(industry-university Cooperative Chemistry Program Symposia) F. A. Cotton (auth.), John P. Fackler Jr. (eds.)-metal-metal Bonds And Clusters In Chemistry And Catalysis-springer Us (1990).pdf

METAL-METAL BONDS AND CLUSTERS IN CHEMISTRY AND CATALYSIS

INDUSTRY-UNIVERSITY COOPERATIVE CHEMISTRY PROGRAM SYMPOSIA Published by Texas A&M University Press ORGANOMETALLIC COMPOUNDS Edited by Bernard L. Shapiro HETEROGENEOUS CATALYSIS Edited by Bernard L. Shapiro NEW DIRECTIONS IN CHEMICAL ANALYSIS Edited by Bernard L. Shapiro

DESIGN OF NEW MATERIALS Edited by D. L. Cocke and A. Clearfield FUNCTIONAL POLYMERS Edited by David E. Bergbreiter and Charles R. Martin METAL-METAL BONDS AND CLUSTERS IN CHEMISTRY AND CATALYSIS Edited by John P. Fackler, Jr. OXYGEN COMPLEXES AND OXYGEN ACTIVATION BY TRANSITION METALS Edited by Arthur E. Martell and Donald T. Sawyer

METAL-METAL BONDS AND CLUSTERS IN CHEMISTRY AND CATALYSIS

Edited by

John P. Fackler, Jr. Texas A&M University College Station, Texas

SPRINGER SCIENCE+BUSINESS MEDIA, LLC

Library of Congress Cataloging-In-Publication Data

Texas A&M University. Industry-Universlty Cooperative Chem1stry Prcgrao Syopcsiuo C7th : 1989 : Texas A&M Un1versityl Metal-oetal bonds and clusters in cheoistry and catalys1s I ed1ted by John P. Fackler, Jr. p. co. -- (Industry-university cooperative chemistry prcgrao syopcs1al "Proceedings cf the Seventh Industry-Unlversity Cccperat1ve Cheoistry Prcgraa, held March 20-23, 1989, at Texas A&M University, College Station, Texas"--T.p. verse. Includes bibliographlcal references. ISBN 978-1-4899-2494-0 ISBN 978-1-4899-2492-6 (eBook) DOI 10.10071978-1-4899-2492-6

1. Meta 1-oeta 1 bends--Congresses. 2- Meta 1 crysta 1s--Ccngresses. 3. Catalysts--Congresses. I. Fackler, John P. II. Texas A & M Unwers1ty. III. Title_ IV. Ser1es. 00461. T4 1989 90-34855 546' . 6--d c20 CIP

Proceedings of the Seventh Industry-University Cooperative Chemistry Program, held March 20-23, 1989, at Texas A&M University, College Station, Texas

© 1990 Springer Science+Business Media New York

Originally published by Plenum Press, New York in 1990 Softcover reprint of the hardcover 1st edition 1990

All rights reserved No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written permission from the Publisher

PREFACE

This book contains a series of papers and abstracts from the 7th Industry-University Cooperative Chemistry Program symposium held in the spring of 1989 at Texas A&M University. The symposium was larger than previous IUCCP symposia since it also celebrated the 25 years that had elapsed since the initial discovery by F.A. Cotton and his co-workers of the existence of metal-metal quadruple bonds. Cotton's discovery demonstrated that multiple bonding in inorganic systems is not governed by the same constraints observed in organic chemistry regarding s and p orbital involvement. The d orbitals are involved in the multiple bonding description. The quadruple bond involves considerable d orbital overlap between adjacent metal centers. Part I of this series of papers focuses upon the impact of this discovery and describes further contributions to the development of the field. Multiple metal-metal bonding now is known to permeate broad areas of transition metal chemistry. The understanding of metal-metal bonding that developed as a result of the discovery of multiple metal-metal bonding awakened a new chemistry involving metal clusters. Clusters were defined by Cotton to be species containing metal-metal bonding. Clusters in catalysis therefore seemed a logical grouping of papers in this symposium. Clusters play an every increasing role in the control of chemical reactions. Part II of this book describes some of the interesting new developments in this field. In Part III the papers examine the role clusters play in describing and understanding solid state materials. Clusters exist throughout solid state chemistry. Modern structural tools now enable us to examine the properties of these materials. Part IV presents papers which relate to the procedures we use today to understand the clusters themselves, the bonding theories and spectroscopy. The Chemistry Department at Texas A&M University is pleased with the support that comes from industrial members of the IUCCP. Without this support, many of the programs in this department would be less healthy. Furthermore, the program reflects the need to maintain the connection between industrial· science and the "ivory tower" of academia. With well established relationships between the research university and modern chemical industries, we can expect to continue to compete effectively with the rapidly developing chemical science elsewhere in the world. We must do this with great skill if we are to remain competitive with our chemical products. Although several persons helped to make this book a reality and to assure the success of the symposium, there are some that were special. First, I want to thank the committee members who helped me choose the topic and the various lectures. They were Arthur E. Martell, Larry M. Cirjak, William J. Kroenke, Brian Kolthammer, Graham Mott, George Vaughn v

and John Smegal. Professor Arthur Martell was a great help to me on various aspects of the programming and planning. His wife, Mary Martell, also gave special attention to the details. My students also helped with transportation and the audio-visual aids. I am especially grateful to Mrs. Carol Dissen who organized the paperwork associated with the symposium and made sure that the manuscripts were properly treated. Her organizational skills made the whole activity painless for the editor. I am indebted to Mrs. Sherri Sanford for her efforts as the book neared completion. I also wish to express my appreciation to Lord Lewis of Cambridge University and Dean Abraham Clearfield of Texas A&M. Lord Jack Lewis presented a beautiful lecture on the environment which, unfortunately, is not included in this book. He also kept other speakers in fine tune by sitting in the first row of the lecture hall for each talk. His questions led to important answers which made the symposium exceptionally lively. Abe Clearfield presented a significant story of the development of the materials research trust underway at Texas A&M University. This paper also is not included in the book. Information about both topics is available from the editor, however. I will be pleased to respond to your written request.

CONTENTS

TWENTY-FIVE YEARS OF CHEMISTRY SINCE THE DISCOVERY OF THE QUADRUPLE METAL-METAL BOND A Quarter-Century of Metal-Metal Multiple Bonds . . . . . . F. A. Cotton The Multiple Metal-Metal Bond: Twenty-Five Years of Synthetic Serendipity and Structural Discovery Richard A. Walton

1

7

Theoretical Studies of Dinuclear Compounds with Multiple Metal-Metal Bonds . . . . . . Bruce E. Bursten and William F. Schneider

19

Uses of Metal Clusters in Homogeneous and Heterogeneous Catalysis . . . . . . . . . . . . . . . . . . . . Donald J. Darensbourg

41

Reactivity of Dinuclear and Tetranuclear Clusters of Molybdenum and Tungsten . . . . . . . . . . Malcolm H. Chisholm

55

Clusters and Their Implications for Catalysis . . . . . . . . . . . Richard D. Adams

75

Metal Clusters in the Solid State . . . . . . . . . . . . . . . . . Robert E. McCarley

91

CLUSTERS IN CATALYSIS Nature of Bimetallic Clusters . . . . . . . . . J. H. Sinfelt

103

Thermochemical Aspects of Organotransition Metal Chemistry. Insights Provided by Metal-Ligand Bond Enthalpies . . . . . . . . . . . . . . . Michel R. Gagne, Steven P. Nolan, Afif M. Seyam, David Stern and Tobin J. Marks

113

Metal Clusters and Supported Metal Catalysts B. C. Gates

127

Boldface denotes symposium speaker

vii

Mechanistic Features of Carbonyl Cluster Rearrangement Brian F. G. Johnson, Adrian Bott, Robert E. Benfield, Dario Braga, Elisabeth A. Marseglia and Alison Rodger Selective Oxidation Chemistry on Soluble Oxides: A Progress Report . . . . . . . . . . . . . V. W. Day, W. G. Klemperer, S. P. Lockledge, D. J. Main, F. S. Rosenberg, R.-C. Wang and 0. M. Yaghi The Study of Clusters of Polylithium Organic Compounds and Structural Studies of Polylithium Organic Compounds . . . . . . . . . . . . . . . Richard J. Lagow

141

161

171

CLUSTERS IN MATERIALS Organometallic Chemical Vapor Deposition of GaAs and Related Semiconductors Using Novel Organometallic Precursors . . . . . , . . . Alan H. Cowley Surface Chemistry of Mixed-Metal Systems D. V. Goodman Organometallic Chemical Vapor Deposition of Aluminum Nitride and Aluminum Metal David C. Boyd, RichardT. Haasch, Kwok-Lun Ho, Jen-Vei Hwang, Roland K. Schulze, John F. Evans, Wayne L. Gladfelter and Klavs F. Jensen

195

205

215

Solid State Carbon-13 NMR of Metal Carbonyls Brian E. Hanson

231

Surface Chemistry of Metal and Semiconductor Clusters . . . . . . . R. E. Smalley

249

BONDING AND SPECTROSCOPY IN CLUSTERS The Electronic Structure of Metal Dimers and Metal Clusters: The Eighteen-Electron Rule vs. Skeletal Electron-Pair Counting Michael B. Hall Experimental Measures of Metal-Metal Sigma, Pi, and Delta Bonding from Photoelectron Spectroscopy Dennis L. Lichtenberger and Roy L. Johnston Formation, Structure and Luminescent Properties of Metal-Metal Bonded Compounds of the Late Transition Metal and Post,Transition Metal Ions Alan L. Balch

265

275

299

ABSTRACTS The Preparation and Characterization of New Heteropolyoxofluorometalate Anion, [Few 17 o56 F6H5 ] 8 Sadiq H. Wasfi viii

311

Facile Exchange of Terminal, Doubly-Bridging, and Quadruply Bridging Carbonyl Ligands in Solution: Crystal Structure and Solution Dynamics of LWM3 (C0) 12H, L=C 5H5 , C5Me 5 and M=Os,Ru Yun Chi, Sue-Lein Wang and Shie-Ming Peng

312

Thermal Constants and Structure of Tin Clusters Richard W. Schmude, Jr., Karl A. Gingerich and Joseph E. Kingcade, Jr.

314

Reactivity and Isomerization of Mo 2 (ALLYL) 4 Reed J. Blau, Ron-Jer Tsay and Su-Inn Ho

315

Surface Coordination/Organometallic Chemistry of Monometal and Bimetallic Electrocatalysts Ginger M. Berry, Michael E. Bothwell, Beatriz G. Bravo, George J. Cali, John E. Harris, Thomas Mebrahtu, Susan L. Michelhaugh, Jose F. Rodriquez and Manuel P. Soriaga Ambient-Temperature Chloroaluminate Molten Salts: Solvents for Chloro Complex Electrochemistry and for Reductive Condensation Syntheses R. T. Carlin and R. A. Osteryoung

316

318

The Maximum Strength of the Chemical Bond between Two Metal Atoms . . . . . . . . . . . K. A. Gingerich

319

Reactions of (~ 7 -c 7 H 7 )M(~ 5 -c 5 H 5 ), M- Ti or Zr, with Carboxylic and Dithiocarboxylic Acids S. A. Duraj, M. T. Adras, R. A. Martuch and S. SriHari

320

Model Hydrodesulfurization Systems: Reactions of Sulfur Containing Molecules on Ni(llO) D. R. Huntley

321

Electrochemical Studie~ of Triangular Niobium Cluster, Nb30 2 (so4 ) 6 .3H2o -, in Sulfuric Acid . . . . . V. Sayers, T. Batten, M. May and V. Katovic

322

Reactivity of Dithioethers toward [Re 2H8 ] 2 - . J. Gregory Jennings and Gregory L. Powell

323

Theoretical Investigations of the Metal-Metal Interactions within the Trinuclear Au 2Pt(CH 2 (S)PH 2 ) 4 Complex . • • . . . . Andrew L. Sargent and Michael B. Hall Structural and Theoretical Studies on Heteronuclear Transition-Metal Clusters Containing the Alkylidyne Ligand . . . . . . . P. Sherwood, M. B. Hall. J. C. Jeffery and F. G. A. Stone

324

325

Electronic Structure and Nature of Bonding in Transition Metal Dimers • • . . . . . . . . . . . . . . . . . Irene Shim

326

Theoretical Calculations on the Interaction of Bridging Carbonyls with Transition Metal Dimers . . . . . Charles Q. Simpson II and Michael B. Hall

327 ix

Bimetallic Hydroformylation Catalysis . . Scott A. Laneman and George G. Stanley 252 cF-Plasma Desorption Mass Spectra of Very . . . . . Large Clusters J. P. Fackler, Jr., C. J. McNeal and R. E. P. Winpenny Systematic Kinetic Studies of Associative and Dissociative Reactions of Substituted Metal Carbonyl Clusters: The Intimate Mechanisms N. M. J. Brodie, Lezhan Chen and A. J. Poe Mixed Pd-Au and Pt-Au Cluster Compounds . Louis H. Pignolet Stability of Small Biclusters of Transition Metals . . . . . . . . with Semi-Conductors J.E. Kincade, Jr., I. Shim and K.A. Gingerich The Topology of the Total Charge Density in Binuclear Transition-Metal Complexes that Formally Contain . . . . . Metal-Metal Bonds . . . . . . . . Preston J. MacDougall and Michael B. Hall

328

329

330 332

333

334

The Effect of Carbonyl Ligands on Osmium and Ruthenium . . . . . . Metal-Metal Bonds . . . . Ann E. Miller and William A. Goddard III

335

Contributors

337

Index . . . .

339

X

A QUARTER-CENTIJRY OF METAlrMETAL MULTIPLE BONDS

F. A. Cotton Department of Chemistry Texas A&M University College Station, Texas 77843

A lot can happen in twenty five years.

Actually, the roots of my

work on metal-metal multiple bonds go back even more than twenty five years.

As early as the late 1950's I wondered about how many instances

of definite, confirmed M-M bonds (even single ones) were to be found in the literature.

Very few, I quickly discovered.

er it would make sense to seek out more.

I then wondered wheth-

However, it was not until I

had acquired the knowledge and equipment necessary for X-ray crystallography that I could actually embark on such a program. happened that chance favored a prepared mind.

It also

Although I undertook a

structural study of "CsReC1 4 ," without anticipating that it would turn out to be Cs 3 [Re 3Cl 12 ], the interpretat ion of that structure in terms of metal-metal bonds, but particularly double bonds, was almost automatic ta me.

1

Chance, or should one say serendipity, continued to write the scenario for another few months.

As recounted in detail elsewhere,2 we 2"discovered" compounds containing the quadruply-bonded Re 2 Cl a ion 3while attempting to explore further the chemistry of the Re 3Cl 12 ion. 2We reported our preparat ion of Re 2 Cl a compounds and the assignment of a bond of order 4 in 1964 and it is on that basis that one can consider 1989 as the twenty fifth anniversary of the discovery of a consistent pattern of multiple bonding between metal atoms. Reports of compounds that were entirely misformulated at the time reported but can now known to contain quadruple bonds antedate 1964 by various periods.

Chromium(II) ace tate , Cr2(02CCH3)4 was described in

Metal-Metal Bonds and Clusters in Chemlstry and Catalysls Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

1844.

3

4 5 The earliest reports ' of compounds with Mo-Mo quadruple bonds

appear to be those of several molybdenum(!!) carboxylates (whose structures were all incorrectly formulated) and "MoBr 2 (NMe 2 )•NHMe 2 , which we now know to be Mo 2 Br 4 (NHMe 2 ) 4 .

We also now recognize that the first

compounds containing Pt-Pt single bonds within a square-prismatic 6

arrangement of ligands were reported in 1905. The first compound con3taining the Tc 2 Cl 8 ion was reported in 1963 but without any suggestion of its true nature. 7 Finally, of course, it is now well known that chemists in the Soviet Union began reporting compounds that were later 228 shown to contain Re 2 Cl 8 or Re 2 Br 8 ions as early as 1952. The early growth of the field was slow, but soon revealed itself to be exponential, as shown in Fig. 1, for close to two decades.

In the

past few years activity has leveled off at about 140 publications per year. It is interesting to see which elements have been most actively studied, once the nature of the bonding was recognized. marized in Fig. 2.

This is sum-

Rhenium, of course, had the early lead, but around

1970 it was surpassed by molybenum.

Molybdenum is truly extraordinary

in its facility for forming metal-metal bonds of many kinds, including a great variety of cluster compounds in which there are bonds of order one, or

thereab~ut.

