highly dispersed, this decomposition; becomes significant even when just a little Ti02 is present, resulting in lower H202 selectivity. But Ti02 can also catalyze other reactions and in this way reduce the yield of the desired products. In practice the necessity of producing pure titanium silicate catalysts and using them under conditions in which no deterioration can take place cannot be overemphasized. In some reports, the presence of the two forms of titanium has been described as arising from two types of sites identified as isolated titanium framework sites and Ti02 particles, and the reactions have been attributed to the catalytic activity of one or the other phase (Huybrechts et al., 1992). Since it is possible to obtain pure phase titanium silicates, it seems preferable to identify the Ti02 phase as an impurity. Impurities. Apart from Ti02, many other impurities can be present, and some of them can modify the products of a catalytic reaction. One of the most common is A13+, which can be introduced with the Si02 raw material. During crystallization, A13 is inserted into the structure and imparts acidic properties +

294

BRUNO NOTARI

to the catalyst, with the consequence that decomposition of H202is greatly enhanced (Kim et al., 1993), and acid-catalyzed reactions-absent with pure titanium silicates-proceed on the impure materials. Titanium silicates have also been synthesized with the intentional addition of A13+ to modify the properties of the materials (Section III.A.2). Another common impurity is Fe3+, the activity of which for H202 decomposition is well known. Fe" can originate from the raw materials used to produce the catalyst or can be introduced with the reactants, accumulating during extensive use. Alkalies have a negative impact on the synthesis of titanium silicates and, if added after the synthesis, transform active catalysts into inactive ones (Bellussi et al., 1991; Clerici et al., 1992, 1993). In some cases, catalysts deactivated by alkalies can be reactivated with acid treatment (Khouw et al., 1994). A clear difference should be made between alkalies present before the synthesis, which prevent the formation of the titanium silicates, and alkalies introduced after the synthesis, which inactivate an active material, but can be removed to allow the regeneration of the catalytic activity. Crystallite dimensions. Crystallite dimensions plays a role in determining the rates of reactions, and their control is of hndamental importance not only for the catalytic activity, but also for the selectivity, since with low rates of the desired reaction, the relative importance of secondary reactions may be greater. The effects of crystallite dimensions have been demonstrated for 1-butene epoxidation (Clerici et al., 1993) and for phenol hydroxylation (van der Pol et al., 1993), and they are significant for many reactions carried with liquid phase reactants. The results obtained for phenol hydroxylation with samples of different dimensions (Table 11) are useful to clarify this point. The results show that crystallites in the range 0.14.3 pm have the highest activity and selectivity. The authors demonstrated that the higher activity of the small particles is a consequence of reducing pore-diffusion limitations. TABLE I1 lnfluence of Particle Size vf TS-1 on Catalytic PerJbrmance"

Sample 1 2 3 4

Average catalyst particle size (Pm) 0.24.3 0.34.5 5.0-6.0 10.0-1 I .7

conversion after I h

Selectivity after I hh

Yield'

(%)

(%)

(%)

50

95 93

93 92 40

H202

44 6 2.5

15 8

18

"Adapted from van der Pol and van Hooff (1992b). Moles of dihydroxybenzene/mol of reacted 100. "Moles of dihydroxybenzenes/moI of H 2 0 2 added at complete H z 0 2 conversion X 100.

HzOzX

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

295

Catalysts of such small dimensions can hardly be handled in practice. A procedure for binding 0.14.3-pm particles into agglomerates of 20-30 pm that can be easily handled while maintaining the same catalytic properties has been a key step in the development of industrial processes using titanium silicates (Bellussi et al., 1986). Reaction conditions. Reaction temperature is of the utmost importance since titanium peroxo compounds decompose above 323 K. Solvents also have a remarkable effect in reaction rates and selectivities, as will be discussed below.

A.

BECKMANN REARRANGEMENT

The transformation of oximes into lactams (the Beckmann rearrangementt and particularly the industrially important transformation of cyclohexanone oxime into caprolactam, a key intermediate in the manufacture of nylon-has been investigated with TS-I and TS-2 catalysts and the results compared with those obtained with silicalite- 1, silicalite-2, ZSM-5, ZSM- 1 1, AI-TS-2, and fumed SiOz. The reaction proceeds at 600 K on silicalite-1 (Sato et al., 1989), but higher selectivity and better overall performances are obtained when TS-I (Thangaraj et al., 1992a) or TS-2 (Reddy, J. S. et al., 1993) is used as the catalyst. The A13+-containing materials ZSM-5, ZSM-I I , and AI-TS-2 give much poorer performance, very likely as a consequence of their acidity, which promotes the decomposition of the oximes into ketones, followed by side reactions and the formation of high-boiling products that cause rapid catalyst deactivation.

B. MTBE SYNTHESIS The synthesis of methyl t-butyl ether (MTBE) from isobutylene and methanol on TS-I has been investigated. This reaction is catalyzed by acids and the industrial production is carried out with sulfonic acid resin catalysts. It has been reported that at 363-383K the reaction proceeds in the presence of the acidic HZSM-5, but also on TS- I , which is much more weakly acidic. However, the characterization of the catalysts used is not completely satisfactory; for instance, the IR spectra reported do not show the 960-cm- band that is always present in titanium-containing silicas. It is therefore possible that the materials with which the reaction has been studied are not pute-phase TS-I. The catalytic activity for MTBE synthesis is, in any case, an interesting result, and further investigations with fully characterized catalysts are expected to provide a satisfactory interpretation of these results (Chang et al., 1992).

'

296

BRUNO NOTARI

C. OXIDATION REACTIONS Titanium silicates are catalysts for various oxidation reactions. Most of those investigated have been carried out with H202 and hydroperoxides as the oxidants, and a few have been investigated with ozone and oxygen as oxidants.

1. Oxidation with Ozone A TS-1 that contains vanadium has been reported to be a highly active catalyst for the ozonization of 3,6-dichloropyridine-2-carboxylicacid. V-Silicalite has lower activity, and V-Al-silicalite is inactive (Komatowski et af., 1993). 2. Oxidation with Oxygen The oxidation of methanol with oxygen has been used to characterize titanium-containing catalysts. Pure silicalite is inactive, and TS- 1 catalyzes the oxidation of methanol to give a mixture of products: CHJOH + 0

2

-+

H-CHO 61%

+ COK02 + H-COOCHJ 18%

(1 1)

15%

However, 2% Ti02 dispersed on SiOz is also a catalyst for the oxidation of methanol, and therefore this reaction does not allow one to discriminate between framework titanium and Ti02. It has been found that in the presence of TS-I containing A13+, the only reaction product is dimethyl ether formed by the dehydration of methanol: 2CH,OH * CHJOCH~+ H20

(12)

The rate of the dehydration catalyzed by the acidic centers associated with AI3+ is much higher than the rate of oxidation catalyzed by the titanium centers. Under these reaction conditions, equilibrium favors dimethyl ether, which is less reactive than methanol toward oxidation; therefore, the ether appears as the only reaction product (Deo et af., 1993). The oxidative dehydrogenation of aqueous ethanol in the presence of oxygen (in air) has also been investigated. At 30% conversion of ethanol, TS-1 is highly selective for the production of acetaldehyde, with limited amounts of other products such as ethylene, diethyl ether, and C 0 2 being formed (Hari Prasad Rao et af., 1991). Under similar conditions, ZSM-5 and Fe-ZSM-5 catalyze the dehydration of ethanol to diethyl ether, whereas Ti02/Si02 gives a mixture of ethylene, acetaldehyde, and diethyl ether, as shown in Table 111. The ammoximation of cyclohexahone with NH3 and 0 2 , which had been investigated with silica catalysts (Armor et af., 1979, 1980, 1982), has been investigated with TS-1 as the catalyst (Dreoni et af.,1991). However, the results have not been considered satisfactory for industrial exploitation, especially when

297

MICROPOROUS CRYSTALLINE TITANIUM SILICATES TABLE I11 Oxidation of 10% Aqueous Ethanol at 30% Conversion" Product distribution Catalyst

SdMh ratio

CH2=CH2

MeCHO

s-I TS- I S-l + Ti02 ZSM-5 Fe-ZSM-5

3,000 37 38 42 41

25 4.6 17.4 0. I 0.3

22 88 46.2 0. I

MeCOOH

Et2O

COz

Others

-

51 1.5

-

2.0 4.6 2.7 0.3 6.0

0.1 -

-

33.1 99.6 93.5

1.1 0.5 -

0.1

"Hari Prasad Rao et al. (1991). h M = Ti, Al, Fe.

compared with the results obtained with HzOZas the oxidant (Section V.C.3.e). The hydroxylation of benzene and the oxidation of hexane by 0 2 and Hz in the presence of palladium-containing titanium silicates, likely proceeding through the intermediate formation of H202, have also been reported (Tatsumi et al., 1992; Ferrini et al., 1992). 3.

Oxidation with Peroxides and Hydrogen Peroxide

Many different oxidation reactions with peroxides and H202 as the oxidants and titanium silicates as the catalysts have been reported. A most relevant case is that of propylene oxide which, as mentioned, is produced from propylene and EBHP on Ti02-SiOz catalyst by the Shell process. Another process, by ARCO, uses propylene and TBHP with homogeneous catalysts based on molybdenum compounds. Together the two processes account for the production of 1 million tons per year. The scope of the oxidation reaction with hydroperoxides has been expanded with the recent development of Ti02-SiOz aerogel materials, which are very active and selective in the epoxidation of many different olefins: cyclo-hexene, cyclo-dodecene, norbornene, and limonene are transformed to the corresponding epoxides with excellent selectivities with respective to both the hydroperoxide and the olefin. On the other hand, these catalysts are inactive or undergo rapid deactivation when H2O2 is used as the oxidant. This has been attributed to the hydrolysis of Ti-0-Si bonds, followed by migration of Ti'" from the network to the adsorbed H 2 0 layer where both condensation to Ti-0-Ti and formation of soluble titanium peroxo species takes place (Hutter et al., 199%). Crystalline titanium silicates on the contrary have a very low rate of hydrolysis. This is due to the multiple bonding of Ti'" with -04- units, also to the hydrophobic nature of the crystals which screens out the bulk of the water from

298

BRUNO NOTARl

the pore system, and thus contributes to the stability of the structure. However, the same crystalline structure requires that both the reactant and the oxidant have dimensions smaller than 5.6 A, the dimensions of the pore system in pentasil zeolites. Ti-beta, having pores of 7 A, reduces this limitation somewhat. The acid properties associated with the presence of A13+ can be, if necessary, modified by suitable treatments. Both H202and hydroperoxides are industrially important oxidants. An accurate evaluation of advantages and disadvantages requires an accurate analysis of every specific case, in view of the different technical problems and economic constraints that the use of one or the other entails. The reactivity of H202 is so high that it can easily oxidize many primary reaction products, and these reactions become more likely as the reaction temperature is increased. Some of these reactions are influenced by reactant shape selectivity and by restricted transition-state shape selectivity. a. Alcohols and Glycols. Alcohols undergo oxidation with H202 in the presence of titanium silicates to produce the corresponding aldehydes or ketones. However, in the presence of other oxidizable molecules such as olefins, the rate of alcohol oxidation is remarkably decreased, to the point that alcohols can often be used as solvents. This is particularly true for methanol, the alcohol having the lowest rate of oxidation. The rates of disappearance of the primary alcohols methanol, ethanol, I-butanol, and 1-octanol at 318 K are presented in Fig. 18. The relative reaction rates calculated from the 50% H202 conversion times are the following: methanol, I ; ethanol, 12; 1-propanol, 8.5; I-butanol, 6.5; 1-octanol, 2.8; and 2-methyl-1-propanol, 2.5.

rA

0 0

c

0.5

Y

0.4

0

0.3

cc

0.2 0.1

0 0

1

2

3

4

5

6

7

Time / min FIG. 18. Conversion data for H202oxidation of primary alcohols catalyzed by TS-I. (A) Ethanol, (0)methanol, ( X ) I-butanol, ( + ) I-octanol. (From Romano ef al., 1990.)

299

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

0 \

0 2

4

6

8 10 12 14 16 18 20 22 24 262830

Time / min FIG. 19. H 2 0 2 oxidation of I-propanol catalyzed by TS-I. ( A ) Acid, ( + ) aldehyde,

(*)

acetal,

(0) ester. (From Romano e/ a / . , 1990.)

The primary alcohol that reacts fastest is ethanol, and the one that reacts most slowly is methanol. The difference in the reaction rates between 2-methyl1-propanol and 1-butanol has been taken as evidence that, under these conditions, the reaction takes place inside the restricted environment of the catalyst. Methanol is oxidized to formaldehyde and dimethyl formate. Under the experimental conditions, neither formic acid nor its esters were detected. The other primary alcohols are oxidized to give aldehydes and, in a subsequent reaction, carboxylic acids. Acetals and esters are also formed. Profiles of the reaction products in the oxidation of 1-propanol are shown in Fig. 19. The aldehyde selectivity is high at low conversions, but acetal and ester are also formed. The acid is obtained only at longer reaction times. The ester is evidently not formed from the acid, and is probably the product of a fast oxidation of a hemiacetalic intermediate. The reaction sequence is represented in Scheme 1. Secondary alcohols are selectively oxidized to ketones. No other products have been detected, and the H202 decomposition is limited. R \ /H C / \ R OH

-

+ H202

R\ /C = O

+ 2H20

(13)

R'

Secondary alcohols react at much higher rates than primary alcohols having similar molecular dimensions. Conversiodtime plots for several secondary alcohols are reported in Fig 20. 2-Pentanol reacts much faster than 3-pentanol. The ratio of reactivities calculated from data at 50% H 2 0 2 conversion is 12 : 1. Because in term of diffusion rates and chemical behavior these two alcohols are similar to each other, the results are explained by restricted transition-state selectivity, a steric influence of the catalyst pores. Cyclohexanol is oxidized at a very low rate, and this is best

300

BRUNO NOTARI

explained by shape-selective catalysis, i.e., restriction of the transport of the cyclohexanol in the pores, which implies that the reaction takes place inside the pores (Romano et al., 1990). The kinetics of secondary alcohol oxidation has been investigated with various reactants. When 2-octanol is oxidized with acetone solvent in the presence of excess reactant, the reaction is first-order with respect to H 2 0 2 and zero-order with respect to the alcohol. Typical results are presented in Fig. 2 1. Temperature changes in the range 303-343 K influence the activity, but not the selectivity; yields of ketone of 95% or higher at total H 2 0 2 consumption are

0.6 0.5 0.4

0.3 0.2 0.1 0 0

8

0

10

20

30

40

50

60

70

Time / min

FIG. 20. Conversion data for H202 oxidation of secondary alcohols catalyzed by TS-1. ( 0 )2-Butano1, ( + ) 2-pentano1, (A) 3-pentano1, (1) cyclohexanol. (From Romano ef al., 1990.)

30 1

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

';c

-1.5-

7 v

-c

-2 -2.5

-3 -3.5

0

20

40

60

80

100

120

140

160

180

Time / min FIG. 21. H 2 0 2 oxidation of 2-octanol catalyzed by TS-I: first-order kinetics; x represents conversion. (From van der Pol er a/., 1993b.)

obtained at all these temperatures. From the first-order rate constants, an activation energy of 7 1 kJ/mol was found for the oxidation of 2-octanol (van der Pol et af., 1993b). When 2-butanol was oxidized in methanol, a different rate expression was found. In the presence of excess H 2 0 2 , the reaction is first-order with respect to the alcohol, and in the presence of excess 2-butanol, the reaction is zero-order with respect to H 2 0 2 . To explain these results, it is assumed that when methanol is used as the solvent, it is more strongly adsorbed at the titanium center than 2-butanol, and that the coordination of H z 0 2 is stronger in the presence of methanol than in the presence of other alcohols. These conditions give rise to the observed kinetics, first-order in 2-butanol and zero-order in H 2 0 2 (Maspero et al., 1994). By analogy, the results obtained in the oxidation of 2-octanol in acetone can be explained if it is assumed that the coordination of the alcohol at the titanium center is strong (as it has no competition from the solvent) and that the rate of reaction is therefore limited by the availability of H 2 0 2 . This gives rise to the observed rate expression, zero-order in 2-octanol and first-order in H 2 0 2 . More complex rate expressions were observed when the reaction was carried out in the absence of solvents or when other solvents, such as H 2 0 or tert-butyl alcohol, were used. Complex rate expressions arise because a number of effects come into play, including preferential sorption in the hydrophobic zeolite, competition for adsorption at the titanium center, and competitive diffusion and reaction. It is therefore difficult to obtain unambiguous answers to mechanistic questions from kinetics.

302

BRUNO NOTARl

Investigations of alcohols with 6-9 carbon atoms have shown that the chain length influences the rate of oxidation of 2-alcohols. The rates decline in the order 2-octanol> 2-heptanol > 2-hexanolB 2-nonanol. A similar sequence was found for the 1-alcohols: 1-octanol > 1-heptanol > 1-hexanol > 1-nonanol. However, the reactivities of primary alcohols are much lower than the reactivities of secondary alcohols. While an increase in reactivity of 2-alcohols with increasing chain length can be expected on the basis of chemical reactivity, the decrease beyond CBmust have another origin, which may be reactant shape selectivity in the TS-1 catalyst. The 2-alcohol generally react faster than the 3-alcohol (Van der Pol et al., 1993b). Glycols undergo oxidation with H202 and titanium silicates, but it is also possible that some of the reactions observed proceed as noncatalytic reactions once the primary oxidation products are formed. Ethylene glycol is oxidized to glycolic acid: CH2-CH2 I I OH OH

+ H202

// -+

CH2-C I OH

0

+ H20

\

(14)

OH

With a 1 : 1 molar ratio of ethylene glycol and H202, the selectivity to glycolic acid was 86% at 44% conversion of the glycol and 99% conversion of the H202. Small amounts of glyoxylic acid (3%) and formic acid were also detected. Propylene glycol is oxidized to hydroxyacetone with high selectivity: CHj-CH-CH2

I

I

+ H202

OH OH

-+

CHj-C-CH2

II

I

+ 2H20

(15)

0 OH

With a molar ratio of propylene glycol to H202 of 2.5, the selectivity to hydroxyacetone at 32% conversion of the glycol was 94%, and the selectivity based on H202 was 85%. Small amounts of acetic acid and formic acid were detected. The initial oxidation proceeds with high selectivity for the secondary alcohol group. Further oxidation affords oxidative cleavage products rather than pyruvic acid, as is observed when the oxidation of hydroxyacetone is carried out with O2 and noble metal catalysts. More complex is the oxidation of 2,3-butandiol (I): the first oxidation product, acetoin (11), can be obtained at 333 K with 96% selectivity at 62% conversion if the solvent is water. When methanol is used as the solvent, the reaction rate is decreased, but the product distribution is the same. When acetone is used as the solvent, a larger amount of butane-2,3-dione (111) is obtained, as summarized in Table IV. C-C-C-C

I

I

OH OH I

C-C-C-C II I 0 OH 11

C-C-C-C

II II 0 0 111

CH3COOH IV

303

MICROPOROUS CRYSTALLINE TITANIUM SILICATES TABLE IV Oxidation of Butane-2.3-diol with H202 Catalyzed by TS-I‘ Selectivity (%)

Diol/HzOz ratio

I I I 4 4 4 4 4

Time

Conversionh

Solvent

(h)

(%)

I1

111

Hz0 Hz0 H2O Hz0 CHiOH CHiOH acetone acetone

3 5 6 4 2.5 4 2 4

40 60 62 16

98 96 96 96 94 91 47 48

2 4 4 4 6 9 53 52

11

16 10 15

~

~

~~~~~

~

“Sheldon and Dakka (1992). ’At a molar ratio of 4 : I , the maximum conversion is 25%.

Acetic acid (IV) is formed when acetoin or butane-2,3-dione is oxidized in acetone, even at low temperatures. The use of acetone significantly changes the relative rates of the different oxidation reactions. The possibility that acetone takes part in promoting the oxidation of acetoin deserves further investigation. b. Oiejins and Diolefins. The oxidation of the lower olefins with H 2 0 2 catalyzed by pure crystalline titanium silicates selectively produces the corresponding epoxides. With some olefins, the reaction can proceed further to give glycols and eventually C=C cleavage products. Olefin oxidation can be carried out with a polar solvent so that a single reactant phase including H 2 0 can be obtained. Methanol or methanol-water mixtures are convenient because, in the presence of olefins, the rate of methanol oxidation is negligible. Ethanol can also be used, but its own oxidation should be evaluated. Ethylene is oxidized to give ethylene oxide, and propylene is oxidized to give propylene oxide (PO): CHI

\

CH =CH2

+ H202

-

CH,

\

CH -CH2 \

/

+ H20

(16)

0

In a subsequent reaction with the solvent, PO produces small amounts of propylene glycol or glycol monoethers. The rate of formation of PO from propylene, H 2 0 2 , and TS-I in methanol at 313 K is shown in Fig. 22. The selectivity to PO is 85%, but it can be improved by silanization of the catalyst, which reduces the hydrolytic activity, or by addition of small amounts of sodium acetate to the reaction mixture. With these modifications, at 97%

304

BRUNO NOTARI

Oo

zo

40

m

M)

100

izo

140

ie

Time I rnin FIG. 22. H 2 0 2 epoxidation of propylene catalyzed by TS-I in methanol-water at 313 K: (1) propylene oxide; (2) H202; (3) propylene glycol and glycol ethers. (From Clerici ef a/., 1991b.)

H202conversion, the selectivity based on PO reaches 97%, and glycol formation is reduced to 3% without impairing the overall catalytic activity. This result is interesting in view of the reported negative effect of sodium acetate on the catalytic activity for n-hexane oxidation (Clerici, 1991a). Clearly, the concentration of the sodium acetate is critical; small amounts reduce the activity of SiOH groups that catalyze hydrolysis of the epoxide initially formed, while larger amounts interfere with the catalytic activity of the titanium centers. When ethanol is used as the solvent, it is oxidized to acetaldehyde at a rate close to that of propylene oxidation, showing that both reagents compete for the catalytic centers. The oxidation of propylene can also be carried out in other solvents, such as methyl acetate, acetonitrile, or tert-butyl alcohol, but the rates are lower than the rate in a methanol-water mixture. The reaction in tert-butyl alcohol is hrther complicated by the fact that, in the presence of TS- 1, the alcohol and H202 react to give TBHP (Maspero et al., 1989). The oxidation of the different butenes gives the corresponding epoxides with high selectivity; the by-products glycol and glycol ethers can be reduced to very low concentrations, as in the case of propylene oxidation. An interesting feature is the retention of stereochemical configuration: cis-2-butene gives exclusively the cis-epoxide, and trans-2-butene gives exclusively the trans-epoxide (Clerici et al., 1989b, 1989~). The sequence of reactivity for the butenes is cis-2-butene (1 6) > 1-butene (6) > isobutylene (4.7) > trans-2-butene (1 .O). A higher reaction rate for the cis isomer was also observed in the epoxidation of 2-hexenes, as was retention of stereochemical configuration (Tatsumi et al., 1990a). In the epoxidation reactions with H202 or organic hydroperoxides catalyzed by group IV-VI metals, the active oxidizing species act as electophiles. The reactivity is determined by the electron density at the double bond, which

305

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

increases with alkyl substitution because of the electron-donating properties of alkyl groups. The sequence of reactivity is isobutylene > cis-2-butene > trans-2-butene > 1-butene. It is reasonable to consider that in titanium silicate-catalyzed reactions the oxidizing species also acts as an electrophile. The different order of reactivity of the C4 olefins in the presence of titanium silicates relative to that observed with soluble catalysts must therefore arise from the fact that alkyl substitution at the double bond is responsible not only for inductive effects, but also for increases in the size and the steric requirements of the molecules. Since the rates of difhsion of the different butenes cannot be the cause of the different reaction rates, a restricted transition-state selectivity must be operating. The presence of fluorides, either HF or NH4F, decreases the rate of 1-butene epoxidation. Fluoride ions coordinate strongly at the titanium centers (the enthalpy of formation of TiF4 is - 1550 kJ/mol) and in this way inhibit the interaction with the reagent molecules. The oxidations of c5-C~ olefins, allyl chloride, allyl alcohol, and allyl methacrylate with H202 give the corresponding epoxides. Data characterizing oxidations camed out in methanol solvent at an olefin concentration of 0.90 M, in the presence of TS-I (6.2 g/L), are given in Table V. As discussed for the epoxidation of propylene, as well as in the case of allyl alcohol epoxidation, the selectivity can be modified by treatments of the catalyst. Addition of Na2C03 drastically reduces the catalytic activity, while with addition of sodium azide a satisfactory level of catalytic activity is maintained with 88% selectivity to the epoxide glycidol. When aluminum-containing TS- 1 is used, no glycidol is formed; only products arising from secondary solvolysis of glycidol are obtained, the solvolysis being attributed to the acidic character of aluminum-containing TS-1 (Hutching et al., 1995). TABLE V Oxidation of CrC8 Olefns and Ally1 Compounds in Methanol"

Olefin I-Pentene I -Hexene Cyclohexane I-Octene Ally1 chloride Allyl alcohol Allyl methacrylate a Tatsumi

Concentration of H202 Temperature Time, I (M) (K) (min) 0.18 0.18 0.18 0.17 0.18 0.14

298 298 298 318 318 318 338

60 70 90 45 30 35 -

H202 Selectivity conversion based on H202

(%)

(%)

(min)

94 88 9 81 98 81 -

91 90

5 8

h

-

91 92 12 77

5 7

et al. ( 199 I ) and Clerici et al. ( 1993b). Not determined.

16

-

3 06

BRUNO NOTARI

Evidence of variables that influence the relative rates of reaction of olefins and alcohols was obtained from experiments with compounds that have both olefinic and alcoholic functions and by the competitive oxidation of mixtures of olefins and alcohols. The data of Table VI show that when the double bond has no substituents, as in ally1 alcohol, but-3-en- 1-01, or 2-methylbut-3-en- 1-01, only the epoxide is formed; but when the double bond has substituents, the epoxidation rate is decreased and ketone and aldehyde products from the oxidation of the OH group are formed. This effect is more pronounced with a greater degree of substitution. Since the double bond and the OH group are part of the same molecule, the difference must arise from the different abilities of the reactants to coordinate and react at the titanium center; restricted transition-state shape selectivity is a possibility. The terminal double bond, sterically less hindered, interacts strongly with titanium, preventing coordination of the competing OH TABLE VI Oxidation of Unsaturated Alocohols in the Presence of TS-I" Product yield [mol (mol Ti)-'] Reactant

Ketonelaldehyde

Epoxide 19 16

30

-OH

"Tatsumi et al., 1993.

31

95

37

4

7

27

43

65

44

141

98

94

18

10

75

17

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

307

group. Because of steric hindrance, this interaction is weaker in substituted olefins, allowing the OH group to undergo oxidation. The steric interactions between the reagents and the titanium centers embedded in the zeolitic structure are demonstrated by the relative rates of reaction of two closely related molecules: C

\

c

/

C

\

c

/

O

but-2-en- 1-01

H

C

\

c

/

C

\

c

/

OH

I C 3-methylbut-2-en- 1-01

Both undergo oxidation with epoxide and aldehyde formation, but the former gives an epoxidatiodalcohol oxidation ratio of 3, whereas the latter gives an epoxidatiodalcohol oxidation ratio of only 0.2. That this difference is due to the steric requirements inside the pores of the catalyst is demonstrated by the fact that the relative rates of epoxidation and alcohol oxidation of the same molecules on a large-pore Ti02/SiOz catalyst are almost the same (Tatsumi et al., 1993).

The epoxidation of cis- and trans- 1-hydroxy-3,7-dimethylocta-2,6-diene with H202 and TS-1 is chemoselective at the 2-position and stereoselective: no epoxidation takes place at the 6-position, and reactant molecules retain their structure in the products. The interesting results have been described as hydroxy-assisted epoxidation. The role of the OH group in the reaction is confirmed by the fact that in the epoxidation of cyclopent-2-en- 1-01, the product with the 0 cis to the OH group is favored over the one in trans by a factor of 9 : 1. The same steric preference was found in the epoxidation of cyclo-hex2-en-1-01 (Kumar et al., 1995). The competitive oxidation of olefins has also been investigated in the presence of alkanes. As discussed below, alkanes are oxidized by H202 in the

308

BRUNO NOTARI

presence of titanium silicates. When the oxidation of 1-octene was carried out in the presence of n-hexane under conditions that would lead to the oxidation of each if it were used separately, it was observed that 1-octene is preferentially oxidized by a factor greater than 40. (Huybrechts et al., 1992). Cyclohexene is oxidized very slowly in the presence of TS-1; little if any epoxide could be obtained under conditions of rapid oxidation of 1- and 2-alkenes to the corresponding epoxides. This low reactivity has been ascribed to the molecular dimensions of cyclohexene, which cannot enter the channel system of TS-1. Evidence for this suggestion was obtained by elution chromatography; when TS- 1 was loaded in a chromatographic column and a mixture of cyclohexene and 2-hexene injected, the retention time for cyclohexene was much less than that of linear 2-hexenes, despite the higher boiling point of cyclohexene (Tatsumi et al., 1990a). Cyclohexene can be oxidized by H 2 0 2 if Ti-beta is used as the catalyst (Corma et al., 1994). A comparison of the performances obtained with the two catalysts, TS-1 and Ti-beta, for the oxidation of cyclohexene and other olefins is given in Table VII. The results confirm the previously reported low reactivity of cyclohexene when the catalyst is TS-1 and indicate that Ti-beta is active for the oxidation of cyclohexene and other bulky olefins. However, for cyclohexene and the linear olefins, the major reaction products formed in the presence of Ti-beta are glycols and glycol ethers, whereas in the presence of TS-I, epoxides are predominantly formed. Also in this case, the epoxides initially formed in the presence of Ti-beta undergo secondary reactions catalyzed by the acidic centers associated with the aluminum in the material, as previously seen for ally1 alcohol and for the epoxidation of 1 -butene on aluminum-containing TS-1 (Bellussi et al., 199la). A different product composition was observed for cyclododecene, TABLE VII Selective Oxidation of Olefns Catalyzed by TS-I and Ti-Beta” TS- I H202

Olefin 1-Hexene Cyclohexene I -Dodecene Cyclododecene

TON”

conversion

(s-I)

(%)

50

98 83 26

I 110

5

Ti-Beta Selectivity

conversion (%)

EP

GJ

GE‘

(s-I)

96

-

100

23 34

4 -

12 14 87 20

77 66

-

Selectivity

H202

TON

80 80 80 41

E

G

GE

12

8

80

-

-

100

-

100

-

80

20

-

“Coma er al. (1994b). ’TON = turnover number. ‘E = epoxide. d G = glycol. ‘GE = glycol ethers.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

309

which in the presence of Ti-beta gives a large amount of epoxide. This result indicates that the reaction of cyclododecene does not proceed inside the pore structure, since in that case the same secondary reactions with glycol formation that are observed for the other alkenes would occur. Thus, for each catalyst the cyclododecene reaction likely takes place on the outer surfaces of the crystals, giving, with some difference in rates, a similar product distribution for each. The fact that the reactivity of cyclododecene is higher than that of cyclohexene in the presence of TS-1 is surprising, because its dimensions are less favorable than those of cyclohexene; reactions on the outer catalyst surface are very likely responsible for these results. In some cases, oxidation of double bonds does not stop at the epoxide, but proceeds further to oxidative cleavage of the double bond. It was reported that the reaction of a-methyl styrene with H202 in the presence of TS-1 or TS-2 produces a-methyl styrene epoxide (15%), a-methyl styrene diol (10 4 0 % ) and acetophenone (40-60%) (Reddy, J. S. et al., 1992). However, results similar to those obtained with titanium silicates were obtained for many other catalysts, such as HZSM-5, H-mordenite, HY, A1203, HGa-silicalite-2, and fumed SO?. These materials have much different properties and differ significantly from titanium silicates; thus, the results cast some doubt on the role of the catalyst in this reaction. Furthermore, the oxidation of styrene is reported to proceed with C=C cleavage and formation of benzaldehyde, in contrast to previous reports of the formation of phenylacetaldehyde with 85% selectivity (Neri et al., 1986). The reinvestigation of this reaction has shown that styrene epoxide is formed as the initial product, but it rapidly undergoes isomerization to phenylacetaldehyde. In the presence of solvent methanol, however, addition of the solvent to the epoxide produces 32.8% of 2-methoxy-2-phenylethan01,decreasing the phenylacetaldehyde yield. In all cases the reaction is accompanied by further oxidation to benzaldehyde (Kumar et al., 1995). The oxidation of butadiene, diallyl carbonate, or diallyl ether gives the products of monoepoxidation with selectivities of 85% and higher when the ratio diene/H202 is 2.5 (Table VIII). The diepoxides are formed in larger amounts when the diene/H202 ratio is 1. TS-1 does not catalyze oxidation reactions with TBHP or other hydroperoxides (Romano et al., 1990; Tatsumi et al., 1991), but other titanium silicates have been found to be active. Interesting results have been obtained with Ti-beta. In the oxidation of I-octene with TBHP catalyzed by Ti-beta, together with the epoxide substantial amounts of glycol, glycol ethers, octanal, and octan-2-one are formed, and even heptanal deriving from the oxidative cleavage of the double bond. However, when the acidity of Ti-beta is neutralized by exchange with alkali or alkaline-earth metal ions, the selectivity to epoxide is increased to 90-100%. Similar high epoxide selectivities have been reported for reactions of olefins and TBHP in the presence of Ti-beta when acetonitrile was

310

BRUNO NOTARI TABLE VIIl Oxidation of Polyunsaturated Compounds in ihe Presence of TS-1" H202 conversion

Yield based on H 2 0 2

Selectivity (%)

("/)

(%)

Monoepoxide

Diepoxide

Alkeneb

Solvent

T (K)

Butadiene Diallyl Carbonate Diallyl ether

t-BuOH CHiOH

293 338

98 95

85 50

85 93

15

CHiOH

338

96

60

90

4

a

5

Romano el a/. (1990b). * DieneM202 ratio = 2.5.

used as the solvent (Corma et af., 1995; Sat0 et al., 1995). Very likely, the weakly basic acetonitrile preferentially coordinates to the acid sites of the catalyst, thus preventing acid-catalyzed ring opening or rearrangements of the epoxide initially formed. Ti-MCM-41 is also an active catalyst with TBHP and olefins; norbomene gives the corresponding epoxides with 90% selectivity in the presence of Ti-MCM-41. However, when H 2 0 2 is the oxidant, conflicting results have been reported for the oxidation of 1-hexene (see Section M E ) . c. Phenols. Phenol is oxidized to give catechol and hydroquinone in almost equal yield and with high selectivity . The reaction is schematically represented as follows:

After the initial discovery of the catalytic activity of TS-1 for this reaction (Esposito et al., 1983,1985), other titanium containing zeolites such as TS-2 have been found to be effective catalysts; indeed, this reaction has become a standard for measurement of the activity and selectivity of different titaniumcontaining zeolites. Suitable solvents are H20-methanol mixtures or H20-acetone mixtures. The best results, 94% selectivity based on phenol and 84% based on H202at 30% conversion of the phenol, can be obtained only when all the reaction conditions are optimized. The temperature must be controlled within few degrees of 363 K, the addition of H202 must be gradual, and the catalyst must be a pure phase (Romano et al., 1990). The results are strongly influenced by the dimensions of the catalyst crystallites, the range 0.2-0.3 pm giving the best performance, as discussed above. In a number of papers it has been reported

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

311

that benzoquinones were formed as major products in the oxidation of phenol, especially at low conversions of H202, but that they were not present in the final reaction mixture when all the H202 had been consumed (Thangaraj et al., 1990, 1991a, 1991b; Reddy, J. S . et al., 1990, 1991d, 1992a). However, it has recently been shown that an error in the analytical method was responsible for these results (van der Pol et al., 1993a). The oxidation of phenol with H202 catalyzed by TS-1 obtained by secondary synthesis by reacting Tic& with preformed ZSM-5 has led to conflicting results. In one case, the catalytic activity was found to be identical to that of TS-1 (Kraushaar et al., 1988, 1989); in another, low activity and formation of tarry products were reported and ascribed to the residual acidity of the starting material and/or the presence of extra-framework Ti02 (Huybrechts et al., 1991b). The oxidation of phenol has also been investigated with Ti02 deposited on silicalite as the catalyst (Ferrini et al., 1990, Kooyman et al., 1992). The performances of TS-1 and TS-2 catalysts for this reaction were shown to be identical when the titanium contents and crystallite dimensions were equal (Tuel et al., 1993a). Because the oxidation of phenol is sensitive to the purity of the titanium silicate catalyst, it has been used as a test reaction to evaluate the purity of the catalytic materials. A standard material called EURO TS-1 has recently been prepared and evaluated in several laboratories (Martens et al., 1993). The high selectivity by which dihydroxybenzenes can be obtained from phenol and H202 when titanium silicates are used as catalysts has led to the development of a new process and the construction of a plant capable of producing 10,000 tons of catechol and hydroquinone per year. This plant, built by Enichem in Ravenna, Italy, includes a catalyst production and regeneration unit. Having started operations in 1986, this facility is working with excellent technical and economical results. The advantages of the new process relative to previous technologies include low H202 decomposition and high phenol conversions, with the formation of < 10% high-boiling by-products. These characteristics minimize the power requirements and make the process economics very competitive (Notari, 1988). Phenol derivatives have also been oxidized; anisole undergoes substitution both in the ortho (30%) and the para position (70%) (Romano et al., 1990). d. Aromatic Hydrocarbons, Alkanes, and Cycloalkanes. Aromatic hydrocarbons are oxidized by H202with titanium silicate catalysts (Esposito el al., 1983, 1985; Thangaraj et al., 1990). Saturated hydrocarbons can also be oxidized (Clerici et al., 1989, 1991;Huybrechts et al., 1990; Tatsumi et al., 1990b). The reactions take place readily at 323 K, but many investigations have been carried out at 353-373 K. The oxidation of these hydrocarbons is quite unusual, considering their chemical inertness. More surprisingly, methanol can be used as

3 12

BRUNO NOTARI

a solvent without undergoing oxidation itself. This uncommon result is probably the consequence of the hydrophobic nature of titanium silicates and the adsorption of the nonpolar hydrocarbons in preference to the more polar methanol. Benzene is oxidized to phenol with high selectivity at low conversions, but at higher conversions reaction of the phenol gives dihydroxybenzenes, as stated above. When silicates containing titanium and aluminum are exchanged with alkali metals and used under carefully selected conditions (40 wt% catalyst based on benzene, 293 K, 24 h reaction time), benzene can be oxidized to phenol at almost total conversion with a selectivity of 95% (Nemeth et af., 1 993). Little information is available characterizing the titanium-containing phases present in these catalysts, but from the preparation procedure it can be inferred that they likely contain highly dispersed Ti02 in the preformed zeolitic structure (Section II1.G). Toluene is oxidized to cresols (ortho : meta : para ratio = 5 : 1 : 4) and not to the benzyl derivatives, despite the low dissociation energy of the benzylic C- H bond. p-Xylene undergoes oxidation to 2,5-dimethylphenol. Ethyl benzene undergoes oxidation both in the aromatic ring and in the side chain, producing ethyl phenols (40%), acetophenone (56%) and 1 -phenyl ethanol (4%). 4-Methylethylbenzene undergoes oxidation in the aromatic ring, giving 2-methyl-5-ethyl phenol (3 IYo), and in the side chain, giving l-ethanol4-methylbenzene (29%) and 1-ethanone-4-methylbenzene (40%) (Romano et af., 1990b; Khouw et al., 1994). Cumene does not undergo oxidation at a measurable rate. I-Butylbenzene undergoes oxidation mainly in the side chain, with traces of aromatic ring oxidation, producing 1-phenyl-1-butanol, 1-phenyl-3-butanol, and the corresponding ketones (Clerici, 1991). Alkanes are oxidized to alcohols and ketones. Linear alkanes are oxidized to secondary alcohols and ketones, with good selectivity based on hydrocarbons and H 2 0 2 (Table IX). For C5 and higher alkanes, the oxidation at the 2-position is favored over that at the 3- and 4-positions. The efficiency of H 2 0 2 utilization is influenced by the purity of the titanium silicates; pure phases give efficiencies of 80% or more, whereas with materials containing TiO2, almost 30% of the H202 is lost by decomposition to O2 (Huybrechts et al., 1992). The reaction proceeds through the oxidation of the hydrocarbon to the alcohols, which are subsequently oxidized to ketones. This sequence is indicated by the fact that with increasing conversion. the selectivity for ketones increases at the expense of that for alcohols. The absence of isomerization products was confirmed by the formation of only 2-hexanone when 2-hexanol was the reactant; the equivalent statement pertains to the 3-compounds (Parton et al., 199I). Conflicting results have been reported for the products obtained in the oxidation of branched hydrocarbons. For 2-methylpentane, reaction at C2 with

313

MICROPOROUS CRYSTALLINE TITANIUM SILICATES TABLE IX Oxidation of n-Alkanes in 95% Methanol" Selectivity based on H202 Hydrocarbon

(%I

Propane n-Butane n-Pentane n-Hexane n-Heptane n-Octane n-Decane

35 69 82 86 75 63 56

Product distribution (mot%) 213 ratio'

4.5 2.6 1.9 2.6 1.1

2-01 66.2 55.0 34.3 32.1 33.7 30.1 11.5

3-01

16.1 25.9 29.2 20.5 20.5

4-01

%one

3-one

4-one

6.2 12.5 36.2

33.8 45.0 47.4 39.8 28.1 32.8 16.5

2.1 2.0 2.8 3.0 4.5

trace 1.o 10.8

"Clerici (1991a) and Huybrechts er a/. (1991a). *Represents the moles of oxygenated products obtained per 100 moles H202 reacted. 'Ratio between 2- and 3-compounds.

the formation of the tertiary alcohol as the major or exclusive reaction product has been reported (Parton et al., 1991; Clerici, 1991a); reaction at C4 with formation of 2-methylpentan-4-one has also been reported (Tatsumi et ai., 1990). In 3-methylpentane oxidation the preference for the tertiary carbon decreases in favor of oxidation at C2, as in the linear alkanes. As shown by other examples, the C-H bond strength is only one of the factors determining the course of the reaction; steric factors are also important. As was mentioned in Section V.C.3 .b, when competitive oxidation of I -octene and n-hexane is carried out, the alkene is preferentially oxidized. Correspondingly, alkenes react at lower temperatures than alkanes. It is therefore surprising that under noncompetitive reaction conditions, the rate of oxidation of n-hexane is higher than that of 1-octene (Huybrechts et al., 1992). One possible explanation for this observation is that the reaction conditions were different (Clerici et al., 1993b). At 373 K titanium peroxo compounds decompose, thereby giving rise to radical chain reactions that are negligible at lower temperatures. Thus there could be a different mechanism for low- and high-temperature oxidations made more complex by secondary uncatalyzed oxidation of initial products (Spinact et al., 1995). e. NHj and Nitrogen Compounds. Titanium silicates catalyze the oxidation of ammonia by H202; and if the reaction is camed out in the presence of a ketone, the corresponding oxime is formed. When the ketone is cyclohexanone, cyclohexanone oxime is formed (Roffia et al., 1987):

314

BRUNO NOTARI TABLE X Catalysis of Cyclohexanone Ammoximation’

Catalyst None Si02 (amorphous) Silicalite Ti02/SiO2 Ti02/Si02 TS- I

Cyclohexanone Reactant Catalyst Ti H2O~/cyclohexanoneConversion Oxime selectivity Oxime yield content (%) molar ratio (”/) (%) (YOon H202)

0 0 I .5 9.8 I .5

53.7 55.7 59.4 49.3 66.8 99.9

I .07 I .03 1.09 I .04 I .06 1.05

0.6 1.3 0.5 9.3 85.9 98.2

0.3 0.7 0.3 4.4 54.0 93.2

~~~

a

Roffia et al. (1990). Reaction at 353 K; reaction time, 1 h.

The performances of different titanium-containing catalysts are summarized in Table X. Many titanium-containing materials, even Ti02/Si02, are active for this reaction, but the oxime yields based on HzOz are always much lower than with TS-1, which reaches 93.2% at total conversion of cyclohexanone. Two sequences of reactions have been proposed to interpret the experimental facts: 1. The preliminary condensation of the ketone with ammonia, followed by the catalyzed oxidation of the imine to the oxime: \

C=O

+ NH3

,C=NH + H202

\ +

\

\ +

C=NH

+ H2O

C=NOH

+ H2O

(19) (20)

2. The catalytic oxidation of ammonia to NH20H, followed by the noncatalyzed condensation of NHlOH with the ketone: NH3 + H202 \

/C=O

+ NHzOH

--t

NHzOH \

+

+ H20

,,C=NOH

+ H20

(21) (22)

The first sequence was proposed on the basis of spectroscopic evidence and by-product formation (Thangaraj et al., 1991; Reddy, J. S. et al., 1991; Tvaruzkova et al., 1991, 1992). The fact that TS-1 is an efficient catalyst for the selective oxidation of ammonia to NH20H in the absence of ketones strongly favors the second pathway (Mantegazza et a)., 1991). Further evidence was

315

MICROPOROUS CRYSTALLINE TITANIUM SlLICATES

obtained from the observation that ketones such as cyclododecanone and tertbutyl cyclohexanone, which are too large to be adsorbed in the pore structure of TS- 1 , give high yields of the corresponding oximes, consistent with the hypothesis that the catalytic oxidation of ammonia to NH2OH takes place inside the pores, whereas the noncatalyzed condensation of N H 2 0 H and the ketone takes place in solution (Zecchina et al., 1992). When a sample of TS- 1 interacts with a stoichiometric amount of H 2 0 2 , a strong band appears at 25,000 cm-' in the UV-visible spectrum; upon addition of a stoichiometric amount of ammonia, the band shifts to 27,500 cm-' and slowly declines. This sequence has been interpreted as evidence of formation of a mixed complex in which both NH3 and the peroxo group are coordinated to the Ti'". This mixed complex could be the precursor of N H 2 0 H (Geobaldo et al., 1992). Primary amines are oxidized to the corresponding oximes. The sequence of reactions closely parallels the sequence observed with other mono oxygen donors, i.e., oxidation to alkyl hydroxylamines (V) followed by oxidation to alkylnitroso compounds (VI) which, via a prototropic shift, rearrange to the oximes (VII): \

,CH-NH~

)CH-NHOH V

)CH-NO VI

-

\

(23)

,C=NOH VII

The major difference with respect to other oxygen donors is the high selectivity to the oxime, for many (but not all) of the amines, and particularly the limited formation of other oxidation products such as nitro compounds, imines, and alkylnitroso dimers, which easily form in the presence of other oxygen donors by reaction of alkylnitroso compound (VI). In the absence of a catalyst, lower conversions and selectivity are observed (Table XI). TABLE XI Oxidation of Primav Amines Cata1,vzed by TS- I"

Amine

Solvent

CH3NH2 CH3NH2 n-C3H7NH2 i-C3H7NH2 i-C3H7NH2 i-C3H7NH2 CdiiNHz C6HiiNHz CGHsCHzNH2

CH3OH CH3OH CHIOH CHxOH I-BUOH~ I-BuOH CHIOH I-Bu-OH CH30H

Conversion

Oxime selectivity

Catalyst

(%)

(%)

(YO)

TS-I none TS-I TS-I TS-I TS-2 TS-I TS-I TS-I

40 3 32 38 29 31 3 3 20

88 0 73 I7 14 84 33 32 82

90 0 86 88 85 90 8 8 55

"Reddy, J. S. ef al. (1993b). *f-BuOH = tert-butyl alcohol.

HzOz efficiency

316

BRUNO NOTARl

The constrained environment in which electrophilic oxidation occurs has a strong influence on the exclusion of bulky product formation, but it is also important that the oxidation potential of titanium peroxo compounds is adequate, as demonstrated by the fact that the same reactions carried out in the presence of V-silicalite, which has the same pore structure, give rise to imine and nitro compounds as major reaction products (Reddy, J. S. et al., 1994). The oxidation of aniline with H202 gives rise to a number of products, some of which are accounted for by the reactivity in solution of the primary products. A possible reaction scheme is the sequential oxidation of aniline (VIII) to phenyl hydroxylamine (IX) to nitrosobenzene (X). The condensation of unreacted VIII with X results in the formation of azobenzene (XI), while reaction of IX with X produces azoxybenzene (XIII). Only when excess H202 is used does nitrobenzene (XII) form. Many titanium silicates are active for this reaction, Ti-beta and Ti-HMS being the most active. The results demonstrate that in catalysis by TS-I and by medium-pore zeolites, the reaction is limited by diffusion. TS-48, which has been found to be inactive in other oxidation reactions, is an active catalyst for the oxidation of aniline (Gontier et al., 1994; Sonawane et al., 1994). Tertiary amines are oxidized to the corresponding nitrogen oxides. Tosyl hydrazones of ketones and aldehydes and imines are oxidized to the corresponding carbonyl compounds. Reactions have been carried out with small molecules and also with molecules that would not diffuse into the pore structure of the titanium silicates. As in the case of C=C bond cleavage, it is possible that these reactions take place on the outer surface of the catalyst crystals. Toluene p-sulfonyl hydrazones (tosyl hydrazones) undergo C =N oxidative bond cleavage with H202 in the presence of TS-1, to give high yields of the corresponding carbonyl compounds; very likely, the reaction proceeds through

N

m SCHEME 2

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

317

the intermediate formation of oxaziridines followed by hrther oxidation to carbonyl compounds (Kumar et al., 1993). f. Su[fides. Sulfides undergo oxidation with silicates to form sulfoxides and sulfones: R \ /

s

-

R \

H202

R'

/

s=o

-

H202

R

R'

O \

H202

catalyzed by titanium

/

s4

(24)

\

R'

0

Typical results for different sulfides in acetone at reflux temperature are given in Table XII. Under similar conditions, diphenyl sulfide, C~HS-S-C~HS, was found to be unreactive, indicating that the reactions listed in Table XI1 take place inside the pore structure, which is not accessible to bulky molecules such as diphenyl sulfide. The selectivity to sulfoxides (Table XII) is the result of a competition between sulfides and sulfoxides for the catalytic site: dimethyl sulfide competes effectively, and the selectivity to sulfoxide is 97% with only 3% sulfone produced. The other reactant molecules give larger amounts of sulfones. However the reaction of dimethyl sulfide was camed out at 298 K, whereas the other sulfides reacted at the reflux temperature of acetone; the temperature difference may explain part of the differences shown in Table XII.

VI.

Catalytic Sites

The catalytic activity of titanium silicates must be ascribed Ti" sites, because pure crystalline silicas are totally inactive. As was discussed in Section 111, Ti" is present in the crystalline structure at random. Very likely, the random distribution that is obtained in the precursor reagents is maintained in the solid. Being dilute, each Ti" is expected to be surrounded by OSi" groups and isolated from other Ti" ions by long 0-Si-0-Si-0 sequences. It has been TABLE XI1 Oxidation Oj'SuiUlfideswilh H202 Catalyzed by

TS-2"

Selectivity (%) Conversion Reactant

(%)

Sulfoxide

'Reddy, R. S. et al. (1992). Reaction at 298 K

Sulfone

318

BRUNO NOTARI

proposed that Ti", in this state of dispersion and tetrahedral coordination, has properties different from those of other materials having Ti" sites with octahedral coordination and are not isolated from each other. It has been proposed that isolated Ti" sites have a low activity for H202decomposition (Notari, 1988). However, isolation alone is not sufficient to explain all the observed properties. The hydrophobic environment prevailing inside the catalyst pores and the multiple Ti-0- Si bonds that allow interaction with reactants but prevent complete hydrolysis of the Ti" sites in titanium silicates must also play a role in stabilizing the dispersed Ti" sites.

Vii.

Reaction Mechanism

Investigation of mechanisms of reactions catalyzed by titanium silicates has been limited to oxidation reactions with H 2 0 2 as the oxidant, as described below. As was previously discussed, elements different from titanium and silicon in the catalyst materials change their properties. Catalytic activity of doubly substituted materials such as Ti-beta, H[Al,Ti]-MFI and -MEL, and H[Fe,Ti]-MFI and -MEL is considered separately because the acidic properties associated with the added element affect the composition of the reaction products. Mechanistic information is difficult to obtain when the catalytically active titanium centers are present in a dilute matrix of silica. Only few techniques can be applied, and the available information does not allow discrimination between possible mechanisms. Consequently, it is necessary in this discussion to rely on analogies with the known chemistry of titanium compounds. Once the crystalline structure and the distribution of titanium therein are established, one of the first issues that must be considered concerns the interactions of the isolated titanium sites with water, H202, and other reactants. Adsorption of a number of compounds on the titanium sites leads to an increase in the coordination of titanium which, as previously noted, is reversible. By reaction with water, Ti-0-Si bonds are hydrolyzed, forming TiOH and SiOH groups. The hydrolysis process cannot be extensive, because Ti'" species separated from the matrix would undergo rapid sintering, as is observed for Ti'" dispersed on amorphous silicas. Therefore, in titanium silicates the Ti" must maintain a number of bonds to the crystalline lattice (Deo et al., 1993). To simplify the graphic presentation, hydrolyzed groups are indicated as TiOH and Ti(OH)2. This latter group can be considered equivalent to a titanyl group, Ti=O. Spectroscopic properties of the Ti(OH)2and Ti=O may be different, but their chemical properties are substantially identical; therefore, their notation will be used interchangeably. The scheme shown in Eq. (25) indicates the hydrolysis of the Ti-0-Si bond and further interactions with water.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

3 19

OF CRYSTALLINE TITANIUM SILICATES A. ACIDITY

In Section 11, the acidity of TiOz-SiOz was discussed and it was shown that no significant acidity is present in the Si02-rich region. This is very likely due to the fact that, in the presence of excess Si02, Ti" assumes a tetrahedral coordination with the consequence that no charge imbalance exists and no Brcansted acidity is created. The same is true for crystalline titanium silicates that contain small amounts of Ti" and, as demonstrated in Section 111, assume tetrahedral coordination. The chemical behavior is as expected; many acidsensitive compounds like the epoxides can be obtained in high yield without undergoing major hydrolysis or solvolysis. The fact that the limited solvolysis observed for propylene and ally1 alcohol can be reduced by silanization or by the addition of controlled amounts of bases is consistent with the hypothesis that the solvolysis is due to silanol groups and disappears when they are transformed or neutralized. Further evidence comes from the observation that if the same epoxidation is carried out with titanium silicates containing traces of trivalent elements, hydrolysis takes place (Section V). Furthermore, TS-1 is inactive for acid-catalyzed hydrocarbon reactions; but when A13 is present, TS- 1 becomes active for xylene isomerization (Reddy, J. S. er al., 1994). As for Ti02-Si02. the adsorption of water, ammonia, pyridine, and pyrrole indicates that Ti" in titanium silicates behaves as a Lewis acid; and the fact that these adsorbates are removed by simple evacuation leads to the conclusion that Ti" is a fairly weak Lewis acid (Bittar et al., 1992; Liu et al., 1994). The change in Ti" coordination that is observed as a result of adsorption has no effect as far as the development +

320

BRUNO NOTAM

of protonic acidity is concerned, since the molecules that are coordinated are neutral. To produce the charge imbalance in the solid that is at the origin of acidity, charged 0 2 -groups are necessary. Acidity in crystalline titanium silicates has been observed only when a titanium-containing zeolite interacts with H 2 0 2 , but this is due to the formation of peroxo compounds, as discussed below. B. TITANIUM PEROXO COMPLEX The addition of H202 to titanium silicates brings about the formation of a titanium peroxo complex, which, as indicated in Section I, has been proposed as the active species for the transfer of oxygen from the oxidant to the reactant. From the chemistry of Ti" compounds in the liquid phase, it is known that H202 acts as a bidentate ligand and displaces other ligands to form the very stable sidebonded (or q2-02)peroxo species XIV. Weakly bonded neutral molecules of the solvent S are also coordinated at the Ti" center. The alkyl hydroperoxides form peroxo complexes displacing only one ligand to give the end-on hydroperoxo complex XV. In some cases, the alkyl hydroperoxo complex may adopt the sidebonded structure XVI, displacing a weakly bonded neutral molecule of solvent:

The stability of the peroxo complexes formed with H202, measureL -y the association constant, is much higher than the stability of hydroperoxo complexes. The strong repulsion between formally unshared electrons in planar H202 can be reduced by transition-metal ions such as Ti", as they accept electron density from the filled antibonding orbitals of H202 interacting with the empty metal d orbitals of appropriate symmetry. It is for this reason that even hydroperoxo complexes may prefer the side-on configuration that can provide the added stability (Conte et al., 1992). The formation of a peroxo complex between H202 and a titanium silicates has been demonstrated in several ways, the most convincing being the appearance of an absorption band in the UV-visible spectra at 26,000 cm- when H202 is added to a titanium silicate. A band at the same frequency is present in the UV-visible spectra of the peroxo complex [TiFs(02)l3-, and the absorption has been attributed to a charge-transfer process 0:- -+ Ti4+ (Geobaldo et al., 1992). The stability of these complexes is limited to a temperature of 333 K they decompose rapidly at 373 K (Huybrechts et al., 1991). The thermal stability of the peroxo complex formed on TS-1 is markedly increased in the presence of

'

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

32 1

bases, the decomposition temperature being shifted to 523-673 K. Stable species are formed which have been characterized by both physical chemical methods and catalytic activity tests. Under carefully controlled conditions the ratio "active O"/Ti approaches unity, indicating that every Ti" site in the solid has been transformed to a peroxo complex. The increased stability of the titanium peroxo complex in the presence of bases could be the reason for the observed deactivation of these catalysts caused by alkalies (Clerici et al., 1993). In the [TiFS(O2)]'- ion, the peroxo group is bonded to Ti" side-on, and therefore this could also be the structure of the complex formed on titanium silicates. However, the possibility of a hydroperoxo species bonded end-on cannot be ruled out, because the side-on structure requires a deeper degree of hydrolysis to give the Ti(OHh group, whereas the hydroperoxo can form on a TiOH group, which is more easily obtainable in a material resistant to hydrolysis. The two forms can be represented as follows:

The titanium peroxo complex under neutral conditions oxidizes alkenes, giving the epoxides with no evidence of acid-catalyzed reactions. But when the oxidation reaction is complete and the epoxides are exposed to an excess of H 2 0 2 , a hydrolytic reaction of the epoxides is observed. The rate of this reaction is similar to the rate obtained when the same epoxides are exposed to 0.1 M formic acid. By contrast, silicalite-1 is inactive and unaffected by the presence of H202. The addition of NaOH reduces the hydrolytic activity of the system TS- 1 H 2 0 2 , indicating that an acidic species is responsible for the hydrolysis. The formation of protonic acidity by H202 has been ascribed to the interaction of the titanium peroxo complex with a donor hydroxyl moiety of a molecule such as H 2 0 coordinated on Ti", resulting in the formation of a cyclic structure. The stabilization provided by the cyclic structure would make the dissociation and the protonic acidity possible. When alcohols, and particularly methanol, are used as solvents, the coordinated OH group could be that of the alcohol (Bellussi et al., 1992):

+

The fact that the acidic properties and the hydrolysis reaction are absent during epoxidation must mean that other donor molecules like alkenes are more strongly coordinated at Ti" and prevent the formation of the complex responsible for generating protonic acidity. The formation of an acidic species could therefore also be explained as the result of the transformation of a

322

BRUNO NOTARI

hydroperoxo group into a side-bonded peroxo group, which becomes possible when more strongly coordinating molecules are absent:

This process would be stabilized by alkalies, and the overall process could be represented as follows [Eq. (28)]:

In the presence of excess alkalies these reactions could be inhibited by the formation of Si-0-Na groups (Khouw et al., 1995). C. MECHANISTIC PROPOSALS Many characteristics related to the particular structure of the material contribute to the final outcome of H202 oxidation reactions catalyzed by titanium silicates: The resistance of the titanium center to extensive hydrolysis. The thermal decomposition of the titanium peroxo compounds, with the consequence that different mechanisms can be operative at low and high temperatures. The selective adsorption of reactants, owing to the hydrophobic nature of titanium silicates. Reactant shape selectivity effects related to the dimensions of reactant molecules and catalyst pores, including restricted transition-state shapeselectivity effects as well as chemical and stereochemical selectivity. These effects were illustrated above. The proposals advanced to give a representation of the mechanism by which the 0 is transferred from the titanium peroxo complex to the reactant molecules can be classified as concerted mechanisms and radical mechanisms. Concerted mechanisms have been proposed on the basis of work carried out with soluble MeV', Wv’ and Ti” peroxo compounds. The experimental evidence is consistent with the hypothesis that these compounds act as oxidants in stoichiometric epoxidations and that the reactions involve electrophilic attack of the peroxo compound on the organic molecule or, what is equivalent, a nucleophilic attack of the organic molecule on the peroxidic oxygen, in a “butterfly” transition state. The reaction product is formed and, after desorption, the peroxo compound is regenerated by reaction of Ti” with H202;this accounts for the catalytic nature of the reaction (Amato et al., 1986). The same type of mechanism

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

323

has been proposed for the surface titanium peroxo complex formed on titanium silicates (Notari, 1988). In the case of alkene oxidation, the mechanism of oxygen transfer from the titanium peroxo complex to the alkene would be as follows:

A modification of the mechanism that involves the hydroperoxo titanium complex and one solvent molecule has been proposed that involves the formation of a stable cyclic titanium peroxo complex (Clerici et al., 1993). In this case, the two peroxo oxygens are not equivalent, and thus two intermediates would be possible:

+

\

/

,c=c,

The metallacycle mechanism can also be considered a concerted mechanism. It is analogous to the one proposed for metal peroxo complexes and is based on the assumed formation of a cyclic intermediate that includes the peroxo group, the reactant molecule, and the metal ion (Mimoun, 1982, 1987; Huybrechts et al., 1992). For alkene epoxidation, the sequence of events would be represented as follows [Eq. (31)]:

The radical mechanism has been proposed to explain the oxidation of saturated hydrocarbons. In the previous mechanisms, the electron density of the double bond or the aromatic ring is considered essential for the attack on the peroxidic oxygen. This condition is absent in saturated hydrocarbons, and considering their inertness, their oxidation probably requires a homolytic mechanism, proceeding through radical intermediates. By analogy with vanadium

324

BRUNO NOTARI

peroxo compounds, it has been proposed that the titanium peroxo complex gives rise to a radical by breaking of one Ti-0 bond. The radical thus formed abstracts a hydrogen atom from the saturated hydrocarbon, producing a radical that recombines with the OH group bonded to the titanium and forms an alcohol, regenerating the catalytic center (Mimoun, 1987; Huybrechts et al., 1990): /

-

'$ o. /

'0'

The radical mechanism has also been proposed as a general mechanism for oxidation of alkenes and aromatics, but several objections have been raised because of the absence of products typically associated with radical reactions. In classical radical reactions, alkenes should react also at the allylic position and give rise to allyl-substituted products, not exclusively epoxides; methyl-substituted aromatics should react at the benzylic position. The products expected from such reactions are absent. Another argument was made against the radical mechanism based on the stereoselectivity of epoxidation. Radical intermediates are free to rotate around the C-C bond, with the consequence that both cis- and trans-epoxides are formed from a single alkene isomer, contrary to the evidence obtained with titanium silicates (Clerici et al., 1993). Recent evidence seems to indicate, however, that radical reactions inside a zeolite do not necessarily follow the same pathways as classical radical reactions. The oxidation of the hydrocarbon ethylcyclopropane with H2O2catalyzed by TS-1, which should proceed through radical intermediates, gives only the alcohol (27%) and the ketone (73%) resulting from hydrogen abstraction at the carbon linked to the cyclopropane ring, and not the products that would be expected from the rearrangement of the radical intermediate:

Furthermore, isopropylcyclopropane gives the tertiary alcohol (50%), and unidentified products (50%), but not products with the rearranged structure. These results indicate that radicals formed on titanium and inside a zeolite can

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

325

be very short-lived, which prevents the rearrangements expected from classical radicals; alternatively, their movements may be restricted so that no rearrangement can occur (Khouw et d., 1994). From the evidence described above concerning the uncommon behavior of radical reactions in hydrocarbon oxidation at a titanium center inside a zeolite, the possibility that peroxo radicals are also involved in oxidation reactions of other compounds should be considered. In contrast to the mechanism discussed above for the hydrocarbons, in this case both the peroxo and the reactant molecule would be coordinated at the Ti" center, and the reaction would therefore take place between two coordinated species. The initial coordination of reactants has indeed been proposed to explain the selective oxidation of alkenes in the presence of saturated hydrocarbons. It was argued that, owing to the hydrophobic nature of titanium silicates, the concentration of both hydrocarbons inside the catalyst pores is relatively high and hence the alkenes must coordinate to Ti'". Consequently, the titanium peroxo complex will be formed almost exclusively on Ti'" centers that already have an alkene in their coordination sphere, and will therefore oxidize this alkene rather than an alkane which may be present in the catalyst (Huybrechts et al., 1992). Objections to this proposal are based on the fact that the intrinsically higher reactivity of alkenes with respect to saturated hydrocarbons is sufficient to account for the selectivity observed (Clerici et al., 1992). But coordination around the titanium center of an alcohol molecule, particularly methanol, is nevertheless proposed to explain the formation of acidic species, as was previously discussed. In summary, coordination around TitV could play a more important role than it does in solution chemistry as a consequence of the hydrophobicity of the environment where the reactions take place. For the oxidation of alcohols, two equally valid kinds of intermediates have been considered for the abstraction of the hydrogen atom that leads to the hydroperoxo and radical formation of aldehydes or ketones-namely, intermediates: H

H

In both cases, the preliminary coordination of the reactants is considered to explain the differences in rates of oxidation of 2- and 3-alcohols. The strainof-coordination bond angles of the alcohol in the transition state are affected by the length and size of the remaining groups and hence by the position of the OH group on the chain. It seems more difficult to account for this effect with an external nucleophilic attack from a noncoordinated molecule (Maspero et al., 1994).

326

BRUNO NOTAN

A radical mechanism involving species coordinated at the same Ti" center is schematically represented as follows for alkene epoxidation:

According to this hypothesis, the results are modified from what would be expected from classical radical reactions. The interest in this hypothesis is that, with the sole exception of saturated hydrocarbons, it could apply to all the compounds that can be coordinated at the Ti" center, such as alkenes, aromatics, alcohols, and sulfides. According to this hypothesis, the weak Lewis acidity of Ti" would help to bring the reactant into its coordination sphere. The initial coordination of the reactant would explain the oxidation of methylsubstituted aromatics in the aromatic ring and not in the side chain, even with a radical-type mechanism. A crucial point in this mechanism is the formation of the peroxo radical, which requires the reduction of the Ti4+ to Ti3+-a process that is easier with highly dispersed Ti" and should therefore be favored in titanium silicates in which the dispersion of Ti" is the highest ever observed. Radical reactions have been proposed to explain the mechanisms of many oxidations catalyzed by metals (Mimoun, 1987), but until now they were considered incompatible with the experimental evidence available for titanium silicate-catalyzed reactions. The recent results indicate that the rationalization of the observed facts with radical mechanisms is as plausible as that with other mechanisms. Indeed, for the oxidation of saturated hydrocarbons, the radical mechanisms are judged to account for the observations better than other mechanisms. Further investigations are needed to clarify these issues.

VIII.

Summary

The discovery of the new titanium silicates and of their catalytic properties in

H202 oxidation reactions has had a major impact in catalytic science and its industrial applications. One 10,000todyear plant for the production of catechol and hydroquinone has been operating since 1986 with excellent results. Moreover, successfir1 tests conducted on a 12,000-todyear pilot plant for cyclohexanone ammoximation (Notari, 1993b) could be followed soon by an industrial-size plant that would greatly simplify the synthesis of caprolactam. Both these examples are clear indications of the potentials of the new oxidation chemistry made possible by the new materials.

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

327

Because of stringent environmental constraints, it is difficult for many innovative chemistries to find industrial application. The new chemistry described here is applied successfully because it offers both ecological and economic advantages. Availability and cost of raw materials (especially H202), integration with other productions, value of by-products, price trend expected for products in the future, investments required, and risks involved in the new technology, all play a vital role in the evaluation. There seems to be no doubt that oxidations with H 2 0 2 or hydroperoxides could be advantageous in many other industrial processes of both basic and fine chemicals. The impact of this subject on the science of solid-state chemistry and catalysis has also been significant. Structures are now synthesized that were considered impossible just a few years ago. Oxidation reactions with zeolites were almost unknown before the discovery of titanium silicates, and the demonstration of the properties of these catalysts has stimulated the search for new materials with analogous catalytic properties. The development of this subject has been more successful than almost anyone would have predicted. With wide-pore titaniumcontaining crystalline silicas, with the possibility of eliminating the undesirable reactions due to the acidic properties of aluminum-containing titanium silicates, with the new Ti02-Si02 aerogel materials, with new materials incorporating titanium in microporous crystalline structures such as ALPO, and with materials incorporating other elements such as vanadium in a crystalline silica structure, it has been possible to carry out oxidations of even large molecules, using H 2 0 2 andor hydroperoxides, thus overcoming the most severe limitations of the smaller-pored TS-1 and TS-2 and expanding the range of oxidation reactions. Further developments are to be expected in this subject, driven by the combination of scientific and technological opportunities. REFERENCES Amato, G., Arcoria, A., Ballistreri, F. P.. Tomaselli, G.A., Bortolini, O., Conte, V., Di Furia, F., Modena. G., and Valle, G., J. Mol. Cafal.37, 165 (1986). Anpo. M., Nakaya, H., Kodarna, S., and Kubokawa, Y., J. Phys. Chem. 90, 1633 (1986). Armor, J. N., U.S. Pat. 4,163,756 (1979). Armor, J. N., J. Amer. Chem. SOC. 102, 1453 (1980). Armor, J. N., and Zambri, P. M., J. Catul. 73, 57 (1982) and references therein. Barrer, R. M., in Olson, D. and Bisio, A., Proc.6th Int. Conf. Zeolites, Reno, 1983, Butterworths Ltd., U.K. 870 (1984). Behrens, P., Assmann, S., Felsche, J., Vetter, S., Schultz-Ekloff, G., and Jaeger, N. I., Proc. 6th Int. Conf. X-ray Absorption Fine Structure, York, U.K., p. 552. Elsevier, Amsterdam, 1990. Behrens, P., Felshe, J., Vetter, S., Schultz-Ekloff, G., Jaeger, N. I., and Niemann, W., J. Chem. Soc.. Chem. Commun. 678 (1991a). Behrens, P., Felshe, J., and Niemann, Cafal. Today 8,479 (1991b). Bellussi, G., Clerici, M. G., Buonomo, F., Romano, U., Esposito, A,, and Notari, B., U.S. Pat. 4,701,428 (1987).

328

BRUNO NOTARI

Bellussi, G., Giusti, A,, Esposito. A., and Buonomo, F., Europ. Pat. 226.257 (l988a) Bellussi, G., Clerici, M. G., Giusti, A., and Buonomo, F., Europ. Pat. 226.258 (1988b). Bellussi, G., Clerici, M. G., Carati, A., and Esposito, A., Europ. Pat. 266.825 (1988~). Bellussi, G.. Carati, A. Clerici, M. G., Esposito, A., Millini, R., and Buonomo, F., Belg. Pat. 1.001.038 (1989). Bellussi, G., Carati, A., Clerici, M. G., and Esposito. A,. Stud. Surf: Sci. Curul. 63,421 (1991a). Bellussi, G., and Fattore, V., Stud. Surf: Sci. Cutul. 69, 79 (1991b). Bellussi, G., Carati, A., Clerici, M. G., Maddinelli, G., and Millini, R., J. Cutul. 133, 220 (1992). Bellussi, G., and Rigutto, M. S., Stud. Surf: Sci. Cutul. 85, I77 ( I 994). Biaglow, A. I., Gorte, R. J., Srinivasan, S., and Datye, A. K. Cutul. Lett. 13, 313 (1992). Bittar, A,, Sayari, A., Trong On, D., Bonneviot, L., and Kaliaguine, S., Proceedings 9th Int. Zeolite Conf., Montreal 1992, (R. von Ballmoos et ul., Eds.). Butterworth-Heinemann, 1993. Blandy, C., Gervais, D., Pellegatta, J., Gilot, B., and Guiraud, R.. J. Mol. Cutul. 64, L1 (1991). Blasco, T., Camblor, M. A,. Coma, A., and Perez-Pariente, J., J. Am. Chem. SOC. 115, 11806 (1993). Blasco, T., Coma, A,, Navarro, M. T., and Perez-Pariente, J., J. Curul. 156, 65 (1995). Boccuti, M. R., Rao. K. M., Zecchina, A., Leofanti, G., and Petrini, G., Stud. Surf: Sci. Curd. 48, 133 (1989). Bonneviot, L., Trong On, D., and Lopez, A., J. Chem. SOC.,Chem. Commun. 685 (1993). Bordiga, S., Boscherini, F., Coluccia, S., Genoni, F., Lamberti, C., Leofanti, G., Marchese, L., Petrini, G., Vlaic, G., and Zecchina, A., Cutul. Lett. 26, 195 (1994). Buonomo, F., Bellussi, G., and Notari, B., U.S.Pat. 4,578,161 (1989). Camblor, M. A., Coma, A., Martinez, A., and Perez-Pariente, J., J. Chem. Soc.. Chem. Commun. 589 (1992). Camblor. M. A,, Coma, A,, and Perez-Pariente, J., J. Chem. Soc.. Chem. Commun. 557 ( 1993a). Camblor. M. A,, Coma, A,, and Perez-Pariente, J., Zeolites, 13, 82 (1993b). Carati, A., Contarini, S., Millini, R., and Bellussi, Symp. on Synthesis and Properties of New Catalysts, Boston 1990, Mat. Res. SOC.Ext. Abstract (EA-24), 47 (1990). Chang, K.-H., Kim, G.-J., and Ahn, W.-S., fnd. Eng. Chem. Res. 31, 125 (1992). Chapman, D. M., and Roe, A. L., Zeolites 10, 730 (1990). Chapus, T., Tuel, A,, Ben Taarit, Y., and Naccache. C., Zeolites, 14,349 (1 994). Chen, C. Y., Li, H. X.,and Davis, M. E., Microporous Muter. 2, 17 (1993). Chen. M. C., Chu, S. J., Chang, N. S., Chen, P. Y., Chuang. T. K., and Chen, L. Y . , Stud. Surf Sci. Cutul. 38, 253 (1988). Clerici, M. G., and Romano, U., Europ. Pat. 230.949 (1987). Clerici, M. G., and Bellussi, G., It. Pat. Appl. 21492 A (1989a). Clerici, M. G., and Bellussi, G., Europ. Pat. 315.247 (1989b). Clerici, M. G., and Bellussi, G., Europ. Pat. 315.248 (1989~). Clerici, M. G., Appl. Cutul. 68, 249 (1991a). Clerici, M. G., Bellussi, G. and Romano, U., J. Cutal. 129, 159 (1991b). Clerici, M. G., Ingallina, P., and Millini, R., Proc. 9th Int. Zeol. Conf. Montreal, 1992, (R. von Ballmoos et ul., Eds.), p. 445. Butteworth-Heinemann, 1993a. Clerici, M. G., and Ingallina, P., J. Cutal. 140, 71 (1993b). Conte, V., Di Fwia, F., and Modena, G., in “Organic Peroxides” (W. Ando, Ed.), p.559. Wiley, New York, 1992. Coma, A., Navarro, M. T., and Perez-Pariente, J., J. Chem. Soc.. Chem. Commun. 147 (1994a). Coma, A., Camblor, M. A., Esteve, P., Martinez, A., and Perez-Pariente, J., J. Cutul. 145, 151 ( 1994b).