It enjoys its place as front runner in multiple

bond formation in part because it has two arenas in which to play: quadruple bonds and triple bonds, of different types structurally. The quadruple bonds are found in compounds that have (at least in a formal sense) a fourfold axis while the triply bonded compounds are characteristically based on threefold symmetry.

Tungsten has poten-

tially the same advantage, but does not compete with molybdenum because of the relative instability and reactivity of its quadruple bonds. Rhenium, incidently, also forms extensive series of both quadruple and triple bonds, but both in the compass of fourfold symmetry. Turning to other features of Fig. 2, we note that rhodium is the other member of the top four, and has been for over a decade. A pedant could argue that the Rh 2 4+ species do not merit mention here at all because they do not contain metal-metal multiple bonds.

Another pedant

might try to justify their inclusion because some are (and others probably could be) the parents of Rh 2 5+ species in which there is a bond order of 1.5 based on the loss of an antibonding electron on going from

2

150 140 130 120

110

I

100 90

~ 0

80

ti

70

~ m

60

50 40

I

~

LL

0

ffi

N = e xp LQI57(Y-1952)]

~~.~ 1965

1970

1975

1985

1980

YEAR Fig. 1. The approximate number of publications on all types of M-M multiple bonds each year, 1964-1988.

200

Mo

w

100 80

Rh Re

60

40

Ru pt

Os

20

10

Tc 5

To 65

Fig. 2.

"66-'70

71-75

76-80

"81-85

86-90 (est)

The approximate number of publications dealing with different M-M bonds. Note logarithmic scale.

3

Rh 2 4+ to Rh 2 5+

As a real-world, practical chemist, I justify their

inclusion because their electronic structures (and molecular structures) are simply logical extensions of those in which the M-M bond orders are higher, as shown by the following progression: Mo 2 (0 2 CR) 4

0 2:rr4cS2

bond order

4

Ru 2 (0 2 CR) 4

0 2:rr4cS2cS* 2:rr* 2

bond order

2

Rh 2 (0 2 CR) 2

0 2:rr4cS2cS* 2:rr*,

bond order

1

As Fig. 2 shows, work on ruthenium, osmium and platinum species has been growing steadily. Technetium has a somewhat restricted role because of the nuisance factor (i.e., its low-level radioactivity) involved in working with it.

Western chemists have almost completely avoided it and

the small literature that exists is largely produced by Soviet chemists. The element chromium has not been included in Fig. 2, partly because it is hard to determine a temporal point of departure.

Prior to

the present era a number of papers dealt with Cr 2 (0 2 CR) 4 L2 compounds. However, it was with the discovery of the supershort Cr-Cr bonds (<2.0 A) in 1977, 9 that the study of Cr-Cr multiple bonds acquired an enormously greater interest.

These super-short bonds are fully com-

parable in electron density with their Mo and W cogeners, and, indeed, surpass them in this respect.

This extraordinarily high bond electron

density has lead to very formidable problems in calculating their electron
V2

,

Nb 2

,

Mo 2 and Ru 2 , where electron correlation problems are what

Texans call humongous. Until very recently, chromium was the only element in the first transition series to form M-M bonds analogous to those of its heavier congeners.

In general terms, we can understand the reluctance of the

other first series elements to do so.

Their d orbitals are small and

thus interatomic overlaps are relatively poor while intraatomic d-d interactions are strong, thus making it energetically costly to reach the appropriate valence states. Recently, however, we have obtained a 4+ 11 4+ Co 2 analog to certain Rh 2 species and it contains the shortest Co-Co bond ever reported, 2.265A.

This encourages us to believe that

with judicious choice of ligands we may succeed in forming other M-M bonds in the first series, such as V=V 4+ triple bonds, Fe-Fe 4+ double 6+ bonds, and even Mn•Mn quadruple bonds.

4

The region of the d-block in which multiple M-M bonds along 3- or 4-fold molecular axes may be formed has also been extended in other directions. The first Ir-Ir 4+ compound that is isostructural with a rhodium analog has been made 12 and compounds containing Nb•Nb 4+ and Ta=Ta4+ triple bonds have been synthesized and found to be extremely stable. 13 To sum up, it is clear that over a quarter of a century an entirely new and increasingly important aspect of transition metal chemistry has been discovered and explored.

Moreover, recent work shows that in

addition to continued evolution of such chemistry for some 8 to 10 elements about which a great deal is already known, there are quite attractive and realistic possibilities for extending the scope to other elements. In conclusion, it is fitting for me to thank the National Science Foundation for unfailing support of our work in this field (as well as others) over all of the past quarter of a century.

A new grant,

recently awarded, will continue that support into the future.

Certain

phases of the work have also been supported, in the years since 1972, by funds from The Robert A. Welch Foundation.

Finally, in recent years

Texas A&M University has provided important facilities and services through the Laboratory for Molecular Structure and Bonding.

REFERENCES 1.

J. A. Bertrand, F. A. Cotton and W. A. Dollase, J. Am. Chern. Soc., 1963, 85, 1349.

2.

F. A. Cotton and R. A. Walton, Multiple Bonds Between Metal Atoms, John Wiley & Sons, New York, 1982, pp 4 et

3.

~-

E. Peligot, C. R. Acad. Sci., 1844, 19, 609; Ann. Chim. Phys., 1844, 12, 528.

4.

E. W. Abel, A. Singh and G. Wilkinson, J. Chern. Soc., 1959, 3097; T. A. Stephenson, E. Bannister and G. Wilkinson J. Chern. Soc., 1964, 2538.

5.

D. A. Edwards and G. W. A. Fowles, J. Less-Common Metals, 1962,

6.

H. Blondel, Ann. Chim. Phys., 1905, !L 110.

7.

J. D. Eakins, D. G. Humphreys and C. E. Mellish, J. Chern. Soc.,

B.

V. G. Tronev and S. M. Bondin, Dokl. Akad. Nauk SSSR, 1952, 86, 87.

~.

512.

1963, 6012.

5

9.

F. A. Cotton, S. A. Koch and M. Millar, J. Am. Chern. Soc., 1977, 99, 7273.

10. SeeR. D. Davy and M. B. Hall, J. Am. Chern. Soc., 1989, 111, 1268 for the most recent calculations. 11. F. A. Cotton and R. Poli, Inorg. Chern., 1987, 26, 3652. 12. F. A. Cotton and R. Poli, Polyhedron, 1987,

~.

1625.

13. F. A. Cotton, M. P. Diebold and W. J. Roth, J. Am. Chern. Soc., 1987, 109, 5506.

6

THE MULTIPLE METAL-METAL BOND: TWENTY-FIVE YEARS OF SYNTHETIC SERENDIPITY AND STRUCTURAL DISCOVERY Richard A. Walton Department of Chemistry Purdue University West Lafayette, IN 47907 INTRODUCTION The development of the field of multiple bond chemistry had as its genesis the studies in the early 1960's by Cotton and co-workers that established the existence of salts of the 1rReX4]"ln anions (X= Cl or Br) that differed in nuclearity (n = 2 or 3). 1 •2 The structural characterization of the [Re3X12l3- (!) and [Re2Xs] 2 - (!) anions, and the recognition that these species (and closely related ones) contain Re-Re bonds of

2

Q

Re

Q

Cl or Br

order 2 and 4, respectively, were the seminal discoveries that provided the impetus for the flood of later work which demonstrated that metalmetal multiple bond chemistry is an important feature in the chemistry of many of the transition elements. Multiple bonds between metal atoms are not only of relevance to the field of metal cluster chemistry,3• 4 but they constitute one of the important classes of multiple bonds in general, the others being multiple bonds between the main group elements, and multiple metal-ligand bonds.5 Multiple metal-metal bonds of orders 4, 3.5, 3, 2.5 and 2 are all well documented. Their place in the scheme of things is represented below.

Metal-Meta/ Bonds and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press. New York, 1990

7

o=o

M=o

M

N=:N

M:=N

M=M

RN=NR

M=NR

M=M

RC=:-CR

M=CR

M:.:.:M

R2 c=cR2

M=CR2

M=M

M

The close relationships that can exist in some instances between these different species is illustrated by the metathesis reaction between W2(0~-Bu)6, a complex which contains a W=W bond, and aliphatic acetylenes to give (~-Bu0)3W=CR.7 As we commemorate the twenty-fifth anniversary of the quadruple bond, as first recognized in [Re2Clsl2- ,*I am pleased to have this opportunity to reflect upon some of our own contributions in the area of metal-metal multiple bonding. In fact, my own involvement began way back in the fall of 1965 when I arrived fresh off-the-boat (literally!) in Boston Harbor, at the doorstep of M.I.T., to begin a 12-month post-doctoral stint with Professor F. A. Cotton. At the time, I never dreamed that this would lead to the stimulating collaboration and friendship that has lasted for almost twenty-five years.

In hindsight, my involvement in this area of research was also something of an accident. I had decided to avoid working on any problem relating to the recently discovered quadruple bond since a large proportion of Al's group was already gainfully employed in this area and I judged the opportunities for quick and useful new results to be somewhat limited (some misjudgement!). Apparently, my "escape" was to be in carbonyl chemistry, but after several dismal and abortive attempts to do something useful, and fearing Al's displeasure (post-docs were of course expected to be productive very quickly!) I thought I had better find something to do that worked. Accordingly, I decided to try to clarify some of the rather messy chemistry that had been reported on the reactions of the trinuclear halide Re3Cl9 with bidentate ligands. While a certain amount of headway was made in this project,B the most useful development from my point of view was discovering the effectiveness of 2,5-dithiahexane (CH3SCH2CH2SCH3, dth) as a ligand, in spite of its rather objectionable odor. As a consequence, I began to look at the reactions of dth with the salt (n-Bu4N)2Re2Cls very early in 1966, and discovered that the course of the reaction depended, to a large extent, upon the reaction conditions. Mild conditions with methanol/cone. HCl as solvent afforded Re2Cl6(dth)2,9 whereas prolonged reflux in this solvent, or preferably in acetonitrile, gave red-black dichroic crystals of Re 2cl 5 (dth) 2 ,9-ll a formally mixed-valent complex whose structure is represented in Fig. 1. *It's also the twenty-sixth anniversary of the metal-metal multiple bond, since the Re-Re bond in [Re3Cl12l3- was the first such bond to be explicitly recognized (J. A. Bertrand, F. A. Cotton and W. A. Dollase, ~ Am. Chern. Soc., 1963, 85, 1349); although many years before, Linus Pauling had speculated on the existence of metal-metal bonds of order greater than unity (L. Pauling, Chemical & Engineering News, 25:2970 (1947)).

8

Fig. 1.

The structure of RezCls(dth)z. (Refs. 10 and 11).

The structure determination of this unusual molecule revealed the presence of a very short Re-Re bond (2.293(2)A) and a stagrered geometrical arrangement of [ReCl4] and [Re(dth)zCl] units. 10 • 1 It was remarkable that so strong a metal-metal bond would be preserved in such an unsymmetrical structure. Based upon these structural features and the observed paramagnetism of this complex (one unpaired electron), we suggested that it be considered as the "zwitterion" Cl 4 ReiiiReii(dth)zCl i.e. cl 4Re-te(dth)zCl. In this way we could view this complex as possessing a Re-Re triple bond, since the a 2 ~ 4 bonding configuration of the [RezClg]2- parent is retained, while the three remaining ".5" electrons are non-bonding and are localized on the two Re atoms (one on the Re(III) center, two on the Re(II)). This was the first metal-metal triple bond, albeit an unusual one. At the time, this result not only demonstrated to me the difficulty of predicting the reaction products of the octahalodirhenate(III) anions, but also the indispensible role that X-ray crystallography would continue to play in structurally characterizing molecules such as Re2Cls(dth)2. The synthetic "serendipity factor", which is still prevalent in our explorations of the reactions of compounds that contain multiple metalmetal bonds, continues to add to the excitement and challenge of this chemistry. Incidentally, our orig!nal suggestion that RezCls(dth)z could best be represented formally as Cl4Re-Re(dth)zCl, arose out of the anticipation that the alternative Re(IV)-Re(I) formulation (i.e. non-zwitterionic) would be less likely because of the great disparity in oxidation numbers. While this is only a formalism, nonetheless, the Re(IV)-Re(I) case iz certainly not untenable since this would simply entail making one of the components to the triple bond a Re(IV)~Re(I) dative contribution (i.e. This point will be considered again a little later. Suffice it Re~ Re). to say, that if we use the Re(IV)-Re(I) oxidation state formalism, then RezC1 5 (dth)z would constitute the first example of an intramolecular disproportionation of a metal-metal multiple bond.

9

THE TRINUCLEAR HALIDES Re3X9 REVISITED. Following the end of my post-doctoral year at M.I.T. and return to England in the fall of 1966 (to a faculty position at the University of Reading), I became involved primarily in studies of the reaction chemistry of the early transition metal halides of Group 4-6 and largely abandoned the field of multiple bond chemistry for the time being. There were however a couple of exceptions. First of all, Al and I got involved in a collaborative study of the Raman spectra of multiply bonded dimetal complexes ([Re2XsJ 2 -, Re2(02CCH3)4X2 (X= Cl or Br), [Mo2Clg] 4 - and Mo2(02CCH3)4). These were the first measurements of the v(M-M) modes of metalmetal multiple bonds. Although my contribution to this work was completed by early 1969, the results of this study were not published until mid1971.12 Our second re-entry into multiple bond chemistry involved studies of the reactions of Re3Cl9 with pyridine and substituted pyridines. I had been intrigued by the observation that while Re3Cl9 (and Re3Brg) reacted with most a-donors (L) to give dark red-purple colored complexes of stoichiometry Re3X9L3, reaction with pyridine gave a green product.l3 We were able to show that this was the reduced rhenium(!!) complex [Re3Cl6(py)3ln and proposed 14 that it had a "polymer of trimers" structure, the polymerization occuring through Re-Cl-Re bridges i.e. { [Re3Cl3]Cl6/3(py)3ln (1). Similar products were obtained with the use of substituted pyridines 14 and with Re3Brg.lS Alternatively, these complexes may actually be hexanuclear clusters, i.e. Re6X12(py)6• perhaps with the type of trigonal prismatic structure (4) that has recently been encountered in the technetium clusters~[Tc6Cl12l2- and [Tc 6 cl 14 ]3- _16,17 The properties of [Re3X6(py)3ln are consistent with this proposal. In actuality, our studies of the Re3Cl9/pyridine systems, which had commenced in 1968, were still underway when I returned to the U.S. in March of 1969, to assume my present position at Purdue University. In abandoning my student (D. G. Tisley) at that time, I was confident that he would see the project through to completion, which he did. Some little

4

3 QRe

10

Qc1

or Br

Q)py

Fig. 2.

The structure of ReJClg(dppm)J.

(Ref. 20)

time later, we began to investigate the reactions of ReJClg and ReJBrg with phosphines to see whether these ligands might also bring about reduction of the ReJ9+ core of these trinuclear halides. It had already been established that phosphines (PRJ) react with ReJXg under mild conditions to give the maroon-purple adducts ReJClg(PRJ)J,lB and in one instance, namely, PRJ = PEt2Ph, its crystal structure had been determined.l9 The very long Re-P bonds (ca. 2.70 A) are consistent with weak binding of the phosphine ligands due to steric crowding of the ligands about the Re atoms which leads to substantial non-bonding repulsions between the chloride and phosphine ligands. Incidentally, recent X-ray crystal structure determinations on ReJClg(PMeJ)J and ReJClg(dppm)J (dppm = Ph2PCH2PPh2) 20 have substantiated the results and conclusions of the earlier structure result. The structure of ReJClg(dppm)J. in which there are three monodentate dppm ligands, is shown in Fig. 2. When Re3Clg is reacted with the monodentate phosphines PEt3, P-n-PrJ or PEt2Ph in refluxing acetone or ethanol for period~ of several days, the triply bonded dirhenium(II) complexes Re2Cl4(PR3)4 are obtained in yields of up to 50%.2 1* When the phosphines PMePh2 and PEtPh2 are used, the complexes Re2Cl5(PR3)3 are formed, again in yields up to 50%. 21 In the analogous reactions between ReJBrg and P-n-Pr3 or PEt2Ph, the triply bonded complexes Re2Br4(PR3)4 are formed.T 5 Once we recognized the identity of these products, more logical synthetic routes could be devised starting from [Re2XsJ 2 - and proceeding through Re2X6(PR3)2, viz.,

This sequence of reactions involves the reduction in the metal core from Re 26+ to Re 25+ to Re24+, with an accompan2in~ change in electronic configuration from u2~462 to u2~ 4 626*1 to u2~ 4 6 6* i.e. the Re-Re bond order *We have shown recently that when PMe3 is used, the complex Re2Cl4(PMe3)4 is formed.