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

329

Corma, A., Navarro, M. T., Perez-Pariente, J., and Sanchez, F., Stud. SurJ Sci. Caral. 84, 69 ( I 994c). Corma, A., Esteve, P., Martinez, A., and Valencia, S., J. Caral. 152, 18 (1995). Dam, C. B., Khouw, C. B., Li, H. X., and Davis, M. E., 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 491 (1993). Dam, C. B., Khouw, C. B., Li, H. X.,and Davis, M. E., Microporous Mater. 2,425 (1994). De Castro-Martins, S., Tuel, A., and Ben Taarit, Y., Zeolites 14, 130 (1994). Deeba, M.. Keweshan, C. F., Koermer, G. S., Kuznicki, S. M., and Madon, R. S., in “Catalysis of Organic Reactions” (J. R. Kosak and T. A. Johnson, Eds.), p. 383. Dekker, New York, 1994. Deo, G., Turek, A. M., Wachs, I. E., Huybrechts, D. R. C., and Jacobs, P. A,, Zeolites. 13, 365, (1993).

Doolin, P. K., Alerasol, S., Zalewski, D. J., and Hoffmann, J. F., Catal. Lett. 25, 209 (1994). de Ruiter, R., Jansen, J. C., and van Bekkum, H., in “Molecular Sieves” (M. L. Occelli and H. A. Robson, Eds.), p. 167. van Nostrand-Reinhold, Princeton. N.J., 1993. Dreoni, D. P., Pinelli, D., Trifiro. F., Tvaruzkova, Z., Habersberger, H., and Jiru, P., Catal. Lert. 11, 285 (1991).

El Hage-Al Asswad, J., Nagy, J. B., Gabelica, Z., and Derouane, E. G., “8th Int. Zeolite Conf. Amsterdam, July, 1989,” Recent Research Reports, p. 475. El Hage-A1 Asswad, J., Ph.D. Thesis, Notre Dame de la Paix. Namur, Belgium, 1990. Esposito, A,, Taramasso, M., and Neri, C., U S . Pat. 4,396,783 (1983). Esposito. A., Neri, C., and Buonomo, F., US. Pat. 4,480,135 (1984). Esposito, A,, Taramasso, M., Neri, C., and Buonomo, Brit. Pat. 2. I 16.974 (1985). Esposito, A., Maspero, F., and Romano, U., Proc. DGMK Conf. “Selective Oxidations in Petrochemistry,” Sept. 1992, p. 195. Goslar, Germany, 1992. Fernandez, A., Leyrer, J., Gonzalez-Felipe, A. R., Munuera, G.. and Knotzinger, H., J. Catal. 112, 489 (1988).

Ferrini, C., and Kouvenhoven, H. W., Stud. Surf: Sci. Cutal. 55, 53 ( I 990). Ferrini, C., and Kouvenhoven, H. W., Proc. DGMK Conf. “Selective Oxidations in Petrochemistry,” Sept. 1992, p. 205. Goslar, Germany, 1992. Flanigen, E. M., Bennett, J. M., Grose, R.W., Cohen, J. P., Patton, R. L., Kirchner, R. M., and Smith. J. V., Nature, 271, 512 (1978). Geobaldo, F., Bordiga, S., Zecchina, A,, Giamello, E., Leofanti, G., and Petrini, G., Catal. Lett. 16, 109 ( 1992).

Gontier, S . , and Tuel, A., Appl. Carol. 118, 173 (1994). Gontier, S., and Tuel, A., J. Catal. 157, 124 (1995). Guth, J. L., Kessler, H., Higel, J. M., Lamblin, J. M., Patarin, J., Seive, A,, Chezeau, J. M., and Wey, R., in “Zeolite Synthesis” (M. L. Occelli and M. E. Robson, Eds.), ACS Symp. Ser. 398, 176 (1 989).

Hari Prasad Rao, P. R., Thangaraj. A., and Ramaswamy, A. V., J. Chem. Soc., Chem. Commun. 1139 (1991).

Hari Prasad Rao, P. R., and Ramaswamy, A. V., J. Chem. Soc., Chem. Commun. 1245 (1992). Hari Prasad Rao, P. R., Belhekar, A. A,, Hedge, S. G., Ramaswamy, A. V., and Ratnasamy, P.. J. Catal. 141. 595 (1993a). Hari Prasad Rao, P. R., Ramaswamy, A. V., and Ratnasamy, P., J. Catal. 141, 604 (1993b). Hartman, P., Mineralogical Mag., Vol. 37, N. 287, 366, Sept. 1969. Holderich, W. F., Hesse, M., and Naumann, F., Angew. Chem. Int. Ed. Engl. 27, 226 (1988). Holderich, W. F., Stud. SurJ Sci. Catal. 49, 69 (1989). Hutchings, G. J., and Lee, D. F., J. Chem. Soc.. Chem. Commun. 1095 (1994). Hutter, R., Dutoit, D. C. M., Mallat, T., Schneider, M., and Baiker, A,, J. Chem. Soc., Chem. Commun. 163 (1995a).

330

BRUNO NOTARI

Hutter, R., Mallat, T., and Baiker, A., J. Cutul. 153, 165 (1995b). Hutter, R., Mallat, T., and Baiker, A., J. Cutal. 153, 177 (1995~). Hutter, R., Mallat, T., and Baiker, A,, J. Catal. 157, 665 (1995d). Huybrechts, D. R. C., De Bruyker, L., and Jacobs, P. A,, Nature 345, 240 ( 1 990). Huybrechts, D. R. C., Parton, R. F., and Jacobs, P. A., Stud. SurJ Sci. Catul. 65,225 (1991a). Huybrechts, D. R. C., Vaesen, I., Li, H. X.,and Jacobs, P. A. Cutal. Lett. 8,237 (1991b). Huybrechts, D. R. C., Buskens, P. L., and Jacobs, P. A., J. Mol. Cutul. 71, 129 (1992). lone, K. G., Vostrikova, L. A., and Mastikhin, V. M., J. Mol. Cutal. 31, 355 (1985). Jacobs, P. A,, Proc. DGMK Conf. in “Selective Oxidations in Petrochemistry,” Sept 1992, p. 171. Goslar, Germany, 1992. Jehng, J. M., and Wachs, I. E., Catal. Lett. 13, 9 (1992). Jentys, A., and Catlow, C. R. A., Curd. Lett. 22,251 (1993). Khouw, C. B.. Li, H. X..and Davis, M. E., ACS Meeting Washington, 1992, Div. Petr. Chem. Preprints, p. 1 136 (1992). Khouw, C. B., Li, H. X., and Davis, M. E., ACS Meeting San Francisco 1992 Preprints. Khouw, C. B., Dam, C. B., Li, H. X., and Davis, M. E.. in “Catalytic Selective Oxidations” (S. T. Oyama and J. W. Hightower, Eds.), ACS Symposium Series N. 523, p. 273 (1993a). Khouw, C. B., Dam, C. B., Li, H. X.. and Davis, M. E., 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 769 (1993b). Khouw, C. B., Dam, C. B., Labinger, J. A,, and Davis, M. E., J. Catal. 149, 195 (1994). Khouw, C. B., and Davis, M. E.. J. Cutul. 151, 77 ( I 995). Kim, G. J., and Cho, B. R. Catul. Lett. 22, 259 (1993). Kooyman, P. J., van der Waal, P., Verdaasdonk, P. A. J., Jansen, K. C., and van Bekkum, H., Curd. Lett. 13, 229 (1992). Kornatowski. J., Sychev, M., Baur, W. H., and Rozwadowski, M., 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 486 (1993). Kraushaar, B., and van Hooff, J. H .C., Cufal. Lett. I, 8 I (1 988). Kraushaar, B., and van Hooff, J. H. C., Catul. Lett. 2, 43 (1989). Kraushaar, B., Ph.D. Thesis, Tech. Univ. Eindhoven, The Nederlands, 1989. Kresge, C. T.. Leonovicz, M. E., Roth, W. J., and Vartuli, J. C., U.S. Pat. 5,098,684 (1992a). Kresge, C. T., Leonovicz, M. E., Roth, W. J., Vartuli, J. C., and Beck, J. S., Nature 359, 710 (1992b). Kumar, P., Hegde, V. R., Pandey, B., and Ravindranathan, T., J. Chem. Soc.. Chem. Commun. 1553 (1993). Kumar, R., Pais, G . C. G., Pandey, B., and Kumar, P., J. Chem. SOC. Chem. Commun. 1315 ( 1995). Kumar, S. B., Mirajkar, S. P., Pais, G . C. G.. Pandey, B.. Kumar, P., and Kumar, R.. J. Curd 156, 163 (1995). Kuznicki. S. M., U S . Pat. 4,853,202 (1989). Kuznicki, S. M., US. Pat. 4,938,939 (1990). Kuznicki. S. M., Thrush, K. A,, Allen, F. M., Levine, S. M., Hamil, M. M., and Hayhurst, D. T., ACS Meeting, New York (1991a). Kuznicki, S. M. and Trush, K. A. U S . Pat. 4,994,191 (1991b). Kuznicki, S. M., US. Pat. 5,Ol 1,591 (1991~). Kuznicki, S. M., Thrush, K. A,, Allen, F. M., Levine, S. M., Hamil, M. M., Hayhurst, D. T., and Mansour, M., in “Molecular Sieves” (M. L. Occelli and H. E. Robson, Eds.), p. 427. Van Nostrand-Reinhold, Princeton, N.J., 1993. Leofanti, G., Genoni. F., Padovan, M., Petrini, G., Trezza, G., and Zecchina, A,. Stud. SurJ Sci. Catul. 62, 553 (1991). Liu, 2.. Tabora, J., and Davis, R. J.. J. Cutul. 149, 117 (1994a).

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

33 1

Liu, Z., and Davis, R. J., J. Phys. Chem. 98, 1253 (1994b). Liu. Z., Crumbaugh, G.M., and Davis, R. J., J. Catul. (1996), in press. Lok, B. M. T., Marcus, B. K., and Flanigen, E. M., International Appl. WO 85/04856 (07.1 1.85). Lopez, A,, Tuiler, M. H., Kessler, H., Guth, J. L., and Popa, J. M., in “X-ray Absorption Fine Structure” (S. S. Hasnain, Ed.), p. 549. Ellis Herwood, New York, 1991. Lopez, A,, Tuiler. M. H., Kessler, H., Guth, J. L., Delmotte, L., and Popa, J. M., J. Solid State Chem. 102,480 (1993). Mantegazza, M. A,, Padovan, M., Petrini, G., and Roffia, P., It. Pat. Appl. A00 1915 (1 99 I). Martens, J. A., Buskens, P., Jacobs, P. A., van der Pol, A. J. H. P., van Hooff, J. H. C., Ferrini, C., Kouvenhoven, H. W., Kooyman, P. J., and van Bekkum, Appl. Cutul. A 99, 71 (1993). Maschmeyer, T., Rey, F., Sankar, G., and Thomas, J. M., Nature 378, 159 (1995). Maspero, F., and Romano, U.,Europ. Pat. 190.609 (1986). Maspero, F., and Romano, U., It. Pat. 22482 A/89 (1989). Maspero, F., Chim. Ind. 75, 291 (1993). Maspero, F.. and Romano, U., J. Cutal. 146, 476 (1994). Miller, J. B., Johnson, S. T., and KO, E. I., J. Cutal. 150, 31 1 (1994). Millini, R., Previde-Massara, E., Perego, G., and Bellussi, G., J. Catal. 137, 497 (1992). Mimoun, H., Angew. Chem., Int. Ed. Engl. 21, 734 (1982). Mimoun, H., Catal. Toduy 1,281 (1987). Mirajkar, S. P., Thangaraj, A., and Shiralkar, V. P., J. Phys. Chem. 96, 3073 (1992). Muiioz-Paez, A., and Munuera, G., Stud. Surf: Sci. Cutal. 63, 627 (1991). Munuera, G., Gonzalez-Elipe, A. R., Femandez, A,, Malet, P., and Espinos, J. P., J. Chem. SOC., Faraday Trans. 85(6), 1279 ( 1989). Nemeth, L. T., Hyatt, E. M., and Malloy, T. P., U S . Pat. 5,233,097 (1993). Neri. C., Buonomo, F., and Anfossi, B., Europ. Pat. 100.118 (1984a). Neri, C., Anfossi, B., Esposito, A,, and Buonomo, F., Europ. Pat. 100.119 (1984b). Neri, C.. and Buonomo, F., Europ. Pat. 102,097 (1986). Neumann, R., Chave, M., and Levin, M., J. Chem. Soc., Chem. Commun. 1685 (1993). Notari, B., Stud. SurJ Sci. Cutul. 37, 413 (1988). Notari, B., Stud. Sut$ Sci. Catul. 60, 343 (1991a). Notari, B., Stud. Surf: Sci. Catul. 61,243 (199lb). Notari, B., Cutul. Toduy 18, 163 (1993a). Notari, B., 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 761 (1993b). Padovan, M., Leofanti, G., and Roffia, P., Europ. Pat. 3 I 1.983 (1989). Padovan, M., Genoni, F., Leofanti, G., Petrini, G., Trezza, G., and Zecchina, A., Stud. SurJ Sci. Cutal. 63,43 1 ( I99 I). Parton, R. F., Huybrechts, D. R. C., Buskens, Ph., and Jacobs, P. A,, Stud. Surf: Sci. Cutul. 65, 47 (1991). Pei, S., Zajac, G. W., Kaduk, J. A., Faber, J., Boyanov, B. I., Duck, D., Fazzini, D., Morrison, T. I., and Yang, D. S., Curd. Lett. 21, 333 (1993). Perego, G., Bellussi, G.,Como, C., Taramasso, M., Buonomo, F., and Esposito, A,, Stud. SurJ Sci. Catal. 28, 129 (1986). Petrini, G.. Cesana, A,, De Alberti. G.,Genoni, F., Leofanti, G., Padovan, M., Paparatto, G., and Roffia, P., Stud. Surf: Sci. Cutal. 68, 761 (1991). Popa, J. M., Guth, J. L., and Kessler, H., Europ. Pat. 292,363 (1988). Ramaswamy, A. V., Sivasanker, S., and Ratnasamy, P., Proc. 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 765 (1993). Reddy, J. S., Kumar, R., and Ratnasamy, P., Appl. Catul. 58, L1-L4 (1990). Reddy, J. S., and Sivasanker, S., Cutul. Lett. 11, 241 (1991a). Reddy, J. S., Sivasanker, S., and Ratnasamy, P., J. Mol. Cutul. 69, 383 (1991b).

332

BRUNO NOTARl

Reddy, J. S., Sivasanker, S., and Ratnasamy, P., J. Mol. Curd 70, 335 (1991~). Reddy, J. S., and Kumar, R., J. Curd. 130, 440 (1991d). Reddy, J. S., Sivasanker, S., and Ratnasamy, P., J. Mol. Curd 71, 373 (1992a). Reddy, J. S., and Kumar, R., Zeolites 12, 95 (1992b). Reddy, J. S., Khire, U. R., Ratnasamy, P., and Rajat Mitra, B., J. Chem. SOC.,Chem. Commun. 1234 (1992~). Reddy, J. S., Ravishankar, R., Sivasanker, S., and Ratnasamy, P., Curd. Lett. 17, 139 (1993a). Reddy, J. S., and Jacobs, P., J. Chem. Soc., Perkin Trans. 122,2665 (1993b). Reddy, J. S., Kumar, R., and Csicsery, S., J. Card. 145, 73 (1994a). Reddy, J. S., and Sayari, A., Curd Len. 28, 263 (1994b). Reddy, J. S., and Sayari, A,, J. Chem. SOC.,Chem. Commun. 23 (1995). Reddy, K. M., Kaliaguine, S., and Sayari, A., Curd. Lerr. 23, 169 (1994a). Reddy, K. M., Kaliaguine. S., Sayari, A., Ramaswamy, A. V.. Reddy, V. S., and Bonneviot, L., Curd Lett. 23, 175 (1994b). Reddy, R. S., Reddy, J. S., Kumar, R., and Kumar, P. J., Chem. Soc.. Chem. Commun. 84 (1992). Roffia, P., Padovan, M., Moretti, E., and De Alberti, G., Europ. Pat. 208.31 1 (1987). Roffia, P., Padovan, M.,Leofanti, G., Mantegazza, M., De Alberti, G., and Tauszik, R. G., Europ. Pat. 267.362 (1988). Roffia, P., Paparatto, G., Cesana, A., and Tauszik, G., Europ. Pat. 301,486 (1989). Roffia, P., Leofanti, G., Cesana, A., Mantegazza, M.,Padovan, M., Petrini, G., Tonti, S., and Gervasutti, P., Stud, SurJ Sci. Cutul. 55, 43 (1990a). Roffia, P., Mantegazza, M., Cesana, A., Padovan, M., Petrini. G., Tonti, S., Gervasutti, V., and Varagnolo, R., Chim. Ind. 72, 598 (1990b). Romano, U., Esposito, A,, Maspero, F., Neri, C., and Clerici, M. G., Stud. SurJ Sci. Cutul. 55, 33 (1 990a). Romano, U., Esposito, A., Maspero, F., Neri, C., and Clerici, M. G., Chim. Ind. 72, 610 (1990b). Sankar, G., Rey, F., Thomas, J. M., Greaves, G. N., Corma, A,, Dobson, B. R., and Dent, A. J., J. Chem. SOC.Chem. Commun. 2279 (1994). Sato, H., Hirose, K., Kitamura, M., and Nakamura, Y .. Stud. S u r - Sci. Cutul. 49, 1213 (1 989). Sato, T., Dakka, J., and Sheldon, R. A., J. Chem. Soc., Chem. Commun. 1887 (1994). Scarano, D., Zecchina, A., Bordiga, S., Geobaldo, F., Spoto, G . , Petrini, G., Leofanti, G., Padovan, M., and Tozzola, G., J. Chem. SOC., Furuduy Trans. 89(22), 4123 (1993). Schraml-Marth, M., Walther, K. L., Wokaun, A., Handy, B. E., and Baiker, A., J. Non-Crysr. Solids 143,93 (1992). Schultz, E., Ferrini, C., and Prins, R., Curul. Lett. 14, 221, (1992). Seebach, D., Hugerbuehler, E., Naef, R., Schnurrenberger, P., Weidman, B., and Zueger, M., Synthesis 138 ( I 982). Seiyama, T., “Metal Oxides and Their Catalytic Actions,” Kodansha, Tokyo, 1978. Semno, D. P., Li, H. X.,and Davis, M. E. J. Chem. Soc., Chem. Commun. 745 (1992). Sheldon, R. A., van Dom, J. A., Schram, W. A. and De Jong, A. J., J. Cutul. 31,438 (1973). Sheldon, R. A., J. Mol. Cural. 7, 107 (1980). Sheldon, R. A., Srud. SurJ Sci. Cutul. 55, 1 (1990). Sheldon, R. A., and Dakka, J., Proc. DGMK Conf. “Selective Oxidations in Petrochemistry,” Sept. 1992, p, 215. Goslar, Germany, 1992. Shilun, Q.,Wenqin, P., and Shangqing, Y., Srud. SurJ Sci. Curd. 49A, 133 (1989). Shirmann, J. P., and Delavarenne, S. Y., “Hydrogen Peroxide in Organic Chemistry,” Ed. Documentation Industrielle, Paris, 1980. Skeels, G. W., and Flanigen, E. M., in “Zeolite Synthesis” (M. L. Occelli and H.E. Robson, Eds.), ACS Symposium Series N. 398, p. 420 (1989) Skeels, G. W., 206 ACS Meeting Chicago 1993 Div. Petr. Chem. Preprints, p. 484 (1993).

MICROPOROUS CRYSTALLINE TITANIUM SILICATES

333

Sohn, J. R., and Jang, H. J., J. Catal. 132, 563 (1991). Sonawane, H. R., Pol, A. V.. Moghe, P. P., Biswas, S. S., and Sudalai, A,, J. Chem. Sac.. Chem. Commun. I2 I5 ( I 994). Spinace, E. V., Pastore, H. 0.. and Schurchardt, U., J. Caral. 157, 631 (1995). Srinivasan, S., Datye, A. K., Hampden-Smith, M., Wachs, I. E., Deo, G., Jehng, J. M., Turek. A. M.. and Peden, C. H .F., J. Cutul. 131, 260 (1991). Sulikowski, B., and Klinowski, J., Appl. Catal. A 84, 141 (1992). Tanabe, K., in “Catalysis Science and Technology” (J. R. Anderson and M. Boudart, Eds.), Vol. 2, p. 23 I . Springer-Verlag, New York/Berlin, I98 I . Tanev, P. T., Chibwe, M., and Pinnavaia, T. J., Narure 368, 321 (1994). Taramasso, M., Perego, G., and Notari, B., Proc. 5th Int. Conf. Zeolites, Naples, 1980 (L. V. C. Rees, Ed.), P. 40. Heyden & Sons, London, 1980. Taramasso, M., Perego, G., and Notari, B., U.S. Pat. 4,410,501 (1983). Taramasso. M., Manara, G.. Fattore, V., and Notari, B., U.S. Pat. 4,666,692 (1987). Tatsumi, T., Nakamura, M., Yuasa, K.. and Tominaga, H., Chem. Len. 297 (1990a). Tatsumi, T., Nakamura, M., Negishi, S., and Tominaga, H., J. Chem. Sac.. Chem. Commun. 476 ( I 990b). Tatsumi, T., Nakamura, M., Yuasa, K., and Tominaga, H., Cutal. Len. 10, 259 (1991). Tatsumi, T., Yuasa, K., and Tominaga, H., J. Chem. Soc., Chem. Commun. 1446 (1992). Tatsumi, T., Yako, M., Nakamura, M., Yuhara, Y., and Tominaga, H., J. Mol. Curd. 78, L41 ( I 993). Thangaraj, A., Kumar, R., and Ratnasamy, P., Appl. Card. 57, LI-L3 (1990). Thangaraj, A,, Kumar, R., Mirajkar, S. P., and Ratnasamy, P., J. Carul. 130, 1 (1991a). Thangaraj, A,, Kumar, R., and Ratnasamy, P., J. Curd. 131, 294 (1991b). Thangaraj, A,, Sivasanker, S., and Ratnasamy, P., J. Catul. 131, 394 (1991~). Thangaraj, A,, Kumar, R., and Ratnasamy, P., J. Carul. 137, 252 (l992a). Thangaraj, A,, Eapen, M. J., Sivasanker, S., and Ratnasamy, P., Zealires 12, 943 (1992b). Thomas, C., Ind. Eng. Chem. 41, N. I I , 2564 (1949). Thomas, J. M., and Greaves, G. N., Science 265, 1675 (1994). Tielen, M., Geleen, M.. and Jacobs, P. A,, Proceed. Int. Symp. Zeolite Catalysis, Siofolk, 1985, Acta Phys. Chem. Szegedensis, p. I (1986). Toba. M., Mizukami, F., Niwa, S., Sano, T., Maeda, K., Annila, A., and Komppa, V.. J. Mol. Curul. 91,277 (1994). Tonti. S., Roffia, P., Cesana, A., Mantegazza, M., and Padovan, M., Europ. Pat. 314.147 (1989). Trong-On, D., Bittar, A., Sayari, A., Kaliaguine, S., and Bonneviot, L., Curd. Lert. 16, 85 (l992a). Trong-On, D., Bonneviot, L., Bittar, A,, Sayari, A,, and Kaliaguine, S., J. Mol. Cutul. 74, 233 (1 9926). Tuel, A. Diab, J., Gelin, P., Dufaux, M., Dutel, J. F., and Ben Taarit, Y., J. Mol. Catal. 63, 95 (1 990). Tuel, A. Moussa-Khouzami, S., Ben Taarit, Y., and Naccache, C., J. Mol. Curd. 68, 45 (1991). Tuel, A,, and Ben Taarit. Y., J. Chem. Soc., Chem. Commun. 1578 (1992). Tuel, A., and Ben Taarit, Y., Appl. Carol. A 102, 69 (1993a). Tuel, A., and Ben Taarit, Y., Appl. Caral. A 102, 201 (1993b). Tuel, A,, and Ben Taarit, Y., Microporous Materials 1, 179 (1993~). Tuel, A., and Ben Taarit, Y., Zeolites 13, 357 (1993d). Tuel, A,, Ben Taarit, Y., and Naccache, C., Zeolites 13,454 (1993e). Tuel, A., and Ben Taarit, Y., Appl. Catal. A 110, 137 (1994a). Tuel, A,, and Ben Taarit, Y., Zeolires 14, 594 (1994b). Tuel, A,, and Ben Taarit, Y., J. Chem. Sac., Chem. Commun. 1667 (1994~).

334

BRUNO NOTARI

Tvaruzkova, Z., Habersberger, K., Zilkova, N., and Jiru, P., Appl. Catal. A 79, 105 (1991). Tvaruzkova, Z., Habersberger, K., Zilkova, N., and Jim, P., Catal. Lett. 13, 117 (1992). Uguina, M. A., Ovejero, G., Van Grieken, R., Serrano, D., and Carnacho, M., J. Chem. Soc., Chem. Commun. 27 (1994). Ulagappan, N., and Krishnasarny, V., J. Chem. Soc., Chem. Commun.373 (1 995). Valtchev. V., and Minton, S., Zeolites 14, 697 (1994). van der Pol, A. J. H. P., and van Hooff, J. H. C., Appl. Catal. A 92, 93 (1992a). van der Pol, A. J. H. P., and van Hooff, J. H. C., Appl. Catal. A 92, I 13 (1992b). van der Pol, A. J. H. P., Verduyn, A. J., and van Hooff, J. H. C., Appl. Catal. A 96, L13 (1993a). van der Pol, A. J. H. P., and van Hooff, J. H. C. Appl. Catal. A 106, 97 (1993b). Vednne, J. C., Dufaux, M., Naccache, C., and Imelik, B., J. Chem. Soc., Faraday Trans. 74, 440 ( I 978). Wulff, H., US.Pat. 3,642,833; 3,923,843; 4,021,454; 4,367,342; Brit. Pat. 1,249,079 (1971). Young, D. A,, U.S. Pat. 3,329,481 (1967). Zecchina, A,, Spoto, G., Bordiga, S., Padovan, M., Leofanti, G., and Petrini, G., Stud. Surf Sci. Catal. 65, 67 1 (199 I a). Zecchina, A., Spoto, G., Bordiga, S.. Ferrero, A., Petrini, G., Leofanti, G., and Padovan, M., Stud. SurJ Sci. Catal. 69, 25 1 (1991b). Zecchina, A., Spoto, G., Bordiga, S., Geobaldo, F., Petrini, G., Leofanti, G., Padovan, M., Mantegazza. M., and Roffia, P., in Proc. 10th Int .Congr. Catal., Budapest 1992 (Guczi er al., Eds.), p. 719. Akademiai Kiado, Budapest, 1993. Zhang, W., and Pinnavaia, T. J., Catal. Lett. (1996). in press.

ADVANCES IN CATALYSIS, VOLUME 41

Structural and Mechanistic Aspects of the Dehydration of Isomeric Butyl Alcohols over Porous Aluminosilicate Acid Catalysts KIRILL ILYCH ZAMARAEV Boreskov Instiiuie of Caialysis Russian Academy of Sciences Novosibirsk 630090. Russia

AND

JOHN MEURIG THOMAS Davy Faraday Research Laboraioiy Royal Insiiiuiion of Greai Britain 2I Albemarle Sireei, London WIX 4BS. United Kingdom

1.

Introduction

Seldom in the study of heterogeneous catalysis does it prove possible to ( 1 ) specify precisely the concentration and nature of the active sites, (2) test whether these sites are of comparable strength and are distributed in a spatially and chemically well-defined manner, and (3) explore the structural and mechanistic features of the system using a wide range of complementary techniques, many of them in situ. Even rarer are situations in which both the access to the active sites and the shape of the reactants may be systematically and subtly varied, so that one is able to compare the performance of the active site in a crystalline environment with an essentially identical one embedded in an amorphous solid. Over the past five years, we and our colleagues have undertaken an extensive study of the acid-catalyzed dehydration of the four isomeric butyl alcohols. In so doing, we compared the performance of crystalline, molecular-sieve acid catalysts (HZSM-5)in a range of crystal sizes (so as to vary diffusion path and active-site concentration) with that of amorphous aluminosilicate (AAS) gels in which the pore size is significantly larger. Our results, which permit the 335 Copynght 0 1996 by Academic Press, Inc. All nghu of reproduction in any form reserved.

336

KlRIL ILYCH ZAMARAEV A N D JOHN MEURIG THOMAS

identification of reaction pathways and key intermediates, reveal the interesting catalytic consequences of pore confinement. They also point the way to a more rational design of new microcrystalline solid acid catalysts of the kind increasingly required in the petrochemical industry (1). Traditionally,the same overall mechanisms of acid catalysis invoking carbenium ions have been assumed to prevail both in heterogeneous (2) and in liquid homogeneous (3) systems. But these mechanisms do not adequately take into account the fact that adsorbed, rather than free, carbenium ions are formed in the pores of solid catalysts. Consequently, a quantum-chemical model that demonstrates how the interaction of carbenium ions with the sites of their adsorption can influence the reaction mechanism has been formulated by Kazansky ( 4 ) , taking double-bond-shift reactions in olefins as a particular example. According to this view, adsorbed carbenium ions are best regarded as transition states rather than reaction intermediates, a notion that had also been proposed earlier by Zhidomirov and one of us (5). Although theoretical and computational advances now afford powerful insights into the mechanisms of heterogeneous catalysis, especially on acidic, zeolitic solids (6a-d), experimental studies (7, 8) still hold sway. This we hope to demonstrate here by reference to the wide range of techniques-spectroscopic, kinetic, and analytical-that we have brought to bear in our studies of the catalytic dehydration of butyl alcohols. Dehydration of butyl alcohols over HZSM-5 and AAS acid catalysts provides unique opportunities to elucidate experimentally how confinement of the reagents, intermediates, and products inside the pores of the catalyst influences the reaction pathways. Indeed, in both HZSM-5 and A A S the dehydration reaction proceeds over the same active site, viz.

\

/ \ /

/

\

-Al \

H I 0

Si-

/

(henceforth written as /Al-O(H)-Si\) with carbenium ions functioning as reaction intermediates (2, 9). These catalysts have profoundly different pore structures: The diameter of the channels in HZSM-5 is very close (ca. 5.5 A) to that of the four distinct reactants, to that of the supposed key intermediate/ transition states (i.e., carbenium ions), and to that of the reaction products (butenes and ethers). The pore diameter of our AAS, on the other hand, is much larger and estimated to be 50 t 5 A. Moreover, the critical molecular diameters of the reactants range from a value that is smaller than 5.5 A for n-butyl alcohol to one that is significantly larger in the case of ferf-butyl alcohol.

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

337

TABLE I Characterization of Aluminosilicate Samples Sample

Crystallite size (pm)

SVAI

N x IO-*’ Sites/g”

42 35

HZSM-5 1

3and5 4

NaHZSM-5 2 AAS


35

2.3 2.8 2.8

0.54

20

3.3‘

1omd

1.1

Per gram of dehydrated sample. ’Aggregates of 0. I-pm crystallites. ‘Na/AI = 0.35.dAverage pore diameter ca. 50 A.

II. Characterization of Catalysts HZSM-5 and NaHZSM-5 samples with different crystallite sizes were used in our studies (Table I). They were synthesized using organic templates (8a-d), and the crystallite sizes were determined by scanning electron microscopy (Fig. 1). The AAS catalyst contained 15% A1203. As seen from Fig. 2, the catalytic activity per unit site is the same for all our acid catalysts.

FIG.1. Scanning electron micrograph (84for HZSM-5 sample 4 of Table I.

338

KIRIL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

1.o

1.0

[B]/10~0rWorg&

2.0

3.0

[Ll.oono /10‘0 altos ggt

[ L I , ~ ~ / I O ’ ~rites g ; ~

FIG.2. Relation (86) between the activity of ZSM-5and AAS in isobutyl alcohol dehydration to butene at 397 K and the number of (a) Bransted acid sites [B]; (b) strong Lewis acid sites, [LIstmng; and (c) weak Lewis acid sites, [LIwrak.(A) NaHZSM-5 sample 2; (a)HZSM-5, sample I ; (0) HZSM-5, sample 3; (0)HZSM-5, sample 4; (0)amorphous aluminosilicate, AAS.

From N2 adsorption isotherms, the surface area (SBET) of the ASS is 680 m2 g-’ and its pore volume is 0.46 cm3 g-’, which is to be compared with a pore volume of 0.17 cm3 g-’ for the HZSM-5 samples. The average pore diameter of A A S is 50 A, whereas in ZSM-5there is an intersecting network of straight and zigzag channels (average diameter 5.5 A), the cavities at the intersections being ca. 9 A in diameter. A. THEACTIVESITE:ITSNATURE, CONCENTRATION, AND LOCATION

Dehydration of butyl alcohols on all our catalysts takes place over strong Brclnsted acid sites, rather than over Lewis acid sites. We know this to be so because the activity of all the catalysts is directly proportional to the concen\ / groupings and does not tration of the bridging hydroxyl TAI-O(H)-SiT correlate with the concentration of strong and weak Lewis acid sites (see Fig. 2, the data of which refer specifically to the dehydration of isobutyl alcohol). Further proof that Lewis acid sites are not catalytically active, whereas Br~nsted ones are, comes from the fact that preadsorbed acetonitrile (which affects Lewis but not Brernsted sites) has no effect on the rate of dehydration, whereas increasing amounts of preadsorbed pyridine (which becomes attached to Brernsted sites) progressively decreases the rate of dehydration of isobutyl \ / alcohol (see Fig. 3 of ref. 86). The concentration of YAl-O(H)-SiT (Br~nsted)sites was measured by four distinct methods-by ‘HNMR spectroscopy, (quantitative) IR spectroscopy, by “poisoning” titration with pyridine, and from direct analysis of Al content+ach in good agreement with the others (86).

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

=#I L

b

339

7 20-

20

a

(I)

b

Time / mln

Time / mln

FIG.3. Kinetics of n-butyl alcohol consumption (a) and dehydration (b) in HZSM-5 (flow microreactor, sample I , 399 K): (A)Water; (0)di-n-butyl ether, n-butene; ( X ) unreacted n-butyl alcohol.

From the solid-state *H NMR spectra of tert-butyl alcohol sorbed within samples of HZSM-5 catalysts, it was concluded that the geometry of the isolated molecule is unperturbed when bound to the active site (Sg, h). Since the critical dimension of the tert-butyl alcohol (ca. 6.8 A) significantly exceeds the channel diameter of the catalyst (ca. 5.5 A), it follows that this reactant molecule must be accommodated at channel intersections [where there is a free diameter of ca. 9 A (lo)],and that, therefore, the Brmsted active sites are themselves located at such intersections, as illustrated in Figs. 4 and 5 . The Brmsted sites (bridging hydroxyls), which are the active sites in HZSM-5 and AAS, exhibit the same average acid strength. This we know from the observed shift in OH stretching (IR) frequency when CO is adsorbed at low temperature (see Fig. 3 of ref. 8j): The magnitude of the shift is (11) directly proportional to the Brensted acid strength. Although the IR absorption peak is considerably broader in AAS than in HZSMJ, reflecting a wider distribution of acid strength, the shifts of the average position of the absorption are the same for the two types of catalyst. This close similarity in acid strength is also borne out by independent measurement of enthalpies of protonation of pyridine: on HZSM-5 the value is (12a) 172, and on AAS (12b) it is 177 kJ mol-’.

111.

Kinetic Studies

Dehydration kinetics of the four alochols were followed using two distinct types of catalytic reactors: a “static” FTIR spectrometer cell, in which the concentration of alcohol adsorbed by the catalyst was adjusted to be less than or equal to the concentration of the active sites; and a flow microreactor, which allowed the escaping products (and reactant) to be identified by gas chromatography. Kinetic measurements conducted with the FTIR cell refer to the

340

KlRIL ILYCH ZAMARAEV AND JOHN MEURlG THOMAS

FIG.4. Computer graphic illustrations of the cross-sectional views of the four isomeric butyl alcohols in relation to the 5.5 A aperture of ZSM-5.

products formed and trapped within the pores of the catalyst. The flow microreactor readily lends itself to transient-flow kinetics experiments in which the flow of butyl alcohol is interrupted periodically and replaced by He only. A.

KINETICSTUDIES USINGA STATIC FTIR CELL

Fourier-transform IR spectra show that the main products of butyl alcohol dehydration, when they are adsorbed on HZSM-5in quantities smaller than or equal to the number of the active sites, are water and butene oligomers, the

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

34 1

FIG. 5. Schematic representation of HZSM-5 porous structure (a) and of molecules of butyl alcohol, of reaction intermediates. and of products in HZSM-5 pores (b); polar and nonpolar fragments of the molecules are shown as white and dashed areas, respectively.

latter being prodominantly C S species (8d). No butenes or other olefins are observed (8). The kinetics of adsorption and dehydration of the butyl alcohol were measured in situ via the time-dependences of the line intensities at 1460-1470 cm- (CH deformation vibrations) and 1640 cm-’ (deformation vibrations of adsorbed H20), respectively. The first spectrum could be recorded 25 s after admission of alcohol to the catalyst. For all the zeolite samples of various crystallite sizes (Table I) at 296 K, the adsorption was complete within 25 s for n-, sec- and isobutyl alcohols. The dehydration process of these alcohols in the zeolitic pores was, however, slower. For a given alcohol (n-, sec-, or iso-) the kinetics of water elimination were identical for catalysts of different crystallite sizes. This firmly establishes the absence of any diffision limitation for dehydration for these three alcohols. The picture is different for the bulkier tert-butyl alcohol: the kinetics of its adsorption at room temperature are markedly retarded with increasing crystallite sizes of HZSM-5; and they exhibit a I”* dependence (t is the time after adsorption) (see Fig. 3c of ref. 8d). This is symptomatic of a diffisioninfluenced process. The diflusion coeficient for tert-butyl alcohol in HZSM-5 was directly measuredto b e D = 2 X 10-”cm2s-’at296K. Note that the size of the tert-butyl alcohol molecule (5.4 X 6.8 X 6.8 A) distinctly exceeds the aperture of ZSM-5 channel, so that a serious mutual perturbation of the geometry of the tert-butyl alcohol molecule by the zeolite, and vice versa, operates during difision. (The translational movement of the rerr-butyl alcohol molecule inside HZSM-5 channel must resemble that of a rabbit inside a snake.) For most of the time, adsorbed alcohol molecules reside \ / at 7Al-O(H)-Siy active sites near channel intersections, where they are conveniently accommodated (Section 1I.B) and dehydrated (8d, g, h).

342

KlRIL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

In summaty, for n-, sec-, and isobutyl alcohols there is no diffusion limitation for the dehydration, whereas for the bulkier tert-butyl alcohol, dehydration in channels of HZSM-5 under certain conditions is influenced by diflusion. For all four alcohols in the zeolitic catalysts with small enough crystallite sizes-when diffusion limitations also disappeardehydration kinetics are well approximated by the exponental function, a fact that is explicable in terms of the unimolecular decay of molecules of butyl alcohol adsorbed on identical active sites. With isobutyl alcohol, for example, the rate coefficient k may be written

k= 2

X

(-

109exp

80 2 ;*GI

mol-'

where the rather small pre-exponential factor corresponds to an activation entropy of - 70 J mol- K- I . All this implies that dehydration of adsorbed butyl alcohols proceeds via a complicated rearrangement involving a reaction coordinate that entails a concerted movement of many atoms.

'

B. KINETICSTUDIES WITH A FLOWMICROREACTOR In contrast to studies with the FTIR cell, here we focus on the rates of formation of the product species (butenes and dibutyl ethers) as well as the consumption of the reactants. Full details have been published elsewhere (&?a&, j , k). Typical results, exemplified by the dehydration of n-butyl alcohol over HZSM-5 (Sj), are shown in Fig. 3, a characteristic feature being the gradual (ca. 8 min at 399 K) saturation of the catalyst with the alcohol. At saturation there are 8 X lo2' mol of adsorbed butyl alcohol per gram, which is more than 3.5 times the concentration of active (Brensted acid) sites. Under the same conditions, our amorphous AAS sample adsorbs 3 X 10'' rnol of n-butyl alcohol per gram, which exceeds the active site concentration by ca. 2.7 times. Thus, under conditions of our flow experiments, butyl alcohol is present in two forms: (i) hydrogen-bonded to the catalyst active site and (ii) nonbonded to the active site. Knowing the pore volumes, we arrive at the concentration of adsorbed n-butyl alcohol at 399 K for HZSM-5 and AAS, respectively, of 4.7 X lo2' mol cmT3and 6.5 X 10'' mol ~ m - Liquid ~ . n-butyl alcohol at 298 K contains 6.6 X lo2' mol cmT3. Thus, the density of the reactant inside the zeolite pores is close to that in the liquid, whereas pores of AAS are more or less empty. This means that conditions for dehydration of the same alcohol over the same (acid) active sites are dramatically different in HZSM-5 and AAS. Not surprisingly, under identical reaction conditions, HZSM-5 and AAS exhibit considerable differences in catalytic properties, notwithstanding their almost identical active sites.

343

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

For instance, the activation energy for butene formation from n-butyl alcohol is 140 t 10 kJ mol-I on HZSM-5 and only 95 2 10 kJ mol-’ on AAS. At 378 K, 94% ether plus 6% butene are formed over HZSM-5, whereas 43% ether and 57% butene are formed over AAS. Bearing in mind that butyl alcohol molecules, as well as those of intermediates and products of their dehydration, have dimensions closely similar to the diameter of the zeolite channels, we infer that a liquid-like packing of butyl alcohol molecules and other reaction participants occurs in the channels (as schematized in Fig. 5). We opine that some specijic ordering of the adsorbed species in the catalyst channels may be induced by hydrogen bonding and hydrophobic interactions between them. The time taken to fill the zeolitic channels with alcohol (ca. 8 min, Fig. 3a) is insufficient to reach steady-state values for the rates of formation of butene (WB) and ether (WBq0). The gradual growth of WB and W B ~to~ O steady-state values within ca. 30 min (Fig. 3b) points to the formation and accumulation of some reaction intermediates from which butene and ether are later formed. In contrast to the initial rates of butene and ether formation, the initial rate of water in the transient period exceeds its steady-state value. This formation (WHz0) reflects a larger concentration of intermediates (from which H 2 0 is formed) during the initial transient regime than during the steady-state one. Transient kinetic phenomena of another type were observed in the so-called “purging” experiments, whereby we switched from feeding the flow reactor with a helium-butyl alcohol mixture to one with pure helium and then back to the previous helium-butyl alcohol. A typical response of a catalyst to such purging is given in Fig. 6, referring to the dehydration of sec- and isobutyl alcohols over HZSM-5. For sec-butyl alcohol, the rate of butene formation initially increases by a factor of about 10 upon purging and then drops to zero. Return (Sk) to the b

’

dosorption

’

Time / min

FIG.6. Rate constants for dehydration of butyl alcohols into butene over HZSM-5 under steadystate conditions and subsequent desorption of butene when the helium-alcohol flow is switched to pure helium flow.(a) sec-Butyl alcohol at 369 K; ( 0 )sample I; (A) sample 2; (0)sample 4. (b) Isobutyl alcohol at 397 K; (0)sample 2. The horizontal lines on the extreme right denote the reaction rate constants after returning to the previous helium-butyl alcohol flow.

344

KIRlL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

previous helium-butyl alcohol flow results in a return to the butene rate of formation at its previous, steady-state level (Fig. 6a). For isobutyl alcohol, no increase in the rate coefficient is observed at the initial stage of purging; and after return to the initial alcohol-helium flow, the rate coefficient reaches only a half of the previous level (Fig. 6b). The kinetic effects (8a-d,j, k) observed during purging for various butyl alcohols and catalysts, as well as the phenomenology of isotopic exchange mentioned later, are quantitatively describable in terms of the reaction mechanism discussed in the next section.

IV. Pathways of Butyl Alcohol Dehydration All steady-state and transient kinetic data for the dehydration of all butyl alcohols over HZSM-5 and AAS catalysts may be rationalized in terms of the reaction mechanism presented in Scheme 1.

REACTIONINTERMEDIATES

A.

The key reaction intermediate is the species

where R = C4H9. (The actual structure of C4H9 fragment is discussed in Section V.) The presence of this intermediate constitutes the main difference between our mechanism and the consecutive-parallel one (Scheme 2) suggested previously (9, 13) for dehydration of alcohols. Other reaction intermediates are R

\ /

R

H

0

37

and

\ /

R

0

37

where alochol and ether, respectively, are hydrogen-bonded to the bridging \ / of the active site. hydroxyl /AI-O(H)-S~T Note that, at steady-state, the dehydration kinetics were always zero-order with respect to alcohol in the feeding flow, signifying that, under these conditions, only a very small fraction of the active sites are free. As schematized in

345

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

+ ROH

OH

%ROH

/f77

111

% SCHEMEI

Fig. 5 , most of them are converted to R

\ /

H

R

\ /

R

p

8

z,z, 3 and

where the relative proportions of each species depends strongly on the type of butyl alcohol, the catalyst (i.e., HZSM-5 or AAS), and the reaction temperature.

-

-

-

A alcohol, E -ether, W water, B butene or other olefin.

2A R E

\ 2B

i") + 2W SCHEME 2

+

B

+

346

KIRlL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

The amounts of these species were estimated from the transient kinetics in purging experiments. Thus, for isobutyl alcohol in both HZSM-5 and AAS, the

77r species prevail. For sec-butyl alcohol, the R

\ /

H

0

27 species prevails in HZSM-5, but comparable amounts of R\

/R

0

37

and

77r

coexist in AAS. Such differences readily explain the difference in behavior, typified by Figs. 5a and b, of two alcohols during purging. With n-butyl alcohol, all the three species coexist in comparable amounts in both HZSM-5 and AAS. The overall reaction scheme for dehydration is identicalfor all butyl alcohols and for both our catalysts, but the relative amounts of various reaction intermediates and the relative values of the rate coeficients for various reaction steps can be dramatically diflerent. Thus, for a given temperature and given catalyst loading in the reactor, and for a given gas-flow rate through the reactor and concentration of butyl alcohol in the gas flow, the observed reaction rates and selectivities with respect to various reaction products can be crucially different for different butyl alcohols and different catalysts (i.e., crystalline HZSM-5 or AAS).

B. REACTIONSTEPS In Scheme 1, steps I, V, and VII are the processes of reversible adsorptiondesorption of the alcohol reactant and butene and ether products onto or from \ / the TAl-O(H)-SiT site. Steps I1 and 111 are reversible interconversions of R

\ /

0

H

R

\ /

0

R

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

347

That such conversions are indeed reversible under steady state conditions follows from the fact that the steady-state rates of butene formation over both HZSM-5 and AAS catalysts remain the same when n-butyl alcohol is substituted for di-n-butyl ether in the flow that feeds the microreactor (8j). Pathway IV, which converts chemisrobed ROR to butene and chemisorbed ROH is, in principle, plausible. But in our kinetic experiments it cannot be detected against the background of the combination of steps I, V, and 111. It is shown, therefore, with the dashed line in Scheme 1 as a possibility that awaits confirmation or rejection. Note that the rate coeficients k determined by our kinetic studies with the static FTIR reactor for all four butyl alcohols are the true rate coeficients for the forward step of stage I1 of Scheme 1, i.e., k = k+". But under the steadystate conditions of the flow microreactor, the observed reation rate, WBuOH, of butyl alcohol dehydration is less than or equal to the product (k+,N) of the rate \ / coeficient kill and of the number, N, of the /Al-O(H)-Siy active sites in the catalyst (8j). This is because the fraction, qROH, of the active sites that is covered with the hydrogen-bonded alcohol ROH is less than or equal to unity. Note also that the rate coefficients in Scheme 1 (k2 among them) measured in our static FTIR and flow GS kinetic measurements should not necessarily coincide because of the solvation effects that may become important in the latter (but not in the former) case (vide infra).

C. SECONDARY REACTIONS Step VI in Scheme 1 is an irreversible secondary side reaction-oligomerizaOR

tion of butene formed from one n b site in the reaction with another. In our static FTIR experiments (Section IILA), we observe that butene oligomers are the final reaction products. In our flow experiments they are seen gradually to poison the active sites of the catalyst, the poisoning being more rapid at elevated temperature. Poisoning via oligomerization is much more pronounced for the HZSM-5 catalysts than for AAS; and its rate increases both with increasing size of zeolite crystallites and during purging (8a). The reasons for these effects are simple. First, during their diffusion from the inside to outside of the zeolite crystallites, butene molecules inevitably must OH

pass through channel intersections, where & groups active in oligomerization are located. In AAS with much larger pore apertures, such collisions are more readily avoided. Second, the probablity that a butene molecule will meet an OH

& group when escaping from the zeolite particle itself increases with increasing crystallite size. Third, during purging, the concentration of the adsorbed

348

KIRlL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

alcohol in the catalyst pores decreases, thereby increasing the fraction of the OR

OH

z b and z b sites that are no longer protected from involvement in poisonous butene oligomerization by hydrogen-bonded alcohol molecules or by adjacent physisorbed molecules of alcohol. Thus, the role of the alcohol in the reaction in question is twofold: On the one hand, it is the reagent; and, on the other, it is a protecting agent that reduces the degree of the undesirable side reaction of butene oligomerization. The poisoning is particularly fast for tert-butyl alcohol dehydration over \ H Z S M J . Lndeed it is so rapid at 333 K that the active ~ A l - O ( H ) - S i ~ sites are blocked and converted into \

BUR I /

-A1 -0 -Si /

\

before the steady-state regime of the dehydration reaction is established. Consequently, dehydration of tert-butyl alcohol in the steady-state regime proceeds on the external surface ( 8 d ) rather than inside the channels of the HZSM-5 crystallites. The other secondary reaction that we observed is isomerization of butene formed in reaction + V (and, perhaps, in the supposed reaction IV). Table I1 shows the isomeric distribution of butenes that are formed from n-butyl alcohol upon dehydration at 423 K in the HZSM-5 zeolites with different crystallite sizes. It is seen that, for Sample 1 (with the smallest crystallite size), the isomeric distribution is very far from equilibrium (Table 11). However, with the increase of the crystallite size or by inserting rather bulky Na' ions into the TABLE II Isomeric Disiribuiion of Burenes Formed upon Dehydraiion of n-Buiyl Alcohol at 423 K

I

t-2

c- 1

58.0 55.0 36.1

23.6 25.5 41.0

18.4 19.5 22.4

NaHZSM-5 2

20.1

53.2

26.8

AAS

38.6

31.5

29.9

9.3

60.9

29.8

Sample HZSM-5 1

3 4

Equilibrium"

"From Ref. 14.

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

349

zeolitic channels, the isomeric distribution tends to the equilibrium value, although it does not quite reach it. The reason for this is again simple. Butene isomerization proceeds on the free \ / . ~ A l - O ( H ) - S i ~ sites that are still present in small amounts. The larger the size of the zeolite crystallite, the higher should be the probability for a butene molecule to collide with such sites and to isomerize over them before it desorbs to the gas phase. \ / The concentration of the active /Al-O(H)-Si\ sites available for isomerization increases during our purging experiments (Section 1II.B). Therefore, in these experiments, we observe as reaction products not only various butene isomers but also various isomers of dibutyl ether and even of butyl alcohol. In particular, for n-butyl alcohol dehydration in HZSM-5, we detected the following in the product stream during purging: di-n-butyl ether; n,sec-butyl ether; di-sec-butyl ether, and sec-butyl alcohol. Apparently, isomeric scrambling takes place first among the butenes and then (through the pathways of Scheme 1) among the oxygenates. In AAS, approximately equal amounts of each of the butene isomers are formed (Table II), values that are very far from their equilibrium distibution. Note that, in agreement with the conclusion concerning the reversibility of steps I1 and I11 of Scheme 1 (Section IV.A), nearly the same isomeric distribution was obtained over AAS for butenes when di-n-butyl ether was used as the reagent in place of n-butyl alcohol. No isomeric scrambling among oxygenate species was observed during purging experiments with AAS. This amorphous catalyst has a more open pore structure that favors neither secondary oligomerization nor secondary isomerization. Thus, secondary reactions of butene isomerization and oligomerization \ / \ / proceed over free /AI-O(H)-Si\ sites and /Al-O(R)-Si\ sites, respectively. The critical sizes of butene molecules are close to those ofthe HZSM-5 channels so that butene molecules collide frequently with these sites lining the walls of the zeolite during their diflusion out of the crystallite. In ASS, on the other hand, butene molecules are not stericallyforced to collide with these sites as frequently as in HZSM-5. They therefore have a lower probability of isomeriring or oligomerizing before escaping from the pores of AAS.

\

V. The Nature of the /Al-O(R)-SiT

/

Reaction Intermediate

We have characterized (8e-i) this intermediate using solid-state I3C CP/MAS NMR, *HNMR, and two dimensional (2D)J-resolved I3C NMR spectroscopy in the course of dehydration of isobutyl alcohol and tert-butyl alcohol in HZSM-

350

KlRIL ILYCH ZAMARAEV AND JOHN MEURlG THOMAS

5, the concentration of adsorbed butyl alcohols being about half that of the \ / active 7Al-O(H)-Sii site. By way of example (8e), Fig. 6 shows variations in the I3C CPMAS NMR spectrum of isobutyl alcohol selectively labeled with I3C in the CH2 groups, i.e.,

cH3

c H3

>CH-*CH,-OH

The spectrum of Fig. 7A refers to the adsorbed isobutyl alcohol, with the line at 74.9ppm being characteristic of the carbon atom that is bound to the oxygen

I vv

>

PPM

""

FIG. 7. Variation of the I3C CP/MAS NMR spectrum of CH3 CH--*CH*-OH adsorbed in CH3 HZSM-5 upon successive heating of the sample for a certain time at various temperatures: (A) AAer adsorption at 298 K; (B) 70 min at 343 K; (C) 40 min at 398 K; (D) 60 min at 413 K (inserted is the contour plot of the 2D J-resolved "C MAS NMR spectrum in the vicinity of the signal at 30.5 ppm); (E) 60 rnin at 448 K, (F) 60 rnin at 448 K.

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

35 1

atom. Upon heating the sample for 30 min at 373 K, where we know from our previous kinetic studies that the alcohol molecules are dehydrated, we observe the line at 73.0 ppm (Fig. 7B) from the isobutyl silyl ether (IBSE) intermediate that is formed in reaction + I1 of Scheme 1: CH3\

/CH3 CH

I

*CH2

I

\

-A1 /

0

/ \ /

Si-

\

Further heating results in the selective transfer of the 13C label from the CH2 group of IBSE into the CH groups of two nonequivalent isobutyl fragments (lines at 30.5 and 31.5 ppm) and to a lesser extent into the CH3 group (the line at 19.2 ppm) of IBSE. The lines at 30.5 and 3 1.5 ppm were attributed to the CH group according to their 2D J-resolved "C MAS NMR spectra ( 8 i ) (insert, Fig. 7D). Molecular rearrangements that explain ( 8 i ) how the I3C label can be transferred to the CH3 and CH groups of IBSE are shown in Scheme 3. Unambiguously, they must involve the isobutyl carbenium ion (IBCI) as the reaction intermediate or transition state. This ion is shown in the upper left comer of Scheme 3. Note the selectivity in the transformation of the su posed IBCI inside HZSM-5. As suggested by the preferential transfer of the IpC label to the CH rather than to the CH3 group, adsorbed IBCI prefers to transform via pathway I1 rather than pathway I of Scheme 3 ; i.e., pore confinement in HZSM-5 kinetically favors the formation of linear, rather than branched, C4 carbenium ions. This conclusion is hrther supported (8e) by our solid-state *H NMR studies, according to which the transformation of isobutyl cation ion into tert-butyl cation is not favored inside HZSM-5, even though such rearrangement is favored both thermodynamically and kinetically in solutions. The skeleton of a carbenium ion bears a positive electric charge, and the surface atomic layer of the zeolitic channels is composed of negatively charged oxygen atoms. Ostensibly, a flat linear carbenium ion adheres more strongly to the channel walls in HZSM-5 than do the non-flat skeletons of the branched carbenium ions. This can be the driving force for the energetic preference for the formation of a linear ion. OR

Thus, for the dehydration of isobutyl alcohol, the 7 7 f / species has the structure of the isobutyl silyl ether, this ether being the true reaction intermediate, since 100% of the isobutyl alcohol reactant is dehydrated (8e) into it via reaction + I1 of Scheme 1. In the absence of the additional flow of isobutyl alcohol molecules, it is stable up to 398 K, but decomposes spontaneously upon

352

KIRlL ILYCH ZAMARAEV AND JOHN MEURlG THOMAS

-

Pathway I

\

0-

/ \

0

/ \ /

/% P i A

*\'

\

/ \

/ \ /

AOP\ 'O A

I

cH2

0- /0\ / \

\

/

Si\ A A Pathway I1

CH C H - & ~ - C H ~ \

3-+ 0I \

0

/ \ /

B'\ A A

tl

B SCHEME 3

fiu-ther heating to give butene oligomers. As shown in Scheme 3, the IBSE structure rearranges reversibly into the IBCI structure through which butene product is expected to form. For tert-butyl alcohol, we have observed (Sf)with I3C CPMAS and 2H NMR spectroscopy, the formation of tert-butyl silyl ether (TBSE)at temperatures as low as 296 K. In these experiments, 13C and 2H-labeledtert-butyl alcohols were used:

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

CH3, CH,-*CH-OH, /

CH,

*CH3 \ CHI-C-OH, /

CH3

and

353

CD, \ CD3-C-OH /

CD3

However, we found that TBSE is a rather stable species, decomposing only upon heating above 373 K. Thus, at 296 K it behaves as a side intermediate species, through which only a small fraction of tert-butyl alcohol molecules dehydrate. The main reaction stream bypasses the TBSE structure, proceeding presumably through the tert-butyl cation ion as the key intermediate (Scheme 4). It is interesting that-regardless of where the I3C label is initially placed (i.e., into the CH-OH or a CH3 g r o u p t a f t e r keeping tert-butyl alcohol within the HZSM-5 sample at 296 K for 15 h, a complete scrambling of the 13C label is observed among various positions in the TBSE intermediate and butene aligomers formed 8'( in HZSM-5 channels from butene under the conditions of our NMR and FTIR studies. Moreover, if D20and (CH3)3COH are simultaneously adsorbed at 296 K in HZSM-5, the deuterium is transferred from the D 2 0 molecule to methyl groups of the alcohol. Similarly, in the course of (CD3)&OH dehydration in HZSM-5, deuterated water is formed as a reaction product. Thus, scrambling of deuterium atom between water molecules and methyl groups of tert-butyl alcohol molecule is facile. Undoubtedly, such scrambling phenomena are possible only with the participation of carbenium ions as reaction intermediates or transition states. Molecular rearrangements that rationalize the scrambling of deuterium atoms are shown in Scheme 5. OR

T!T

Thus, reaction intermediate for dehydration of butyl alcohols can exist in two forms, i.e., butyl silyl ether (BSE) and adsorbed butyl carbenium ion (BCI). Our NMR and kinetic data imply the existence of reversible transformations between BSE, BCI, and adsorbed butene (Bua& that ar shown in Scheme \ / 6. BUads is butene that is hydrogen-bonded to /AI-O(H)-Si\. It has been

SCHEME 4

354

KIRIL ILYCH ZAMARAEV A N D JOHN MEURIG THOMAS

L

CHs-h-CH2D

+

+ 0‘ + DHO

I

I

CH3-C-CH2D

+ OD

I

I

OH

SCHEME 5

observed earlier on HNaY zeolite with IR spectroscopy by Pauhhtis et al. (5a) and on HZSM-5 and HY zeolites with I3C NMR spectroscopy by Lazo et al. (5b) at lower temperatures. BCI is a more labile chemical structure centrally involved in the main reaction stream. BSE is a more stable structure, but is sometimes still active enough to participate in the main reaction stream (the case of isobutyl alcohol). But sometimes it is too stable to be in that stream, as is the case for tert-butyl alcohol. Then it plays the role of the side intermediate or, under certain experimental conditions, may even become the deadend of the reaction similar to butene oligomerization. Bu,ds may be a precursor of butene production in the reaction stream.

VI.

Concluslons

The reaction mechanism for the catalytic dehydration of all four butyl \ / alcohols over the same Brensted acid yAl-O(H)-Siy sites in HZSM-5 zeolites with different crystallite sizes and in amorphous aluminosilicate (AAS) catalyst may be rationalized by one all-embracing scheme. Reaction pathways are found to be identical for all the butyl alcohols and the catalysts studied, with respect both to the main reaction stream and to the side reactions. However, depending on the particular structure of both the reactant (i.e., n-, sec-, iso-, or tert-butyl alcohol) and catalyst (i.e., amorphous AAS or crystalline HZSM-5, of specific crystallite size), the observed reaction rates and selectivity

B“

r

BC I SCHEME 6

‘“ads

355

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

toward various products can be dramatically different, even under similar reaction conditions. All these differences are, however, well understood in terms of the proposed and tested reaction mechanism. Three important general conclusions may be drawn from our studies. The first OC4b

refers to the nature of the key reaction intermediate nb , which clearly exhibits carbenium-ion properties, such as scrambling of carbon and hydrogen atoms over its skeleton. In this crucial respect, dehydration of butyl alcohols over solid HZSM-5 and AAS catalysts mechanistically resembles homogeneous acid-catalyzed reactions, whereby carbenium ions in their classical forms serve m4H9

as key reaction intermediates. But the nb intermediate does not coincide exactly with a classical carbenium ion for the following reasons: OC4H9

1. T/fT can exist in three different states: in an ion pair involving a butyl \ / carbenium ion and the yA1-0- - S i y residue of the solid catalyst; as a butyl \ / silyl ether; and as butene hydrogen-bonded to the ~ A l - O ( H ) - S i ~ group. Interconversions between these states (Scheme 6) as well as intramolecular \ / rearrangements in the BCI//Al-O--Si~ ion pair of the type shown in Schemes 3-5, proceed with finite rates that vary with the temperature. (We note in passing that catalytic chemists may fine-tune those rates in reaction engineering by, e.g., making some or all of them lower than the rates of formation andor decomposition of . In this way, reaction selectivity with respect to a desired product can be optimized.) OC4H9

2. Interaction of a carbenium-ion state of nb with the wall of the catalyst pore can kinetically favor reactions that are not favored for carbenium ions in solutions. For example, pore confinement in HZSM-5 favors the formation of linear C4H: (Section V), although branched C4H: are favored in solution both kinetically and thermodynamically. Our conclusions support the ideas of Derouane et al. (16) who, by analogy with enzymes, proposed that confinement effects must be important for catalysis with zeolites. We would expect peculiarities (1) and (2) to be also inherent in carbeniumion-type intermediates of other reactions (such as the cracking, alkylation, and isomerization of hydrocarbons) occurring in the pores of solid acid catalysts. OR

The knowledge of the properties of the key reaction intermediate nb alerts one to what may be done to improve the performance of acid catalysts in converting various organic feedstocks into desired products. One may, for

356

KIRIL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS \

example, try to design catalysts where the /Al-O(H)-Siy

/

active site is

located in pockets the size and geometry of which match best those of the particular carbenium ion that leads to the formation of the desired product. In particular, we may look forward to designing a catalyst that would selectively convert a mixture of butene isomers into only one isomer (e.g., isobutylene) ( I 7) via the hydration of the butene feedstock into alcohols followed by their subsequent selective dehydration into the desired olefin only. The second conclusion that we draw is that, in the catalyst pores, reaction mixtures that are conventionally treated as gases may, in fact, exist in a liquidlike form. We believe that for catalysts with fine pores, this indeed must be a general phenomenon if only because of the ubiquity of capillary condensation. This being so, it follows that reactions in catalyst pores should demonstrate certain effects that can be neglected in gases but not in liquids. Among these are the cage effect, the influence of the dielectric properties of the liquid on reaction rates, solubility effects, and so on. Solubility effects may explain the formation of immiscible liquid phases by mixtures of polar and nonpolar substances, as well as segregation in catalyst pores of gaseous mixture of an organic compound with H2 or 0 2 , into a mixture of the gaseous H2 or O2 with the organic liquid where solubility of H2 or O2 is rather poor. Recently, such segregation of reagents in catalyst pores has been reported ( I d ) in hydrogenation and hydrodearomatization processes. It seems that such solubility effects may exert quite dramatic effects on the overall reaction kinetics. Deliberate, premeditated design of the porous structure can help to utilize the positive effect (e.g., easier separation of reaction products) as well as to suppress the negative effect (e.g., decrease of the reaction rate) of such segregation. It is interesting that, for reaction mixtures consisting of molecules with dimensions close to the cross-sections of the catalyst pores (as in the case of butyl alcohol dehydration in HZSM-5), the reacting mixture may be envisaged as a liquid with dimensions less than three. This, in turn, introduces additional factors with respect to the unanalyzed peculiarities of mass-transfer kinetics in the catalyst pores. The third conclusion that we draw is that a reactant may play a twofold role in catalysis, acting not only as a reactant but also as an agent that protects the active sites from becoming involved in undesirable size reactions. Once the importance of such an effect is recognized and understood, one may formulate other ways of protecting active sites from undesirable reactions, e.g., by intentionally adding to the reaction mixture some special substance (other than the reactant) that serves to protect active sites but is not consumed during the course of reaction. Such an approach resembles the strategy of choosing an appropriate solvent composition when carrying out homogeneous catalytic reactions in solution.

DEHYDRATION OF ISOMERIC BUTYL ALCOHOLS

357

ACKNOWLEDOMENTS The authors are grateful to the Russian Academy of Sciences, the Royal Society, and the Science and Engineering Research Council for their financial support of the research summarized in this account. This review was prepared during the tenure of a Kapitza Fellowship by one of us (KIZ) at the Royal Institution. REFERENCES 1. Sanefield, C. N., “Heterogeneous Catalysis in Practice.” McGraw-Hill, New York, 1980. 2. Rabo, J. A., Ed., in Zeolite Chemistry and Catalysis: ACS monograph. Amer. Chem. SOC.,

Washington, D. C., 1976. 3. Olah, G. A., and Schlosberg, R. H., Eds. “Carbonium Ions.” Vols. I-IV. Wiley, New York, 1968. 4. Kazansky, V. B., Ace. Chem. Res. 24,379 (1991). 5. Zamaraev, K. I., and Zhidomirov, G. M., in Proc. 5th Intern. Symp. on Relations between Homogeneous and Heterogeneous Catalysis, Novosibirsk, 1986. (Yu. 1. Yermakov and V. A. Likholobov, Eds.), Vol. 1, p. 23. VNU Science Press, Utrecht, 1986. 60. Zhidomirov. G. M., and Kazansky, V. B., Adv. Caral. 34, 131 (1986). 66. Sauer, J., Chem. Rev. 89, 189 (1989). 6c. Jackson, R. A,, and Catlow, C. R. A,, Mol. Simulation I, 207 (1988). 6d. van Santen, R. A,, de Bruyn, D. R., den Ouden, C. J. J., and Smit, B., in “Introduction to Zeolite Theory and Modelling” (H. van Bekkum, E. M. Flanigen, and J. C. Jansen, Eds.), Introduction to Zeolite Science and Practice, Studies in Surf. Sci. Catal. Vol. 58, p. 317. Elsevier, Amsterdam, 199 I . 7a. Liengme, B. V., and Hall, W. H., Trans. Faraday SOC.62, 3229 (1966). 76. Lombardo, E. A,, Sill, G.A,, and Hall, W. K., J. Cafal. 22, 54 (1974). 7c. Kemball, C., Leach, H. F., Sundrich, B., and Taylor, K. C., J. Catal. 27,416 (1972). 7d. Kemball, C., Leach, H. F., and Moller, B. W., J. Chem. Suc. Faraday Trans. I624 (1973). 8a. Makarova, M. A,, Williams, C., Romannikov, V. N., Zamaraev, K. I., and Thomas, J. M., J. Chem. SOC.Faraday Trans. 86, 581 (1990). 86. Williams, C., Makarova, M. A,, Malysheva, L. V., Paukshtis, E. A., Zamaraev, K. I., and Thomas, J. M., J. Chem. SOC.Faraday Trans. 86, 3473 (1990). 812.Makarova, M. A,, Williams, C., Thomas, J. M., and Zamaraev, K. I., Cafal. Leff. 4, 261 ( 1990). 8d. Williams, C., Makarova, M. A., Malysheva, L. V., Paukshtis, E. A., Talsi, E. P., Thomas, J. M., and Zamaraev, K. I., J. Catal. 127, 377 (1991). 8e. Stepanov, A. G.,Romannikov, V. N.. and Zamaraev, K. I., Catal. Left. 13,407 (1992). ’8 Stepanov, A. G.,Zamaraev, K. I., and Thomas, J. M., Catal. Left. 13, 407 (1992). 8g. Stepanov, A. G.,Maryasov, A. G.,Romannikov, V. N., and Zamaraev, K. I., Proc. 10th In. Congr. Catal., Budapest, 1992 (L. Guszi, F. Solymosi, and P. Tetenyi, Eds.), Part A, p. 621. Akademiai KiadoElsevier, Budapest, 1993. 8h. Stepanov, A. G.,Maryasov, A. G.,Romannikov, V. N., and Zamaraev, K. I., Magn. Res. Chem. 32, 16 (1994). 8i. Stepanov, A. G.,and Zamaraev, K. I., Caral. Left. 19, 153 (1993). 8j. Makarova, M. A., Paukshtis, E. A., Thomas, J. M., Williams, C., and Zamaraev, K. I., J. Cafal. 149, 36 (1994). 8k. Makarova, M. A,, Williams, C., Zamaraev, K. I., and Thomas, J. M., J. Chem. Soc. Faraduy Trans. 90, 2147 (1994). 81. Makarova, M. A., Paukshtis, E. A,, Thomas, J. M., Williams, C., and Zamaraev, K. I., Caral. Today 9 61 (1991).

358

KIRIL ILYCH ZAMARAEV AND JOHN MEURIG THOMAS

8m. Thomas, J. M., and Zamaraev, K. I., Top. Curd 1, 1 (1994). 9u. Yue, P. L., and Olaofe, O., Chein. Eng. Res. Dev. 62, 81, 167 (1984). 9b. Aronson, M. T., Gorte, R. J., and Fameth, W. E., J. Cutul. 98, 434 (1986); ibid. 105, 455 (1987). 9c. Kalvachev, Yu.,Bezouhanova, C., and Lechert, H. Zeolites 11, 7 3 (1991). IOU. Olson, D. H., Kokotailo, G. T., Lawton, S. L., and Meier, W. M., J. fhys. Chem. 85, 2238 (1981). lob. Lermer, H., Draeger, M., Stefen, J., and Unger, K. K., Zeolites 5, 131 (1985). IOc. van Koningsveld, H., Jansen, J. C., and van Bekkum, H., Zeolites 10, 235 (1990). 11. Paukshtis, E. A., and Yurchenko, E. N., Usp. Khim. 52, 426 (1983 (in Russian). 12u. Paukshtis, E. A., Soltanov, R. I., and Yurchenko, E. N., React. Kinet. Cutul. Lett. 19, I19 (1982). 126. Cardona-Martinez, N., and Dumesic. J. A., J. Curul. 125, 427 (1990). 13u. Knozinger, H., and Kohne, R., J. Cutul. 5, 264 (1966). 13b. Knozinger, H., Angew. Chem., Int. 7 , 791 (1968). 13c. Jacobs, P. A., Tielen, M., and Uytterhoeven, J. B., J. Cutal. 50, 98 (1977). 13d. Moravek, V., and Kraus, M., Collect. Czech. Chem. Commun. 51, 763 (1986). 14. Zhorov, Yu. M., “Isomerisation of Olefins.” Khimia, Moscow, 1977 (in Russian). 150. Paukshtis, E. A.. Malysheva, L. V., Kotsarenko, N. S.,and Karakchiev, L. G., Kinet. Kutul. 21, 455 ( I 980) (in Russian). 1.56. Lazo, N. D., Richardson, B. R., Schettler, P. D., White, J. L., Munson, E. J., and Haw, J. F., J. fhys. chem. 95,9420 (1991). 16. Derouane, E. G . , Andre, J-M., and Lucas, A. A., J. Cutul. 110, 58 (1988). 17. Natarajan, S., Wright, P. A., and Thomas, J. M., J. Chem. Soc. Commun. 1861 (1993). 18. Ostrovskii, N. M., Bukhavtsova, N. M., and Duplyakin, V. K., React. Kine!. Cutul. Lett. 53,253 ( I 994).

ADVANCES Ih' CATALYSIS. VOLUME 41

Thermal and Catalytic Etching Mechanisms of Metal Catalyst Reconstruction TA-CHIN WE1 AND JONATHAN PHILLIPS The Pennsylvania State Universiv Department of Chemical Engineering 133 Fenske Laboratory Universiry Park, Pennsylvania 16802-4400

1.