11

p

Fig. 3

The structure of the Re 2c1 4 P4 core of Re2Gl4(PEt3)4. (Ref. 24).

decreases from 4 to 3.5 to 3. These reactions terminate at a stage (Re2X4(PR3)4 or Re2X5(PR3)3) that is dependent upon the halide (X - Gl or Br) and the basicity of the phosphine. The preceding discoveries were important for several reasons. First, they demonstrated the ability of the dirhenium core to undergo redox changes without disruption of the dimetal unit. This was later substantiated by our discovery that the Re2X4(PR3)4 complexes could in turn be oxidized reversibly to [Re2X4(PR3)4]+ and [Re2X4(PR3)4] 2+, both by chemical and electrochemical means,22,z3 thereby reversing the aforementioned bond order change. The structure of the Re2X4P4 cores in the species [RezX4(PR3)4]n+ (n = 0, 1 or 2) remains essentially unchanged throughout the series, as shown by crystal structure determinations on Re2Gl4(PR3)4 CPR3 = PEt3 or PMe2Ph), [Re2Cl4(PMezPh)4]PF6 and [Re2Gl4(PMe2Ph)4](PF6)2.2J,Z 4 The structure of the RezGl4P4 core, as present in Re2Gl4(PEt3)4, is shown in Fig. 3. The aforementioned redox changes have some interesting consequences as we shall see later. A second point of interest relating to the conversion of Re3X9 to Re2X4(PR3)4, via Re3Xg(PR3)3, concerns the structure of the pyridine derivatives [Re3X6(py)3ln discussed earlier. If these complexes have a hexanuclear cluster structure Re6Xlz(py)6, closely akin to that shown in~ (vide supra), then we can consider them as 3D-electron clusters and perhaps formulate them as containing the cluster unit ~. by analogy with the bonding treatment accorded the hexanuclear technetium clusters like [Tc6Gllz] 2 - by Wheeler and Hoffmann. 1 7 In support of this contention, we can cite the conversion of fRe3Gl6(py)3]n to triply bonded RezGl4(PEt3)4 upon its reaction with PEt3. 21 While we have as yet been unsuccessful in our attempts to isolate Re6Xlz(PR3)6 as intermediates in the conversion of Re3Xg(PR3)3 to RezX4(PR3)4, we think there is good reason to believe they exist. Indeed, there would then be a close analogy to the relationship between quadruply bonded MozGl4(PR3)4 and the tetranuclear clusters of the type Mo4Gla(PR3)4 in which the Mo 4 8+ core can be represented as in 6.25

EXPLOITING THE REDOX ACTIVITY OF THE Re 24+ GORE. In the previous section it was pointed out that the Rez4+ core is very easily oxidized in two one-electron steps to Re 25+ and Re 26+. The extent to which this redox behavior can be utilized in the reduction

12

0

5

0

0

-

0

6

of organic molecules has recently become of interest to us. However, one unfortunate property of these multiply bonded dirhenium complexes is the ease with which the dimetal unit is cleaved in the presence of ~-acceptor ligands (such as CO and isocyanides) and even some u-donors.26 This is especially true in the case of [Re2XsJ 2 -, Re2X6(PR3)2 and Re2X4(PR3)4 (PR3 = monodentate phosphine) but much less so when a bidentate phosphine ligand such as Ph2PCH2PPh2 is present. The latter type of ligand can form an intramolecular bridge between the two metal atoms and so enhance the stability of the unit. A molecule that nicely fits the bill is Re2Cl4(~­ dppm)2 and its bromide analogue; these complexes can serve as excellent reagents in dirhenium chemistry.27 Although Re2Cl4(~-dppm)2 reacts with CO and isocyanide ligands to produce a variety of interesting products,2B-32 in no instance have we been able to bring about the reductive coupling of CO or RNC. In the case of the reaction between CO and Re2Cl4(~-dppm)2 for example, the complexes Re2(~-Cl)Cl3(~-dppm)2(CO) (an A-frame-like molecule), Re2(~-Cl)(~­ CO)Cl3(~-dppm)2(CO) and [Re2(~-Cl)(~-CO)Cl2(~-dppm)2(C0)2l+ are formed in sequence, but even in the presence of other reducing agents and electrophilic reagents (to activate the CO ligands) we have been unable to reductively couple the CO's. However, to our surprise, nitriles are reductively coupled quite readily. The reactions of Re2X4(~-dppm)2 (X= Cl or Br) with nitriles initially give the his-nitrile complexes [Re2X3(~-dppm)2(NCR)2]PF6, 33 but upon further reaction with excess nitrile under reflux, the green, paramagnetic complexes [Re2X3(~-HN2C2R2)(~-dppm)2(NCR)]PF6 (X= Cl or Br; R =Me, Et, i-Pr, ~-Bu or Ph) are formed in quite good yield.34,35 The structures of these complexes are as represented in Fig. 4. The fivemembered metallacycle ring can best be rationalized in terms of contributions from a singly deprotonated diimine ligand RC(~N-)C(=NH)R and a triply deprotonated enediamine ligand RC(N2-)=C(NH-)R.34,35 These reactions are the first where the reductive coupling of nitriles has occured at a multiply bonded dimetal core in which the nuclearity of the metal unit is preserved. In the case of the [Re2Br3(~-HN2C2Me2)(~­ dppm)2(NCMe)]+ cation (Fig. 4) the Re-Re distance is 2.666(l)A. We can envision the reactions as first giving the dicationic species [Re2X3(~­ HN2c2R2)(~-dppm)2(NCR)]2+ (these are formally Re26+ or Re28+ derivatives depending upon the ligand formulation) which then undergo a one-electron reduction under the reaction conditions to afford the monocations (correspondingly formulated as Re2 5+ or Re2 7+).

13

C2

Br2

Cl Fig. 4

Structure of the [Re2Br3(~-HN2C2Me2)(~-dppm)2(NCMe)]+ cation. (Ref. 35).

Whereas the complexes Re2X4(~-dppm)2 provide a stable dimetal template at which to reductively couple nitriles, the reaction course is quite different when the stabilizing effect of the dppm bridges is removed. This is illustrated by the reactions of ~-Re2Cl4(dppbe)2 (dppbe = 1,2-C6H4(PPh2)2), a complex that contains chelating dppbe ligands and no bridging groups to enhance its stability. It reacts with hot propionitrile in the presence of cone. HCl to afford the Re(V) imido complex Re(NCHzEt)Cl 3 (dppbe) in ca. 50% yield. Similar behavior occurs with other nitriles, and also when a mixture of Cn-Bu4N)2Re2Cls, dppbe, RCN and a few drops of cone. HCl is used in place of ~-Re2Cl4(dppbe~~· although the reaction still probably proceeds via ~-Re2Cl4(dppbe)2. One plausible mechanism for these reactions involves the disproportionation of the Re2 4 + core of ~-Re2Cl4(dppbe)2 to give Re(I) and Re(III) species, the former providing the four electrons necessary to reduce the nitrile. The precedence for such a disproportionation originates with the 2,5dithiahexane complex Re2Cl5(dth)2, which was the beginning of our story (vide supra). That such a disproportionation is not only feasible but also very likely, is supported by the increasing number of examples of this type of behavior in multiple bond chemistry, as we shall now discuss.

INTRAMOLECULAR DISPROPORTIONATION AT METAL-METAL MULTIPLE BONDS. Our interest in the complex Re2Cl5(dth)2 has been revived of late by two developments. First, as we mentioned above, because of the role that intramolecular disproportionation might play in the conversion of

14

a-Re2Cl4(dppbe)2 into the imido complexes Re(NCH2R)Cl3{dppbe) upon reaction with organic nitriles in the presence of cone. HCl. Second, as a consequence of our recent discovery of a new class of dirhenium complexes that can be formulated as 'mixed-valent' Re(IV)-Re(II) species. While we were investigating the reactions of the quadruply bonded his-acetate dirhenium{lll) complexes Re2{02CCH3)2X4L2 (X = Cl or Br; L = H20, py, DMF or Me2SO) with monodentate tertiary phosphines in alcohol solvents {ROH; R = Me, Et, n-Pr or i-Pr) we isolated alkoxide complexes of stoichiometry Re2X4(0R)2(PPh3)2 in the case of PPh3. 37 With other phosphines {PMe3, PMe 2Ph and PMePh2) the reactions resulted in the reduction of the dirhenium core and the formation of the now familiar triply bonded compounds Re2X4(PR3)4. 37 However, rather than Re2X4(0R)2(PPh3)2 having a structure that resembles those of the well characterized Re(Ill}-Re(III) derivatives Re2X6(PR3)2 (X= halide), 38 they are in reality the Re(IV)Re(II) species (R0)2X2ReReX2(PPh3)2. This has been established in the case of Re2Cl4(0Et)2{PPh3)2 by a crystal structure determination (Fig. 5). The very short Re-Re bond (2.23l(l)A) and an eclipsed rotational geometry is in accord with the retention of a Re-Re quadruple bond and, formally, with one component of this bond being dative in character in the sense Re(l)~Re(2} i.e. Re*==Re. This is similar to the situation that would hold in the case of Re2Cls{dth)2, if we formulate it as a Re(IV)-Re(I) dirhenium complex (vide supra). These are now other examples of well-defined intramolecular disproportionation reactions occuring at multiple metal-metal bonds. A more recent example is that in which the reaction between the dimolybdenum(II) complex Mo2(0-l-Pr)4{HO-l-Pr)4 and Me2PCH2CH2PMe2(dmpe) in hexane gives Mo2(0-i-Pr)4(dmpe)2, a complex that has the structure (Pr-i-0}4MoMo(dmpe)2.39 This interesting molecule has a staggered rotational geometry and is best formulated as containing a Mo-Mo triple bond uniting d2 Mo{IV} and d 6 Mo{O) atoms.39

C2

Cl4

C4

15

A final development that should be mentioned is the recent synthesis and characterization of Re2Cl5(dto)2 (dto ~ 3,6-dithiaoctane), 4 0 a close relative of Re2Cl5(dth)2. The reaction of (n-Bu4N)2Re2Cl8 with dto in ethanol proceeds via (n-Bu4N)Re2Cl7(dto); this has been structurally characterized Interestingly, if a similar intermediate is formed in the reaction between [Re2Cl8]2- and dth, then (n-Bu4N)Re2Cl7(dth) could react with a further equivalent of dth in one of two ways. If attack is at the rhenium atom that does not already contain a coordinated dth ligand, we can expect Re2Cl6(dth)2 to be formed. Attack at the other rhenium could lead to Re2Cl5(dth)2, as has been suggested by Heyen and Poweu. 4 0


ACKNOWLEDGEMENTS I would like to thank the National Science Foundation for the support of much of our work in this area, and my talented coworkers whose contributions are cited throughout this article.

REFERENCES 1.

2.

3. 4. 5.

6. 7.

16

F. A. Cotton and R. A. Walton, "Multiple Bonds Between Metal Atoms", J. Wiley, New York (1982). This source provides a detailed perspective of the early discoveries in the field: see Chapter 1, pp 1-35. F. A. Cotton, J. Chern. Ed., 60:713 (1983). This article also gives a useful historical overview of the development of multiple metal-metal bond chemistry, including some early discoveries in the field of metal cluster chemistry. F. A. Cotton and M. H. Chisholm, Chemical & Engineering News, 60:40 (1982). There are many excellent reviews on metal cluster chemistry. See, for example, G. Schmid, Structure and Bonding (Berlin), 62:51 (1985). W. A. Nugent and J. M. Mayer, "Metal-Ligand Multiple Bonds", J. Wiley, New York (1988). This provides an excellent source of material on transition metal oxo, nitride, imido, alkylidene and alkylidyne complexes. M. Akiyama, M. H. Chisholm, F. A. Cotton, M. W. Extine, D. A. Haitko, D. Little and P. E. Fanwick, Inorg. Chern., 18:2266 (1979). M. L. Listemann and R. R. Schrock, Organometallics, 4:74 (1985).

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

F. A. Cotton and R. A. Walton, Inorg. Chern., 5:1802 (1966). F. A. Cotton, C. Oldham and R. A. Walton, Inorg. Chern., 6:214 (1967). M. J. Bennett, F. A. Cotton and R. A. Walton, J. Am. Chern. Soc., 88:3866 (1966). M. J. Bennett, F. A. Cotton and R. A. Walton, Proc. Roy. Soc., 303:175 (1968). W. K. Bratton, F. A. Cotton, M. Debeau and R. A. Walton, J. Coord. Chern., 1:121 (1971). R. Colton, R. Levitus and G. Wilkinson, J. Chern. Soc., 4121 (1960). D. G. Tisley and R. A. Walton, Inorg. Chern., 12:373 (1973). H. D. Glicksman and R. A. Walton, Inorg. Chim. Acta, 19:91 (1976). See for example, V. I. Spitsyn, A. F. Kuzina, A. A. Oblova and S. V. Kryuchkov, Russ. Chern. Revs., 54:373 (1985). R. A. Wheeler and R. Hoffmann, J. Am. Chern. Soc., 108:6605 (1986). See ref. 1, p 275. F. A. Cotton and J. T. Mague, Inorg. Chern., 3:1094 (1964). S.M.V. Esjornson, P. E. Fanwick and R. A. Walton, unpublished results (1988). J. R. Ebner and R. A. Walton, Inorg. Chern., 14:1987 (1975). P. Brant, D. J. Salmon and R. A. Walton, J. Am. Chern. Soc., 100:4424 (1978). F. A. Cotton, K. R. Dunbar, L. R. Falvello, M. Tomas and R. A. Walton, J. Am. Chern. Soc., 105:4950 (1983). F. A. Cotton, B. A. Frenz, J. R. Ebner and R. A. Walton, Inorg. Chern., 15:1630 (1976). T. R. Ryan and R. E. McCarley, Inorg. Chern., 21:2072 (1982). R. A. Walton, A.C.S. Symp. Ser., 155:207 (1981) and references cited therein. R. A. Walton, Polyhedron, in press (1989). A. C. Price and R. A. Walton, Polyhedron, 6:729 (1987) and references cited therein. F. A. Cotton, K. R. Dunbar, A. C. Price, W. Schwotzer and R. A. Walton, J. Am. Chern. Soc., 108:4843 (1986). F. A. Cotton, K. R. Dunbar, L. R. Falvello and R. A. Walton, Inorg. Chern., 24:4180 (1985). L. B. Anderson, F. A. Cotton, K. R. Dunbar, L. R. Falvello, A. C. Price, A. H. Reid and R. A. Walton, Inorg. Chern., 26:2717 (1987). P. E. Fanwick, A. C. Price and R. A. Walton, Inorg. Chern., 27:2601 (1988). T. J. Barder, F. A. Cotton, L. R. Falvello and R. A. Walton, Inorg. Chern., 24:1258 (1985). D. Esjornson, P. E. Fanwick and R. A. Walton, Inorg. Chern., 27:3066 (1988). D. Esjornson, D. R. Derringer, P. E. Fanwick and R. A. Walton, Inorg. Chern., submitted for publication. M. Bakir, D. Esjornson, P. E. Fanwick, K. S. Jones and R. A. Walton, unpublished results (1989). A. R. Chakravarty, F. A. Cotton, A. R. Cutler and R. A. Walton, Inorg. Chern., 25:3619 (1986). See ref. 1, p 49. M. H. Chisholm, J. C. Huffman and W. G. VanDer Sluys, J. Am. Chern. Soc. ,109:2514 (1987). B. J. Heyen and G. L. Powell, Polyhedron, 7:1207 (1988).