Introduction

Metals can undergo two types of etching processes in single gases and gas mixtures: thermal etching and catalytic etching. Since the differences between these two types of etching are not clearly delineated in most studies, some confusion has arisen. Thus the first goal of the present review is to provide clear definitions of the two types of etching so that the processes can be readily distinguished. The mechanism of etching, particularly that of catalytic etching, is still in dispute. Therefore; the second goal is the presentation of a systematic and critical discussion of the various proposed etching mechanisms. The third goal is to demonstrate that the mechanism of catalytic etching is virtually identical to that of other types of etching. In particular, it will be demonstrated that plasma etching, etching in low-earth orbit, and catalytic etching all occur via related mechanisms. The first three goals focus on scientific aspects of etching. In contrast, the fourth goal is to demonstrate that etching is technologically significant. There are numerous areas of technology in which etching is critical. For example, the impact of catalytic etching on the properties of metal gauzes, used as catalysts in a number of large scale industrial processes, has been a recognized problem for more than 70 years. Catalytic etching results in the loss of valuable metal, active catalyst surface, and mechanical strength. Potential positive applications of catalytic etching exist as well. For example, some previously discovered catalytic etching phenomena may lead to novel plasma processes which could be of value in integrated circuit processing. Indeed, at present one difficulty with using copper as an interconnect material in integrated circuits is the fact that there are no suitable technologies for dry etching copper. Yet, copper is known to etch catalytically in hydrogen-oxygen mixtures. This fact suggests some new technological strategies for dry etching copper. 359 Copyright 0 1996 by Academic Press. Inc. All rights of reproduction in any form reserved.

360

TA-CHIN WE1 AND JONATHAN PHILLIPS

Thermal etching also has technological significance. For example, current crude oil refining technology practice involves the treatment of reforming catalysts, used in continuous beds, with chlorine. This treatment redisperses the metal via a type of thermal etching, as described below. Catalyst particle “cracking,” another form of thermal etching, is being adopted to assist in the redispersion of precious metals used in automobile exhaust abatement catalysts. This technology may permit less expensive metals to replace rhodium in this application. A.

PHENOMENOLOGY AND DEFINITION

Thermal etching is defined to be any metal surface reconstruction that can occur in the absence of chemical reactions involving gas-phase species. By this definition, reconstruction that occurs in the presence of reacting mixtures, but which mechanistically could occur in the absence of any reaction between gas phase-species, is still clearly thermal etching. In general, the phenomenology of thermal etching is the reconstruction of the surface to produce facets and/or pits with little or no mass change. However, when molecular gas-phase species interact with metal to form volatile species significant loss of mass can occur. Catalytic etching is defined to be any surface reconstruction that can take place only in an environment in which there are reactions between gas-phase species. Metal reconstruction is the result of interactions between species formed due to these reactions and the metal surface. Catalytic etching is normally characterized by severe corrosion at temperatures far below the melting temperature of the metal. Because it often involves significant metal volatilization, catalytic etching can be a major industrial problem. Generally, catalytic etching results in significant loss of mass, although in special cases it leads to the redeposition of the metal as particles without significant net metal loss. In all cases, catalytic etching leads to the production of highly irregular surface structures, which are not minimum-energy configurations. Some related processes are difficult to classify. One example is the formation of blisters and pits in metals which sometimes accompanies treatment of metals, either simultaneously or sequentially, with hydrogen and oxygen (embrittlement). Limitations of the classification scheme with respect to some reconstruction processes are discussed below.

B. MECHANISMS The above definitions are somewhat different fiom those given in earlier publications. In earlier studies, thermal etching has been defined to be etching that takes place in “nonreactive” or “mildly reactive” gases or under vacuum. Catalytic etching has been defined to be any etching process that takes place in

THERMAL AND CATALYTIC ETCHING

36 1

“reactive” gas environments. The revised definitions given here are believed to better reflect the significant differences in the mechanisms of the two types of etching revealed in recent studies. In particular, a significant shift in understanding catalytic etching has occurred in the last ten years. A number of mechanisms leading to reconstruction result from the interactions of single gases with metals, and all are considered thermal etching. First, reconstruction results from the evaporation of metal as the temperature approaches the melting point. Second, surfaces can reorganize into lower energy configurations via localized surface diffusion without metal volatilization or any loss of mass. Third, reconstruction sometimes results from evaporation of compounds that form due to interactions between gas-phase molecules and metal surface atoms. This includes the evaporation of metal oxide species in an oxygen atmosphere and the formation and evaporation of metal carbonyls in a gas environment containing significant amounts of carbon monoxide. Such processes can take place at temperatures far below the metal melting temperature. Fourth, supported (catalyst) particles sometimes “crack” as a result of strains induced by metal/gas interactions. There is a history of debate regarding which of the above processes is primarily responsible for thermal etching. It is argued in the present review that all of these processes take place, and that the dominant mechanism is a function of the particular system. In contrast to thermal etching, for which many mechanisms exist, recent investigations suggest that one particular mechanism explains the vast majority of reconstructions that can occur only in the presence of reacting gas mixtures. That is, catalytic etching generally results from the interaction of free radicals, formed via gas-phase reactions, with the metal surface. Homogeneously formed free radicals interact with the metal to form volatile, metastable metal-containing intermediates. This is consistent with the finding that etching often occurs near known explosion limits. In some cases, the volatile intermediates interact in the gas phase to produce metal particles. These particles are then redeposited on the metal surface, leading to the production of particle-covered surfaces. In other cases, the volatile intermediates “wash out” of the reactor, leading to rapid weight loss. The above model of catalytic etching is not universally accepted. Several older mechanistic models exist. For example, it was suggested that localized temperature gradients, induced by surface reactions, might lead to uneven rates of diffusion or volatilization, and hence catalytic etching. Evidence from investigations outside the field of catalysis tends to support the more recent model. Specifically, a great deal of work outside of catalysis shows that free radicals are responsible for many etching processes. For example, several studies designed to demonstrate the existence of free radicals show that methylene radicals will cause volatilization of a number of metals. Also, modern research into the mechanism of etching in plasmas (dry etching for integrated

362

TA-CHIN WE1 AND JONATHAN PHILLIPS

circuit processing) indicates that free radicals produced in the plasma are often the sole agent responsible for metal etching. Thus, for many metal-etching applications the plasma simply serves as a convenient means to generate free radicals. Finally, a number of studies of material damage to spacecraft in low earth orbit indicate that free radicals, particularly oxygen atoms, are responsible for much of the damage/etching of polymer and metal components. The theory that homogeneously formed free radicals are responsible for catalytic etching is consistent with the conclusions reached in these other fields, namely, that free radicals can interact with metals to form volatile intermediates.

11.

Thermal Etching

Early studies of the etching of a number of metals, particularly silver, copper, platinum, and tungsten, have been thoroughly reviewed by Shuttleworth in 1948 (I), Moore in 1963 (2), and Flytzani-Stephanopoulos and Schmidt in 1979 (3). Little of the early work is repeated here; rather, the goal of the following review is to focus on the significant contributions and clarify fundamental concepts. A.

CLASSICAL THEORIES

Two theories of thermal etching emerged from early studies: namely, that thermal etching results either from (i) enhanced evaporation (kinetic control) of metal or metal complexes (e.g. metal oxides) from specific surface planes or from (ii) surface migration of species to reorganize the surface into a minimum energy state (thermodynamic control). In some cases it was suggested that both processes worked simultaneously, although the relative importance of each is not generally resolved. B. EARLY EXPERIMENTAL INVESTIGATIONS

Experimental investigations conducted before about 1970 were characterized by the use of a variety of optical microscopy techniques and thus were limited to relatively large samples, such as foils, wires, and gauzes. A number of general conclusions can be drawn about the phenomenology of the process of thermal etching from a careful consideration of the early work. First, the temperature at which facet formation is found to occur roughly relates to the volatility and/or the melting temperature of the metal. Second, the nature of the facet pattern is a function of the composition of the gas phase. Indeed, in many gases, facets do not form at all. Third, under low-pressure and vacuum conditions the samples tend to form pitted surfaces more readily or to be characterized by

THERMAL AND CATALYTIC ETCHING

3 63

“hill-and-valley structures.” Fourth, when evaporation is “inhibited,” surfaces are relatively smooth. 1. Unsupported Metals under Vacuum or in a Single-Component Gas The metal first and most frequently studied in conjunction with etching is platinum. One of the earliest and most complete investigations of platinum weight loss was conducted by Hulett and Berger (4), although qualitative observations were made as early as 1879 (5). Hulett and Berger observed measurable weight loss from platinum foils and wires following electric heating in flowing air at temperatures as low as 800°C. No weight loss was measured in experiments conducted at lower temperatures or when the experiments were “conducted in vacuo” (no details given), in agreement with earlier work (6). The authors also collected platinum crystals, 80 to 200 pm in diameter, on relatively cool surfaces about 1.5 cm from wires of platinum heated electrically in air. They conjectured that stable, volatile compounds of platinum form in the presence of oxygen at high temperatures but are unstable and decompose at lower temperatures. No information regarding the structure or appearance of the surface of the metal is available in these early reports. In 1902 Rosenhain (7) reported that grain boundary grooves form on platinum following high-temperature treatment. Later, Rosenhain and Humphrey (8) noted the same phenomenon for steel heated in air at elevated temperatures. The first report of “striations” came from Rosenhain and Ewen in 1912 (9) for silver heated in air (Fig. 1). Leroux and Raub (10) performed a detailed study of the etching of silver and silver alloys in oxygen and hydrogen at elevated temperatures. They were the first to propose that thermal etching is kinetically controlled. Specifically, they suggested that faceting results from differences in the rate of evaporation from different planes and that evaporation rates are influenced by adsorbed species. investigations of other systems supported the model of facet formation resulting from enhanced evaporation of metal oxides from selected surface planes. For example, Elam (11) suggested a similar explanation for the formation of facets on copper heated at elevated temperatures (950°C) in air at low pressures. Studies conducted in the 1950s of platinum surfaces treated in oxygen also indicate that the rates of evaporation and facet formation are enhanced for platinum heated in oxygen at temperatures exceeding 1200°C (12-14). Gwathmey and Benton (Z5-17), studied the behavior of copper spheres in different environments and found that the nature and velocity of the transformations is a function of the composition of the gas phase. The finding is consistent with etching resulting from the evaporation of volatile species. According to a second class of model, thermal etching is driven by a need to reduce total surface free energy. According to this theory, faceting will take place even in the absence of any net weight loss. The first to suggest a model of

364

TA-CHIN WE1 A N D JONATHAN PHILLIPS

FIG. I . Optical micrographs (300X) from an early study of thermal etching showing the striation structure on silver heated in air (9).

this variety was Johnson (18), who noted that the method of heating was important in determining the final structure of the surface. Direct current produced faceted surfaces on tungsten, tantalum, nickel, molybdenum, platinum, and iron surfaces. Thermal gradients were also shown to produce faceted surfaces. On the basis of these results, Johnson concluded that facets form via the surface migration of ionic species. Others who studied the effect of directcurrent heating on faceting supported the suggestion of a thermodynamic driving force for faceting (19, 20). On the basis of other etching experiments with silver, Chalmers et al. (21) suggested an explanation for thermal etching involving both of the earlier models. They concluded that thermal etching results from two processes. Specifically, grain boundary grooving was postulated to be thermodynamically driven, with the surface rearranging itself in the grain boundary region via a

THERMAL AND CATALYTIC ETCHING

365

process of atomic diffusion, tending toward the minimum energy configuration. On areas separate from the grain boundaries, they suggested (in agreement with the earlier workers) that facet formation results from a kinetic process in which silver oxide-like species evaporate at a higher rate from particular locations on the surface. The two models were tested by Hondros and Moore (22), who concluded that etching is in fact a kinetically controlled process. They heated a silver foil in air in a closed silver container to reduce net evaporation to zero. No weight loss and no evidence of the formation of facets or significant grain boundary grooving was detected. In contrast, in flowing air in a glass reactor, etched surfaces formed very quickly. They argued that if thermodynamics were driving the etching process, then faceted surfaces should have formed both with and without net evaporation. Moreover, they found that surfaces which were originally faceted by heat treatment in flowing oxygen became smooth after heat treatment in the closed silver box. On the basis of this finding, they raised a question regarding the true equilibrium surface stucture. Hondros and Moore (22) also found that heating under vacuum or in nitrogen did not lead to the formation of facets. The latter produced smooth surfaces, and the former led to pit formation. From these data and the studies of earlier workers, the authors concluded that the mechanism of silver etching is the preferential evaporation of silver oxide from certain surface planes. Rhead and Mykura (23) tried with limited success to repeat the results of Moore. They found that limited faceting took place when net evaporation was suppressed; however, the extent of the faceting was found to be far less than in the case of heating the foils in open air. Recent investigations of the thermal etching of silver (24) suggest that evaporation (kinetic model) correctly describes etching processes for silver under vacuum. In contrast, under oxygen a drive to attain equilibrium (thermodynamic model) appears to accurately explain most observations. Specifically, it was found that under ultra-high vacuum conditions hill-and-valley structures form only when free evaporation is permitted. When evaporation was suppressed, no facets formed. This modern work may also explain the earlier experimental discrepancies. It was shown that as the temperature is raised and evaporation increased, faceting disappeared even in the free evaporation case. This was attributed to a reduction in surface anisotropy and hence differences in relative evaporation rates which generally accompany increases in temperature. Experiments conducted by the same group for the thermal etching of silver in an oxygen atmosphere suggest that evaporation of metal plays little role in the thermal etching process. The thermodynamic model appears to best explain the observations. That is, identical faceted surfaces formed both in the case of suppressed and free evaporation.

366

TA-CHIN WE1 AND JONATHAN PHILLIPS

Winterbottom (25) found that in any gas, at least at very low pressures (
THERMAL AND CATALYTIC ETCHING

367

crystallites formed on PtiRh (90/10) gauzes after 2600 h at 1125°C in a reaction mixture which was 12% NH3 and 13% CH4 in air at 1 atm were generally similar to those found prior to use of the sample as a catalyst, except that the original facets were larger. In contrast, other early experimental work appeared to support the suggestion that true catalytic etching does take place during HCN synthesis. Indeed, Garton and Turkevich (34) noted that heating of platinum catalysts in any individual gas present during reaction did not result in the formation of large-scale facets. Yet these workers found faceting under reaction conditions. A number of other factors may account for this observed influence of the reaction mixture and mitigate against the conclusion that true catalytic etching occurred. The techniques used by Garton and Turkevich may not have had sufficient resolution to allow observation of small-scale faceting. Other early investigators (12, 14) also failed to note faceting on platinum after oxygen treatments. Yet modem experiments with electron microscopes have clearly shown that small-scale facets form on platinum after treatment in a number of gases, including oxygen and nitrogen, even at temperatures as low as 600°C (35). Notwithstanding the uncertainty regarding the experimental results, the mechanistic models presented by Garton and Turkevich are of interest. Specifically, they suggested that the differences in structure observed following treatment in reaction mixtures may arise from one of two factors: (i) extreme surface temperature increases on the catalyst surface under reaction conditions, or (ii) unknown intermediates that are present only during reaction. In a sense, the latter suggestion is related to the modem model of catalytic etching: Catalytic etching generally results from homogeneously formed free radicals which induce metal atom volatilization. Other work supports the suggestion that extreme surface temperature increases are responsible for the formation of very large facets. It has been shown (details follow) that in both ethylene/oxygen and hydrogedoxygen mixtures there is true catalytic etching of platinum foils, as evidenced by great weight loss and the formation of highly irregular structures. This etching is observed only for a narrow range of reactant stoichiometries and temperatures. In any single gas, product or reactant, at any temperature, there is no weight loss and only small facets form. However, in some reaction mixtures, but under conditions of no weight loss, large facets form. The finding that large-scale facets formed on platinum in NH3/C&, C2I&/O2,and H2/02 reaction mixtures suggests that this is not a chemical process, but instead is related to surface heating. Such heating may enhance the rate of diffusion, thus allowing the material to more readily attain a structure close to the true equilibrium shape. In sum, the question regarding the role of exothermic surface reactions (surface heating) enhancing the rate of large-facet formation cannot be answered on the basis of the information now available.

368

TA-CHIN WE1 AND JONATHAN PHILLIPS

C. EARLYTHEORY

Extensive theoretical work in support of both models of thermal etching was produced before 1970. The theoretical basis for the thermodynamic model was the concept of reduction in total surface energy by the preferential formation of low-energy, low-index planes. The true equilibrium shape of a crystal is the shape with the lowest surface energy, as noted by Curie (36) and Gibbs (37). The thermodynamic models provide no information regarding the process of surface rearrangement. There was also a great deal of modeling of transport processes in support of the kinetic model of thermal etching. The basis of these models is that differences in chemical potential lead to mass transport via a number of mechanisms. It is important to note that these models treat the surface as a continuum and do not involve atomic-level mechanisms. Thermodynamic Model

1.

The means to determine the minimum-energy shape for a crystal of fixed volume was developed by Wulff (38), who showed that the equilibrium shape can be determined if the surface tension, y, at all crystallographic orientations is known. As illustrated in Fig. 2, on a polar y plot of the surface tension as a function of orientation, the inner envelope of the planes drawn perpendicular to and at the ends of the radius vectors gives the equilibrium shape of a crystal of constant volume. Faceting in the equilibrium crystal shape is due to cusps in the polar y plot.

a

FIG.2. A schematic Wulff construction for an equilibrium crystal shape using the polar y plot of the surface tension. (a) The equilibrium shape is that found from the inner envelope of tangents to the y plot. (b) An ECS with (001) facets produced by cusps in the y plot (39).(Reprinted from frog. Surf Sci., Volume 39, E. H. Conrad, Page 65, Copyright (l992), with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 IGB, UK.)

THERMAL AND CATALYTIC ETCHING

369

The cusp points on the Wulff plot generally correspond to low-index planes. This is because the surface tension of a solid is primarily determined by the strength of the bonding of the individual surface atoms. Atoms in the low-index planes form the greatest number of bonds and hence have less energy than atoms in less densely packed planes. Thus, the equilibrium shape generally consists almost entirely of low-index planes. One of the major concerns addressed in early studies of thermal etching is the possibility that a surface with a random average orientation on a crystal may be spontaneously restructured to create facets without any change in the average orientation of that surface. In brief, the answer put forth originally by Herring (40, 41), and further developed by others (42-46), is that energetics can favor the spontaneous decomposition of a random surface into a faceted surface in which all of the facet surfaces correspond to surfaces present on a true equilibrium crystal. That is, a randomly oriented surface will often decompose into two or more low-index planes. The total surface area will be increased, suggesting an increase in total surface energy. Yet the net surface energy will be decreased because of the relatively low surface tension of the low-index facets. The requirement for a surface of a given macroscopic orientation f i B to break up into two new surfaces of orientation iiA and tic is that the total surface energy be reduced.

where yI and A , are the surface tension and area of the surface of orientation 6,. Herring (40, 41) showed that this stability can be treated in terms of tangent sphere construction in the polar y plot. As shown in Fig. 3. the surface with orientation fiB is unstable since the sphere constructed through the origin and tangent to the y plot at point B is pierced by the y plot. On the other hand, the

a

b

FIG. 3. According to Herring’s original theory, minimization of surface energy can lead to faceting of a high-index plane into low-index planes. (a) Herring’s tangent sphere construction for determining the stability of a surface. (b) Cross section of the surface with Orientation AB is faceted into two planes with orientation iiAand tic (3). (Reprinted from Prog. Sur- Sci., Volume 9, M. Flytzani-Stephanopoulos and L. D. Schmidt, Page 83, Copyright (1979), with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 IGB, UK.)

370

TA-CHIN WE1 AND JONATHAN PHILLIPS

surface of orientation fiC is at least metastable since its tangent sphere lies entirely inside the y plot. On the basis of the tangent sphere construction, Herring (40) identified three possible types of equilibrium crystal shapes. (1) If the cusps are very pronounced. and no tangent sphere can be drawn inside all the cusps, the equilibrium crystal shape is a polyhedron. (2) If the cusps are mild such that no tangent sphere passes outside the y plot, the equilibrium crystal shape consists of flat surfacesjoined by round regions, with no sharp corners and edges. (3) If some tangent spheres are inside while others are outside the y plot, the equilibrium crystal shape is composed of flat surfaces and curved regions, with sharp edges. 2 . Kinetic Model As discussed above, a thermodynamically unstable surface will reduce its total surface energy by forming facets. From the point of view of kinetics, gradients in the chemical potential on a nonequilibrium surface will drive the movement of surface materials toward equilibrium. The transport mechanisms are the same as those that can operate during sintering (47): (a) surface diffusion, (b) bulk diffusion, (c) evaporation-condensation, and (d) plastic or viscous flow. Mullins (48-50) published a series of papers detailing the mass-transfer processes which leads to the flattening of corrugated surfaces and thermal grooving. Assuming a nearly flat crystal surface, with surface properties independent of orientation, Mullins showed that the equation -dy(x' ')

- -(am4

dt

+ bo3 + cw2 + do)y(x,t )

describes the decay of the sinusoidal surface profile in a frequency factor proportional to inverse facet length under the simultaneous action of the above four mass-transfer mechanisms. In Eq. (2), the constants have the following definitions: a =

(y

+y

y s "a2

kT

,

b=

(Y + Y'w"n kT

'

where D, and D,, are the surface and bulk diffusivities, respectively; v is the number of atoms per unit surface area; po is the vapor pressure of solid; is the atomic volume of solid; M is the atomic weight of solid; T is absolute temperature; and p is the viscosity of the solid. It is evident from Eq. (2) that surface diffusion should dominate small-scale facet formation (large w), with bulk diffision contributing only for sufficiently large facets (small w ).

THERMAL AND CATALYTIC ETCHING

37 1

The time evolution of facet size for different kinetic processes was also described by Mullins (48, 50). For each transport process alone, the facet size L increases with time according to a power law,

L

- t"

(4)

where n depends on the transport mechanism; n = 1/4 for surface diffusion, n = 1/3 for bulk diffusion, and n = 1/2 for evaporation-condensation. The kinetic models all allow for evaporation-condensation to be a significant mechanism in surface reconstruction. In particular, as noted earlier, it was frequently suggested that a metal or metal oxide would evaporate preferentially from certain planes, leading to a surface (presumably equilibrium) consisting of planes with the lowest evaporation rates. Net weight loss was anticipated. Yet, no evidence of weight loss is available from the early literature. D.

RECENT EXPERIMENTAL INVESTIGATIONS

The character of thermal etching studies changed markedly after the mid1970s. First, virtually all experimental work is now done with electron micro-

scopes. Second, the primary focus of experimental work shifted from relatively large single crystals (foils, wires, and gauzes) to supported particles. Those few investigations of the thermal etching of large-scale systems conducted in the last 20 years were done as controls for catalytic etching studies (51). During the preceding decade the theoretical research has also shifted, possibly as a result of detailed experimental findings. The emphasis is now on microscopic-level reconstruction of surfaces. Advanced surface diffraction and imaging techniques allow detailed characterization of surface morphology at an atomic level. These studies show that metal surfaces contain high concentrations of atomic steps, usually one atom in height, separated by well-ordered terraces. Statistical mechanical theories were developed to explain how atomic-scale processes can lead to the formation of these structures. Much of the important recent work on thermal etching is focused on the impact of high-temperature treatments in single-component gas atmospheres on the structure of supported metal particles. Sintering, redispersion, and changes in particle shape are all effected by thermal etching. The work on the impact of single-gas treatments on supported particles can be conveniently divided into the following topic areas: (1) The impact of thermal etching on the equilibrium shapes of supported particles. It is clear that particle shape and surface structure are a function of the identity of the gas. (2) The impact of thermal etching on metal transport and hence particle dispersion. Metal can be transported by a number of different gases. In some

372

TA-CHIN WE1 AND JONATHAN PHILLIPS

cases this phenomenon has been harnessed to redisperse used, highly sintered catalysts. (3) Treatments in some gases that lead to the gross modification of particle structure. For example, treatments of supported platinum particles in oxygen lead to particle fragmentation. 1. Equilibrium Impact on Particle Shape The equilibrium shape of particles has always been a subject of interest in catalysis. The general theory is the same as that for larger structures: Particle shape will be governed by the tendency to minimize the total Gibbs free energy. However, in the case of small particles, the energy associated with intersections between surface planes becomes significant (52-54). The expression for Gibbs free energy is

where the first sum is over the phases present, the second over the surfaces and interfaces, and the third over the edges. For particles less than 10 nm in diameter, the number of different edges, corners, and faces on the surface can vary dramatically with minor changes in particle shape (55). This in turn will dramatically influence the free energy. Thus, it is not apparent that small particles will have the same shape as a large crystals of the same material. Some of the earliest investigations of the influence of gases on particle shape were performed by Boudart and Dumesic (56, 57). They studied the impact of various gases on the exposed surfaces of iron particles used as ammoniasynthesis catalysts, finding that, after treatment in H2/N2 mixtures, the catalytic activities of the particles increased. They concluded that this increase resulted from the reconstruction of the particles to favor an increased number of (1 11) planes. It was shown in this and other investigations (58, 59) that the activity of iron for ammonia synthesis is related to the fraction of the surface consisting of (1 1 1) planes. It has long been known that catalytic activity for some reactions is a strong function of the identity of the surface crystal planes (17). Thorough investigations of the influence of gases on particle shape were reported by Schmidt et al. (60). They found that the shapes of platinum particles (20-200A in diameter) supported on silica or alumina were effected by the composition of the gas phase when the particles were heated to 600°C at atmospheric pressure. In hydrogen, the particles are inevitably cubic. In contrast, in all other gases, the particles tend to be rounded. The exposed crystal planes also depend on the gas. In hydrogen the (100) face is dominant whereas in nitrogen the (1 11) face is dominant with the (100) face also being present. Schmidt et al. also noted that the shapes formed at 600°C in the various gases were stable at temperatures up to 450°C in other gases. Thus, the modifications

THERMAL AND CATALYTIC ETCHING

373

of the surface could have a strong influence on catalytic activity at temperatures reached in many practical catalytic processes. Lee et al. (61)measured the equilibrium shape of clean platinum by monitoring the changes in the shape of a series of micrometer-sized platinum droplets during annealing at 1200°C in lo-’ Torr of oxygen. Consistent with the work of Schmidt, their results showed that the equilibrium particle shape is influenced by what gas is present. As shown in Fig. 4, at equilibrium, the clean particle shape is nearly spherical, with distinct (100) and (1 1 1) facets. The facets occupy only 16% of the surface. However, if the particle is contaminated with carbon, the particle is a cubooctahedron with large (1 1 l), (loo), and (1 10) facets. A phenomenon related to thermal etching is particle “reconstruction.” Particles containing impurities (e.g., alumina, sulfur) undergo restructuring to expose different surface planes following high-temperature treatment. The driving force is thermodynamics. The impurities change the identities of the low-index planes (62-64). This phenomenon is not strictly thermal etching, as gas-surface interactions are at most a secondary influence. 2.

Volatile Species Transport

There are several important examples illustrating the ability of gases to transport metal in supported catalyst particles and in some cases to redisperse the metal. These processes are clearly thermal etching, not catalytic etching, because reactive conditions are not required. A single gas is responsible for the observed particle reconstruction. The first example of significant metal transport in a single-component gas was reported in a 1964 patent (65). The patent describes reactivation of platinum catalysts supported on refractory oxides, following deactivation during the hydroforming of naphthas. The process consists of three steps, oxidation (which burns off carbonaceous materials), treatment in gaseous chlorine (at a maximum temperature of 7OO0C), and reduction. Most of the sintering that occurred during operation could be reversed by treatment of the particles in halogen gases, including chlorine, fluorine, and bromine; chlorine was preferred. Particles that had grown in average size from 25 A to more than 200 A during operation as catalysts could be redispersed by treatment in flowing halogen gas such that their final size was 50 A. A corresponding increase in catalytic activity was observed as well. Other studies include reports of similar halogen treatments used in the reactivation of a number of noble metal and noble metal alloy catalysts (66-69). Foger and Jaeger (70, 71) studied the mechanism of platinum redispersion by chlorine, finding that redispersion occurs by a four-step process. First, one of a number of platinum chloride species forms. For successhl redispersion it is critical that conditions be selected such that P-PtCl, forms. This process is influenced by temperature, concentration, and support materials. Second, the volatile P-PtC12 is transported in the vapor phase. Third, P-PtCl, is readsorbed.

314

TA-CHIN WE1 AND JONATHAN PHILLIPS

FIG.4. Scanning electron micrographs of the equilibrium shapes of platinum particles show that both the gas phase and impurity in the metal can influence equilibrium shape. (a) A clean Pt particle is nearly spherical with distinct (100) and (1 1 1 ) facets after treatment in lo-’ T o r of oxygen at 1200°C. (b) A carbon-covered Pt particle is cubo-octahedral (61).

THERMAL AND CATALYTIC ETCHING

375

Fourth, the PtC12 species is further chlorinated to form Pt4+ species which are strongly bound to the surface. This process leads to a completely new spatial distribution of platinum. After reduction there is an entirely new distribution of platinum particle sizes. One caveat is that metal is lost as volatile species and removed from the reactor. Operating conditions must be selected with care. A second example of metal transport by single-component gas involves the formation of metal carbonyls that are unstable in carbon monoxide. Shen et al. (72) discovered that supported nickel particles are not stable in carbon monoxide. Under certain conditions, CO combines with the metal in the particles to form volatile metal carbonyls, such as nickel carbonyl. These volatile species carry the metal out of the reactor, resulting in a rapid net loss of metal. In some cases, metal is not carried out of the reactor, but new metal particles form at pore mouths, blocking them and effectively deactivating the catalyst. The authors found that a relationship exists between net metal transport of any type and the temperature and pressure of carbon monoxide (Fig. 5). At high temperatures, the particles are stable, except in high-pressure flowing carbon monoxide. As the temperature is lowered, the pressure at which metal will be volatilized becomes progressively lower. The authors attributed this behavior to the relative stability of the volatile metal carbonyls. The stability of these species decreases as temperature increases. Thus, for a net transport of metal to take place as temperature is increased, the rate of formation of these species must increase. The rate of formation evidently is proportional to the carbon monoxide pressure. The phenomenon of metal transport via the creation of volatile metal carbonyls is familiar to workers using carbon monoxide as a reactant. It is often found that carbon monoxide is contaminated with iron pentacarbonyl, formed by interactions between carbon monoxide and the walls of a steel container. Thus, it is common practice to place a hot trap between the source of the CO and the reaction vessel. Iron carbonyl decomposes in the hot trap and never reaches the catalyst that it would otherwise contaminate or poison. Transport of a number of transition metals via volatile metal carbonyls is common. For example, Collman et al. (73) found that rhodium from rhodium particles supported on either a polymeric support or on alumina could be volatilized to form rhodium carbonyls in flowing CO. 3. Particle Splitting Perhaps the most thoroughly studied example of the gross structural change taking place in catalyst particles as a result of treatment in single-component gases is the “splitting” of supported platinum particles in oxygen atmospheres. Many workers, using a variety of techniques including transmission electron microscopy, chemisorption measurements of surface area, and X-ray diffraction studies of particle size distribution, demonstrated that brief periods of oxygen

376

TA-CHM WE1 AND JONATHAN PHILLIPS

k o(kk) FIG.5 . Several metals have been shown to be volatilized by carbon monoxide. A map of the safe (no metal loss) operating conditions (0)and unsafe (0)operating conditions for Ni/AI20, catalysts. The equilibrium curves for various equilibrium partial pressures of Ni(C0h were calculated by using thermodynamic data from the literature (72).

treatment at elevated temperatures (>500”C) of thoroughly reduced platinum particles supported on refractory oxides led to an increase in the total metal surface area (74-82). The change in the total surface area of the particles occurs after relatively brief exposure to oxygen (ca. 2 h or less). Longer treatments in oxygen may lead to a gradual sintering of the supported platinum particles. Sintering is anticipated as thermodynamics strongly favors a net reduction in surface area, a process generally understood to take place either by coalescence growth or by Ostwald ripening. The majority of workers attribute the unexpected phenomenon of increasing surface area to the “splitting” of partially oxidized particles during the early stages of treatment. Apparently there is a strain, due to the mismatch of oxidized and reduced structures in a particle, that leads to particle splitting. This

THERMAL AND CATALYTIC ETCHING

377

suggestion is consistent with findings that the change in particle structure takes place only for relatively large particles in which strains would be relatively large (81, 82). A second model has been postulated to explain increased particle dispersion following oxygen treatment. Several workers suggested (78, 83) that particle redispersion takes place via a process of surface migration of platinum oxide species followed by the nucleation of new particles. Yet, simple calculations indicate it is highly unlikely that redispersion occurs via a process of vaporphase transport by Pt02. Even at 700"C, the highest temperature used in these types of experiments, the vapor pressure of PtOz (84) is far too low to result in significant platinum transport. Two other assumptions in the model have been questioned as well (82).The first is that the formation of new particles is due to the trapping of the mobile species at highly reactive sites on the alumina surface. The second is that new particles form following collisions between migrating PtO, species. The first assumption is questioned because it is not clear why particles do not nucleate and grow on all reactive sites on the support surface during the original production of the catalyst. The second assumption is questioned because the cross sections of the migrating species are probably far smaller than those of the stationary particles. It seems likely that migrating species rarely collide. A third objection is that the general theory of growth by atomic migration (85-89) is the intraparticle atomic transport or Ostwald ripening theory, and that theory indicates that atomic migration inevitably leads to particle growth. Sintered catalyst particles can also be redispersed in hydrogen atmospheres. For example, Hu et al. (90, 91) showed that Nb205-promoted Rh particles formed RhNb04 during calcination at 700°C (Fig. 6a). Subsequent reduction of the RhNb04 in hydrogen at 773 K led to the formation of Rh and NbOz particles (Fig. 6b). Moreover, the size of the Rh particles (48 A) was far less than that of the original RhNb04 particles (139 A) (92). This process of particle splitting resulted in a dramatic increase in rhodium particle surface area A similar redispersion process was observed for V205-promotedRh catalysts (92). Particle splitting is not merely of academic interest. Recently, it has been demonstrated that three-way automotive exhaust catalysts can be made from supported palladium alone, rather than from rhodium and platinum. The opportunity results in part from improvements in technology that permit palladium to be redispersed by a particle-splitting process during lean-bum periods. Possibly this redispersion results from palladium particle splitting.

E. MODERNTHEORY In contrast to early theory in which the surface is treated as a continuum, most modern theories take the atomistic nature of the crystalline surface into account.

378

TA-CHIN WE1 AND JONATHAN PHlLLlPS

b

a 0 Nb205,@RhZO3

0 RhNbOq

0 RhNbOq

ORh QDNb02

I Oxidation 773 K

4 82.

an m

.'*

I

Oxidation 973 K

4 I

Oxidation 1173K

3.

@ Nb2O5

ORh2O3

I Reduction 773 K

JI

NbO2

I Oxidation 773 K

JI I

Reduct ion 773 K

\L

NbO2

-#d%3%L FIG.6. Modeling of redispersion by particle splitting. (a) Model for the formation of RhNb04 during the calcination process. (b) Structure changes of RhNb04 particle during treatment in Hzor 0 2 (91).

In general, such models are based on the assumption of a classical, central-force lattice model for interatomic bonding. The atomic arrangement of any particular plane can be generated by separating the lattice model along the imaginary plane with the desired spatial orientation. For example, in the terrace-ledge-kink (TLK) model (93-99, a crystalline surface is represented by smooth segments of a particular low-index crystal plane (terrace) with an appropriate density of monatomic steps (ledges) containing a certain density of offsets (kinks). The basic idea of the classical theory of thermal etching remains valid in these atomistic studies. That is, the equilibrium surface morphology is determined by the minimization of surface energy (Wulff plot). However, in these atomistic models, the formulations of surface energy are described in terms of interatomic forces. Atomistic thermodynamic models that allow a calculation of the equilibrium crystal shape at different temperatures are discussed below; modern kinetics studies of thermal etching are also reviewed.

THERMAL AND CATALYTIC ETCHING

1.

379

Thermodynamic model

Gruber and Mullins (96) were the first to use the atomistic model to discuss the influence of temperature on the surface energy due to configurational entropy contributions. They developed the insight that the thermal evolution of the surface free energy is governed by the wandering of steps. Since then, statistical mechanical theories describing the surface energy in terms of atomic behavior have been extensively developed ( 9 9 , and calculations of equilibrium crystal shape over a wide range of temperatures have been performed. Jayaprakash et al. (98, 99) presented an interacting TLK model, focusing on the effect of long- and short-range interactions between distinct facets and those between facets and curved surfaces. In this model the 2D statistics of steps is reduced to the 1D quantum fermion system, and the variation of the surface free energy (per projected area in the low-index plane) with angle 0 and temperature T is described by the equation

where yo(T) is the surface tension per unit area of the low-index surface, /3( T ) is the free energy per unit length for forming an isolated single-layer-height step, h is the step height, and g(T)ltan 813 is the free energy cost per unit area due to stepstep interactions. The first exact calculations of equilibrium crystal shape at nonzero temperature in two dimensions were done by Rottman and Wortis (100) and Avron et al. (101). Later, Wortis and co-workers (102, 103) presented a three-dimensional model for a simple cubic lattice within the mean-field approximation with nearest-neighbor (NN) and next-nearest-neighbor (NNN) interactions. They exploited the thermodynamic conjugacy of the Wulff plot and the equilibrium crystal shape to give interfacial phase diagrams. Sharp edges or points on the crystal shape correspond to first-order phase transitions, while smooth joining of curved and faceted regions corresponds to second-order phase transitions. As illustrated in Figs. 7 and 8, the equilibrium crystal shape at nonzero temperature consists of facets and smoothly curved surfaces. When N" interactions are attractive, only second-order transitions occur. When NNN interactions are repulsive, first-order transitions and tricritical behavior also occur.