17

THEORETICAL STUDIES OF DINUCLEAR COMPOUNDS WITH MULTIPLE METAL-METAL BONDS

Bruce E. Bursten and William F. Schneider Department of Chemistry The Ohio State University Columbus, OH 43210 Introduction The last twenty-five years have been witness to the birth and growth of metal-metal multiple bond chemistry. During this time a wide variety of compounds containing multiple bonds of order up to four have been synthesized, and a wealth of data concerning their structure, properties, and reactivity has been collected. 1 An exciting aspect of research in this field is the central role that both qualitative and quantitative electronic structure theory have played in developing a rational framework for understanding the physical structures and properties of these complexes. Approximate molecular orbital theory was applied to these systems almost as soon as they were discovered, and its use has progressed in unison with experimental work, so that the two are intimately intertwined. The synthesis of a new compound containing a metal-metal multiple bond is almost always combined with an interpretation of its properties and reactivity in terms of some form of molecular orbital theory. Conversely, the metal-metal multiple bond problem provides a good test of the usefulness of an electronic structure method. This synergism between theory and experiment that has existed in the metal-metal multiple bond field could not have developed without the advent of theoretical methods that are flexible enough to treat very large molecules. For instance, the prototypical quadruple-bond-containing complex, the octachlorodirhenate(III) dianion (1), contains 288 electrons, Cl

1 ,,....,_Cl

Cl

iz-

I ....,...,_Cl

Rci"-·-4-R{

Cl~~Cl Cl/1Cl 1

which is far too many to be treated rigorously even by the largest ab initio programs of today. While various modifications of the traditional Hartree-Fock and configuration interaction procedures have evolved that

Metal-Meta/ Bonds and Clusters in ChemiStry and Cata/ysrs Ediced by J.P. Fackler, Jr. Plenum Press, New York, 1990

19

allow one to handle problems of this size with a reasonable degree of accuracy, they are expensive and have only recently become generally available. A fortuitous occurrence of the early 1970's was the development of local density functional methods that were capable of modelling the important interactions in inorganic molecules at reasonable computational expense. The first such method to be applied extensively to metal-metal multiple bonds was the SCF-Xa-SW method of Slater and Johnson. 2 While the approximations inherent in this particular method are severe, they do allow one to carry out calculations on very large molecules, and, in the case of metal-metal multiple bonds, the method yields a reasonable representation of metal-ligand and metal-metal interactions. [Re 2Cl 8 ] 2 - (1) was the first inorganic species unambiguously identified as containing a multiple bond between two metal atoms. 3 The crystal structure of 1 reveals two rhenium atoms separated by only 2.24 A, each surrounded by four chloride ligands in essentially a square planar arrangement. The two square planar units are oriented in an eclipsed fashion, to yield an overall (idealized) D4h symmetry. The eclipsed arrangement and short Re-Re separation are counterintuitive from a steric point of view, but by using basic crystal field and symmetry arguments

E

(a)

(b)

(c)

(d)

a9(z2)

Xx

M-L<J

D

D D Xx

M-X<J

D M

Figure 1.

20

x•... M .••• x x-' ""x

M-X<J

D

X,,,,M,,,,L L,.

""x

Qualitative molecular orbital scheme for a cationic transition metal (a), a D4h ML4 fragment (b), a D4h MX4 fragment (c), and a D2 h ML2 X2 fragment (d).

Cotton was able to propose a qualitative description of the bonding between the two metals which explained the structure. 4 The rhenium orbitals necessary for metal-ligand bonding are subtracted from the metal valence atomic orbital set, and then the remaining metal-based orbitals are combined to form rhenium-rhenium bonding and antibonding molecular orbitals of a, ~. and 6 symmetry with respect to the Re-Re vector. The eight metal-based electrons are just enough to fill all the bonding orbitals, and a net quadruple bond results. This elementary description of the bonding in octachlorodirhenate(III) has been remarkably successful and has formed the basis for all further theoretical treatments. The separation of metal-based orbitals into those involved primarily with the ligand bonding and those involved principally with the metal-metal interactions, while simple, is an extremely useful starting point from which to investigate all the different sorts of multiple bonds which have been discovered. The qualitative correctness of this type of analysis has been borne out repeatedly in SCF-Xa-SW and other electronic structure calculations. In this paper, we will give an overview of the bonding in dinuclear complexes that contain multiple metal-metal bonds. Our approach will be one in which we emphasize the role of the ligands in selecting the orbitals available for metal-metal bonding. D4h M2X 8 Complexes For an historical perspective, we will first consider the analysis of the interaction of two square planar d4 ML4 fragments 5 in greater detail. A bare transition metal atom has a total of nine atomic orbitals (AOs) at accessible energies for bond formation. These AOs are of s, p, and d symmetry and, for most common metal oxidation states, have the energetic ordering nd < (n+l)s < (n+l)p (Figure la). The ligand environment determines how these AOs are ultimately partitioned between metal-ligand, metal-metal, and non-bonding molecular orbitals (MOs). Suppose the ligands L interact with the metal in a a-only fashion. The low energy a donor orbitals of the ligands mix with the higher energy metal-based AOs to form bonding MOs of mostly ligand character and antibonding MOs of mostly metal character. In the case of our D4h fragment, the filled ligand donor orbitals are of the proper symmetry (a 18 + eu + b 18 ) to interact with the metal-based sand dz2 (a 18 ), Px and Py (eu), and ~2-y2 (b 18 ) metal AOs (Figure lb). 6 These metal AOs are raised in energy while the remaining ones are, to a first approximation, unaffected. The a 18 interaction with the dz2 AO occurs through the torus of that orbital and results in only a weak hybridization and destabilization of it. Thus, the dzz remains available for further bonding. The net result of the aonly ligand interaction is to remove all but three sets of metal-based orbitals, those of a 18 (d.z), e 8 (~z• dy.), and b 28 (~) symmetry, from the frontier region. These are the orbitals that can take part in further bond formation with a second ML4 fragment. The ligands involved in homoleptic multiply bonded compounds are frequently ~ donors of low to moderate strength, which modifies their effect on the ordering of the frontier metal orbitals of ML4 . For instance, halide ligands are common in the multiple bond chemistry of group 6 and 7 metals. Qualitatively one expects the low energy halide lone pair ~ orbitals to interact with and further destabilize the metal AOs. A simple symmetry analysis reveals that, for halide ligands, there is one combination of ~ orbitals of proper symmetry (b 28 ) to interact with the ~ metal orbital and one of proper symmetry (e 8) to interact with the ~ •• dy. set. The net destabilization of each will be a function of the

21

particular ligand and metal, but a reasonable net ordering is a 18 (d.z) < e 8 (d,.., ely.) < b 28 (d,cy) (Figure lc). 7 The stronger the w-donating ability of the ligands, the greater the splitting in energy of these orbitals. In actual compounds with the M2X8 stoichiometry, the ligands are bent slightly away from one another on opposite sides of the molecule, so that the M-M-X angles are greater than 90" (103.7" in 1 3 ). This bending is due to nonbonded repulsions between the ligands, and results in a more complex mixing of the ligand and metal orbitals. The electronic consequences are minor, however, and it is equally informative to treat the MX 4 units as two square planes. The interaction of four weak w-donor ligands with a central metal leaves four metallic frontier orbitals available for metal-metal bond formation. These orbitals can combine with their partners of the same symmetry on a second fragment to generate a set of bonding and antibonding molecular orbitals, as diagrammed in Figure 2. Since the orbital energies in each fragment are identical, a symmetric bonding arrangement occurs, such that each MO has equal contributions from each metal. From basic overlap arguments, the anticipated ordering of energy levels is a 18 (u) < eu (w) < b 28 (o) < b 1u (o*) < e 8 (w*) < a 2u (u*). 4 In a typical d 4 -d4 M2X8 molecule, such as [Re 2 Cl 8 ] 2- (1), there are eight metal-based electrons (four from each fragment) to be assigned to the metal orbitals, and these exactly fill all the bonding orbitals and leave the antibonding ones empty. The a 18 (u) and eu (w) MOs are cylindrically symmetric and invariant to rotation about the metal-metal axis. The b 28 (o) MO is not invariant to rotation, however, and in order to maximize the net bonding

E

a 19(a)

X ....,X I ,...,X IM·~M··· X

x/1 x/1 X

Figure 2.

22

X

Qualitative molecular orbital diagram of the interaction of two MX4 fragments to generate the four components of an M-M quadruple bond.

in this level, the two square planar fragments must orient themselves in an eclipsed conformation. The combination of these components results in a formally quadruple bond between the two Re atoms. This qualitative analysis is supported splendidly by quantitative calculations. The first of these, by Norman and Kolari, were Xa-SW calculations on the octachlorodimolybdate(II) tetraanion, 8 which is isostructural with 1. They found that the occupied energy levels could readily be partitioned into four groups: the Cl s orbitals; four Mo-Cl u bonding levels, composed mostly of Cl p contributions with an admixture of Mo d.,.z-yz, d.z, and d,.z,yz (20-28%); 14 Cl 'II' lone pair levels, which also contain a small amount of Mo orbital mixing; and the metal-metal bonding orbitals in the order u <,.. < 6. The u level was found among the chlorine lone pairs, although the SCF-Xa-SW treatment is suspected of overestimating this metal-metal interaction. 9 All three of the metal-metal bonding levels contain sizable chlorine contributions, with the ,.. orbital affected the most. The lowest unoccupied orbitals include the 6*, ,..•, and u*, along with two Mo-Cl antibonding orbital of mostly d.,.2-y2 character. These results are all as expected from our qualitative considerations of the metal-ligand and metal-metal interactions, summarized in Figures 1 and 2. Similar results were found for [Re 2Cl 8 ] 2- by both non-relativistic 7 • 10 and relativistic 11 Xa-SW calculations. The MOs of this ion cleanly separate into metal-metal bonding, metal-ligand bonding, and ligand nonbonding, except for the Re-Re u bonding MO, which is spread among several a 11 orbitals. A similar spreading is observed in the SCC-DV-Xa results. 12 The inclusion of relativistic effects has its greatest valence level impact on this u interaction because of its large 6s contribution. 13 Two of the bonding a 11 orbitals move to lower energy, and the angular contributions adjust so that a larger fraction of the metal-metal bonding is housed in the lower one. The 'II' and 6 levels are analogous to those in [Mo 2Cl 8 ] 4 - and are little changed by relativistic effects. Thus, our qualitative consideration of metal-ligand and metal-metal interactions leads to an electronic structure picture that merges nicely with the more quantitative molecular orbital calculations. Perturbations on D4h MzX8 Complexes Now that we have built up a good semi-quantitative description of the multiple-bonding in the highly symmetric quadruply bonded D4h M2X8 molecules, it is worthwhile to consider the effects on the metal-metal bonding of perturbing this basic structure. One possible perturbation involves a lowering of the spatial symmetry through modification of the coordination sphere. Various types of ligands, such as monodentate or bidentate phosphines or alkoxides, can take the place of halides in M2X8 complexes to form new multiply bonded compounds, including the Group 6 D2 d M2L4X4 molecules (2), built from two square planar ML2X2 units, the semistaggered homologues of these that contain bidentate phosphines (3), and the pseudo-C 4v L4MMX4 molecules (4). Another possible perturbation involves the formal oxidation or reduction of the metal core unit by adding or subtracting electrons. The result is a population of antibonding levels or depopulation of bonding levels and concomitant alteration of molecular geometry. All these effects can be analyzed simply in terms of metalligand interactions tuning metal-metal interactions.

Du M2L4X4 Complexes. Examples of this type of molecule include Mo 2 (PMe 3) 4 Cl 4 14 and Mo 2 (PMe 3) 4 (0CH2 -,!:-Bu) 4 15 , in which two neutral u donor and two anionic u and,.. donor ligands are coordinated to each metal. A general ~-MLzX2 fragment (Figure ld) differs from the ML4 one (Figure 23

X

I~ \~

L

X

X

L

~ p

P

' ' l. .~x J\x;l M--M

M:::':__M··''

x~lL L~lX

p~_,l

4

3

2

lb) in that two u-only donor ligands are replaced by two u and ~ donor ligands. The ~ (b 18 ) orbital is destabilized by ~ interaction, but by a lesser amount than in MX 4 because there are only half as many ligand ~ orbitals to interact with. We also see that only one of the dzz (b 28 ) and dy. (b 38 ) orbitals is oriented properly to receive ~ electron density from the ligand set, so that the degeneracy of these two is lifted. Thus, if the ~ donating ligands are oriented along the y axis, the dy. orbital is destabilized with respect to the ~•• and the former is hybridized away from the metal more than the latter. The metal-based electrons are now divided among four non-degenerate fragment orbitals of mostly d metal character. Combination of two MLzX2 fragments to generate 2 results in metalmetal interactions to form u, ~. and 6 type MOs. The non-degenerate u and 6 components are strictly analogous to those in M2X8 . The metal-metal ~ bond has inherent asymmetry: the "~•" orbital on one fragment combines with the "dy•" of the other, and vice versa, to form a degenerate pair of ~ molecular orbitals, each of which is unequally distributed between the two metal centers. Unlike the M2L8 molecules, the degeneracy of these two is broken if one rotates one fragment about the metal-metal axis, but two singly degenerate bonding orbitals remain, so that the net bond order is not reduced. The electronically preferred orientation is not apparent from this symmetry analysis, but it is clear that a net quadruple bond is possible between two MLzX2 fragments. Xa-SW calculations on Mo 2 (PH 3 ) 4 Cl 4 16 • 17 reveal the expected u < 71' < 6 ordering of metal-metal bonding orbitals, as shown on the right side of Figure 3. As in the calculations mentioned previously, the u bond is lower in energy than is found experimentally. Between the u and ~ levels are a series of chloride lone pair and metal-phosphine u bonding orbitals. Since Cl is a weak ~ donor ligand, it splits the ~. and dy. orbitals only weakly (right side, Figure 3), so that there is very little asymmetry in the resultant ~ bond. In fact, the Mo-Mo separations in Mo 2 (PMe 3 ) 4 Cl 4 14 (5) and its rotationally-constrained analog Mo 2 (dppm) 2 Cl 4 18 (6) are nearly identical, suggesting that the net bonding between the metals is the same in either orientation. On the basis of LCAO-HFS calculations, Ziegler has concluded that the geometrical preferences of 5 result from steric rather than electronic factors. 19 The asymmetry introduced in the ~bonds by

I ~MCJI -~Cl M~~Mo.• Cl

PMe3

~ICI CI;1PMe3

MeJP

5

24

6

going from a D 4h to a Dzd molecule does not have a large effect on the overall bonding in Mo 2 (PMe 3 )Cl 4 . Xa-SW calculations have also been performed on Dzd Mo 2 (0H) 4 (PH 3 ) 4 as a model for Mo 2 (PMe 3 ) 4 (0CH2 -~-Bu) 4 . 17 The molecular orbital diagram is presented on the left side of Figure 3, and is quite similar to that found for Mo 2 (PH 3 )Cl 4 • The most notable differences are a reduced HOMO-LUMO gap, due to the greater Mo-Mo separation and reduced ~ overlap, and a much greater splitting of the dxz and dy. fragment orbitals, leading to a more asymmetric metal-metal ~ bond. The greater asymmetry in the ~ bond may add an additional electronic barrier to rotation about theM-Maxis. The net influence of the alkoxide ligands is to interact with the metal-based AOs more strongly than do chlorides, and thus to weaken somewhat the metal-metal bond. "Semi-Staggered" M2 L4X4 Complexes. The introduction of bidentate phosphine ligands with ethylene backbones to M2 (PR3 ) 4Cl 4 molecules results in the formation of bridged compounds 3. 20 The two ML2Cl 2 fragments are twisted relative to one another, with the degree of twist dependent on the specific molecule. The u bond is not directly affected by this rotation. However, the symmetry of the metal-metal ~bond is broken, resulting in two singly degenerate ~ bonding orbitals. As discussed above, the chloride ligands perturb the metal-based dxz and dy. orbitals only slightly, so that the asymmetry introduced into the metal-metal ~ bonding levels is small and the two should remain energetically similar. Stronger ~ donor ligands, such as alkoxides, create a greater separation of the metal AOs and could significantly reduce their bonding interaction; such compounds

(eV)

10e(11')

·2.0

(eV) ·1.0

-3.0 ·2.0

·3.0

5b2(11)

5tJ:! (yz)

-4.0

5b1 (11) ~-I'•F'T'"T'~(8)

7a 1 (a) -4.0

5b1 (xz) 7a 1 (z9

-5.0

-6.0

-5.0

-7.0 -6.0

7a 1 (a)

·7.0

4a1 (a)

HO -8.0

PH

1,..-PH•I.J>H

Mo...i-Mo

9_.,,PH,IPH}:.I Mo...i-Mo

:l Cl.,1PH, H,P"I HO OH'IPH, H,J>'Cl Figure 3.