2. Kinetic Model Binh et al. ( 1 04, 105) performed experimental investigations of surface diffusion under ultra-high vacuum and found that free evaporation must also be considered for temperatures higher than about 0.65 T, (the melting point) for clean surfaces. They presented a thermal grooving model (106-108) in which free evaporation was considered simultaneously with surface diffusion. In this

380

TA-CHIN WE1 AND JONATHAN PHILLIPS

e(h Wulff Plott

b

Crystal Shape (equatorial (perspective) plane)

-@-

T=O

E l

OCTCTR~

T,


T

R1

e3


FIG.7. Detailed models of surface free energies based on quasi-chemical metal-metal interactions allow detailed Wulff plots, and hence particle shapes, to be predicted as a function of temperature. (a) Interfacial phase diagram for simple cubic lattice model with nearest-neighbor and next-nearest-neighbor attraction. (b) Representative Wulff plots and equilibrium crystal shape of (a) (103).

case, Mullins’ theory (48, 50) is found to be invalid. As shown in Fig. 9, the groove profile evolution is not steady-state anymore. The groove depth as a function of time no longer follows the t”4 rule, but rather, increases more slowly and tends asymptotically to a constant value. This is in agreement with experimental results. Recently, the kinetics of faceting was studied using a Monte Carlo simulation ( I 09). Evaporation-condensation and surface diffusion were considered for the simple cubic crystal with nearest- and next-nearest-neighbor interactions. It was

38 1

THERMAL AND CATALYTIC ETCHING

a

rough

F

rough

T3

n

0

d4

PI2

Crystal Shape (perspective)

0 Tu
FIG. 8. (a) Interfacial phase diagram for simple cubic lattice model with nearest-neighbor attraction and next-nearest-neighbor repulsion. (b) Representative Wulff plots and equilibrium crystal shape of (a) (103).

found that, in contrast to Mullins’ theory (48, SO), a power law cannot describe facet growth. The exponent n depends on surface temperature, crystallographic orientation, interatomic potential, and the operating mechanism. In all cases, three time regimes can be distinguished in the development of facets: (1) disordering of surface atoms as surface energy and surface roughening increase, (2) microfacet nucleation, and (3) coarsening of microfacets into macrofacets by thermal fluctuations. Faceting is accelerated as temperature increases, but thermal roughening dominates at high temperatures. Surface diffusion is the dominant process at short times and for small facets, but evaporation-condensation becomes important at long times and for large facets. Growth and etching

382

TA-CHIN WE1 AND JONATHAN PHILLIPS

a

b xlo.2

5 -only

I

)

-

surface diffusion

+

free evaporation

-

0

0.5

1

m(Bt )1/4

x 1 0.5

o

1

2

3 (Bt)’/4

4

5 ~10-5

FIG. 9. (a) Groove depth evolution curves. (b) Groove width evolution curves. When free evaporation and surface diffusion are considered simultaneously, the groove profile evolution increases more slowly and tends asymptotically to a constant value (108).

promote faceting under conditions close to equilibrium, but they induce kinetic roughening under conditions far from equilibrium. Some cases of etching processes are difficult to define as either catalytic etching or thermal etching, as exemplified by extensive investigations of the oxidation of CO on Pt(ll0) by Ertl’s group (110-116). Faceting is never observed as a result of exposure of the Pt( 110) surface to either CO or 02. However, in certain oxygedC0 mixtures, a stationary spatial pattern is observed. This consists of facets of (430) orientation with a lateral periodicity of 10-20nm (115, 117). When the supply of either of the reactants was stopped, the (1 10) surface was restored. The fact that faceting takes place only in the presence of a gas mixture suggests catalytic etching. However, no chemical reaction in the gas phase is required, and the final structure is simply a faceted surface, suggesting thermal etching. Thus, this is an appropriate example to discuss at the interface of the thermal/catalytic etching sections of this review. A Monte Carlo simulation of the faceting of Pt(ll0) in the presence of 02 and CO was carried out by Imbihl et al. (118) to account for the CO-induced 1 X 1 * 1 X 2 phase transition and the enhancement of O2 adsorption at step sites. A Langmuir-Hinshelwood mechanism, CO diffision by site hopping, and a single attractive interaction energy between horizontally neighboring rows were all considered in the simulation. The key step in the faceting of Pt( 1 10) is the CO-induced 1 X 1 1 X 2 phase transition. The mass transport in this process, which involves 50% of the surface atoms, creates steps. The sticking coefficient of oxygen adsorption is higher at step sites than at terrace sites. This preferential adsorption leads to facet growth. However, the faceting process

-

383

THERMAL A N D CATALYTIC ETCHING I1101

t

Number of Cycle5

IlfOI

f

i

m

1000 5- 2

z- -

3000

Q

3

1 1500 / 7300

increases roughness and thus creates a surface with higher energy. This leads to a competition with thermodynamic forces which tend to produce a flat surface, as such surfaces have the minimum surface energy. That is, while kinetic instabilities work in favor of steep facets, thermal reordering tends to reduce the surface profile. In effect, this leads to the selection of facets with a certain size and orientation depending on the reaction conditions. As shown in Fig. 10, the kinetic roughening leads to the development of facets as observed experimentally. A steady state is reached after 4000 Monte Carlo cycles. When the reacting gases are stopped after 4000 Monte Carlo cycles, thermal reordering then leads to the reappearance of the flat surface, as shown by the profile recorded after 6000 Monte Carlo cycles.

111.

Catalytic Etching

Catalytic etching is the name given to the gross morphological changes of metal surfaces (e.g., catalysts) that occur in reactive gas atmospheres. Changes include the formation of nonequilibrium surface structures such as pits, excrescences, and particles; dramatic weight loss; deep groove formation; and even “metal wool” formation (119). It has been shown that while nonequilibrium structures form on the surfaces of foils in reactant mixtures, there is no significant change in the surface structure in any single-component product or reactant gas (35). Entire metal films have been removed under reaction

3 84

TA-CHIN WE1 AND JONATHAN PHILLIPS

conditions, whereas in each of the separate product and reactant gases there has been no detectable loss of metal (120). The nature of the etching is also a fimction of the reaction conditions. For example, pitting and fuzing of platinum result from ammonia oxidation, whereas in the same reactor deep grooves form during propane oxidation (27). In both processes, massive metal loss occurs. This type of gross structural transformation is properly considered catalytic etching. In contrast, during HCN synthesis (31, 121) broad facets andor excrescences form on the metal surface, but there is little if any metal loss. The formation of pits has often been associated with hydrogentoxygen reactions (122, 123). In general, processes that lead to facet formation, without metal loss and without the formation of clearly nonequilibrium structures, are best characterized as thermal etching processes. Historically, catalytic etching has been a poorly understood phenomenon, and it has not always been clearly distinguished from thermal etching. Moreover, the largest number of studies have focused on platinum (3, 27, 34, 124-130) although the catalytic etching of silver (122, 131, 132), copper (16, 28, 133), nickel (134), and some platinum alloys (51, 121, 135-138) has also been investigated. A.

MODELS

Two principal theories exist regarding the mechanism of catalytic etching. One school of thought is that catalytic etching takes place primarily by the same mechanisms that account for thermal etching. That is, the etching results from surface diffision of species such as platinum oxide or the evaporation of similar species. Essentially the only difference between such theories and the theories of thermal etching is the inclusion of temperature gradients. It is postulated that these temperature gradients, which are produced by an uneven rate of surface reaction, drive the structure toward a nonequilibrium configuration (2, 3, 139-141). The second theory of catalytic etching is that free-radical species produced by homogeneous reaction processes are responsible for etching (35, 120, 142). The radicals interact with the metal in the surface to form volatile, metastable intermediates. The free-radical species are present only in reaction mixtures, thus accounting for the fact that the etching is clearly not the weighted sum of the actions of the individual gases. Models of catalytic etching that are similar to models of thermal etching are open to a number of criticisms. For example, it is unclear how the volatile metal oxide transport model explains the different platinum surface structures formed in the same reactor (see above). Nor is it clear why significant metal loss occurs during ammonia oxidation (143, 144), but not during HCN synthesis (31, 32, 1 4 3 , even though the oxygen concentrations are similar in the two cases. Furthermore, it is not clear why significant metal loss or etching has never

THERMAL AND CATALYTIC ETCHING

385

been detected in pure oxygen or pure ammonia atmospheres (34, 146), or why weight loss is not observed in pure oxygen atmospheres. The second model appears to be generally more consistent with the available data regarding catalytic etching as well as with a general understanding of the chemistry of free radicals. For example, free-radical formation is a fhction of the reaction mixture; thus the character of the etching is expected to vary with the reaction mixture. There is no need to postulate the existence of uneven rates of reaction or any temperature gradient. In fact, catalytic etching has been observed to take place in environments in which it is not likely that either kind of gradient exists on the catalyst surface (35,120). Moreover, the fact that free radicals can volatilize metals has been long known. In addition to the above models it is clear that “pitting” often occurs as a consequence of subsurface bubble formation. In particular, it is clear that bubbles, leading to pitting, often result from subsurface interactions between hydrogen and oxygen to form water (122, 123). This process is also known as “embrittlement,” and it has been extensively studied (147-150). Although this phenomenon is often associated with reaction mixtures, it is not clear that it is true catalytic etching. In fact, it has been repeatedly shown that the same process takes place if hydrogen and oxygen are flowed over the surface either sequentially or concurrently. B. EARLYINVESTIGATIONS It has been recognized for about a century that metal surfaces can be grossly modified in the pressure of reactive mixtures. Yet there are surprisingly few studies in which adequate data and controls are available. In many of the early investigations, qualitative and subjective evaluations were made of surface structure following reaction treatments. For example, Bone and Wheeler (131) described a silver surface as “frosted” in a 3 : 1 H2 : O2 mixture over a narrow temperature range (497-547°C). In Langmuir’s study of CO oxidation, he noticed that the platinum filament underwent a progressive change to become a better catalyst (151) and that the change in the structure of the surface was brought about by the reaction (152). The “qualitative” character of these speculations can be attributed to the fact that electron microscopes were not used for characterization of catalytic etching until 25 years ago. The inadequacy of control experiments cannot be attributed to the lack of availability of electron microscopes. Yet only rarely were control experiments conducted to demonstrate that no single reactant or product gas was responsible for the observed surface modification. Furthermore, few investigations were conducted in which weight change was measured. Even more rare were investigations designed to determine the influence of variations in reaction parameters on the nature of the etching process.

386

TA-CHIN WE1 A N D JONATHAN PHILLIPS

A good example of an early report on catalytic etching-a report that contains valuable observation but lacks appropriate controls-is that by Armstrong and Hilditch (134) concerning the etching of nickel catalysts during the hydrogenation of natural oils. They reported that hydrogenation at temperatures of ca. 200°C in the presence of finely divided, but unsupported nickel catalysts invariably led to the formation of fine colloidal suspensions of nickel in the liquid product. This nickel, which significantly darkened the liquid, was so fine that it could not be removed by ordinary filtration. In contrast, no nickel was found in the liquid phase in contact catalysts that were inactivated due to poisoning. It was also reported that even supported nickel catalysts led to the formation of colloidal suspensions of nickel in the liquid product; however, in this case the amount of nickel was far less, and it could be filtered out by passing it through a bed of the original support material. These results led the workers to suggest that catalysis actually leads to the removal of surface nickel atoms, primarily due to local heating which takes place at the reaction site. Furthermore, during the catalytic process, the nickel atom is temporarily part of a liquid- or gas-phase intermediate. Once the catalytic process is complete, the authors postulated that the “free” nickel atom readsorbed onto the bulk nickel, adsorbed onto the inert support, remained as “nickel sol” in the liquid, or continued to act as a catalyst. It was claimed that this model explained several observations, such as the differences between unsupported and supported nickel. The supported metal has a greater surface area upon which the metal can readsorb, so it tends to leave fewer atoms in the product liquid. The model also explains the observation that the reaction vessel became coated with a thin film of nickel after lengthy use. This postulated etching mechanism is similar to the recent model discussed above, whereby etching results from free-radical-surface interactions. A second good example of early work that lacked adequate controls is the extensive set of studies of the catalytic etching of copper conducted by Gwathmey et al. (28, 133, 153, 154). They found that in hydrogedoxygen mixtures copper restructured far more dramatically than it did in any singlecomponent gas. Although these early reports contain a great deal of valuable information, there was clearly no attempt to conduct systematic control studies, nor to evaluate the influence of reactant gas composition. In one extensive study conducted by Gwathmey’s group (28), it was shown that there is preferential faceting and roughening of particular crystal planes of copper single crystals and polycrystals when they are in hydrogedoxygen mixtures at about 450°C. The focus of the study was on the roughening of Cu( 100) and (1 10) surfaces in reaction mixtures containing low oxygen concentrations (ca. 2.5%). In reaction mixtures a large increase in surface area was noted, as well as the apparent formation of rough (not faceted) surfaces, which led the authors to the conclusion that there is a fundamental difference between

THERMAL AND CATALYTIC ETCHING

387

the restructuring that takes place in reaction mixtures and that which takes place during thermal etching. In addition to the focus on the roughening of certain crystal planes in gas mixtures, two additional observations were recorded, which are particularly intriguing given the recently proposed gas-phase free-radical model of catalytic etching. First, a considerable amount of copper was found on the walls of the glass reactor, many centimeters from the copper single-crystal sample, even when the experiment was carried out at only 400°C. In contrast, heating in hydrogen, even at IOOO”C,did not produce any evidence of metal loss. The explanation given for the copper transport under reaction conditions was similar to that given for hydrogen embrittlement (148-151, 155). Water is produced by reaction between dissolved hydrogen and oxygen in “pores.” The water thus formed “expands,” causing metal to be “expelled.” Even the authors note that this explanation is not entirely satisfactory given that the samples are generally single crystals. Second, “red powder,” which could easily be removed with a cloth, was found to form and cover the sample surface after many days of reaction at temperatures as low as 300°C. The powder formed only if the oxygen concentration was greater than 5%. Unfortunately, no analysis of the composition or phase of the powder was reported. Some of the clearest and most unambiguous reports of catalytic etching are associated with the industrial use of woven noble metal screens. These screens are used as catalysts in two high-volume, high-temperature reactions: the oxidation of ammonia (the first step in the production of nitric acid) and the formation of hydrogen cyanide. For example, in early reports Gillespie and Kenton (146) showed that during the oxidation of ammonia, Pt/Rh (10-20%) screens are severely etched, as evidenced by gross restructuring occurring within minutes. They reported that the wires in the woven screens nearly doubled in diameter (Fig. 11) due to the formation of metal nodules on the surface. They also reported a slow net volatilization of the metal, such that on average approximately 0.01 troy oz. of noble metal was lost per ton of nitric acid produced. Even this rate of noble metal loss is economically significant. These authors suggested that the mechanism of etching is the formation of precious metal oxides (primarily of platinum) under reaction conditions and the subsequent volatilization of these oxides. Although it was noted that the same rate of metal loss and the same gross transformation did not take place in a pure oxygen environment, no attempt was made to reconcile these observations with the etching model. The need to reduce metal losses during ammonia oxidation led to the development of a number of specialized technologies. The first was the substitution of alloy gauzes for pure platinum gauzes. Handforth and Tilley (143) demonstrated that platinum gauzes containing 10-20% Rh lost far less metal than pure platinum screens and were actually somewhat more active than pure

FIG.1 I . Used WRh gauze (10% Rh) catalyst with rearranged surface structure (20X). Loss of catalytic metal and catalyst gauze strength during ammonia oxidation is a serious industrial concern (146).

THERMAL AND CATALYTIC ETCHING

389

platinum screens. A number of other alloys, both platinum and palladium, were tried; but all were rejected either because they had reduced catalytic activities or because they lost metal and mechanical strength even faster than pure platinum. Several other methods are described by Connor (159, including the use of downstream filters and the use of gold screens downstream of the reaction zone which interact with volatilized platinum and rhodium to form metal alloys. Even recently there have been investigations of metal loss during ammonia oxidation, and new technologies are periodically introduced to improve the catalytic screens. For example, Fierro et al. (157) recently studied the long-range transport of platinum and rhodium metal from alloy screens used for ammonia oxidation. Johnson Matthey, which manufactures many of the industrial catalysts, recently developed a “knitted” (rather than woven) screen technology which it claimed enhances screen strength, allowing use of screens with less of the expensive rhodium and enabling plants to run for longer periods without downtime (158). There were few early studies of structural change in screens during HCN synthesis. In one study, however, it was claimed, counter to later literature (31, 32, 145, 159, 160), that during industrial HCN synthesis at approximately 1000°C there is significant metal loss (147). This is attributed to the volatilization of platinum oxide. It is clear that if the temperature is high enough, platinum oxide will volatilize, causing net metal loss. Platinum oxide has a fairly high vapor pressure at elevated temperatures (161), and it has been shown to transport significant quantities of platinum (14, 121, 162, 163). This mechanism of platinum metal loss is not in dispute; only the temperature at which it becomes significant.

C. MODERN INVESTIGATIONS It was not until the early 1970s that systematic studies were conducted of the relationships between reaction gas composition and the character of catalytically etched surfaces. Many studies focused on systems of industrial interest. Thus, ammonia oxidation and hydrogen cyanide synthesis catalyzed by platinum and platinum-rhodium wires and foils were investigated extensively. Etching in other reaction mixtures was examined as well (e.g., ethylene/oxygen and hydrogedoxygen) in hopes that these systems would provide general insight into etching. Two major models of the mechanism of catalytic etching emerged. According to one model, catalytic etching results from volatile metal oxide transport. The final structures are influenced by temperature gradients as well. According to a second model, etching results from interaction between homogeneously formed free radicals and the metal surface. The interactions are postulated to result in the formation of volatile metal complexes. In turn, these lead to metal loss, metal particle formation, etc. In the following discussion, the

390

TA-CHIN WE1 AND JONATHAN PHILLIPS

examples are divided according to which of these two models was presumed to explain most observations. In some cases, the metal oxide model was found to be insufficient, yet no alternative was proffered; these examples are included in the metal oxide transport section. 1. Metal Oxide Transport Model Schmidt and Luss (33) were among the first to use electron microscopy to characterize morphology of alloy screens (Pt/Rh 90/10) after prolonged use as either ammonia oxidation or HCN catalysts. They found that in both reactor types there was extensive reconstruction of the catalyst screens. The original wires were smooth, but after lengthy use the surfaces were covered with highly faceted crystal protrusions. Only under ammonia oxidation conditions was there extensive metal loss. In commercial applications similar amounts of oxygen enter both reactors, although the partial pressure of oxygen leaving the HCN reactor, while not negligible, is lower than that leaving the ammonia reactor. The authors pointed out the difficulty of settling on a single mechanism of etching. They noted that the absence of metal loss during the higher-temperature HCN process and the absence of similar restructuring in any single-component gas indicate that the vapor-phase transport of platinum oxide is a possible, but not a certain, mechanism. In a classic study, Schmidt et al. (139) found that the nature of the etch structures formed on a platinum wire surface during ammonia oxidation was a function of both the temperature and the reacting gas composition. In pure oxygen streams at temperatures less than 1200”C, the surface remained smooth. At higher temperatures only a limited number of pits formed. In pure ammonia there was limited pitting, and rounded facets formed. In contrast, in reaction mixtures, severe restructuring of the wire surfaces took place. In excess ammonia, curved surfaces were found to cover the entire sample over a wide range of conditions. At temperatures greater than 1000°C extensive pitting took place. In excess oxygen fewer pits were formed, but rounded facets were found in all cases. The authors used this information to produce a phase diagram (Fig. 12) of the types of reconstruction observed as a function of temperature and gas-phase composition. Schmidt et al. (139) postulated that in the presence of excess oxygen, platinum was transported as volatile oxides through the gas phase and boundary layer. This mechanism could not adequately explain the reconstruction observed far into the excess ammonia regime. It was suggested that under these and other conditions, other volatile platinum species formed. Moreover, these species might decompose by reaction in the boundary layer, leading eventually to the platinum “replating” itself. Other groups, although accepting many aspects of the platinum oxide transport model, also concluded that it cannot explain all the observations. For

THERMAL AND CATALYTIC ETCHING 1500

39 1

r

1300f Pits

h

0 e I-

gOOlz. . E 700

'I

excesa 4

/ Sloichiomelric region

excess NH3

'\

b .

'.

.

example, Pielaszek (164) pointed out that the model does not explain increased metal loss under reaction (ammonia oxidation) conditions. He also believed that etching resulted from the preferential removal of platinum from high-index planes, leaving faceted, low-index structures. He suggested that the cauliflowerlike structures that covered the surface of the gauze probably formed by the redeposition of free platinum atoms from the gas phase. In a later study by the Schmidt group (27), electron microscopy was used to characterize morphological changes in microspheres (<0.6 cm in diameter) of Pt, Rh, Pd, and Pt-Rh alloy in a number of reaction environments; the reactions were ammonia oxidation, ammonia decomposition, and propane oxidation. No other experimental techniques, such as weight-loss measurements, were employed. After prolonged exposure to reaction mixtures of ammonia and air at temperatures less than 727"C, the surfaces of the spheres were reconstructed to favor specific crystal planes. The structure of the facets was found to be a function of the reaction mixture, temperature, and metal (Fig. 13). In the same reaction mixtures, as well as in pure ammonia at higher temperatures

392

TA-CHIN WE1 AND JONATHAN PHILLIPS

FIG. 13. The structure of Pt spheres changes dramatically as a function of the gas composition and temperature. Micrographs of Pt spheres in reaction mixtures. (a) After exposure to 10% NH, in air at 1350°C. the Pt sphere appears pitted. (b) Irregular crystalline structures build up on Pt sphere exposed to 10% CO in air at 1000°C. (c) Deep wavy channels form on Pt sphere used in 2% C3Hs oxidation. (d) Carbon particles deposited on platinum used in 14% C,Hs in air mixtures at 727°C (27).

(ca. 1330"C), the surfaces of the spheres were found to be unfaceted but covered with small pits. In propane/air mixtures (temperature not available) clearly nonequilibrium surfaces formed. For example, in excess he1 mixtures the surfaces were covered with small spheres about of 0.3 pm in diameter. These were presumed to be carbon. In excess oxygen mixtures deep wavy channels formed. In CO/air mixtures at 977°C irregular, rounded (perhaps particulate) surfaces formed Although there was clear evidence that catalytic etching took place, there were few control experiments and no quantitative characterization of the process. Two different explanations were offered for the observations, but both relied on the assumption that metal is transported as platinum oxide. The first is based

THERMAL AND CATALYTIC ETCHING

3 93

on the idea that reaction rates vary from one position to another, and thus the oxygen partial pressure is high near areas of high catalytic activity (where platinum oxide formation is relatively rapid) and low near areas of low activity (where platinum oxide formation is thus relatively slow). The presumed gradients in PtO, concentration on the surface lead to a net metal transport via surface diffusion. According to the second model, platinum oxide forms preferentially at certain locations on the surface and is volatilized, leading to net metal loss. Other workers ( I 65) used X-ray photoelectron spectroscopy (XPS) to examine the influence of ammonia oxidation on the surface composition of alloy gauzes. Afier several months on stream, the surface was covered by the same types of highly faceted structures noted by others. As illustrated in Fig. 14, XPS analysis provides evidence that the top microns, and in particular the top 100 A of the surface, were enriched in rhodium. This enrichment was attributed to the preferential volatilization of platinum oxide. The rhodium in the surface layers was present in the oxide form. Other probes confirm the enrichment of the surface in rhodium after ammonia oxidation (166). Rhodium enrichment has been noted by others (164, 1 6 9 , and it has been postulated that in some cases it leads to catalyst deactivation (168). Anderson (160) investigated the etching of Pt/Rh 90110 and Pt/Rh/Pd 901515 gauzes as a function of pressure and time on stream (no temperature data are available) under ammonia oxidation conditions representative of commercial reactors. Changes in surface area were measured with cyclic voltammetry and changes in gauze structure were observed by electron microscopy. The surface area of the layer of the bed nearest the inlet increased dramatically and quickly (e.g., by a factor of 30 within 20 h). Layers farther from the inlet showed slower surface area increases, and the final surface areas were not as large as those at the inlet. Electron microscopy showed that the increase in surface area is not due to faceting, but rather to the formation of “sprouts” on the surface (Fig. 15), which caused the wire diameter to increase about twofold. Anderson also accepted the model of etching by platinum oxide formation and volatilization, yet the results clearly demonstrate that the surface area increase and etching rate correlate with local reaction rate and not oxygen concentration. No control experiments (e.g., with the sample in pure oxygen) were conducted to test the model. One caveat pertains to the platinum oxide transport model: It does not appear to be able to explain the differences in metal weight loss during ammonia oxidation and hydrogen cyanide synthesis by the Andrussow process. In the Andrussow process a mixture of methane, ammonia, and air is used to maintain a high temperature ( 12OOOC) while generating hydrogen cyanide. Alternative processes require energy input, because HCN synthesis is an endothermic reaction. Thus, in both ammonia oxidation and HCN synthesis, platinum or alloy

394

TA-CHIN WE1 AND JONATHAN PHILLIPS

EUlN

a

et4d

4d 512 Pt

~

"

EBev

'

~

315

3d 312

"

'

'

'

310

'

'

I

305

1 Rh 3d 312 I 7 1 2

C

1

10

102

103

.

104

mn) FIG. 14. Rhodium enrichment of a Pt/Rh gauze (10% Rh) after ammonia oxidation. (a) Photoelectron spectra (Rh 3d and Pt 4 d ) of a moderately oxidized gauze. (b) In-depth analysis of a used gauze. One minute of sputtering is roughly equivalent to 20 A (165).

gauzes are exposed to substantial oxygen partial pressures at elevated temperature. During ammonia oxidation there is considerable weight loss but virtually none occurs during hydrogen cyanide synthesis (31, 32, 145, 159, 160). Furthermore, during HCN synthesis there is massive restructuring of the gauzes, the reconstruction being similar to that following ammonia synthesis. In both cases the wire thickness increases, reducing the aperture between strands for gas

FIG.IS. In a gauze-containing reactor the chemical composition of the gas phase changes as a function of depth in the reactor. This is reflected in the changing character to the catalytic etching. Electron micrographs of Pt/W alloy gauzes after ammonia oxidation: (a) first layer, (b) fifth layer, (c) tenth layer (160).

396

TA-CHIN WE1 AND JONATHAN PHILLIPS

flow. In both cases the surfaces are covered with protuberances (given a variety of names: e.g., “cauliflower,” “highly faceted structures,” “sprouts,” “nodules,” “excrescences”), which are often found to be highly porous networks of small faceted particles (32, 145, 159). One difference between the two etching processes (other than the rate of weight loss) is that during HCN synthesis the gauzes sinter into a solid mass (layers can no longer be separated) and become brittle, whereas following ammonia oxidation the gauzes are separable and maintain some ductility (33. 159). To explain the fact that weight loss occurs during ammonia oxidation but not during HCN synthesis, Knapton (159) suggested that the mechanism of etching is different in the two cases. In the case of HCN synthesis he suggested that active platinum atoms are created by the heat released at the surface by exothermic reactions. These activated atoms diffise on the surface (not in the gas), coming to rest at low-energy sites, normally at the edges of high-index planes. The net result is the creation of surfaces with low-index planes. In sum, according to one postulated mechanism of catalytic etching, particularly of platinum, etching occurs via the transport of volatile metal oxide species. It has always been understood, even by the model’s proponents, that it does not fully explain the observations. It has been repeatedly suggested that other volatile species may exist, or that other unaccounted-for processes are responsible for the observed behavior. Indeed, as discussed below, several recent studies suggest that other volatile species and other unaccounted-for processes may very well explain some or all catalytic etching. 2. Free-Radical Metal Transport Model In recent years a new model of catalytic etching has emerged. This model attributes etching to the interactions between homogeneously generated free radicals and metal surfaces. As discussed below, this model is very similar to models devised to explain “etching” encountered in other (non-reaction gas) environments. Phillips et al. (142) conducted extensive studies of the gross changes that take place in the structure of platinum thin films and foils in ethylene/oxygen (35, 120) and hydrogedoxygen mixtures. The phase diagrams observed as a function of sample temperature and gas phase composition are shown in Fig. 16. Gross structural changes take place over a relatively narrow temperature range. For example, in ethylene/oxygen mixtures. clear examples of metal volatilization are detected only between 500 and 700OC. The nature of the reconstruction is clearly distinguished from that observed in thermal etching. For example. complete stripping of platinum from metal thin films was observed, but only in reactive mixtures. In contrast, in any single-component gas or under vacuum, only limited sintering was detected. The nature of the observations led to the

397

THERMAL AND CATALYTIC ETCHING

I

A

A A

L1-L 700

800 900 1000 Sample Temperature ( O K )

b A 0 0

"

I

500

.

I

600

.

I

700

.

0 0 0

I

800

.

I

900

.

I

1000

Sample Temperature (K)

FIG. 16. Catalytic etching is a function of temperature and reacting gas composition. (a) Platinum was found to etch only over a narrow temperature range in ethylene/oxygen mixtures. (b) Similar behavior was observed in hydrogedoxygen mixtures. The character of the surface found after 40 or more hours at the marked conditions is given by symbols: (A) faceted surfaces; (0) platinum particles on surface; ( 0 )carbon deposits and platinum particles; (0)thin carbon deposits, without platinum particles. (Reprinted with permission from (35, 142). Copyright 1985 and 1988 American Chemical Society.)

development of the free-radical etching model described earlier, as well as to additional work on etching of alloys, sintering of supported particles in reaction mixtures, and etching of metals by plasma-generated free radicals.

398

TA-CHIN WE1 AND JONATHAN PHILLIPS

In an early investigation of the etching of platinum foils in ethylene/oxygen mixtures (35), it was discovered that there are five clearly distinguishable regions of etching in the phase space defined by temperature and the gasphase oxygedethylene ratio. At sample temperatures less than approximately 480°C (region I) the foils simply facet. The facet patterns are indistinguishable from those which form the presence of single-component gases. This conclusion was based on extensive control studies, conducted over a broad range of temperatures. Control studies were conducted not only of etching in a vacuum, but also independently in each the individual gases involved in the reaction, including carrier gas, reactants, and products: Nz,0 2 , C2H4, H20, and COz . These investigations clearly showed that only limited faceting takes place in any individual gas (including oxygen) at any temperature less than 827°C. In contrast, in reactive gas mixtures major transformations of the surface take place. At sample temperatures between approximately 480 and 680"C, in ethylene/oxygen mixtures in which oxygen was present in greater than stoichiometric amounts (O2/C2H4 > 3), the samples became covered with relatively large platinum particles (average diameter approximately 1 pm) (region 11). In excess fuel mixtures (02/C2H4 < 3) over the same temperature range (region IV), the sample surfaces were covered with a thick layer of carbon in which a high concentration of small platinum particles (diameter < 0.1 pm) were embedded (Fig. 17) (35, 169). At temperatures exceeding 677°C in excess oxygen (region 111), the foils were again found to be simply faceted. In excess fuel at the same high temperatures (region V), the foils were faceted but also partially covered with carbon. It was concluded that the structures that formed in regions I1 and IV could not be attributed to the sum of the etching effects of the individual gases. There was a synergism resulting from the mixing of gases. Moreover, this synergism existed only over a limited range of temperature. It was postulated that the synergism resulted from the gas-phase formation of a particular free radical, perhaps methylene (33, 170-1 72), which was formed as part of the principal mechanism for homogeneous ethylene oxidation over a limited temperature range. It was also suggested that the hydroperoxyl radical might be responsible. Similar postulates of free-radical-initiated processes have been made to explain other phenomena, such as the formation of soot, which forms only over the range of temperatures at which a particular free radical is produced (I 73-176). Another study of the same system was conducted, with the focus on thin silica-supported platinum films in an impinging jet flow (120). It was shown (Fig. 18) that in regions I1 and IV all of the platinum directly below the impinging jet was removed from the thin film. Large platinum particles (ca. 1 pm in diameter) were found downstream on top of regions of unetched platinum film several millimeters from the area of impingement of the jet. Many

THERMAL AND CATALYTIC ETCHING

399

FIG.17. The character of etching of platinum foils in ethylene/oxygen mixtures is a function of gas phase composition. Under excess fuel conditions between 500 and 700°C, carbon films containing platinum particles form on top of the foil. TEM micrograph of a carbon film formed after 40 h of treatment at 587°C in a fuel excess 0&H4 mixture (region IV). (Reprinted with permission from (35).Copyright 1985 American Chemical Society.)

platinum particles were also found as far as several centimeters from this area on top of silica in regions where no metal was initially present. In region IV, platinum etching was accompanied by the deposition of thick carbon films. In all , other regions and in the presence of each of the individual gases (N2, 0 2 C2H4, H20,C 0 2 ) , no metal transport and no large particle formation was observed. Only limited sintering, quantitatively identical to that occurring under vacuum at the same temperature, was observed. Again, particular attention was paid to the influence of oxygen, and again no evidence was found for any difference between the influence of oxygen, any other gas, or vacuum. To determine the plausibility of the postulate of hydroperoxyl radical intermediates, a study was conducted of the effect of hydrogedoxygen mixtures on the same types of films in the same impingingjet reactor (142). Detailed kinetics suggested that at low temperatures (ca. 480°C) in either ethylene/oxygen or hydrogedoxygen mixtures the hydroperoxyl radical is the dominant free radical present during combustion. Moreover, it is clear that no hydrocarbon radicals were present under reaction conditions. Thus, if etching were observed in a

400

TA-CHIN WE1 AND JONATHAN PHILLIPS

FIG.18. Etching in ethylene/oxygen mixtures of platinum thin films on Si02 substrates in an impinging jet reactor dramatically illustrate that metal is volatilized under some conditions: Hole formation in (a) region 11 and (b) region IV (see Fig. 16). In all other regions of the phase diagram and in any single product or reactant gas no metal was removed from the films (120).

hydrogedoxygen mixture, it would seem unlikely that any hydrocarbon free radical (e.g., methylene) would be responsible for the etching observed in ethyleneloxygen mixtures.

THERMAL AND CATALYTIC ETCHING

40 1

Studies of etching in hydrogedoxygen mixtures appeared to support the freeradical-induced etching postulate. Again it was discovered that etching occurred only in the presence of reaction mixtures and only over a limited range of temperatures and gas compositions. Specifically, etching took place only at temperatures greater than about 450°C and in mixtures in which the oxygen/ hydrogen molar ratio was greater than about 2/l. The nature of the etching was very similar to that observed in ethylene/oxygen mixtures. Metal was removed only in the area of impingement of the jet, and many large particles of platinum were found downstream. The only significant difference was that a band of large particles (cu. 5 pm in diameter) formed at the edge of the region from which all metal had been removed (Fig. 19a). A band of particle clusters was found just beyond the large-particle band. These clusters were clearly composed of smaller particles about 1 pm in diameter (Fig. 19b). The similar nature of etching in ethyleneloxygen and in hydrogedoxygen mixtures suggested that the same free radical was responsible for etching in both cases. Thus all carbon-containing free radicals were eliminated as candidates. Moreover, recent investigations of the combustion of hydrogen (I 77-183) and of ethylene (183) suggest that at low temperatures the dominant reaction mechanism involves the hydroperoxyl (H02) radical. Earlier suggestions that the methylene radical dominates the oxidation process at low temperatures are no longer considered valid. To explain the particles that formed in both the ethylene/oxygen and hydrogen/oxygen mixtures, it was postulated that they form in the gas phase and that the overall etching process takes place in three steps. First, free radicals are formed homogeneously in a boundary layer adjacent to the surface. Second, these radicals interact with metal atoms in the surface. This interaction results in the formation of volatile intermediates. Third, the metastable, volatile intermediates interact in the gas phase so that metal particles are formed and stable product molecules released. Individual metastable species presumably interact with each other and also with particles formed from multiple collisions. The larger particles interact with each other as well. The third step in the model of particle growth in the gas phase is very similar to that describing the growth of soot (I 73-1 76). The general mathematical analysis of this type of particle growth was first developed by Smoluchowsky (184-186) and is also used to describe other processes, such as particle sintering. Next it was demonstrated that the nature of the sintering of platinum particles in Pt/Si02 is different in ethylene/oxygen mixtures than in any single-component gas (82). Specifically, it was found that sintering was dominated by Ostwald ripening (single-atom transport) under etching conditions and by coalescence (whole-particle transport) growth under all other conditions. Some (187, 188) have also suggested that particle sintering is greatly accelerated under

FIG. 19. Particle structure on catalytically etched Pt thin films exposed in H2/02 mixture is a function of position relative to the impinging jet. (a) Just beyond the area from which platinum is fully removed a ring of large faceted single crystals is found. (b) Beyond the ring of large single crystals lies a ring of particle-agglomerate structures. It is suggested that both structures are initially composed of small particles which formed in the gas phase and precipitated onto the surface. Particles in the outer ring are agglomerates of these small particles. Particles in the inner ring are re-etched agglomerates. (Reprinted with permission from (142). Copyright 1988 American Chemical Society.)