Molecular orbital correlation diagram for Mo 2 (0H) 4 (PH3 ) 4 (left) and Mo 2 Cl 4 (PH 3 ) 4 (right) showing the interactions important for metal-metal bond formation. The energy scale has been offset to make the mid-points of the HOMO-LUMO gaps for the two molecules coincident. (From ref. 17). 25

are not currently known to exist. rotation, such that it disappears staggered. 21 The degree of twist, be correlated with an increase in decrease in the 6 ~ 6* transition

The 6 bond is also diminished by the when the two fragments are exactly and the resultant loss of 6 bonding, can the metal-metal separation and a energy. 22

Asymmetric Metal-Metal Multiple Bonds. All the compounds discussed thus far have been composed of two electronically equivalent square planar metal fragments. In these cases the metal atoms are equivalent in the final compound and the electrons involved in multiple bond formation are derived equally from each fragment. However, it is also possible to combine fragments in which the metal centers are in different formal oxidation states, so that the two do not participate equally in bond formation. Such compounds are rare, but find recent examples in the formally d 3 -d5 Re(IV)-Re(II) dimer (R0) 2X~ReReX2 (PPh 3 ) 2 (7; X- Cl or Br, R -Me, Et, or Pr), 23 and in the formally d -d6 Mo(IV)-Mo(O) dimer (.iPr0)4MoMo(dmpe)z (8; dmpe- 1,2-bis(dimethylphosphino)ethane).z4

I *,p I . ,p

RO

PPh3

.....:..; 4

...... ~

Re--Re

Cl~~ Cl~~PPh3 RO 7

n :: :-

I

OR

,.oR

......~

' M"a-3-Mi,.

'.~pOR . .,.IOR

PJ '

8

We have performed Xa-SW calculations on (H0) 4MoMo(PH3 )) in both the staggered and eclipsed geometries as a model of compound 8. 1 The results are diagrammed in Figure 4. Analysis of the Mo(OH) 4 and Mo(PH3 ) 4 fragments shows that the donor ability of the two ligand sets are considerably different, so that there is an energy mismatch between the orbitals necessary to form the four components of a quadruple bond. The a bond (7a 1 ) becomes slightly polarized towards the Mo(O) center, and the ~ (9e) slightly towards the Mo(IV). Most important, the ligand interaction prevents the formation of a 6 bond. The ~ (2b 1 or 2hz) orbital in the Mo(OH) 4 fragment is strongly destabilized through ~ interactions with the hydroxide ligands and becomes too high in energy to interact with the corresponding orbital (2hz) on Mo(PH 3 ) 4. There is essentially no difference in the MO diagrams of the eclipsed and staggered models (left and right sides of Figure 4, respectively) since no 6 bonding can occur in either. We conclude that a formal triple bond exists between the two metal atoms, which is in accord with the observed Mo-Mo separation.z 4 In the eclipsed conformation, the two electrons that are capable of participating in 6 bond formation are instead localized in the non-bonding 5hz orbital on the Mo(O) center. The preference for a staggered over eclipsed geometry is readily understood in terms of steric interactions, since, without a 6 bond, there is no electronic constraint to rotation. The differing donating abilities of the ligand sets in 8 adjust the available metal orbitals to limit the number of possible bonding interactions. In compound 7, the ligand environments about each Re atom are not as significantly different as in 8, so that the metal-metal bonds will be less polarized. No calculations have been carried out on this molecule, but the Re-Re separation and eclipsed geometry are in accord with a net quadruple bond. 23 While the orbitals on each metal fragment are affected differently by the ligand sets, it appears that, unlike 8, the differences

26

OH

I ~PH'I ~oH M6"-·-M~. . ,I ~1 Hy> Hy> OH OH H~

""'illil

(eV)

-1.0

~...

10e(7t")

-2.0

-3.0

-4.0 4a,

(z2j

-5.0

-6.0

-7.0

7a1 (a)

7a1 (a)

. ~oH . HO-M~-OH

HO~

Figure 4.

Molecular orbital correlation diagram for (H0) 4MoMo(PH 3 ) 4 in both the staggered (left) and eclipsed (right) conformations. Only the interactions important for metal-metal bond formation are included_ (From ref. 4).

are not great enough to prevent formation of the 6 bond. These two examples point out the important role that metal-ligand interactions have in tuning metal-metal bonding interactions_ Complexes With Other Than 8 Metal-Based Electrons. For a molecule that can be described by Figure 2, eight metal-based electrons is the optimal number for maximizing the multiple metal-metal bonding. Oxidation of any of the quadruply bonded molecules discussed so far would result in the removal of an electron from the 6 bonding orbital; reduction would result in the addition of an electron to the 6* antibonding orbital. Either case leads to a reduction of the formal metal-metal bond order and presumably to a lengthening and weakening of the metal-metal bond. Surprisingly, the latter is not always the case. Changes in the oxidation state of the metals that accompany oxidation or reduction result in the contraction or expansion of the metal AOs that take part in bond formation; these variations in the radial extent of metal-based orbitals can have a greater influence on metal-metal bonding than the loss of the 6 bonding component. [Tc 2 Cl 8 ] 3 - contains nine metal-based electrons. Extrapolating from the electronic structure of octachlorodimolybdate(II), one would anticipate a valence electronic configuration of (u) 2 (~) 4 (6) 2 (6*) 1 , and this configuration is confirmed by Xa-SW calculations. 25 The paramagnetic ion has one electron localized in the 6* orbital, to yield a net bond order of 3.5. Calculations reveal no other qualitative differences with the electronic structure of other M2X8 species_ An eclipsed arrangement of the chlorides is maintained, suggesting that the rotational influence of the 6

27

bond has not been destroyed. The quadruply bonded [Tc 2 Cl 8 ] 2 - ion has also been characterized, and, interestingly, the Te-Te bond is longer than that found in [Tc 2Cl 8 ] 3-. 26 The reason, as suggested above, most likely rests in the lower metal oxidation state of [Tc 2 Cl 8 ] 3-, leading to commensurately more diffuse 4d AOs, and hence stronger Te-Te w and 6 bonds. 27 No accurate calculational results have been reported to verify this hypothesis. The D2 d complexes Re 2 Cl 4 (PR 3 ) 4 , which conform to structure 2, contain two d5 rhenium atoms. Extrapolating from the molecular orbital diagram of the isostructural Mo 2 Cl 4 (PR3 ) 4 complexes (Figure 3), the anticipated electronic configuration is (o) 2 (w) 4 (6) 2 (6*) 2 ; the 6* orbital is fully occupied, so that the net bond order is three. This assignment is confirmed by SCF-Xa-SW calculations. 28 Several interesting differences exist between the electronic structures of Re 2 Cl 4 (PH 3 ) 4 and [Re 2 Cl 8 ] 2 -. 11 First, the metal-metal w bond is distributed between four different e levels in the former, compared to two for the latter. The o bonding orbital is considerably lower in energy in Re 2 Cl 4 (PH 3 ) 4 than in [Re 2 Cl 8 ] 2 -, which may indicate a stronger o bond in the Re(II)-Re(II) dimer than in the Re(III)-Re(III) one. The structure of the parent compounds is eclipsed, but for steric rather than electronic reasons; the eclipsed structure maximizes the separation of the very bulky phosphine ·ligands. Despite its smaller metal-metal bond order, the Re-Re separation in Re 2 Cl 4 (PEt 3 ) 4 is approximately the same as in [Re 2 Cl 8 ] 2 -. Again this anomalous result is most likely due to a compromise between the loss of 6 bonding and increased o and w bonding as two electrons are added to the Re-Re molecular orbitals.

Thus far, we have focussed on the combination of two essentially square planar ML4 fragments to form multiple bonds, and have seen how the metal-ligand interactions influence the type and extent of metal-metal bonding. A second structure that is prevalent among the group 6 elements results from the interaction of two ML3 units to form a M2 Ls molecule (9) . 1 • 29 Here there are fewer metal-ligand interactions than in the M2 L8

L

L

L

I \f

M -3- M

I\L

I

L

L

9

case, but the symmetry of the interactions dictates that fewer metal-based orbitals remain available for metal-metal bond formation. Thus, the maximum metal-metal bond order in the M2Ls molecules is less than that in M2La. We can consider the interaction of two ML3 fragments in the same manner as was done for ML 4 . Figures Sa and Sb show the interaction of a bare transition metal atom with three a-only donor ligands L, arranged in a trigonal planar (D 3h) geometry. 30 The ordering of the bare metal atom orbitals is nd < (n+l)s < (n+l)p, as before. In ML3 , the ligand donor orbitals are of proper symmetry (a1 ' and e') to interact with the metalbased a 1 ' (sand dz2) and e' [(d,2-y2, ci,y) and (Px• Py)] AOs. As in ML4 , the dz2 orbital is only weakly destabilized since its overlap with the ligands, through its torus, is poor. The e' interaction dominates the metal-ligand bonding, and both the cl,z-yz and c1,y orbitals are fully utilized in metal-

28

E

(a)

(b)

(c)

0 M-Lo

M-Lo

D M"'+ Figure 5.

L'"'"·M-L L_....

L~7M......_L

Qualitative molecular orbital diagrams for a cationic transition metal (a), a planar (D 3h) M~ fragment (b), and a pyramidal (C 3v) ML3 fragment (c).

ligand bond formation. By contrast, only the dxz~2 orbital is heavily involved in metal-ligand bond formation in ML4 • The result of the a-only ligand interaction is to leave only the a 1 ' (d2 2) and e" (dx •• dy 2 ) metalbased orbitals available for further bond formation. The ligands generally associated with the M2 L6 compounds of Mo and W are either strong u donor (alkyls) or strong u and w donors (amides and alkoxides). The former are adequately treated by our analysis above, but the latter require a little more consideration. The nitrogen atoms in Mo 2 (NMe 2 ) 6 are planar, 31 strong evidence that significant w donation to the metal is occurring. In addition, the amide ligands orient themselves such that their lone pair orbitals are perpendicular to the metal-metal bond axis. In this orientation the lone pairs transform as a 2 ' and e' (under D3h symmetry). The a 2 ' combination is rigorously non-bonding with the metal atom, but thee' can (and does) interact with the dx2-y2, d,y (e') metal orbitals, leading to even greater destabilization of these and further limiting their ability to participate in metal-metal bond formation. The effects on the metal-metal bond itself are expected to be secondary, however, since the orbitals primarily involved in bond formation are unaffected. The alkoxide ligands in Mo 2 (0-~-Bu) 6 are not oriented in a simple fashion, 32 but similar w bonding interactions are doubtless present. Combination of two ML3 fragments generates a set of metal-metal molecular orbitals of axial symmetry and energy u < w < w* < u* (Figure 6).

29

E

a1a (a)

L

I M I\L L Figure 6.

, L

L

L

I \"' M-'-l ....

L

'i.

I L

L

L

,~

!

M I L

Qualitative molecular orbital correlation diagram showing the interaction of two planar Mla fragments to generate a metalmetal triple bond.

No 6 bonding component is possible, since the necessary d orbitals are removed by metal-ligand bond formation. For a typical d3 -d3 ~Ls compound, such as Mo 2 (NMe 2 ) 6 , there are just enough metal-based electrons to fill the bonding molecular orbitals while leaving the antibonding ones vacant. The overall bond order is three, and the absence of a 6 bonding component suggests that there should be free rotation about the metal-metal axis. Sterle interactions would be expected to favor the staggered geometry 9, which is always observed experimentally. The question of rotational barriers about the metal-metal bond in M2L6 molecules has been a subject of some controversy. As with the M2 L8 molecules, the ligands in the M2Ls complexes are found to bend away from theM-Maxis so that, e. g., the Mo-Mo-N angle in Mo 2 (NMe 2 ) 6 is 103". 31 Free rotation about the metal-metal bond assumes that its ~ component is composed solely from the dxz and dy. orbitals. While this must be the case when D3h MLa fragments are involved, this restriction is relaxed as the ligands bend back to their equilibrium positions (Figure Sc). The dxz• dy. set is destabilized through metal-ligand a bonding. Further, the ~-yz and ~ orbitals are now allowed by symmetry to mix into the metal-metal bond and to lend some 6 or 6* character to the ~ component. Extended Huckel calculations show that in the staggered conformation, mixing occurs between the metal ~ and 6* orbital combinations, resulting in an overall weaker ~ bond. Conversely, in the eclipsed orientation, the mixing is between the~ and 6 combinations, resulting in a stronger~ bond. 33 Thus, neglecting steric interactions between the ligands, the eclipsed orientation would appear to be more stable. However, more rigorous calculations employing the Xa-SW, 34 LCAO-HFS, 30 and generalized molecular orbital (GM0) 35 methods all indicate that the mixing of the ~ and 6 orbitals is overestimated by the extended Huckel method, and that the metal-metal ~bond is best described as primarily dx •• dy. in character. 30

These methods all predict nearly free rotation about the metal-metal bond in the absence of ligand-ligand interactions. The qualitative conclusion which was reached above, that the d,a-ya and d,y orbitals are utilized exclusively for metal-ligand bonding, is modified somewhat by the inclusion of obtuse M-M-L angles, but is still basically accurate. Xa-SW calculations have been used to investigate the similarities and differences between alkoxide, amide, and alkyl ligands in Mo 2X6 complexes. 34 • 36 Figure 7 is a comparison of the occupied molecular orbitals of Mo 2 (CH3 ) 6 (10), Mo 2 (NH2 ) 6 (11), and Mo 2 (0H) 6 (12). All three possess OH HO HO

I I /\ OH OH HO

Mo--Mo 3

''

12 fully occupied Mo-Mo a and w molecular orbitals. As the electronegativity of the ligands decreases from hydroxide to amide to methyl, a steady increase is seen in the energy of the predominantly ligand energy levels. 12 has a large separation of the 0-H a, the Mo-O a, the oxygen lone pair, and the Mo-Mo a and w molecular orbitals, with the last being the HOMO. In contrast, the nitrogen lone pair orbitals are as high in energy as the metal-metal bonding ones in 11, so that a lone pair molecular orbital becomes the HOMO. The Mo-N a bonding levels are significantly lower in energy. Finally, in 10 the Mo-C a bonding orbitals are in the same energy range as the Mo-Mo bonding levels, and extensive mixing between the two sets is observed. The amount of 6* mixing into the Mo-Mo w bond is inversely related to the ability of the ligands to "use up" the metal

-4

(aV)

-5

-6

-7

Mo-MoCJ

Mo-MoCJ

-8

-9

D C-HCJ

D Mo-NCJ

-10

D 01t

D

Mo-OCJ -11

Figure 7.

10

11

12

Comparison of the valence level molecular orbital schemes for Mo 2 (CH 3 ) 6 (10), Mo 2 (NH2 ) 6 (11), and Mo 2 (0H) 6 (12). (From ref. 34).

31

d,..y orbitals through 1r interactions. Thus, in 11, where strong Mo-N interactions are present, there is no o• contribution to the Mo-Mo 1r bonding MO, but in 10, in which metal-ligand 1r interaction is minimal, a significant (27.8%) admixture of o· character is present in the corresponding level. These three calculations point out the wide variation in metal-metal character that can be induced by the ligand set. While the basic electronic structures of 10, 11, and 12 all are similar to that proposed qualitatively in Figure 6, each is unique due to the characteristics of the surrounding ligands. d,..2-y2,

1r

Unligated Metal-Metal Multiple Bonds The discussion so far has centered on the influence of the ligand set in moderating metal-metal bonding. A natural extension is to metal dimers that contain no supporting ligands. Can we predict qualitatively the nature of multiple bonding between two bare metal atoms, and are our qualitative notions supported by quantitative calculations? These questions are not merely of theoretical interest; Mo 2 , for instance, has been prepared in the gas phase and its bond length determined to be

5

(eV)

4

3

2

0

-1

-2

-3

-4

-5

Mo

Figure 8.

32

Molecular orbital correlation diagram showing the interaction of two Mo atoms to generate an Mo 2 dimer and the energy level reorderings upon four-electron oxidation. (from ref. 39b).

1.93±0.02 A. 37 This dimer has the shortest Mo-Mo bond known, a strong indicator that the metal-metal bonding is of a high order. Likewise, Cr2 has an extremely short (1.68±0.01 A) bond, 38 much shorter than that observed in any ligated complex of dichromium. In order to develop a qualitative picture of the bonding in group 6 dimers, we can consider a bare, zero-valent molybdenum atom. The ground electronic state is 7 S, corresponding to a 5s 14d 5 configuration. The 4d orbitals are slightly lower in energy than the Ss, but both of these are significantly lower than the Sp, which can be ignored. The Ss and 4d AOs can be combined to obtain a series of bonding and antibonding MOs of rigorous u (s and d 2 z), ,.. (d,.., dy.), and 6 (d,.z-yz, d,.y) symmetry. A strong du interaction is anticipated between the two 5d2 z AOs, with successively weaker ,.. and 6 interactions between the appropriate Sd orbitals. The Ss AOs combine to form su and su* MOs. The ordering of the 6, 6*, su, and su* MOs is difficult to predict; if the su bonding interaction is strong, the anticipated order is d6 < su < d6* < su*, with all the bonding MOs below the antibonding ones. In this case the twelve valence electrons present in Mo 2 would just fill all the bonding MOs, and a net bond order of six would result. The dimer would then best be described as d5 s 1 -d5 s 1 . However, if the su interaction is relatively weak, or if there is extensive mixing between the 4d2 z and Ss AOs, then the ordering could change, such that the 6* level is below the su bonding level. In this case, a bond order of four would result. The 6* becomes doubly occupied, and a paramagnetic d6 -d6 species is predicted. Quantitative Xa-SW calculations support the former view of the bonding in Mo 2 • 39 Figure 8 shows the results of these calculations. 3 ~ The ground electronic configuration is best described as (lu 8 ) 2 (l7ru) 4 (16 8 ) 4 (2u 8 ) 2 , corresponding to a Mo-Mo sextuple bond. The l,..u and 16 8 MOs are composed almost entirely of d orbital contributions. The lu 8 MO is primarily 4d 2 z in composition, but contains a small admixture (16%) of bonding 5p 2 character. No Ss character is present in this MO. The 2u 8 , on the other hand, while primarily Ss in nature, contains contributions from both the 5p 2 and 4d2 2 AOs. Analysis of the overlap density in these MOs shows that the principle Mo-Mo bonding interactions occurs through the lu 8 and l,..u orbitals. Despite their relative energetic orderings, the 2u8 orbital possesses greater bonding character than the 16 8 . The very short Mo-Mo distance in Mo 2 can be attributed to this second u interaction. 3 ~ There is a nice correspondence between this description of Mo 2 and that developed earlier for d 4 -d4 dimers, such as those based on the Mo 2 4+ core. When Mo 2 is oxidized to Mo~+, the separation between the 4d and Ss orbitals increases substantially. sb The d based orbital drop far below the suMO in energy (Figure Be). What remains is a set of MOs of energy and type u <,.. < 6 < 6* < ,..• < u*. This ordering is very reminiscent of that observed in the ligand supported dimers based on the M2L8 structure. The 6 and 6* orbitals are doubly degenerate in Mo 2 4+, but the effect of ligation is to remove this degeneracy through destabilization of one member of each pair. A stable quadruple bond core is left behind. While diatomic molecules are generally the easiest to calculate with ab initio techniques, the presence of so many orbitals that can contribute to metal-metal bonding makes the group 6 diatomics very challenging to treat with rigorous theoretical methods. Of particular interest have been theoretical predictions of the metal-metal bond lengths, a problem that can not be addressed by the Xa-SW method. The large amount of electron correlation present in these sextuple bonds rules out a simple HartreeFock treatment and the inclusion of configuration interaction is essential. A fairly restrictive MCSCF calculation on Mo 2 led to the prediction of a bond length of 2.1 A; 40 no comment was given on the

33

disagreement between this value and the prior experimental determination. 37 " The bond length and spectroscopic properties were calculated with much better agreement to experiment by using the GMO MGSGF-GI treatment. 41 The highest quality calculation, including bondcentered basis functions and 3212 different configurations, yields final natural orbital populations of (la 6 ) 1 · 88 (l:lru) 3 · 78 (16 6 ) 3 • 42 (2a6 )1. 92 (lau) 0 · 08 (16u) 0 · 58 (l1rg) 0 · 22 (2au) 0 · 12 , corresponding fairly closely to that predicted by less rigorous methods. The greater correlation of the 6 orbitals compared to the 2ag is suggestive of the greater importance of the sa bond in decreasing the Mo 2 bond length, just as was concluded from the Xa-SW results. The electronic structure of Gr 2 has been a fascinating area of theoretical dispute. The greater electron correlation of first row metals makes this a harder problem than Mo 2 , in spite of having fewer electrons to consider. A bond length of 3.06 A was calculated via the spin-optimized generalized valence bond (GVB) method prior to the experimental determination, 42 and the molecule was described as a weakly bound antiferromagnetic dimer. These authors threw down the gauntlet to the proponents of local density functional methods, 43 and, after the subsequent determination of an extremely short bond in Gr 2 , 38 had to run the same. 44 Since then, there have been a series of improved theoretical analyses concerning the electronic structure of this molecule. 45 Multiple Metal-Metal Bonding Between Actinide Elements Finally, we will consider briefly an area that we believe will be an exciting one in future electronic structural studies of metal-metal multiple bonding. We have thus far considered only metal-based s, p, and d orbitals to be available for metal-metal bonding. What could be expected if f orbitals were also allowed to interact in the formation of metal-metal bonds? Bond formation will be best facilitated by the

~----~

#!-~~----~

u-·--o-~

~:·----(j) cp: Figure 9.

34

1x( •2_ 3ll, f y( y2- 3x2]

Schematic of the possible diatomic interactions between f orbitals on adjacent atoms. (From ref. 48).

actinide rather than lanthanide metals since the actinide Sf orbitals are considerably more diffuse that the 4f orbitals of the lanthanides. We anticipate this to be an exciting area for synthetic chemists as well; there are currently no known complexes that contain direct metal-metal bonds between two f-element metal atoms, in spite of some genuine attempts to achieve this. 46 Actinide diatomic molecules do exist as transient species in the gas phase, however, and, for example, the bond dissociation energies of Th 2 and U2 have been measured mass spectrometrically as 28S±21 and 218±21 kJ mol- 1 , respectively. 47 Diatomic Interactions Between f Orbitals. Just as diatomic interactions of the d orbitals are determined by the magnetic quantum numbers of the orbitals (viz. a: m1 = 0; ~= m1 - ±1; 6: m1 - ±2), so are those between the f orbitals. The diatomic bonding interactions are shown in Figure 9. Since 1 - 3 for the f orbitals, it is now possible to form bonds of ~ (m1 - ±3) symmetry. Nonrelativistic Xa-SW calculations on U2 and Np 2 show, respectively, the partial and complete filling of the Sf-Sf~ bonding level, with predicted bond orders of six and seven. 48 These calculations were not stable to the inclusion of relativistic effects, however, and the diatomic interactions between the 6d orbitals were energetically close to those between the Sf orbitals. In the absence of a more correct treatment, it is probably best to consider these Xa-SW calculations as exploratory ones on the existence of ~ bonds between metal atoms. Hypothetical U2Me 6 • There is a parallel between the above calculations on actinide diatomic molecules and the transition metal dimers discussed previously: in each case there is an excess of orbitals available for metal-metal bonding, which complicates the theoretical treatment. In the case of actinide dimers, will it be possible to use

-1

(eV)

Uli

-2

·3

Ua

-4

-5

, ';, a 1g (a)

-7

H1C

I u

r

~

H1C H,'t

Figure 10.

H1C

CH1 CH1

. u--u I l" u,c

....~

H,'t

I

en,

CH1 CH1 ,~

I

u

I

CH,

Molecular orbital diagram of the interaction of two UMe 3 fragments to generate a U2Me 6 molecule. (From ref. 49). 35

ligands to remove some of· the orbitals from the metal-metal bonding manifold, as was the case in M2L8 and M2 L6 complexes? Given that there are some similarities between the chemistry of Mo (d 6 ) and U (f6 ), we have explored the possibility of forming multiple uranium-uranium bonds in the hypothetical U(III) dimer U2Me 6 . 49 Calculations on a pyramidalized (G 3v) UMe 3 fragment bear some similarities and differences to the Mo~ fragment discussed previously. The lowest metal-based orbitals are the fu (a 1 ) and f~ (e), analogous to the frontier d orbitals of the transition metal fragment. In the case of UMe 3 , however, the remaining f orbitals are not destabilized by the ligands; both the f~ and fo orbitals reside< O.S eV above the f~. This is consistent with our description of the dichotomy of roles served by the Sf and 6d orbitals in actinide complexes: the 6d orbitals are used to bind ligands (and are thus destabilized by them), while the Sf orbitals are used to house metal-based electrons. 50 The consequences of not removing the other Sf orbitals from the frontier region are evident in Figure 10, which shows the generation of U2Me 6 from two UMe 3 fragments at an assumed U-U bond length of 2.4 A. Because the f orbitals are more strongly directional than are the d orbitals, the U-U u and ~ interactions are as strong as those observed in Mo 2Me 6 . The low energy of the other f orbitals complicates the simple picture, however. The interaction of the fo orbitals (which overlap much better than do do orbitals) leads to a U-U o bonding orbital that is energetically comparable to the U-G u bonds. The net result is the "stealing" of two U-G electrons to form a U-U o bond. While this is favorable from the point of view of the metal-metal bond, it will be catastrophic to the viability of the molecule! It is interesting to note that U(III) complexes with bulky alkyl, 51 amide, 52 and alkoxide 53 ligands have been prepared, and none of them contain a significant U-U interaction. The design of molecules that will be able to support a direct actinide-actinide bond will doubtless be an active and challenging area of future research for both theoretical and synthetic chemists. Acknowledgments We gratefully acknowledge support for this research from the Division of Chemical Sciences, Office of Basic Energy Sciences, U. S. Department of Energy (Contract DE-FG02-86ER13S29), and from the National Science Foundation for a Predoctoral Fellowship (W.F.S.). B.E.B. is a Camille and Henry Dreyfus Foundation Teacher-Scholar (1984-1989). References (1)

Gatton, F. A.; Walton, R. A. Multiple Bonds Between Metal Atoms; Wiley: New York, 1982.

(2)

(a) Johnson, K. H. Adv. Quantum Chern. 1973, l, 143-18S. (b) Slater, J. G. "Quantum Theory for Molecules and Solids: The Self-Consistent Field for Molecules and Solids"; McGraw~Hill: New York, 1974; Vol. 4.

(3)

Gatton, F. A.; Harris, G. B. Inorg. Chern. 1965,

(4)

(a) Gatton, F. A.; Curtis, N. F.; Harris, G. B.; Johnson, B. F. G.; Lippard, S. J.; Mague, J. T.; Robinson, W. R.; Wood, J. S. Science 1964, 145, 1305-1307. (b) Gatton, F. A. Inorg. Chern. 1965, ~. 334336.

36

~.

330-333.

(5)

Throughout this paper, L will refer to a neutral, two-electron donating ligand, such as a phosphine, amine, or thioether. X will refer to an anionic two-electron donating ligand, with possible wdonating ability, such as a halide, alkoxide, or amide. R will refer to an anionic two-electron donating ligand with no w-donating ability, such as an alkyl.

(6)

Albright, T. A.; Burdett; J. K.; Whangbo, M. H. Orbital Interactions in Chemistry; Wiley: New York, 1985.

(7)

Mortola, A. P.; Moskowitz, J. W.; Rosch, N. Int. J. Quantum Chern. Symp. No. 8 1974, ~. 161-167.

(8)

(a) Norman, J. G,; Kolari, H. J. J. C. S. Chern. Comm. 1974, 303-305. (b) idem J. Am. Chern. Soc. 1975, 97, 33-37. (c) idem J. C. S. Chern. Comm. 1975, 649-651.

(9)

Bursten, B. E.; Clark, D. L. Polyhedron 1987, £, 695-704, and references therein.

(10)

Mortola, A. P.; Moskowitz, J. W.; Rosch, N.; Cowman, C. D.; Gray, H. B. Chern. Phys. Lett. 1975, 32, 283-286.

(11)

Bursten, B. E.; Cotton, F. A.; Fanwick, P. E.; Stanley, G. G. J. Am. Chern. Soc. 1983, 105, 3082-3087.

(12)

Trogler, W. C.; Ellis, D. E.; Berkowitz, J. J. Am. Chern. Soc. 1979, 101, 5896-5901.

(13)

Pitzer, K.S. Ace. Chern. Res. 1979,

(14)

Cotton, F. A.; Extine, M. W.; Felthouse, T. R.; Kolthammer, B. W. S.; Lay, D. G. J. Am. Chern. Soc. 1981, 103, 4040-4045.

(15)

Chisholm, M. H.; Felting, K.; Huffman, J. C.; Tatz, R. J. J. Am. Chern. Soc. 1984, 106, 1153-1154.

(16)

Cotton, F. A.; Hubbard, J. L.; Lichtenberger, D. L.; Shim, I. J. Am. Chern. Soc. 1982, 104, 679-686.

(17)

Bursten, B. E.; Schneider, W. F. Inorg. Chern., in press.

(18)

Abbott, E. H.; Bose, K. S.; Cotton, F. A.; Hall, W. T.; Sekutowski, J. C. Inorg. Chern. 1978, 17, 3240-3245.

(19)

Ziegler, T. J. Am. Chern. Soc. 1984, 106, 5901-5908.

(20)

Agaskar, P. A.; Cotton, F. A.; Derringer, D. R.; Powell, G. L.; Root, D. R.; Smith, T. J. Inorg. Chern, 1985, 24, 2786-2791, and references therein.

(21)

Best, S. A.; Smith, T. J.; Walton, R. A. Inorg. Chern. 1978, 17, 99104.

(22)

(a) Campbell, F. L.; Cotton, F. A.; Powell, G. L. Inorg. Chern. 1985, 24, 4384-4389, and references therein. (b) Hopkins, M. D.; Zietlow, T. C.; Miskowski, V. M.; Gray, H. B. J. Am. Chern. Soc. 1985, 107, 510512.

(23)

(a) Chakravarty, A. R.; Cotton, F. A.; Cutler, A. R.; Tetrick, S.M.; Walton, R. A. J. Am. Chern. Soc. 1985, 107, 4795-4796. (b)

~.

271-176.

37

Chakravarty, A. R.; Cotton, F. A.; Cutler, A. R.; Walton, R. A. Inorg. Chern. 1986, 25, 3619-3624. (24)

Chisholm, M. H.; Huffman, J. C.; VanDerSluys, W. G. J. Am. Chern. Soc. 1987, 109, 2514-2515.

(25)

Cotton, F. A.; Kalbacher, B. J. Inorg. Chern. 1977, 16, 2386-2396.

(26)

Cotton, F. A.; Daniels, L.; Davison, A.; Orvig, C. Inorg. Chern. 1981, 20, 3051-?.

(27)

Cotton, F. A.; Davison, A.; Day, V. W.; Fredrich, M. F.; Orvig, C.; Swanson, R. Inorg. Chern. 1982, 21, 1211-1214.

(28)

Bursten, B. E.; Cotton, F. A.; Fanwick, P. E.; Stanley, G. G.; Walton, R. A. J. Am. Chern. Soc. 1983, 105, 2606-2611.

(29)

Chisholm, M. H.; Cotton, F. A. Ace. Chern. Res. 1978, 11, 356-362.

(30)

Ziegler, T. J. Am. Chern. Soc. 1983, 105, 7543-7549.

(31)

Chisholm, M. H.; Cotton, F. A.; Frenz, B. A.; Reichert, W. W.; Shive, L. W.; Stults, B. R. J. Am. Chern. Soc. 1976, 98, 4469-4476.

(32)

Chisholm, M. H.; Cotton, F. A.; Murillo, C. A.; Reichert, W. W. Inorg. Chern. 1977, 16, 1801-1808.

(33)

(a) Albright, T. A.; Hoffmann, R. J. Am. Chern. Soc. 1978, lQQ, 77367738. (b) Dedieu, A.; Albright, T. A.; Hoffmann, R. J. Am. Chern. Soc. 1979, 101, 3141-3151.

(34)

Bursten, B. E.; Cotton, F. A.; Green, J. C.; Seddon, E. A.; Stanley, G. G. J. Am. Chern. Soc. 1980, 102, 4579-4588.

(35)

(a) Hall, M. B. J. Am. Chern. Soc, 1980, 102, 2104-2106. A.; Hall, M. B. Inorg. Chern. 1983, 22, 728-734.

(36)

Cotton, F. A.; Stanley, G. G.; Kalbacher, B. J.; Green, J. C.; Seddon, E.; Chisholm, M. H. Proc. Natl. Acad. Sci. USA 1977, ~. 3109-3113.