THERMAL AND CATALYTIC ETCHING

403

reaction conditions. It is possible that these are also examples of free-radicalenhanced transport. X-Ray diffraction was used to investigate silica-supported platinum particles after treatments in either single-component gases or after treatment in reaction regimes known to result in etching. The X-ray Pt( 111) peak was deconvoluted after various treatments to yield the platinum particle size distribution. It was found that in any single-component gas or under vacuum, the particle size distribution changed as anticipated for coalescence growth. In short, it appears that the smaller particles in the distribution coalesced to form larger particles. The large particles, because of stronger interactions with the support, diffuse much more slowly and are rarely involved in coalescence. In contrast, under reaction conditions known to result in etching, only the larger particles grew. The number of smaller particles decreased, but there was no indication of growth. This type of behavior is consistent with Ostwald ripening. During Ostwald ripening there is atom-by-atom transport from smaller particles to larger ones, driven by the net difference in vapor pressure as a hnction of particle size. These results support the hypothesis of etching by free radicals. Only under reaction conditions is it expected that there will be a mechanism for atomic transport resulting in Ostwald ripening, whereas in other environments only coalescence growth is anticipated. The results of other studies add support to the free-radical hypothesis. For example, the etching of Pt/Rh (904 0) gauzes in hydrogerdoxygen mixtures occurred only over the same range of reaction conditions under which pure platinum is etched (51). Etching was demonstrated by two different phenomena. First, in a simple flow reactor, operated so that rapid mass transfer away from the metal surface was favored, the gauzes underwent dramatic weight loss, but only under the known etching conditions. No weight loss was measurable in any pure gas, or under reaction conditions which were previously shown not to etch pure platinum. Second, in an impinging jet reactor, which does not favor mass transport away from the metal, there was massive restructuring of the surface, but only under conditions also known to etch pure platinum. Very little weight loss was detected. The observed reconstruction of Pt/Rh (90/10) gauzes in hydrogetdoxygen mixtures provides additional insight into the etching process. In the flow reactor in which significant weight loss was detected there was no observable modification of the surface morphology. The surface was striated and lightly faceted both before and after treatment. In the impinging jet reactor the reconstruction took the form of a layer of particles forming on the gauze surface. As shown in Fig. 20, directly under the impinging jet the particles were relatively small (<1 pm in diameter), and only a thin layer formed. Away from the impingingjet larger particles, which appear to be agglomerates of many small particles, were

404

TA-CHIN WE1 AND JONATHAN PHILLIPS

FIG. 20. Micrograph of a 90/10 FVRh gauze after exposure to Hz/02 mixture for 45 h at 602°C. The areas close to the center of the impinging jet (at right) are covered with small particles and large-particle agglomerates (at left) (51).

found. These are very similar in appearance to the “cauliflower” or “sprout” structures reported by others to form on the surfaces of gauzes used in ammonia oxidation. Similar agglomerated particle structures resulted from hydrogen oxidation of platinum in films in the same reactor. The same phenomenological model of particle formation apparently applies in both cases. Relatively small particles form in the gas phase by interactions between metastable volatile intermediates. These particles, as they become larger, precipitate out on the surface. On areas of the surface in which there is a high density of these particles, they tend to agglomerate to form larger, imperfectly sintered aggregates. In contrast, in the flow reactor the density of metastable species is apparently never high enough to favor the formation of particles, and the volatilized platinum simply washes out of the reactor. The above model of particle formation from the Pt/Rh (90/10) gauze might suggest that the particles which cover the surface following etching should be pure platinum. As control studies show that pure rhodium does not etch, the etching process might be expected to volatilize platinum selectively. However, scanning electron microscopyiEDAX results show that the particles have essentially the same composition as the bulk (Fig. 21). This observation requires some explanation. One possibility is that the volatile species which leave the surface

THERMAL AND CATALYTIC ETCHING

405

FIG.21. EDAX maps of the distribution of (a) platinum and (b) rhodium for the cross section of an area of a Pt/Rh sample covered with large agglomerates produced during catalytic etching in a Hz/02 mixture (51).

406

TA-CHIN WE1 AND JONATHAN PHILLIPS

consist of small clusters, and not single atoms. These clusters could contain both platinum and rhodium. In sum, it is clear from the results of these experiments that catalytic etching is not the sum of the etching effects of the individual gases present during the reaction. The results also show that, contrary to earlier suggestions (140, 189, 190), catalytic etching is not a function of temperature gradients. Recently, an alternative model was proposed, whereby the catalytic etching of platinum which takes place in either ethyleneloxygen mixtures or in hydrogen/ oxygen mixtures results from interactions between homogeneously formed free radicals (probably HOz) and the surface. These lead to the formation of volatile, metastable platinum-containing species, which desorb from the surface, leading to significant metal transport over relatively long distances. In some circumstances the platinum species density in the vapor phase becomes so high that vapor-phase reactions leading to the formation of platinum particles take place. On supported catalysts the formation of the metastable species leads to the creation of a high concentration of single-atom species in the catalyst pores and hence to Ostwald ripening. Under other conditions, when there are no free radicals present to create gas-phase platinum-containing species, sintering can take place only by coalescence.

D. RELATEDPHENOMENA There are a number of phenomena related to catalytic etching that support the free-radical metal transport model. Although the model of gas-phase radical/ metal-surface interactions is new to the catalytic community, it has strong precedent in work done in other areas of science. First, in early investigations designed to demonstrate the existence of free radicals, it was shown that the radicals are responsible for metal etching. Second, investigations of materials etching in plasmas, examined primarily because of the importance of this phenomenon in semiconductor fabrication, also indicate that free radicals are always involved in the etching process. Third, studies of materials etching in low earth orbit suggest that it is free-radical species which cause the greatest damage. Each of these topics is addressed briefly below. 1. Etching by Thermally Generated Free Radicals More than fifty years ago it was discovered that free-radicallmetal-surface interactions could lead to gross morphological changes in metal surfaces, due to the long-range transport of metal by metastable radical-metal species (191, 192). For example, early workers found that methyl radicals strongly attack lead, zinc, cadmium, antimony (193-195) and possibly copper (196) films. There were also studies showing that atomic hydrogen could etch a wide

THERMAL AND CATALYTIC ETCHING

407

s = 6 6 z - p Movable furnace

ekTetrarnethyllead

I-- . .

FIG.22. Apparatus for the detection of organic free radicals. Methyl radicals are formed in the heated zone and react with the lead mirror downstream to form volatile tetramethyllead (199).

variety of materials, including arsenic, antimony, selenium, tellurium, germanium, and tin (197). The suggestion that hydrogen atoms can etch many materials was recently repeated (198). Some of these experiments are classics, and are even described in detail in standard chemistry textbooks (199). In a typical case, tetramethyllead, flowing in a glass tube, is heated immediately upstream of a lead metal mirror (a thin film of lead deposited on the inside of the tube). This causes two processes to occur: First, a new lead mirror forms in the heated zone; second, the original lead mirror gradually disappears (Fig. 22). It was postulated that the first process occurs because heating causes tetramethyllead to decompose, leaving lead metal on the reactor wall, and creating gas-phase methyl radicals. The second process takes place because the free radicals formed by the decomposition flow down to the original lead mirror where they react to re-form volatile tetramethyllead. It is also known that metal transport can be initiated by interactions between gas-phase molecular species rather than free radicals and solid metals (Section 1I.D).

2. Plasma Etching Research into the mechanism of etching by free radicals has taken place only in the last few years. The new interest resulted from the recognition that dry etching (etching in the absence of a liquid phase) is crucial to the processing of micron- and submicron-sized circuit elements, where anisotropic etching is crucial. Not surprisingly, given the significance of silicon, the central focus of almost all the recent studies has been silicon etching in halogen plasmas. This research has been very successful, and a great deal of detailed information regarding the mechanism and kinetics of silicon etching in halogen plasmas is now available. The focus of the present review is work done on metal etching, but both semiconductor and metal etching are briefly reviewed.

408

TA-CHIN WE1 AND JONATHAN PHILLIPS

a. Plasma Semiconductor Etching. The most informative studies of silicon etching involve the separate production of different components of plasmas. This approach allows for a controlled study of the etching characteristics of each of the components individually as well as the nature of the synergism that results when the various components are present together. The most common practice is to produce energetic ions (most often, argon) using commercially available ion guns and to introduce reactive free radicals (almost always halogens) in a molecular beam of some type of unstable parent molecule. Shown in Fig. 23 is a modulated-beam mass spectrometer built in an IBM laboratory.

FIG.23. Schematic diagram of a modulated-beam mass spectrometer system. (a) Top view; (b) side view (224).

THERMAL AND CATALYTIC ETCHING

409

Much effort has been expended in attempting to distinguish experimentally between various postulated models of silicon etching. Winters (200) cataloged four different mechanisms that can explain for all systems how ion bombardment coupled with chemical processes leads to an enhanced sputtering rate. The mechanisms are as follows: (i) Ion bombardment induces a chemical reaction (201-204); (ii) ions “detrap” inert halogenated species which form spontaneously owing to the interaction of halogen radicals with the silicon surface (205); (iii) ions physically sputter surface halogenated species; and (iv) the chemical interactions of neutral free radicals with the surface are enhanced because of damage done by ions (206, 207). A fifth model in which the energetic ions excite surface molecules, enhancing the rate of reaction with adsorbed species, was postulated by Gerlach-Meyer (208). Most experimental work on silicon etching points to the following simple conclusions: The rate of silicon etching in a fluorine plasma is determined by the rate of ion-induced chemical reaction [mechanism (i) above], except at very high fluorine pressures when spontaneous etching, that is etching in which the ions play no role at all, begins to dominate (209-212). The advantage of fluorine plasma etching is that it is anisotropic, as only those regions exposed directly to the energetic ions remain clean enough to etch under normal plasma operating conditions. This allows the production of smaller, more tightly packed circuit elements. In chlorine plasmas, mechanism (i) also dominates, but, in contrast to etching in fluorine plasmas, chlorine plasma etching is never dominated by a spontaneous etching process. It is necessary for the energetic ions to implant adsorbed chlorine, leading to the formation of silicon chloride species which readily undergo thermal desorption (213-216). Recent studies of plasma etching have also been driven by the need to improve the production of submicron-sized circuit elements. Efforts have been focused on complex plasma recipes in which more than one gas species is added to the plasma to create multi-radical plasmas (217, 218). Another focus has been on finding plasma systems capable of etching a new generation of semiconducting materials such as Ga-As and other group 111-V materials ( 2 1 e 2 2 3 ) . b. Plasma Metal Etching. A minor subset of all the modem work on plasma etching concerns etching of metals. One of the more surprising results, pointed out by Winters and Cobum (212),is that in many metal etching systems, but not all (200, 224), the rate of etching is suppressed by ion bombardment (Fig. 24). In one of the earliest studies of metal etching by plasmas it was found that the etching rate of silicon was strongly dependent upon the ion energy, whereas the etching rate of aluminum was completely independent of the ion energy. These results certainly suggest that aluminum etching is simply a chemical process. In a series of studies in which the rate of etching as a function of ion-beam intensity and/or free-radical parent molecule intensity was

410

TA-CHIN WE1 AND JONATHAN PHILLIPS

b cu (100) T = 310%

Time (rec) FIG.24. Etch rate of (a) AI(100) and (b) Cu(100). Ion bombardment does not influence the etch rate of aluminum. However, the etch rate of copper is suppressed by ion bombardment (212).

measured, Winters and others showed that many metals etched faster with the ion beam off (212, 224, 225). They suggested that in many cases the principal mechanism of metal etching was one of compound growth-evaporation (226). This result is completely consistent with the older body of research cited above. Copper is a current focus of plasma metal etching studies (227-234). Traditionally, aluminum has been the choice as the main metal material in integrated circuits; but as circuit elements get smaller, future generations of integrated circuits will increasingly require a metal that has a lower resistivity and better ability to cany high current densities than aluminum. It is generally agreed that

THERMAL AND CATALYTIC ETCHING

41 1

copper will be the metal of choice. Thus far, low-temperature dry etching of copper has proved to be an elusive goal, primarily because copper chlorides, which form readily (235, 236), desorb only at temperatures exceeding 150°C. High-temperature processing can lead to such problems as interdiffusion of layers. Given the importance of the problem, it can be anticipated that research will continue until a solution is found. Other metals also may become the focus of research in the near future. There is hope that aligned magnetic systems, formed in all-metal (no semiconductor) sandwiches, may be useful as high-speed, high-density permanent memory devices. Experimental configurations consist of copper and gold layers (237, 238). Investigations of etching of metals have also been conducted in flow-type microwave plasmas. In such a system it is not possible to control separately the concentrations of free radicals and charged species. However, there are concentration and temperature gradients. Moreover, the charged species recombine more rapidly than the free radicals, and generally at the tail end of the afterglow only free radicals are present. Thus, it is often possible through multiple experiments to determine the contribution of each species to the overall etching process. Tin etching in hydrogen plasmas has been examined both recently and in the early years (197, 239) of plasma chemistry. In the recent investigation it was demonstrated that tin foils are drastically restructured in hydrogen plasmas without any concomitant weight loss (240). It was found (Fig. 25) that a standard polycrystalline tin foil restructures to form a porous mass of smallscale tin filaments. There is no weight loss associated with this process, and X-ray diffraction data suggest that there is no phase change either. To explain these phenomena, it was suggested that short-lived volatile tin hydrides form in the hydrogen atom environment of the afterglow region, and these lead to shortrange mass transport, followed by decomposition of the intermediate back into tin atoms and hydrogen. As this process is carried out at relatively low temperatures (<200"C) to avoid melting the tin, there is not sufficient thermal energy available to sinter the tin back into the polycrystalline state. Instead, only relatively small crystals form, leading to the creation of many voids, and hence porosity. Silver etching has also been studied in the afterglow region of a microwave plasma (241). The process was found to be a complex function of the precise plasma conditions. In regions of moderate to low concentrations of oxygen and charged species, it was found that silver foils oxidize at a nearly constant rate, independent of the metal oxide film thickness. At similar temperatures in molecular oxygen (or even atomic oxygen) only thin silver oxide layers can be formed, due to the protective nature of the layer. Silver oxide normally forms a self-limiting diffusion barrier. Apparently, in the afterglow of a microwave

412

TA-CHIN WE1 AND JONATHAN PHILLIPS

FIG.25. Tin foil reconstruction in hydrogen plasma. After treatment in hydrogen plasma, the foil surface become a network of large pores and strings of small, randomly oriented crystallites (240).

plasma a significant negative charge buildup on the metal oxide produces a driving force for diffusion in addition to the chemical potential, namely, a high potential gradient. The most general models of oxide growth include a term for potential-gradient-driven diffusion. Although it can be ignored in simple gasphase oxidation, it cannot be ignored in a plasma environment. Moreover, in the afterglow of a microwave plasma, the relative mobility of electrons is so much greater than that of ions that the negative charge buildup at the surface can be significant. This enhances cation diffusion so that the rate-limiting step to oxide growth is no longer cation diffusion, but rather surface reaction. In regions of concentrations of free radicals and charged species, not only oxidation, but etching takes place. The etching takes place so rapidly that it might better be called metal volatilization. The mechanism for this etching is not clear. One possibility is that oxidation takes place so rapidly that a great deal of charge is trapped in the structure. This could lead to almost explosive decomposition of the structure. More work is required to clarify the mechanism of etching in this region. The etching of platinum in oxygen plasma afterglows is perhaps the most thoroughly studied form of metal etching in a discharge. It was discovered early (242) and later confirmed (243, 244) that platinum loses weight in an oxygen

THERMAL AND CATALYTIC ETCHING

413

discharge. However, in none of those studies was the concentration of any species in the discharge measured. For this reason, the conclusion reached in those studies that platinum is etched by “active” oxygen created in the discharge is considered premature. Recently a correlation has been found between etching rates and the concentration and temperature of oxygen atoms and charged species (245). Platinum is rapidly etched as a result of a synergism between oxygen atoms and electrons (Fig. 26). Oxygen atoms alone do not etch the metal. Perhaps more surprisingly, it was found that electrons rather than ions assist the etching process. As noted above, in semiconductor studies it is generally found that ions are responsible for enhanced etching rates. However, there are a few other instances in which it has been found that electrons enhance plasma processes (246-252). In sum, the mechanism of plasma etching is often similar to the new postulated mechanism for catalytic etching. In virtually all cases fiee radicals play a key role in creating volatile species that remove mass from the surface. Plasma-generated free radicals interact with the surface to form volatile intermediates which then desorb, leading to a net loss of mass from the solid Sometimes radicals form volatile intermediates with surface atoms, but the

FIG. 26. Surface morphology of a platinum sample after 4 h of oxygen plasma treatment. The platinum surface is partially covered by faceted particles 5 pm in diameter (245).

414

TA-CHIN WE1 AND JONATHAN PHILLIPS

process of formation or desorption requires energy input from impinging highenergy charged species. 3 . Etching in Low Earth Orbit

Surfaces of spacecraft in low earth orbits (200-600 km) often undergo damage (253, 254), the severity of which is a function of the type of material. Organic polymers and carbon surfaces are the most severely damaged and may even disappear altogether. Some metal films, such as osmium, are also known to disappear, whereas others, such as silver, are severely oxidized. Painted surfaces lose gloss and become pitted (255). Concern regarding this damage, particularly as it related to the space shuttle, resulted in detailed investigations of the phenomena. Early on it was postulated that the damage was the result of interactions between these materials and highenergy oxygen atoms. which are the most plentiful atmospheric species at the altitudes in question (256). Moreover, it was suggested that these high-energy species oxidized the materials and that the oxidized materials could be readily sputtered by high-energy atmospheric species (impact energy about 5 eV). Laboratory studies of interactions of materials with high-energy oxygen atoms demonstrated that the oxidatiodsputter hypothesis was generally successful in predicting observed behavior (253, 257). There were some exceptions. For example, the flaking of silver surfaces reported from shuttle experiments was not repeatable in the laboratory. However, in no study was the interaction between atomic oxygen and various materials conducted at low impact energy. Nor was any attempt made to explore the synergistic effects of combining interactions of charged and free-radical species. Thus, the possibility that spacecraft damage results from a simple free-radical/surface interaction of the type commonly encountered in both catalytic etching and plasma etching was never tested, nor was the possibility that a synergistic process involving free-radical and charged species of the type seen in plasma etching. In any event, practical solutions to the problem were found. The simplest solution is to coat the surfaces of polymeric composites and other sensitive surfaces with thin films of materials that resist corrosion. Among the coatings that appear to provide protection are anodized aluminum and thin layers of sputtered nickel. E. SUMMARY Rapid metal (or semiconductor) etching involving loss of material is encountered in several areas of technology and science. Similar mechanisms account successfully for these processes. Vapor-phase free radicals interact with the solid surface to form surface complexes which are subsequently volatilized. This

THERMAL AND CATALYTIC ETCHING

415

process leads to net loss of material from the solid and often to gross surface reconstruction as well. There are subtle, but important, differences among the etching processes. In the case of catalytic etching, the surface species formed by free-radical/metal interactions spontaneously volatilize. In contrast, in plasma etching, high-energy charged species often play a synergistic role. In some cases they provide the energy required for volatilization. In other cases they provide the energy required for reactions between free radicals and surface atoms, and the surface complexes thus formed volatilize spontaneously. Material damage in low earth orbit also involves free-radical species; it is certain that oxygen atoms play a crucial role, but the details are not clear. Also, there are some cases in which catalytic etching results from some other processes. For example, pits may be formed as a result of subsurface gas formation. It is notable that virtually no cross-referencing exists in the literature of these various subjects. Notwithstanding the similarity of the phenomenological character of the various processes-a similarity clearly noted by the use of the word “etching” to describe all the related processes-there are few cases of workers’ in one area building on the relevant research conducted in the neighboring fields. Clearly, if duplication of effort is to be avoided, and the greatest crossfertilization of ideas is to take place, a greater awareness of advances in related fields is needed. REFERENCES 1. Shuttleworth, R., Metullurgiu 38, 125 (1948). 2. Moore, A. J. W., in “Metal Surfaces” (W. D. Robertson and N. A. Gjostein, Eds.), p. 155.

3. 4. 5. 6. 7. 8. 9. 10. 11.

12. 13. 14. 15.

16. 17. 18. 19. 20. 21. 22.

American Society for Metals, Metals Park, Ohio, 1963. Flytzani-Stephanopoulos,M., and Schimdt, L. D., Prog. SurJ Sci. 9, 83 (1979). Hullet, G. A,, and Berger, H. W., J. Am. Chem. SOC.26, 1512 (1904). Aitken, C., Nature 31, 268 (1879). Narhwold, R., Weid. Ann. 35, I16 (1888). Rosenhain, W., Proc. R . Soc. A70, 252 (1902). Rosenhain, W., and Humphrey, J. C. W., Proc. R. Soc. A83, 200 (1909). Rosenhain, W, and Ewen, D., J. Inst. Met. 8, 149 (1912). Leroux, J. A. A., and Raub, E., Z. Anorg. Allgem. Chem. 188, 205 (1930). Elam, C. F., Trans. Furuduy Soc. 32, I604 (1936). Lacroix, R., Rev. Met. (furis) 53, 809 (1956). Raub, E., and Plate, W., Z. Metullkd. 48, 529 (1957). Fryburg, G . C., and Petrus, H. M., J. Electrochem Soc. 108, 496 (1991). Gwathmey, A. T., and Benton, A. F., J. Chem. Phys. 8,431 (1940). Gwathmey, A. T., and Benton, A. F., J. Chem. Phys. 8, 569 (1940). Gwathmey, A. T., and Benton, A. F., J. Phys. Chem. 46, 969 (1942). Johnson, R. P., Phys. Rev. 54,459 (1938). Mrowca, A. D., J. Appl. Phys. 14, 684 (1943). Langmuir, I., Actu Metall. 5, 13 (1957). Chalmers, B., King, R., and Shuttleworth, R., Proc. R. Soc. A193,465 (1948). Hondros, E. D., and Moore, A. J. W., Actu Merull. 8, 647 (1960).

416

TA-CHIN WE1 AND JONATHAN PHILLIPS

23. Rhead, G . E., and Mykura, H., Actu Metull. 10, 843, 578 (1962). 24. Bao, X . , Lehmpfuhl, G.. Weinberg, G., Schlogl, R., and Ertl, G., J. Chem. Soc. Furuduy Truns. 88, 865 (1992). 25. Winterbottom, W. L., J. Appl. fhys. 40, 3803; 3810 (1969). 26. Wong, S., Flytzani-Stephanopoulos,M., Chen, M., Hutchinson T. E., and Schmidt, L. D., J. Vuc. Sci. Technol. 14,452 (1977). 27. Flytzani-Stephanopoulos,M., Wong, S., and Schmidt, L. D., J. Cutul. 49, 51 (1977). 28. Leidheiser, H., and Gwathmey, A. T., J. Am. Chem. SOC.70, 1200; 1206 (1948). 29. Cunningham, R. E., and Gwathmey, A. T., Adv. Cuful. 10, 57 (1958). 30. Meelheim, R. Y . , Cunningham, R. E., Lawless, K. R., Azim, S., Kean R. H., and Gwathmey, A. T.,froc. Int. Congr. Cutul. 2, 2005 (1961). 31. Kauffer, C. T., and Leitenberger, W., Chem. Eng. Technol. 25, 697 (1953). 32. Pan, B. Y. K., Ind. Eng. Chem. froc. Des. Dev. 8, 263 (1969). 33. Schmidt, L. D., and Luss, D., J. Cutul. 22, 269 (1971). 34. Garton, G., and Turkevich, J., J. Chim. fhys. 51, 516 (1954). 35. Wu, N. L., and Phillips, J.. J. fhys. Chem. 89, 591 (1985). 36. Curie, P., Bull. Soc. Chim. Fr. 8, 145 (1885). 37. Gibbs, J. W., Truns. Conn. Acud. 3, 343 (1878); reprinted in “The Scientific Papers of J. Willard Gibbs,” Vol. I. Dover, New York, 1961. 38. Wulff, G., Z. Kristullogr. 34, 449 (1901). 39. Conrad, E. H., frog. Surf: Sci. 39, 65 (1992). 40. Herring, C., fhys. Rev. 82, 87 (1951). 41. Herring C., in “Structure and Properties of Solid Surfaces” (R. Gomer and C. S. Smith, Eds.), p. 5 . Univ. of Chicago Press, Chicago, 1953. 42. Mullins, W. W., fhilos. Mag. 6, 1313 (1961). 43. Mullins, W. W., in “Metal Surfaces” (W. D. Robertson and N. A. Gjostein, Eds.), p. 17. Amer. SOC.Metals, Metals Park, Ohio, 1963. 44. Carbrera, N., Surf: Sci. 2, 320 (1964). 45. Hirth, J. P., in “Energetics in Metallurgical Phenomena” (W. M. Mueller, Ed.), Vol. 11, p. I . Gordon & Breach, New York. 1965. 46. Hirth, J. P., in “Structure and Properties of Metal Surfaces,” Honda Memorial Series on Materials Science. No. I , p. 10. Maruzen Co, Tokyo, 1973. 47. Herring, C., in “Physics of Powder Metallurgy” (W. E. Kinston, Ed.), p. 143. McGraw-Hill, New York, 1951. 48. Mullins, W. W., J. Appl. fhys. 28, 313 (1957). 49. Mullins, W. W.. J. Appl. fhys. 30, 77 (1959). 50. Mullins, W. W., Truns. AIME 218, 354 (1960). 51. Hess, J. M., and Phillips, J., . I Cutul. . 136, 149 (1992). 52. Winterbottom, W. L., Acfu Mefull. 15, 303 (1967). 53. Murr, L. E., “Interface Phenomenon in Metals and Alloys.” Addison-Wesley, Reading, Mass., 1975. 54. Cahn, J. W., Actu Metull. 28, 1333 (1967). 55. Van Hardeveld, R., and Van Montfoort, A,, Surj: Sci. 4, 396 (1966). 56. Dumesic, J. A., Topsee, H., Khammouma, S., and Boudart, M., J. Cutul. 37, 503 (1975). 57. Boudart, M., Dumesic, J. A., and Topsee, H., froc. Nufl. Acud. Sci. USA 74, 806 (1977). 58. Brill, R., Richter, E. L., and Ruch, E., Angew. Chem. Int. Ed. Eng. 6, 82 (1967). 59. Spencer, N. D., Schoonmaker, R. C., and Somorjai, G. A., J. Curul. 74, 129 (1982). 60. Wang, T., Lee, C., and Schmidt, L. D., Surf: Sci. 163, 181 (1985). 61. Lee, W. H., Vavloon, K. R., Petrova, V., Woodhouse, J. B., Luxton, C. M., and Masel, R. I., J. Cutul. 126, 658 (1990).

THERMAL AND CATALYTIC ETCHING

417

62. Somojai, G. A., in “Catalyst Design: Progress and Perspectives” (L. L. Hegedus, Ed.), p. 1 I . Wiley, New York, 1987. 63. Strongin, D. R., and Somoqai, G. A,, J. Catal. 118, 99 (1989). 64. Goodman, D. W., Acc. Chem. Res. 17, 194 (1984). 65. Kearby, K. K., Thoru, J. P., and Hinlicky, J. A,, U S . Pat. 3,134,732 (1964). 66. Schmidt, R. A., D’Netto, G . A., Razon, L. F., and Brown, J. R., in “Chemical Instability” (G. Nicolis and F. Baras, Eds.), p. 33. Reidel, Holland, 1983. 67. Birke, P., Engels, S., Becker, K., and Neubauer, H. D., Chem. Tech. 31,473 (1979). 68. Birke, P., Engels, S., Becker, K., and Neubauer, H. D., Chem. Tech. 32, 253 (1980). 69. Delannay, F., Defosce, C., Delman, B., Menon, P. G., and Froment, G. F., lnd. Eng. Chrm. Prod. Rex Div. 19, 537 (1980). 70. Foger, K., and Jaeger, H., J. Catal. 92, 64 (1985). 71. F o p , K., Hay, D., and Jaeger, H., J. Catal. 96, 170 (1985). 72. Shen, W. M., Dumesic, J. A,, and Hill, Jr., C. G., J. Catal. 68, 152 (1981). 73. Collman, J. P., Hegedus, L. S., Cooke, M. P., Norton, J. R.. Colcetti, G . , and Marquardt, D. N.. J. Am. Chem. Soc. 94, 1789 (1972). 74. Flynn, P. C . . and Wanke, S. E., J Cutul. 37, 432 (1975). 75. Ruckenstein, E., and Chu, Y. F., J. Cutal. 59, 109 (1979). 76. Johnson, M. F. L., and Keith, C. D., J. Phys. Chem. 67,200 (1963). 77. Jaworska-Galas, Z., and Wrzyszcz, J., lni. Chem Eng. 6, 604 (1966). 78. Fiedorow, R. M., and Wanke, S. E., J. Catal. 43, 34 ( 1 976). 79. Ruckenstein, E., and Malhotra, M. L., J. Catal. 41, 473 (1981). 80. Gollob, R., and Dadyburjor, D., J. Curd 65, 473 (1981). 81. Lamber, R., and Romanowski, W., J. Catal. 105, 213 (1987). 82. Wu, N. L., and Phillips, J., J. Catal. 113, 129 (1988). 83. Stulga, J. E., Wynblatt, P., and Tien, J. K., J. Catal. 62, 59 (1980). 84. Alcock, C. B., and Hooper, G. W., froc. R. SOC.A254, 551 (1960). 85. Wanke, S. E., and Flynn, P. C., Catal. Rev. Sci. Eng. 12, 93 (1975). 86. Wynblatt, P., and Gjostein, N. A,, Prog. Solid Sfate Chem. 9, 21 (1975). 87. Granqvist, C. G., and Buhrman, R. A., J. Catal. 42, 477 ( I 976). 88. Granqvist, C. G., and Buhrman, R. A., J. Appl. fhys. 47, 2200 (1976). 89. Ruckenstein, E., and Dadybujor, B., J. Catal. 48, 73 ( I 977). 90. Hu, Z., Nakamura, H., Kunimori, K., Asano, H., and Uchijima, T., J. Catal. 112, 478 (1988). 91. Hu, Z., Nakamura, H., Kunimori, K., Yokoyama, Y., Asano, H., Soma, M., and Uchijima, T., J. Catal. 119, 33 (1989). 92. Hu, Z., Wakasugi, T., Maeda, A,, Kunimori, K., and Uchijima, T., J. Catal. 127, 276 (1988). 93. Kossel, W., Nachr. Akad. Wiss. Goettingen 135 (1927). 94. Stranski, I. N., Z. Phys. Chem. 136,259 (1928). 95. Burton, W. K., Cabrera, N., and Frank, F. C., fhilos. Trans. R. SOC. A243, 299 (1950). 96. Gruber, E. E., and Mullins, W. W., J. fhys. Chem. Solids 28, 875 (1967). 97. Wortis. M., in “Chemistry and Physics of Solid Surfaces” (R. Vanselow and R. Howe, Eds.), Vol. VII, p. 367. Springer, Berlin, 1988. 98. Jayaprakash, C., Rottman, C., and Saam, W. F., fhys. Rev. B 30, 6549 (1984). 99. Jayaprakash, C., Saam, W. F., and Teitlel, S., Phys. Rev. Lett. 50,2017 (1983). 100. Rottman, C., and Wortis, M., Phys. Rev. B 24, 6274 (1981). IOI. Avron, J. E., Van Beijeren, H., Schulman, L. A., and Zia, R. K. P., J. Phys. A 15, L81 (1 982). 102. Rottman, C., and Wortis, M., Phys. Rev. B 29, 328 (1984).

418

TA-CHIN WE1 AND JONATHAN PHILLIPS

103. Rottman, C., and Wortis, M., Phys. Rep. 103, 59 (1984). 104. Vu Thien Binh, Piguet, A,, Roux, H., Uzan, R., and Dreschsler, M., SurJ Sci. 44, 598 ’ (1974).

105. Vu Thien Binh, Uzan, R., and Dreschsler, M., Surf Sci. 57, 118 (1976). 106. Vu Thein Binh, Moulin, Y., Uzan, R., and Dreschler, M., Proc. 7th Int. Vac. Congr. and 3rd Int. Conf. Solid Surfaces, 21 3 I (1 977). 107. Vu Thein Binh, Moulin, Y., Uzan, R., and Dreschler, M., SurJ Sci. 68,409 (1977). 108. Vu Thien Binh, Moulin, Y., Uzan, R., and Dreschler, M. Surf Sci. 79, 133 (1979). 109. Vlachos, D. G., Schmidt, L. D., and Aris, R., Phys. Rev. B 47,4896 (1993). 110. Ertl. G., Adv. Catal. 37, 213 (1990). 111. Ertl, G., Science 254, 1750 (1991). 112. Eiswirth, M., and Ertl, G., SurJ Sci. 177, 90 (1986). 113. Gritsch, T., Coulman, D., Behm, R., J., and Ertl, G., Phys. Rev. Lett. 63, 1086 (1989). 114. Jakubith, S., Rotermund, H. H., Engel, W., Oertzen, A. V., and Ertl, G., Phys. Rev. Lett. 65, 3013 (1990). 115. Ladas, S., Imbihl, R., and Ertl, G., Surf Sci. 197, 153 (1988). 116. Ladas, S., Imbihl, R., and E d , G. Su$ Sci. 198, 42 (1988). 117. Falta, J., Imbihl, R., and Henzler, M., Phys. Rev. Lett. 64, 1409 (1990). 118. Imbihl, R., Reynolds, A. E., and Kaletta, D., Phys. Rev. Lett. 67, 275 (1991). 119. Lyubovsky. M. R., and Barelko, V. V., J. Catal. 149, 23 (1994). 120. Wu, N. L., and Phillips, J., J. Appl. Phys. 59, 769 (1986). 121. Schafer, H., “Chemical Transport Reactions.” Academic Press, New York, 1964. 122. Bao, X., Lehmpfuhl, G . , Weinberg, G., Schlogl, R., and Ertl, G., J. Chem. Soc. Faraday Trans. 88,865 (1992). 123. Kiueh, R. L., and Mullins, W. W., Trans. AlME 242, 237 (1968). 124. Genkin, E. S. et al., Dokl. Acad. Nauk 316, 658 (1991). 125. Pareja, P., Amariglio, A,, Piquard, G., and Amariglio, H., J. Catal. 46, 225 (1977). 126. Eyraud, C., Domanski, B.. and Berger, C., Bull. Soc. Chim. Fr. 23, 1765 (1956). 127. Noms, L. F., and Pamvano, G., in “Reactivity of Solids” (Mitchell, De Vries, Roberts, Cannon, Eds.), p. 149. Wiley, New York, 1969. 128. Parsons, C. L., lnd. Eng. Chem. 11, 541 (1919). 129. O d e , I. A. P., “Chemical Reaction Engineering,” Vol. 1, p. 146. Pergamon, New York, 1957. 130. Holmes, A. W., Plat. Met. Rev. 3, 2 (1959). 131. Bone, W. A., and Wheeler, R. V., Philos. Trans. R. Soc. A206, 1 (1906). 132. Graham, P., Trans. R. Soc. 156, 435 (1866). 133. Wagner, Jr. and Gwathmey, A. T., J. Am. Chem. Soc. 76, 390 (1954). 134. Anstrong, E. F., and Hilditch, T. P., Proc. R. Soc. 108, 1 I 1 (1925). 135. Rubel, M., Psonicka, M., and Palczewska, W., J. Mar. Sci. 20, 3639 (1985). 136. McCabe. R. A., and Smith, G. D. W., Plat. Met. Rev. 27, 19 (1983). 137. Psonicka, M., and Dymkowski, T., Pol. J.. Chem. 52, 121 (1978). 138. Psonicka, M., Pol. J. Chem. 54,2283 (1980). 139. McCabe, R. W., Pignet, T., and Schmidt, L. D., J. Catal. 32, 114 (1974). 140. Flytzani-Stephanopoulos,M., and Schmidt, L. D., Chem. Eng. Sci. 34, 365 (1979). 141. Frohberg, G., and Adam, P., Thin Solid Films 25, 525 (1975). 142. Dean, V. W., Frenklach, M., and Phillips, J., J. Phys. Chem. 92, 5731 (1988). 143. Handforth, S. L., and Tilley, N. J., lnd. Eng. Chem. 26, 1287 (1934). 144. Bartlett, R. W., J. Electrochem. Soc. 1114, 547 (1967). 145. Pan, B. Y. K., and Roth, R. G., lnd. Eng. Chem. Proc. Des. Dev. 7, 53 (1968). 146. Gillespie, G. R., and Kenton, R. E., CHEMTECH, 1 (lo), 627 (1971).