(37)

(a) Efremov, Yu. M.; Samoilova, A. N.; Kozhukhovskii, V. B.; Gurvich, L. V. J. Mol. Spectr. 1978, 73, 430-440. (b) Hopkins, J. B.; Langridge-Smith, P.R. R.; Morse, M. D.; Smalley, R. E. J. Chern. Phys. 1983, 78, 1627-1637.

(38)

(a) Michalopoulos, D. L.; Geusic, M. E.; Hansen, S. G.; Powers, D. E.; Smalley, R. E. J. Phys. Chern. 1982, 86, 3914-3916. (b) Bondybey, V. E.; English, J. H. Chern. Phys. Lett. 1983, 94, 443-447.

(39)

(a) Norman, J. G.; Kolari, H. J.; Gray, H. B.; Trogler, W. C. Inorg. Chern. 1977, 16, 987-993. (b) Bursten, B. E.; Cotton, F. A.~ Faraday Soc. 1980, 14, 180-193.

(40)

Wood, C.; Doran, M.; Hillier, I. H.; Guest, M. F. Symp. Faraday Soc. 1980, 14, 159-169.

(41)

Bursten, B. E.; Cotton, F. A.; Hall, M. B. J. Am. Chern. Soc. 1980, 102, 6348-6349.

38

(b) Kok, R.

(42)

Goodgame, M. M.; Goddard, W. A. G., III J. Phys. Chern. 1981, 85, 215217.

(43)

Goodgame, M. M.; Goddard, W. A. G., III Phys. Rev. Lett. 1982, 48, 135-138.

(44)

(a) Delley, B.; Freeman, A. J.; Ellis, D. E. Phys. Rev. Lett. 1983, 50, 488-491. (b) Goodgame, M. M.; Goddard, W. A. G., III Phys. Rev. Lett. 1985, 54, 661-664. (c) Delley, B. Phys. Rev, Lett. 1985, 55, 2090. (d) Painter, G. S. J. Phys. Chern. 1986, 90, 5530-5535.

(45)

(a) Kok, R. A; Hall, M. B. J. Phys. Chern. 1983, 87, 715-717. (b) Walch, S. P.; Bauschlicher, C. W.; Roos, B. 0.; Nelin, C. J. Chern. Phys. Lett. 1983, 103, 175-179. (c) Baykara, N. A.; McMaster, B. N.; Salahub, D. R. Mol. Phys. 1984, 52, 891-905. (d) Sundholm, D.; Pyykko, P.; Laaksonen, L. Finn. Chern. Lett. 1985, Sl-55.

(46)

(a) Cotton, F. A.; Marler, D. 0.; Schwotzer, W. Inorg. Chim. Acta 1984, 85, L31-L32. (b) Cotton, F. A.; Marler, D. 0.; Schwotzer, W. Inorg. Chern. 1984, 23, 4211-4215. •

(47)

(a) Gurvich, L. V.; Karachevstev, G. V.; Kondrat'yev, V. N.; Lebedev, Y. A.; Mendredev, V. A.; Potapov, V. K.; Khodeev, Y. S. Bond Energies, Ionization Potentials and Electron Affinities; Nauka: Moscow, 1974. (b) Gingerich, K. A. Symp. Faraday Soc. 1980, 14, 109-125.

(48)

Bursten, B. E.; Ozin, G. A. Inorg. Chern. 1984, 23, 2910-2911.

(49)

Bursten, B. E.; Novo-Gradac, K. J., unpublished results.

(50)

(a) Bursten, B. E.; Fang, A. J. Am. Chern. Soc. 1983, 105, 6495-6496. (b) Bursten, B. E.; Novo-Gradac, K. J. J. Am. Chern. Soc. 1987, 109, 904-905. (c) Bursten, B. E.; Strittmatter, R. J. J. Am. Chern. Soc. 1987, 109, 6606-6608. (d) Bursten, B. E.; Rhodes, L. F.; Strittmatter, R. J. J. Am. Chern. Soc. 1989, in press.

(51)

VanDer Sluys, W. G.; Burns, C. J.; Sattelberger, A. P. Organometallics 1989, ~. 855-857.

(52)

Andersen, R. A. Inorg. Chern

(53)

VanDer Sluys, W. G.; Burns, C. J.; Huffman, J. C.; Sattelberger, A. P. J. Am. Chern. Soc. 1988, 110, 5924-5925.

1979, 18, 1507-1509.

39

USES OF METAL CLUSTERS IN HOMOGENEOUS AND HETEROGENEOUS CATALYSIS

Donald J. Darensbow:g

DepanrnnentofCherrnstty Texas A&M University College Station, Texas 77843 Molecular metal carbonyl clusters can serve as pivotal metal derivatives for a large variety of chemical processes. These are illustrated In Scheme 1, where a tetranuclear cluster, the majority of which are 3-dimensional and tetrahedral, is Indicated. Source of Dispersed Metals on Oxide Supports (Heterogeneous Catalysis)

Homogeneous Catalysis

Scheme 1

Reactivity Patterns

on Metal Surface

Structural Models

for Chemisorption on Metal Surfaces

Metal clusters may be defined rather liberally as compounds containing more than one transition metal atom directly bonded to each other. However, tetranuclear clusters represent the simplest metal cluster where a pronounced similarity with the metallic state is anticipated. That is, these species possess a 3-dimensional network of metal atoms held together by metal-metal bonds, and where each metal atom forms at least three different metal-metal bonds. Indeed the impetus for examining the role of metal clusters in homogenous catalysis was prompted by Muetterties formulation of the "surface cluster analogy". I Nevertheless. for the most part the work presented herein will involve derivatives of the 2-dimensional trinuclear, triangular metal carbonyl cluster, Ru3(CO) 12· This contribution will attempt to illustrate by examples from our research program in organometallic chemistry how metal clusters have impacted on these various areas. Our entry into defining the mechanistic aspects of catalytic processes involving transition metal carbonyls commenced with an investigation of the water-gas shift reaction (hereafter referred to as WGS).2 The organometallic chemistry pertinent to this process is indicated In skeletal form In Scheme 2. Metal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J.P . Fackler, Jr. Plenum Press, New York, 1990

41

Schfme2

The WGS reaction Is used to Increase the amount of H2 In CO /H2 mixtures obtained from fossil fuels (eq. 1). Presently, catalysts consisting of Iron-chromium oxide mixtures are used at 350-400°C to reduce the CO content to 3-4%. When lower levels of CO (0.1%) are required, a mixture of Cu-Zn oxides which Is sulfur-sensitive but active at 1902600C Is used. For reaction (1). 6.G 0298 = 28.5 kJ/mole, or lfH20ill Is considered, (1)

6.G 0298 =-19.9 kJ/mole. Hence, the free-energy change favoring H2 production decreases with increasing temperature. Soluble group 6 and group 8 metal complexes are attractive, homogeneous, altematlves as WGSR catalysts since they are potentially resistant to poisoning by sulfur and also are active at low temperatures, where very little CO is present at equilibrium. The most thoroughly investigated catalytic system is that derived from ruthenium dodecacarbonyl in aqueous ethoxyethanol solutions in the presence of KOH.3-5 This catalyst system operates between 90-110'C and 0.5-2.0 atm CO, with the mature catalyst providing a tumover number of approximately 3 moles C02/H2 per mole Ru3(COl12 per day. During catalysis the,prtmary ruthenium species In solution are HRu3(COl11- and H3Ru4(COl12-. Since these two clusters are linked together by the equilibrium process indicated in eq. (2). as H2 accumulates the tetranuclear species become predominant. 6

It has been proposed that the catalytic process can be divided into two cycles (Scheme 3) involving separately the tetranuclear and trinuclear ruthenium clusters,

1 H2

CO

42

Ru3(C0)12 " \ H -

C0 2

Schfme3

with the latter species being catalytic more active. Consistent with this implication when the H2 is continuously removed from the reaction mixture, only HRu3(CO) 11- was observed and concomitantly the turnover frequency increased by fourfold over that seen when H2 was allowed to build up in concentration. Formation of H3Ru4(COh2- from the Ru3 units is indicative of some degree of cluster fragmentation. Hence the ubiquitous issue of the involvement of mononuclear catalytically active species presents itself. This feature is generally evident in homogeneously catalyzed reactions involving ruthenium clusters. Related to this question it is of interest to note a recent report in the literature of a mononuclear Ru(II) complex, namely K[Ru(Hedta)(CO)], which is an extremely active homogeneous catalyst for the WGS reaction under very mild conditions. The optimum reaction conditions were determined to be 50'C and 15 atm of carbon monoxide, where a turnover frequency of 350 mol C02/H2 per mole catalyst per hour was found.? This ruthenium catalyst is clearly the most active homogeneous catalyst thus far reported for the WGS reaction, 8 being much more active than the ruthenium clusters and operating at much milder conditions than the heterogeneously catalyzed reactions (400'C and 200 atm CO). Currently much of our research program is directed at understanding the organometallic chemistry of carbon dioxide, one of the products of the WGS reaction, and the incorporation of this knowledge Into catalytic processes utilizing C02 as a source of carbon in the production of useful organic chemicals. In this regard anionic metal hydrides afforded by the sequence of events depicted in Scheme 2 may be used In the reduction of carbon dioxide. The C-H bond-forming process from the reaction of metal hydrides and carbon dioxide is indicated in reaction (3).9 The alternative mode for C02 insertion into a metal-hydride bond leading to the production of metallocarboxylic acids has not been documented. Nevertheless, these species have been identified from the nucleophilic addition of OW ions to metal carbonyls (Scheme 2).

[MtH + C02

,0

[MtO-C-H

(3)

The production of methyl formate from the hydrocondensation of carbon dioxide in alcohols utilizing anionic group 6 metal hydrides as catalysts has been reported under rather mild reaction conditions (loading pressures of C02 and H2, 250 psi each, and 125'Cj.10 Reaction (3) represents a fundamental step in this process, with a subsequent reaction of the metalloformate complex with dihydrogen via a ligandassisted heterolytic splitting mechanism leading to formic acid. HCOOH consequently rapidly reacts with methanol to provide methyl formate (see Scheme 4).

co,-(

~H 2 ~co

Schane4

HCOOH

CO

j HCOOMe J___MeOH +

HP

43

Consistent with Scheme 4 the addition of CO retards alkyl formate production, strongly implying C02 as the primary source of the carboxylic carbon atom in HCOOR This was verified by carrying out reactions in the presence of HC02W(l3co)5- which provided only H 12cooR after short reaction times. However, in the absence of C02 and H2 the anionic metal hydrides were observed to be effective catalyst precursors for converting CO and methanol Into methyl formate. In this connection it is important to recall that a combination of reactions (1) and (4) leads to reaction (5), i.e., the formation of methyl formate from the carbonylatlon of methanol. HCOOMe + HzO

COz + Hz + MeOH

MeOH + CO -

HCOOMe

(4)

(5)

The carbonylatlon process (eq.5) involving group 6 metals as catalysts Is more efficient at producing HCOOMe than the reaction proceeding through carbon dioxide (eq.4).11,12 For example, under comparable reaction condltlons the tumover numbers per day are approximately 15 and 270, respectively for the carboxylation and carbonylatlon catalytic processes employing anionic tungsten carbonyl hydrides as catalysts. A catalytic cycle in accordance with experimental observations is represented in Scheme 5. Intermediates for both processes (Schemes 4 and 5) were in general Identified by in situ FTIR studies using a cylindrical lntemal reflectance cell. 13 The methanol carbonylatlon process is greatly inhibited by carbon dioxide. This inhibition involves the reaction of the cocatalyst, OMe-, with C02 to yield methyl carbonate.

Cf\OH +

co In totally analogous processes anionic ruthenium carbonyl clusters serve as catalysts or catalyst precursors for the hydrogenation of carbon dioxide in alcohols (eq.4) or the carbonylation of alcohols (eq.5) to produce alkyl formates. There are however some subtle differences with regard to reactivity and reaction pathway obseiVed when employing these multinuclear metal carbonyls as catalysts. Table 1 contains a portion of the data obtained for the catalytic hydrocondensatlon of C02 and molecular hydrogen by anionic ruthenium clusters In the presence of methanol to provide methyl formate.14 The conditions for the reaction were moderate pressures of carbon dioxide and hydrogen (250 psi of each at ambient temperature) and a temperature of 125'C. 44

Table 1. Formation of Methyl Formate from the Oxides of Carbon Using Anionic Ruthenium Carbonyl Clusters Catalyst

!Co2

PH2

HR11a(COf1

250

250

4.1

HC02Ril:J(COf0

250

250

5.7

H3Ru4 (COf2

250

250

7.3

R'lJ(COf2

250

250

<0.3

HCO~Il:J(COf0

250

250

HR11a(C0f1 8

!Co

turnovers8

100

4.1

250

106

Moles of methyl formate/mole of catalyst per 24 h..

As noted in Table 1 the HC02RU3(CO) 10- anion has a slightly greater activity for alkyl formate production than the HRu3(COl 11- anion, by about one turnover. This metalloformate derivative has been synthesized independently from either Ru3(COh2 and [PPNII02CHI (eq.6) or C02 insertion into HRu3(COl11- (eq.7), and it has been fully characterized.15 Indeed, the formate ruthenium cluster would be antlclpated to be an

(7)

intermediate in the catalytic cycle originating with HRu3(COJ 11- on the basis of equation 7. However, unlike the mononuclear group 6 metal carbonyl hydrides which readily react with carbon dioxide the reaction of HRu3(CO) 11- with C02 is energetically less favorable, occurring only at moderate pressures (60-400 psi) of carbon diOXide. In fact reaction (7), or analogous process involving an alternative polynuclear metal carbonyl hydride species (vide infra), most likely represents the rate-determining step in the catalytic production of alkyl formates. In either instance, i.e., utilizing HRu3(CO) 11- or HC02Ru3(CO) 10- as catalyst precursor, the metal carbonyl containing species quantitatively recovered at the end of a catalytic run is the tetranuclear cluster, H3Ru4(CO) 12- . This is of course anticipated in the presence of large quantities of hydrogen based on the equilibrium indicated in equation (2). Subsequent catalytic runs employing H3Ru4(CO) 12- as catalyst revealed this species to be more effective at catalyzing the production of methyl formate than the trinuclear clusters. Hence, it was suggested that a tetranuclear species was the catalytically active species in the experiment involving C02/H2 in Table 1. The nature of the Ru4 species may be a formate derivative analogous to the ClRu4(CO) 13- 'butterfly" complex described by Geoffroy and coworkers.I6 The dramatic difference observed in the rates of reactions (3) and (7), both denoting C02 insertion into anionic metal hydrides, reflects the divergence in reactivity expected for terminal vs bridging hydrides. That is, the latter species has to rearrange to a hydrldic terminal hydrogen bound to ruthenium prior to reacting with carbon dioxide. A similar observation has been noted for the protonation process responsible for H2 production for HRu3(CO) 11- in the WGS reaction. 17 It has been demonstrated that this latter process Is greatly influenced by an incoming CO ligand. This most likely occurs by way of an electron-rich HRu3(COh2- intermediate containing a terminal hydride ligand which readily reacts with H20 to afford Ru3(COl12. H2 and the cocatalyst oH- . Presumably in our instance terminal hydride formation is promoted by carbon dioxide. Scheme 6 depicts our current understanding of the carbon diOXide insertion process involving a bridging ruthenium hydride moiety. Consistent with this

45

description the rate of 13co ligand exchange with HRus(CO) 11- Is faster than the rate of C02 insertion. Concomitantly. the insertion process is inhibited by the presence of carbon monoxide. resulting in a decrease in catalytic activity in the hydrocondensation of C02 (Table 1, entry 5). SchemeS

Scheme 6 assumes the involvement of two metal centers in the carbon dioxide insertion reaction. i.e .. intermediate A instead of B. Although species B is possible, as Is required for mononuclear metal complexes, there appears to be less molecular rearrangement required for the formation of A. This is seen upon modelling the decarboxylation process by simply bond-opening one end of the bridging formate ligand in the structure ofHC02Rus(C0)1Q- and rotating about the alternate C-O bond (eq.S).

i

I

-

/c,

~ Ru--Ru A

-

i

o, . .

/c\ ~

(8)

Ru--Ru B

This operation places the formate hydrogen atom much closer to the metal center adjacent to the one bearing the monodentate formate ligand than to the metal center bonded directly to the formate ltgand, 2 .33 A vs 3 . 16 A.