THERMAL AND CATALYTIC ETCHING

419

147. Pine, J. M.. Plat. Met. Rev. 2, 7 (1958). 148. Wyman, L. L., Trans. AIME 104, 141 (1933); 111, 305 (1934); 137, 291 (1940). 149. Rhines, F. N., and Anderson, W. A., Trans. AIME 143, 312 (1941). 150. Ransley, C. E., J. Inst. Metal. 65, 147 (1939). 151. Langmuir, I., Trans. Faraday Sac. 17, 621 (1921). 152. Langmuir, I., Trans. Faraday Sac. 17, 607 (1921). 153. Gwathmey, A. T., and Cunningham, R. E., J. Chim. Phys. 51,497 (1954). 154. Cunningham, R. E. and Gwathmey, A. T., J. Am. Chem. Sac. 76, 391 (1954). 155. Pilling, N. B., Trans. AIME 60, 322 (1919). 156. Connor, H., Plat. Met. Rev. 11, 60 (1967). 157. Fierro, J. L. G., Palacios, J. M., and Tomas, F., Plat. Met. Rev. 34, 62 (1990). 158. Harner, B. T., Plat. Met. Rev. 37, 76 (1993). 159. Knapton, A. G., Plat. Met. Rev. 22, 131 (1978). 160. Anderson, D.R., J. Catal. 113, 475 (1988). 161. Alcock, C. B., Hooper, G. W., Proc. R. Sac. A254, 551 (1960). 162. Nowak, E. J., Chem. Eng. Sci. 21, 19 (1966). 163. Nowak, E. J., Chem. Eng. Sci. 24,421 (1969). 164. Pielaszek, J. Plat. Met. Rev. 28, 109 (1984). 165. Contour, J. P., Mouvier, G., Hoogewys, H., and Leclerc, C., J. Catal. 48, 217 (1977). 166. McCabe, A. R., and Smith, G. D. W., J. Phys. (Paris) C9,483 (1984). 167. Busby, J. A,, and Trimm, D. L., J. Catal. 60,430 (1979). 168. Philpott, J. E., Plat. Met. Rev. 15, 52 (1971). 169. Wu, N. L., and Phillips, J., J. Calal. 113, 383 (1988). 170. Lehner, S., J. Am. Chem. Sac. 53, 3742 (1931). 171. Douglas, J. E., Rabinovitch, B. S., and Looney, F. S., J. Chem. Phys. 23, 3 I ( I 955). 172. Dahlgren, Jr., D.,and Douglas, J. E., J. Am. Chem. Sac. 80, 5108 (1958). 173. Frenklach, M., Ramachandra, M. K., and Matula, R. A,, Proc. 20th Int. Symp. Combustion, 871 (1 984). 174. Frenklach, M., Clary, D. W., Gardiner, Jr., W. C., and Stein, S. E., Proc. 20th Int. Symp. Combustion. 887 (1984). 175. Frenklach, M., Hsu, J. P., Miller, D. L., and Matula. R. A., Combust. FIame 64, 141 ( 1956). 176. Haynes, B. S., and Wagner, W. G., Prog. Energv Combust. Sci. 7, 229 (1981). 177. Dougherty, E., and Rabitz, H.. J. Chem. Phys. 72,6571 (1980). 178. Westbrook, C. K., Dryer, F. L., and Schug, K. P., Proc. 19th Int. Symp. Combustion, 153 (1982). 179. Warnatz, J., in “Combustion Chemistry” (W. C. Gardiner, Jr. Ed.), Chapter 5 (1984). 180. Chinnick. K. J., Gibson, C., and Griffiths, J. F., Proc. R. Sac. A405, 129 (1986). 181. Chinnick, K. J., and Griffiths, J. F., J. Chem. SOC.Faraday Trans. I1 82, 881 (1986). 182. Baulch, D. L., Drysdale, D.D., Home, 0. D.,and Lloyd, A. C., “Evaluated Kinetic Data for High Temperature Reactions,” Vol. 1. Butterworths, London, 1972. 183. Harris, S. J., Weiner, A. M., and Blint, R. J., Combust. FIame 72, 91 (1988). 184. Wagner, H. G., in “Particulate Carbon” (D. Siegla and G. W. Smith, Eds.), p. I . Plenum, New York, 1981. 185. Prado, G., and Lahaye, J., in “Particulate Carbon” (D. C. Siegla and G. W. Smith, Eds.), p. 143. Plenum, New York, 1981. 186. Lai, F. S., Friedlander, S. K., Pich, J., and Hidy, C. M., J. Colloid Interface Sci. 39, 395 ( 1 972). 187. Zhang, Z., Lerner, B., Lei, G.-D., and Sachtler, W. M. H., J. Catal. 140, 481 (1993). 188. Kharas, K. C. C., Robota, H. J., and Liu, D. J., Appl. Catal. 82, 225 (1993).

420

TA-CHIN WE1 AND JONATHAN PHILLIPS

189. Storch, H. H., J. Am. Chem. SOC.56, 374 (1934). 190. Edwards, W. M., Worley, F. L., and Luss, D., Proc. AlChE 65th Annual Meeting. Paper No. 191. 192. 193. 194. 195. 196. 197. 198. 199.

200. 201. 202. 203. 204. 205.

206. 207.

208. 209. 210. 211. 212. 213. 214. 215. 216. 217.

218. 219.

220. 221. 222.

223. 224.

225. 226. 227.

226 ( I 972). Rice, F. O., and Glasebrook, A. L., J. Am. Chem. SOC.56,2381 (1934). Pearson, T. G., Purcell, R. H., and Saigh, G . S.. J. Chem. SOC.409 (1938). Paneth, F., and Lautsch, W., Nature 125, 564 (1930). Paneth, F., and Lautsch, W.. Eer. GesellschuJi 64, 2702 (1931). Paneth, F., and Hofeditz, W., Ber. Gesellschaft 62, 1335 (1929). Hund, D. T.. and Rochow, E. G., J. Am. Chem. SOC.67, 963 (1945). Pearson, T. G., Robinson, P. L., and Stoddart, E. M., Proc. R. SOC.A142, 275 (1933). Hiraoka, H., J. Vac. Sci. Techno/. B 4, 345 (1986). Laidler, K. J., “Chemical Kinetics,” McGraw-Hill, New York. 1965. Winters, H., J. Vuc. Sci. Techno/. A 3, 700 (1985). Tu, Y. Y., Chuang. T. J., and Winters, H. F., Phys. Rev. E 23, 823 (1981). Gerlach-Meyer, U., Coburn, J. W., and Kay, E., SurJ Sci. 103, 177 (1981). Knabbe, E. A., Coburn. J. W., and Kay, E., SurJ Sci. 123,427 (1982). Coburn, J. W., Vacuum, 34, 157 (1984). Dielman. J., and Sanders, F. H. M., Solid Stale Technol. 27, 191 (1984). Donnelly, V. M., and Flamm, D. L., Solid State Technol. 24, I6 1 (198 I ). Flamm, D. L., and Donnelly, V. M., Plasma Chem. Plasma Proc. 1, 3 17 ( I 98 I). Gerlach-Mayer, U., SurJ; Sci. 103, 524 (1981). Flamm, D. L., Donnelly, V. M., and Mucha, J. A., J. Appl. Phys. 52, 3633 (1981). Winters, H. F., and Houle, F. A,, J. Appl. Phys. 54, 1218 (1983). Winters, H. F., J. Vac. Sci. Technol. B 1, 927 (1983). Winters, H. F., and Coburn, J. W., J. Vac. Sci. Technol. E 3, 1376 (1985). Barish, E. L., Vitkavage, D. J., and Mayer, T. M., J. Appl. Phys. 57, 1336 (1985). Vitkavage, D. J., Ameen, M. S., and Mayer, T. M., J. Vac. Sci. A 2,492 (1984). Mayer, T. M., Barker, R. A,, Barker, L. J., and Whitman, L. J., J. Vac. Sci. Technol. 18, 349 (I981). Okano, H., and Horike, Y., Jpn. J. Appl. Phys. 20,2429 (1981). Coburn, J. W., in “Materials Modification by Energetic Atoms and Ions” (K. S. Grabowski, S. A. Barnett, S. M. Rossnagel, and K. Wasa, Eds.), Vol. 268, p. 3. Materials Research Society, Pittsburgh, 1992. Coburn, J. W., Pure Appl, Chem. 24, 709 (1992). Pearton, S. J., Chakabarti, U. K., Katz, A,, Ren, F., Fullowan, T. R., Abernathy, C. R., and Hobson, W. S., in “Materials Modification by Energetic Atoms and Ions” (K. S.Grabowski, S. A. Barnett, S.M. Rossnagel, and K. Wasa, Eds.), Vol. 268, p. 17. Materials Research Society, Pittsburgh, 1992. Komuro, M., and Hiroshima, H., J. Vac. Sci. Technol. E 9, 2656 (1991). Aoyagi, Y., Shinmura, K., Kawasaki, K., Tanaka, T., Gamo, K., Namba, S., and Nakamoto, I., Appl. fhys. Lett. 60, 968 (1992). Beckman, J., and Jackman, R. B., in “Materials Modification by Energetic Atoms and Ions” (K. S. Grabowski, S. A. Barnett, S. M. Rossangel, and K. Wasa, Eds.), Vol. 268, p. 35. Materials Research Society, Pittsburgh, 1992. Tyrell, G. C., Marshall, D., Beckman, J., and Jackman, R. B., J. P h p . (Cond. Mar.) 3, S179 ( 199I). Winters, H. F., J. Vac. Sci. Technol. E 3, 9 (1985). Winters, H. F., J. Vac. Sci. Technol. A 3, 786 (1985). Winters, H. F., Coburn, J. W., and Chuang, R. J., J. Vac. Sci. Technol. E 1,469 (1983). Hampden-Smith, M. J., and Kodas, T. T., MRS Bull. 18, 6, 39 (1993).

THERMAL AND CATALYTIC ETCHING

42 1

228. Shwartz, G. C., and Schaible, P. M., J. Electrochem. SOC.130, 1777 (1983). 229. Howard, B. J., Wolterman, S. K., Yoo, W. J., Gittleman, B. and Steinbruckel, C. H., in “Surface Chemistry and Beam-Solid Interactions” (H. Atwater, F. A. Houle, and D. Lownndes, Eds.), Vol. 201, p. 129. Materials Research Society, Pittsburgh, 1991. 230. Farkas, J., Chi, K. M., Hampden-Smith, M. J., Kodas, T. T., and Dubois, L. H., J. Appl. Phys. 73, 1455 (1993). 231. Hosoi, N., and Ohshita, Y., Appl. Phys. Lett. 23,2703 (1993). 232. Howard, B. J., and Steinbmchel, Ch., J. Vac. Sci. Technol. A 12, 1259 (1994). 233. Ohno, K., Sato, M., and Arita, Y., Jpn. J. Appl. Phys. 28, L1070 (1989). 234. Li, J., Sedel, T. E., and Mayer, J. W., MRS Bull. 19, 8, 15 (1994). 235. Sesselmann, W., and Chuang, T. J., SurJ Sci. 176, 67 (1986). 236. Sesselmann, W., Marinaro, E. E., and Chuang, T. J., SurJ Sci. 178, 787 (1986). 237. Johnson, M., Science 260, 320 (1993). 238. Johnson, M., fhys. Rev. Lett. 70, 2142 (1993). 239. Paneth, F., Matthies, H., and Schmidt-Hebbel, E.,Ber. Gesellshafr 55, 775 (1992). 240. Chou, C. H., and Phillips, J., J. Vac. Sci. Techno/. A 8, 3941 (1990). 241. Chou, C. H., and Phillips, J., J. Vac. Sci. Technol. A 9, 2727 (1991). 242. Fryburg, G. C., J. Chem. Phys. 24, 175 (1956). 243. Chaston, J. C., flat. Met. Rev. 9, 51 (1965). 244. Rubel, M.. Pszonicka, M., Ebel, M. F., Jablonski, A,, Palczewska, W., and Cannon, J. L., Metals 125, 7 (1986). 245. Chou, C. H., and Phillips, J., J. Appl. Phys. 68, 2415 (1990). 246. Coburn, J. W . , and Winters, H. F., J. Vac. Sci. Techno/. 16, 391 (1979). 247. Coad, J. P., Bishop, H. E., and Riviere, J. C., SurJ Sci. 21, 253 (1970). 248. Margoninski, Y., Segal, D., and Kirby, R. E., Sut$ Sci. 53, 488 (1975). 249. Verhoeven, J., and Los, J.. SurJ Sci. 58, 566 (1976). 250. Margoninski, Y., J. Appl. Phys. 47, 3868 (1976). 251. Neave, J. H., and Joyce, B. A,, J. Phys. D 10,243 (1977). 252. Tompkins, H. G., Surt Sci. 62, 293 ( I 977). 253. Leger, L. G., AIAA 21st Aerospace Sciences Meeting, Reno, Paper No. 83-0073 (1983). 254. Whitaker, A., AIAA Shuttle Environment and Operations Meeting (1983). 255. Singh, B., Amore, L.J., Saylor, W., and Racette, G., AlAA 23rd Aerospace Sciences Meeting, Reno, Paper No. 85-0477 (1985). 256. Hedin, A. E., Reber, C. A., Newton, G. P., Spencer, N. W., Brinton, H. C., Mazer, H. G., and Pones, W. W., J. Geophys. Res. 10, 2139 (1977). 257. Slemp, W. S . , Santos-Mason, B., Sykes, Jr. G. F., and Wine, Jr, W. G., AlAA 23rd Aerospace Sciences Meeting, Reno, Paper No. 85-042 I (1985).

This Page Intentionally Left Blank

Index A

AAS (aluminosilicate acid catalysts), dehydration of butyl alcohols over, 335-356 Absorption, by heteropoly compounds, 179- 180, 190-191 Acetoin, synthesis, 302 Acetone, condensation, 158 Acidity crystalline titanium silicates, 3 19-320 heteropoly compounds, 139-150 Acylation, heteropoly compounds, 175-176 Adamantane, oxidation, 203 Addenda atoms, heteropolyacids, 1 18, 120, 12 1 Agostic interactions, I 8 Alcohols absorption by heteropoly compounds, I79 oxidation, 235 titanium silicate-catalyzed, 298-303, 306 Aldehydes, oxidation, 215-218 Alkanes aromatic, 3 12-313 oxidation, 220, 307-308 photoreactions, 235 skeletal isomerization, 176- 177 Alkenes epoxidation, 326 Prins reaction, 156 vibrational spectroscopy, 102-104 butenes, 7 I , 80-9 I cyclic alkenes, 101 2,3-dimethylbut-2-ene, 100-101 ethene. 3 1-74 hexenes, 93-95 2-methylpropene, 97-1 00 pentenes, 9 1-93 propenes, 74-80 Alkylaromatic compounds, industrial acylation, I75 Alkylation, heteropoly compounds, 160-161, 170-174 423

Allyl alcohol, oxidation, 305 Allyl chloride oxidation, 305 thermal decomposition, 80 Allyl methaclylate, oxidation, 305 Aluminophosphates, 256 Aluminosilicate acid catalysts, dehydration of butyl alcohols over, 335-356 Aluminum, etching, 409, 410 Amberlyst, 167 Amines, oxidation, 3 15 Ammonia, oxidation, 313, 387, 389, 393, 394, 396 Anderson structure, heteropolyacids, 123 Aniline, oxidation, 3 I6 Anisole acylation, I76 oxidation, 3 1 1 Ascorbic acid, synthesis, 222

B Beckman rearrangement, TS-I and TS-2 catalysts, 295 Benzene alkylation, 172 hydroxylation, 297 oxidation, 3 12 Benzopinacol, rearrangement, 189 Bulk type I catalysis, 1 14, 116, 1 17, 165, 186-189 BulktypeIIcatalysis, 114, 116, 117,211-215 Butadiene oxidation, 309, 3 I0 vibrational spectra, 95-97 2.3-Butanedio1, oxidation, 302 n-Butane isomerization, I76 oxidation, 3 13 synthesis, 230 1,4-Butanediol, dehydration, 155

424

INDEX

Butane-2,3-dione, synthesis, 302-303 Butanes, oxidation, 220 2-Butano1, oxidation, 301 Butenes formed by butyl alcohol dehydration, 348 isomerization, 187, 188 oxidation, 304 vibrational spectra, 7 I , 90-9 1 But-2-en-I -01, oxidation, 307 Butyl alcohols dehydration over AAS catalysts, 335-337, 355-356 catalyst characterization, 337-339 intermediates, 349-354 kinetics, 339-344 reaction pathways, 344-349,354 oxidation, 301 I-Butylbenzene, oxidation, 3 12 rerr-Butyl hydroperoxide, oxidation with, 209-2 10 Butylidyne, 82 rert-Butyl-4-methylphenol, synthesis, 22 1 2-tert-Butylpheno1, synthesis, 161 tert-Butyl-p-xylene (BPX), synthesis, I70

Carbenium ions, acid catalysis, 336 Catalytic etching, 359, 383-384 definition, 360-361 in low earth orbit, 414-415 models, 359, 360-362 plasma etching, 407-414 thermally generated free radicals, 406-407 1-Chloroprop-2-ene, thermal decomposition, 80 Cobalt ethene vibrational spectra on, 59, 60 oxide-supported metal catalysts, 10, 11 Condensation reactions, heteropoly compounds, 157-158 Copper etching, 359,363,410-41 1 ethene vibrational spectra on, 60, 62 oxide-supported metal catalysts, 10, I 1, 12 Cracking, thermal etching, 360 p-Cresol, alkylation, 171, 22 1 Cumene, oxidation, 312 Cyclic alkenes, vibrational spectra, 101 Cycloalkanes, aromatic, 3 1 1-313 Cvclohexanol. oxidation., 299-300 ---

Cyclohexanone, ammoximation, 296, 3 13-314 Cyclohexene dehydrogenation, 197 oxidation, 305, 308, 309 Cyclohexyl acetate, decomposition, 161 Cyclopent-2-en-l-ol. epoxidation, 307 pCymene, synthesis, 201

D Dawson structure, heteropoly compounds, 121-122, 131 Dealkylation, heteropoly compounds, 160-16I , 170-174 n-Decane, oxidation, 3 I3 I-Decene, epoxidation, 201 Dehydration butyl alcohol, over AAS catalyst, 335-356 with heteropoly compounds, 153-156, 165-168 2,6-Dialkyl-4-rert-butylphenol, alkylation, 17I Diallyl carbonate, oxidation, 309, 3 10 Diallyl ether, oxidation, 309, 310 2,6-Di-rert-butylphenol, dealkylation, 161, I7 I Dienes, vibrational spectra, 95-97 Diethyl ether, infrared spectra, 184-I85 2.3-DimethyIbut-2-ene dehydrogenation, 235 vibrational spectra, 100-101 Diolefins, oxidation, 303-3 10 Dioxygen, liquid-phase oxidation, heteropoly compound catalyzed, 200-203 Diphenyl sulfide, 3 17 I-Dodecene, oxidation, 308, 309 DRIFT (diffise-reflection Fourier-transform infrared spectroscopy), 6, 8 DTA (differential thermal analysis), heteropoly compounds, 127-128

E EELS (electron energy loss spectroscopy), 3 Electrocatalysis, heteropoly compounds, 233, 239-240 Electronic absorption spectra, heteropoly compounds, 135-136 Embrittlement. 385 Epoxidation heteropoly compound-catalyzed, 20 1 with hydrogen &oxide, 304-3 I0

425

INDEX ESR (electron spin resonance) spectroscopy, heteropoly compounds, 137-138 Esterification, heteropoly compounds, 160, 169-170 Esters, decomposition, heteropoly compounds, 160, 169-170 Etching catalytic etching, 359, 383-384 definition, 360-361 models, 359, 360-362 plasma etching, 407-414 thermally generated free radicals, 406-407 in low earth orbit, 414-415 thermal etching, 360-362 definition, 360 early investigations, 362-367 equilibrium particle shape, 372-373 models, 360-361, 362, 368-371, 377-383 particle splitting, 375-377 volatile species transport, 373-375 Ethane, oxidation, 220 Ethanol absorption, 190 dehydration, 167, 183-184, 190 oxidative dehydrogenation, 296 Ethene adsorbed, 14 formation, 183-184 hydrogenation, I oxidation, 226, 303 Wacker process, 226 vibrational spectra, 63-74 on cobalt, 59, 60 on copper, 60, 62 on gold, 60, 62-63, 64 on indium, 59-62, 66 on nickel, 50-57, 66 on palladium, 43-50, 63-64 on platinum, 3 1-43, 64 on rhodium, 57-59, 63-64, 66 on ruthenium, 59, 60, 63-64 on silver, 60, 62-63, 64 Ethyl benzene, oxidation, 312 Ethylcyclopropane, oxidation, 324 Ethylene; see Ethene Ethylidyne, 14, 31, 34, 36, 38, 53, 65, 67, 84 Ethyne, 65 ETS-4, 287,292-293 ETS-10, 287, 292-293 EURO TS-I,31 1

EXAFS (extended X-ray absorption fine structure), 28, 137 Extra-framework titanium, 293

F Faceting, kinetics, 380-382 Foils, etching, 366 Fourier-transform (FT) spectroscopy, 4, 339-342 DRIFT (diffise-reflection), 6, 8 Free radicals, catalytic etching, 396-406 C Glycols, oxidation, titanium silicate-catalyzed, 298-303 Gold ethene vibrational spectra on, 60, 62-63, 64 oxide-supported metal catalysts, 12

H n-Heptane, oxidation, 3 13 Heteroatoms, 118, 120, 121 Heteropolyacids defined, 117 heteroatoms. 1 18, 120, 121 Prins reaction, I56 supported, 149-150 Heteropolyanions, 1 13, 1 17, 1 19-12 1 Heteropoly blues, 191 Heteropoly compounds absorption, 179-180, 190-191 acid-catalyzed reactions heterogeneous, 161-178 liquid phase, 150-16 I acidic properties in solid state, 141-150 in solution, 139-141 catalysis, 114, 116-117, 190-191 as catalyst, 113-116, I 17, 223-232 characterization electronic spectra, 135-136 ESR, 137-138, 192 EXAFS, 136 infrared spectroscopy, 128-I3 I , I94 NMR spectroscopy, 132-135, 149, 185, 192 Raman spectroscopy, 131, 150

426

INDEX

STM, 137 TEM, 137 XPS, 138-139 chemical synthesis with, 221-223 defined, 114 electrocatalysis, 233, 239-240 hybrid catalysts, 223-232 nomenclature, 1 I8 oxidation reactions liquid phase, 200-210 by solid compounds, 2 10-22I photocatalysis, 233-239 pseudoliquid phase, 178-19 I redox properties, in solution, I9 1-1 93 in solid state, 193-200 stability, 127-128 structure, I 18- 1 19 primary, 119-123 secondary, 123-124 tertiary, 124-I26 synthesis, 126-127 terminology, I 17-1 18 Heteropolymolybdates, 191 Heteropolytungstates, 191, 235, 239 Heteropolytungsticacids, 142 Hexa-l,3-diene, vibrational spectra, 97 Hexane isomerization, 176 oxidation, 297, 313 Hexenes epoxidation, 232 oxidation, 305, 308 vibrational spectra, 93-95 HEELS (high-resolution electron-energy-loss spectroscopy), 3 Hybrid catalysts, 223-232 Hydration, with heteropoly compounds, 153-156, 165-168 Hydrocarbons aromatic, oxidation, 3 I 1-3I3 olefin-to-paraffin ratio, 190-I9 1 paraffins oxidation, 220 photoreactions, 235 synthesis from methanol, 168-169 Hydrocarbons, adsorbed group vibration frequency, 17-26 history, 2 low-energy electron dihction, 29 near-edge X-ray absorption fine shucture, 29

nuclear magnetic resonance, 27-28 photoelectron dihction, 29-30 scanning tunnel microscopy, 30 temperature-programmeddesorption, 26-27 ultraviolet photoelectron spectroscopy, 28 vibrational spectroscopy, 102-104 alkenes, branched chain, 97- 101 alkenes, cyclic, 101 alkenes, linear, 31-95, 102-103 dienes, linear, 95-97 X-ray absorption fine structure, 29 X-ray photoelectron spectroscopy, 28 Hydrogenation, adsorbed hydrocarbons butene, 80-81, 87 ethene, 39-40.49-50, 55-56 propene, 79-80, 88, 102 Hydrogen peroxide oxidation catalyzed by heteropoly compounds, 203-209 catalyzed by titanium-containing materials, 256, 296-3 17 titanium peroxo complex, 320-326 Hydrotalcite, 232 Hydroxyacetone, synthesis, 302 I -HydroxyJ,7-dimethylocta-2,6-diene, epoxidation, 307 HZSM-5, 168, 295 dehydration of butyl alcohols over, 335-356

I IETS (inelastic electron-tunneling spectroscopy), 6, 8 Infrared spectra, 102-104 alkenes, branched-chain 2,3-dimethylbut-2-ene, 100-101 2-methylpropene, 97- 100 alkenes, cyclic, 101 alkenes, linear, 102-103 butenes, 71, 80-91 ethene, 31-74 hexenes, 93-95 pentenes, 9 1-93 propenes, 74-80 dienes, linear buta-1,3-diene, 95-97 hexa-LS-diene, 97 diethyl ether, 184-185 heteropoly compounds, 128-131, 159

427

INDEX Infrared spectroscopy, 2-4, 8, heteropoly compounds, 128- I3 I, 194 INS (inelastic neutron scattering), 6-7, 8 Intercalated polyanions, 23 1-232 Iridium ethene vibrational spectra on, 59-62, 66 oxide-supported metal catalysts, 10 Iron, oxide-supported metal catalysts, 10 Isobutane alkylation, 173- I74 synthesis, 176, 230 Isobutene, vibrational spectra, 97 Isobutylene alkylation of p-cresol with, 22 1 hydration, 153-154, I67 separation, 155, 189 Isobutyl propionate, decomposition, 160 Isobutyric acid, dehydrogenation, 197, 2 18-220 Isopolyacids, defined, 117 Isopolyanions, defined, 1 13, I I7 Isopropyl alcohol, photoxidation, 238-239 lsopropylcyclopropane, oxidation, 324

K Keggin anions, I 19- 120 Keggin structure, heteropoly compounds, 119-121, 128-131, 139, 232 Kinetics butyl alcohol dehydration, 339-344 catalytic etching, 379-383 in pseudoliquid phase, 182- 184 thermal etching, 370-371

L Lacunary Keggin anion, heteropolyanions, 121 LEED (low-energy electron diffraction), chemisorbed hydrocarbons on metals, 29

M Mars-Van Klevelen mechanism, 2 I 1 Mesitylene, synthesis, 158 Metal carbonyls, formation, 375 Metal catalysts MSSR, 4, 15-17 vibrational spectroscopy of adsorbed species, 7-12, 30-31 Metallacycle mechanism, 323-324

Metalloporphyrins, 223 Metal oxides, catalytic etching, 390-396 Metal-surface selection rule (MSSR), 4, 15-17 Metal transport, volatile metal carbonyls, 375 Metal wool, 383 Methacrolein, oxidation, 2 15-218 Methacrylic acid, synthesis, 215 Methane, oxidation, 220 Methanol conversion into hydrocarbons, 168- I69 dehydration, 296 oxidation, 296, 299 photooxidation, 234 3-Methylbut-2-en- 1-01, oxidation, 307 Methyl tert-butyl ether (MTBE), synthesis heteropoly compound catalyzed, 158, 160, 177-178, 187 TS- 1 catalyzed, 295 Methylidyne, 65 2-Methylpentane, oxidation, 3 12-313 3-Methylpentane, oxidation, 3 13 Methyl-N-phenylcarbamate,synthesis, 226-227 2-Methylpropene, vibrational spectra, 97-1 00 u-Methyl styrene, oxidation, 309 Mixed oxides, titanium and silicon, 257-267 Monomethylamine, dehydrogenation, I78

N NEXAFS (near-edge X-ray absorption fine structure), chemisorbed hydrocarbons on metals, 29 Nickel catalytic etching, 386 ethene vibrational spectra on, 50-57, 66 oxide-supported metal catalysts, 10-1 I Nitrobenzene, carbonylation, 226-227 NMR (nuclear magnetic resonance) butyl alcohol dehydration, 349-353 chemisorbed hydrocarbons on metals, 27-28 heteropoly compounds, 132-135, 149, 185, I92 Nomenclature, heteropoly compounds, 1 18

0 n-Octane, oxidation, 3 13 I-Octene, oxidation, 305, 308, 309, 313 Olefins

428

INDEX

epoxidation, 232 oxidation, 303-3 10 Olefin-to-paraffi ratio, hydrocarbons, 190- 191 Osmium, spacecraft in low orbit, 414 Oxidation, catalysis by heteropoly compounds, 200-22 1, 225-228 by titanium silicate, 296-3 17 Oxide-supported metal catalysis, 7-12, 30-3 1 Oximes, transformation into lactams, 295 Oxoanions, defined, I 17 Ozone, oxidation with, 296

P Palladium dehydrogenation, 10 1- 102 ethene vibrational spectra on, 43-50, 63-64 heteropolyanion-supported, 229-230 oxide-supported metal catalysts, 9, 10-1 1 particle splitting, 377 Particle shape, etching, 372-373 Particle splitting, catalysts, 375-377 PED (photoelectron diffraction), chemisorbed hydrocarbons on metals, 29-30 n-Pentane, oxidation, 3 13 2-Pentanol, oxidation, 299 Pentenes oxidation, 305 vibrational spectra, 91-93 Peroxo complexes, 320-326 Peroxo radical, 326 Phenol alkylation, I89 oxidation, 3 10-31I synthesis, 312 Phenylacetylene, hydration, I55 Photocatalysis, heteropoly compounds, 233-239 Photoelectron spectroscopy, chemisorbed hydrocarbons on metals, 28 Plasma etching, 407-414 Platinum etching, 363, 367, 387-389 foils, 396, 398-403 plasma etching, 412-413 ethene vibrational spectra on, 31-43,64 faceting, 382, 383 foils, etching, 396, 398-403 heteropolyanion-suppod 230-23 1 oxide-suppoxted metal catalysts, 9, 10-1 1, 14 particle shape, 372-373,374

particle splitting, 375 reactivity, I0 I redispenion, 373, 375 Platinum oxide, transport model, 390-393 Polyacids, defined, 1 I7 Polyanions, intercalated, 23 1-232 Polyatoms, defined, 1 I8 Polycrystalline foils, etching, 366 Polyoxoacids. defined, 1 17, 1 18 Polyoxometalates defined, 117, 1 18 transition-metal-substituted,224-225 Polyoxomolybdates, 193 Polyoxotetramethyleneglycol (PTMG), synthesis, 156-157 Polyoxotungstates, 193 Polyoxovanadates, 193 Polysulfone, 167-168 Polytungstic acids, 161 Pore structure, solid heteropoly compounds, 124-126, 163 Prim reaction, heteropolyacid catalysis, 156 Propane, oxidation, 220, 3 13 2-Propan01,dehydration, 165, 182-183, 186-187, 188 Propenes, vibrational spectra, 74-80 Propionaldehyde, cyclotrimerization, 152, 157, 159 Propylene hydration, 153 oligomenzation, 178 oxidation, 303-304 Propylene glycol, oxidation, 302 Propylene oxide, 303-304 Propylidyne, 74, 84 Pseudoliquid catalysis, 114, 116 Pseudoliquid phase heteropoly compounds, I 78- 19 I kinetics, 182- 184 spectroscopy, 184- 186 Pyridine absorption, 119 desorption from heteropolyacids, 142

R RAlRS (reflection-absorption infrared SP~C~SCOP 3,4.8 Y).

429

INDEX hydrocarbons, adsorbed butene, 80-8 1 ethene, 3 1,42,44,62,66 propene, 75, 76, 102 MSSR, 15 Raman spectroscopy, 6, 8 heteropoly compounds, 13I , 150 MSSR, 15 Redox chemistry, heteropoly compounds, 191-193 Reduction, heteropoly compounds, 193 Reoxidation, 198-199 Rhodium ethene vibrational spectra on, 57-59, 63-64. 66 heteropolyanion-supported,228-229 oxide-supported metal catalysts, 10, 1 1 Ruthenium ethene vibrational spectra on, 59, 60, 63-64 oxide-supported metal catalysts, 11

S

SAPO-5, 256 Semiconductors, plasma etching, 408-409 SERS (surface-enhanced Raman spectroscopy), 6, 8, 62-63 Shell process, 297 Silicalite-I, 254, 314 Silicoaluminophosphates,256 Silicon, etching, 408, 409 Silver etching, 363, 364, 365, 385,411-412 ethene vibrational spectra on, 60, 62-63, 64 oxide-supported metal catalysts, 12 Silver alloys, etching, 363, 365 Site symmetry, adsorbed species, 13-14 Spacecraft, etching, 4 14-415 Splitting, platinum parricles, 375 Stability, heteropoly compounds, 127-128 Steroids, esterification, 223 STM (scanning tunneling microscopy) chemisorbed hydrocarbons on metals, 30 heteropoly compounds, 137 Styrene, epoxidation, 201 Sulfides, oxidation, 3 17

Surface area, solid hetempoly compounds, 124-I26 Surface catalysis, 114, I 16

T TEM (transmission electron microscopy), heteropoly compounds, I37 a-Terpinene, dehydrogenation, 201,204 Terrace-ledge-kink (TLK) model, 378, 379 Tetrahydrofiran cleavage, 156 polymerization with heteropoly compounds, 156-157 synthesis, 155-156 Tetramethyllead, 407 Thermal etching definition, 360 early investigations, 362-367 equilibrium particle shape, 372-373 models, 360-361, 362, 368-371, 377-383 particle splitting, 375-377 volatile species transport, 373-375 Thermal gravimetric analysis, heteropoly compounds, 127 Thermodynamics catalytic etching, 379 thermal etching, 368-370 Ti-beta zeolite, 256, 284-285 reactions, 298, 308, 309,3 16 synthesis, 29 1-292 Ti-HMS, 256,285-287, 292, 3 16 Ti-L. 287,292 Ti-MCM-41, 256, 285-287, 292, 310 Ti-MCM-48, 292 Ti-Mordenite, 287, 292 Tin, etching, 41 1 Ti-omega, 287, 292 Titanium peroxo complexes, 320-326 Titanium silicalite-I, 254-255, 257, 267-282 ~ c t i o n295-297,308-310, ~, 314, 315, 319 synthesis, 288-291 Titanium silicalite-2, 256, 282-284 reactions, 295 synthesis, 291 Titanium silicates, 253-257, 326-327 acidity, 3 19-320 catalytic reactions, 193-195

430

INDEX

Beckmann rearrangement, 295 mechanisms, 318-326 methyl-t-butyl ether synthesis, 295 oxidation, 296-317 catalytic sites, 3 17-318 ETS-4, 287,292-293 ETS-10,287,292-293 extra-framework titanium, 293 peroxo complexes, 320-326 synthesis, 288-293 Ti-beta zeolite, 256, 284-285 reactions, 298, 308. 309, 3 16 synthesis, 291-292 Ti-HMS, 256,285-287. 292, 316 Ti-L, 287, 292 Ti-MCM-4 1,256, 285-287, 292, 3 10 Ti-MCM-48.292 Ti-mordenite, 287, 292 Ti-omega, 287,292 Ti-W, 287, 292 Ti-Y, 287, 292 Ti-ZSM-48, 256,284, 291 TS-I, 254-255, 257. 267-282 reactions, 295-297, 308-310, 314, 315, 319 synthesis, 288-291 TS-2, 256, 282-284 reactions, 295 synthesis, 291 TS-48, 316 Ti-W, 287,292 Ti-Y, 287, 292 Ti-ZSM-48, 256, 284, 291 a-Tocopherol acetate, synthesis, 222 Toluene alkylation, 174 oxidation, 3 12 Tosyl hydrazones, oxidative bond cleavage, 3 16-317 TPD (temperature-programmeddesorption), chemisorbed hydrocarbons on metals, 26 Transmission infrared spectroscopy, 3-4 2,4,6-Triethyl-l,3,5-trioxane,synthesis, 157 1,3,5-Trimethylbenzene,alkylation, 161, 162, 169, 172. 189 1,3-Trioxane,polymerization, 157 TS-I, see Titanium silicalite-I TS-2, see Titanium silicalite-2 TS-48, 316 12-Tungstophosphate,formation, 126

U UPS (ultraviolet photoelectron spectroscopy), chemisorbed hydrocarbons on metals, 28

V Vanadium peroxo compounds, 324 VEELS (vibrational electron-energy-loss spectroscopy), 3, 5, 8 hydrocarbons, adsorbed ethene, 31-32, 35, 42, 44,57, 59, 64,66 propene, 74-76, 78 MSSR, IS, 16 Vibrational spectroscopy, 102-104 absorbed species hydrocarbon ligands on metal clusters, 17-26 metal-surface selection rule, 4, 15-I7 site symmetries, 13-14 symmetry of adsorption complex, 14 alkenes, branched-chain 2,3-dimethylbut-2-ene, 100-101 2-methylpropene, 97- 100 alkenes, cyclic, 10 I alkenes, linear, 102-103 butenes, 7 I , 80-9 1 ethene, 3 1-74 hexenes, 93-95 pentenes, 91-93 propenes, 74-80 dienes, linear buta-l,3-diene, 95-97 hexa-l,S-diene, 97 diffise-reflection Fourier-transform infrared spectroscopy (DRIFT), 6, 8 electron energy loss spectroscopy (EELS), 3 high-resolution electron-energy-loss spectroscopy (HREELS), 3 inelastic electron-tunneling spectroscopy (IETS), 6, 8 inelastic neutron scattering (INS), 6-7, 8 infrared spectroscopy, 2-4 oxide-supported metal catalysts, 7- 12, 30-3 1 Raman spectroscopy, 6, 8 reflection-absorption infrared spectroscopy (RAIRS), 3, 4, 8 surface-enhanced Raman spectroscopy (SERS), 6 transmission infrared spectroscopy, 3-4,8

431

INDEX vibrational electron-energy-loss spectroscopy WEEL), 3, 5, 8 Vinogradovite, 255 Vitamin C, synthesis, 222 Vitamin E, synthesis, 222 Vitamin K, synthesis, 223

W Wacker process, 226 Wulff plot, 368, 378

chemisorbed hydrocarbons on metals, 28 heteropoly compounds, 138-139 X-ray absorption, chemisorbed hydrocarbons on metals, 28-29 XRD (X-ray diffraction) catalytic etching, 403 heteropoly compounds, 127 p-Xylene alkylation, 170- I 7 I , 176 photoreactions, 235 Z

X XANES (X-ray absorption near-edge structure), chemisorbed hydrocarbons on metals, 29 XPS (X-ray photoelectron spectroscopy) catalytic etching, 393

Zeolites, 253-254 reactions, 298 synthesis, 291-292 Ti-beta, 256,284-285 Zorite, 255 ZSM-5, 295, 296 ZSM-I 1,295

This Page Intentionally Left Blank