Thus far we have not quantified the kinetic parameters for the carboxylation or decarboxylation processes involving a bridging hydride moiety as in these cluster derivatives, however, the reaction indicated in eq.9 has been investigated kinetically for mononuclear metal hydrides employing isotopically labelled C02. 18,19 The details of [MtOzCH + l3co 2

46

(9)

one such Investigation is summarized In Scheme 7, where the energetic for the barriers were obtained from reaction (9) and CO ligand substitution in HC02Cr(COls-.20 As would be anticipated based on Scheme 7 the decarboxylation reaction Is retarded In the presence of carbon monoxide. C0 2 Insertion Mechanism

~

e

I ,.;

H

,c, 0

(1)

!

+

HCr(CO);

C02

~

-Cr-

II""' I

(5)

........ -- --.--------------------------------------------- ------

:

:: : ~

l

Scheme7 (3)

(2)

(4)

2

Rxn Coordinate

Retumlng to Table 1 It is apparent from the last entry that HRU3(CO) 11- is an effective catalyst for methanol carbonylation (eq.5). In the presence of excess carbon monoxide HRu3(CO) 11- In hot methanol readily affords Ru3(CO) 12 and hydrogen (vide supra), in addition to an equivalent quantity of OMe-. Indeed a comparably active catalyst system is provided by Ru3(COh2 and one equivalence ofKOMe.12 The proposed methoxycarbonyl intermediate in the methyl formate synthesis (eq.9) is observable by IR (vCO: 2072(w), 2015(s), 1990(s), 1965(m), and 1603 cm-1) and 13c NMR (206.6 ppm@ ambient temperature) spectroscopies In TIIF/MeOH.21,22 Reaction (10) has as well

proved to be a most competent means of preparing 13c-enriched samples of Ru3(COh2. 21 Unlike the mononuclear group 6 M(C0)6 complexes, which In CO/MeOH in the absence of the cocatalyst KOMe lead to no catalytic activity, Ru3(CO) 12 alone in CO/MeOH Is a modestly active catalyst for methyl formate production. That Is, the CO ligand in Ru3(COh2 are substantially electrophilic in character to afford the methoxycarbonyl intermediate In methanol or methanol/THF In the absence of added OMe-. This was verified by 13c NMR, where Ru3(CO) 12 In pure TifF /MeOH was shown to exists In part as the methoxycarbonyl adduct with a resonance at 206.6 ppm at room temperature. Similarly, Ru3(CO) 12 In pure THF /MeOH exchanges CO ligands with 13co at a rate greatly enhanced over that seen in TifF solution. In the remaining portion of this presentation I wish to make a very risky transition (in the vemacular of the horseman "change horses In midstream"), proceeding from homogeneous to heterogeneous catalysis. Nevertheless, the focus will

47

remain on catalytic processes incorporating carbon dioxide. In particular an investigation of supported metal catalysts for the hydrogenation of C02 to methane, methanation (eq.ll). 23 C0 2 + 4Hz

-

CH4 + 2Hz0

( 11)

One of the essential problems to be dealt with in heterogeneous catalysis is the preparation and stabilization of very small metal particles supported on oxide carriers. In this regard metal carbonyl clusters in and on solid metal supports such as silica, alumina, and magnesia, serve as good sources of highly dispersed, low valent metals for catalysis. A large variety of ruthenium clusters have been supported on alumina and activated as described in eq.(l2).24 The infrared v(CO) spectra of all the supported clusters were found to be identical after activation, having a characteristic two band pattern with peaks at 2043 and 1963 cm-1. The structure of these surface species has been proposed to be [Ru(C0)2X2ln. where X is a surface oxygen atom of alumina and n is unknown.24,25

1) THF or hexane Al203

(12)

Catalysts prepared similarly from RuCIJ produce a different surface species upon activation under hydrogen at 200'C followed by exposure to carbon monoxide. The infrared spectrum of this species exhibits a three band pattern with peaks at 2140, 2075, and 2013 cm-1. which matches the ..J(CO) infrared pattern of [RuC12(COl31n or [RuC12(COl41n. The species produced by the two types of supported ruthenium complexes exhibit different reactivities and selectivities as well. It was shown that the ruthenium carbonyl clusters are generally more active catalysts precursors than the ruthenium chloride derived catalyst (Table 2). They are also much more selective, producing methane and water exclusively. The ruthenium chloride derived catalysts, on the other hand, also produce carbon monoxide and water (WGS chemistry).

Table 2. Relative Reactivities for Methanation of C02 using Alumina-Supported Ruthenium Catalysts• Catalyst Precursor RuClj3Hp

180°C 1.0

Ru(C0)5

4.7

Ru3(C0)12

12.6

H4 Ru4(C0\ 2

14.6

CRu6(C0\ 7

21.7

a Relative react1v1ty"" moles of CH 4 produced per total moles of ruthemum

Conditions· C02: H 2= 1:2 5, flow rate 0.25 mL/s; glass reactor 1/4 m. th1ck and 1ft long, w1th ca. 2 g of catalyst; products analyzed by GC-IR

48

Although all the surface species produced by the various ruthenium carbonyl derivative appear to be identical, it is obvious upon inspection of Table 2 that these vary significantly in catalytic actiVity toward methanation. Indeed a correlation is noted between metal cluster size and reactiVity, with the higher nucleartty clusters exhibiting greater reactivity. This is best seen in a plot of metal dispersion as determined from 02 chemisorption measurements and precursor, cluster nuclearity (Figure 1). Since the metal particles are generally greater than 50 A as determined by electron microscopy, it emerges that the degree of metal aggregation upon decarbonylation (or activation) is inversely proportioned to cluster nucleartty. The surface reactions of the activated cluster catalysts were studied by diffuse reflectance FTIR spectroscopy in an effort to identify the surface species which leads to methane production. 26 Figure 2 shows the diffuse reflectance spectra of the surface of an activated catalyst under a mixture of C02:H2 of 1:2.5, as is used in catalysis. Under this atmosphere the original carbonyl bands maintain their band pattern but are shifted slightly to lower energies at 2036 and 1954 cm-1. A formate species is also present on the surface under catalytic conditions as evidenced by the symmetric and antisymmetric ..J(C02) stretches at 1591 cm-1 and 1375 cm-1 and also the C-H stretch at 2904 cm-1. A 150'C, no methane is observed: however, when the sample is heated to 200'C, methane production commences as is seen by the appearance of the band at 3016 cm-1. This band can be assigned to the C-H stretching frequency of gaseous methane which is trapped in the DRIFTS cell, and so is observed in the IR band pattern. Methane production increases with increasing temperature, as can be seen by the amount of methane produced at 250'C in Figure 2.

No. of Ru atoms in catalyst precursor

Figure 1. Plot ofmetal dispersion 98 clusternuclearlty.

I.IRVENUI"18EAS CM-l

Figure 2. IR Spectra of Catalyst under C~ and H2 Mixture.

49

Independent experiments indicate that when a catalyst pretreated with C02 and H2 at 130'C to afford surface formates is evacuated and retreated with hydrogen, methane production occurred without an appreciable diminish of the infrared bands due to the formate species. Hence, hydrogenation of the surface formates does not appear to be the source of methane. This taken along with changes noted in the terminal CO region of the infrared spectra during catalysis leads us to propose a mechanism for methane formation which proceeds via metal carbonyl and RuO(ads) to RuC(surface species). Hydrogenation of RuO(ads) and RuC(surface species) ultimately result in methane and water production. In an effort to prepare even better metal dispersion, we have investigated a higher surface area support, i.e., a faujasite-type zeolite. In addition, the zeolite pores might limit metal crystallite growth upon activation of an intrazeolite metal carbonyl cluster. However, high nuclearity ruthenium carbonyl clusters are difficult to support on such zeolites from solution. Some success had been achieved by sublimation of Ru3(COl12 onto Y-type zeolite.27 In this instance it has been suggested that the ruthenium carbonyl molecules are located inside the supercages of the support. This does not seem possible since the Ru3(CO) 12 cluster appears to be too large (9.2 Al to fit through the zeolite pores (7.6Al. In our investigations we have employed two different procedures for supported Ru3(COl12 on zeolites. The first of these follows the previously reported vapor phase deposition onto the dry NaY support. The second preparation proceed through the Ru(CO)S monomer under high CO pressure. The size of the monomer (6.3 A) should allow penetration of the zeolite pores. It has been shown that the 13 A supercages of the NaY zeolite accommodates three molecules of the structurally similar Fe(CO)s molecule. 28 Therefore, clusterfication to the stable Ru3(CO) 12 upon release of CO pressure can occur inside the zeolite supercages. The location of the Ru3(COl 12 species in the two catalyst precursor preparations was determined to be outside the supercage for the vapor phase synthesis and inside the supercage for the high pressure synthesis (Figure 3) by their size selective reactivity with phosphorus donor ligands (PPh3 vs P(OMe)3). That is, upon addition of solutions of PPh3 (a ligand previously shown to be too large to penetrate the zeolite pores29J to the two differently prepared solid catalysts, only the vapor phase deposited catalyst reacted to provide Ru3(COli2-niPPh3)n derivatives in solution with concomitant little Ru3(CO) 12 remaining on the zeolite support. On the other hand the high pressure synthesized catalyst was found to only react with smaller phosphine ligands, such as PMe3, with no dissolution of the trinuclear cluster. 30

Figure 3. NaY with Ru3(COl12 encapsulated in the bottom-light supercage.

50

The catalytic activities of the two catalyst precursors upon activation under hydrogen or helium at 200'C for 8-10 hrs are reported as plots of percent conversion of C02 to CH4 per unit weight of ruthenium vs temperature (Figures 4 and 5). In both cases selectivity to methane was greater than 95%, with a trace of CO at temperatures at or above 300'C being the only other observable carbon-based product. In each case, the catalytic activity is compared to that of the previously discussed Ru3(COl12/A1203 precursor. As is evident in Figures 4 and 5 the best catalyst precursor for methanation of C02 is the one in which Ru3(CO) 12 was deposited from the vapor phase onto the outer surface of the zeolite. For the zeolite supported catalysts the metal dispersion was greater for the helium activated (>60"Al) than for the hydrogen activated catalyst (see Figure 6). Indeed in the case of the intrazeolite catalyst, helium activation appears to result in little metal aggregation, as is apparent in that the catalyst retains its yellow color upon activation, rather than decomposing to a grey solid (suggestive of the growth of metal particles) which is the case for the H2-activated catalyst. IDO

10 ::>

..,

a:

~

-..

50

~

~ w

40

I

>

~

u

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

B

20

~

200

2~

300

3110

400

TEMPERATURE I"Cl

Catalytic activity of catalyst precursor prepared by VP Figure 4. deposition reported as percent conversion of C02 to CH4 per .001g of Ru versus temperature following activation of the precursor a) under a flow of He (60 ml/min) at 200"C, and b) under a flow of H2 (60 ml/min) at 200"C. The dashed line represents the activity of 1% Ru3(COl12 on A1203 activated under H2 at 200"C.

100

B

cf!BO

""

~50 z

.0

~

~ 40

z

8

..:zo 350

400

Catalytic activity of catalyst precursor prepared by high Figure 5. pressure synthesis reported as percent conversion of C02 to CH4 per .01g of Ru versus temperature following activation of the precursor a) under a flow of He (60 ml/min) at 200"C, and b) under a flow of H2 (60 rnl/min) at 200"C. The dashed line again represents 1% Ru3[CO) 12 in Al203. 51

28

a::"

24

A

"' ow 20

(D

a::

0

(fJ

0

<{

16

0

u (fJ

w _j

B

12

0

:::;: ':' Q

><

10

15

20

25

PRESSURE (mm Hgl

Figure 6. CO chemisorption results for high pressure synthesized catalyst: a)Heactivated at 200'C: blH2-activated at 200"C: c)He-activated followed by catalysis at 300"C for 36 hours: dlH2-activated followed by catalysis at 300"C for 36 hours.

Contrary to the behavior of the vapor phase deposited Ru3(CO) 12 catalyst, which suffered a slight loss in catalytic activity with time, presumably due to a buildup of inactive carbon on metal sites, the behavior of the intrazeolite ruthenium catalyst displayed an increase in activity with time under catalytic conditions. Nevertheless, during this time period the metal dispersion decreased significantly (see Figure 6). For example, an increase in catalytic activity of >300% was exhibited by a helium-activated system after 72 hrs of catalysis at 300"C. This is most likely due to the high volatility of the very stable ruthenium oxides which migrate and agglomerate outside the zeolite supercages,31,32 where the ruthenium has been shown to be catalytically more active. The low activity initially manifested by the highly dispersed ruthenium metal in the intrazeolite catalysts is evidently a mass transport problem, even though the reactants and products are themselves all small entities. Copper has been implicated in the catalytic conversion of C02 to methanol, 33 hence we have investigated copper-ruthenium clusters for the formation of oxygenates in the C02 hydrogenation reaction.34 Cu-Y (73% of the Na+ replaced in NaY by cu+2 by ion exchange) zeolite supported ruthenium carbonyl catalyst precursors have been prepared via either the vapor phase or high pressure method and activated. In either instance no methane is observed for the C02/H2 reaction until a temperature of 350"C is reached. Even at this temperature the methanation activity is fairly low, and there are no indications of methanol production. Hence, the Cu-Y appears to deactivate the ruthenium towards methane production with no concomitant enhancement of catalytic activity for methanol synthesis. Our efforts in this area are continuing.

Acknowledgements. The author is most grateful to the National Science Foundation. whose support has made possible his contribution,s to the research described herein. He is likewise extremely appreciative to all his colleagues mentioned in the references, whose many original contributions have made this such an exciting area of research in which to work. References 84, 959 (1975); .!b!d... 85, 451 (1976).

1.)

E. L. Muetterties, Bull. Soc. Chim.

2.)

D. J. Darensbourg, Israel J. Chern., 15, 247 (1976).

52

Bel~ ..

3.)

R M. Lalne, R G. Rinker, and P. C. Ford, J. Am Chern Soc., 99, 252 (1977).

4.)

C. Ungermann, V. Landis, S. A. Moya, H. Cohen, H. Walker, R. G. Pearson, R. G. Rinker, and P. C. Ford, J. Am. Chern. Soc., 101, 5922 (1979).

5.)

P. C. Ford, Ace Chern. Res.. 14, 31 (1981).

6.)

J. C. Bricker, C. C. Nagle, and S. G. Shore, J. Am. Chern. Soc., 104, 1444 (1982).

7.)

M. M. Taqui Khan, S. B. Halligudi, and S. Shukla, Angew. Chern Int, Ed Engl., '1:7, 1735 (1988).

8.)

R. M. Laine and R. B. Wilson, Jr . .In: "Aspects of Homogeneous Catalysis," Renato Ugo, ed., D. Reidel Publishing Co., Dordrecht (1984).

9.)

(a) D. J. Darensbourg and R. A. Kudaroski, Adv Orgauomet. Chern., 22, 129 (b) D. A. Palmer and R. Van Eldik, Chern. Rey., 83, 651 (1983). (c) D. Walter, Coord Chern. Rey., 79, 135 (1987). (d) A. Behr, Angew, Chern. Int. Engl. (1983).

:M.... 27, 661

(1989).

10.)

D. J. Darensbourg and C. Ovalles, J, Am, Chern Soc , 106, 3750 (1984).

11.)

D. J. Darensbourg, R L. Gray, C. Ovalles, and M. Pala, J. MoL Catal., 29, 285 (1985).

12.)

D. J. Darensbourg, R L. Gray, and C. Ovalles, J. Mol. Catal., 41, 329 (1987).

13.)

D. J. Darensbourg and G. Gibson, in: "Experimental Organometallic Chemistry", A. L. Wayda and M. Y. Darensbourg, Eds., American Chemical Society. Washington, D. C. (1987).

14.)

D. J. Darensbourg, C. Ovalles, and M. Pala, J. Am· Chern. Soc .. 105, 5937 (1983).

15.)

D ..J. Darensbourg, M. Pala, andJ. Waller, Organoruetallics, 2, 1285 (1983).

16.)

G. R Steinmetz, A. D. Harley, and G. L. Geoffroy, Inorg. Chern., 19, 2985 (1980).

17.)

J. C. Bricker, C. C. Nagel, A. A. Bhattacharyya, and S. G. Shore, J. Am. Chern. ~107,

377 (1985).

18.)

D. J.Darensbourg, M.Y. Darensbourg, L. Y. Gob, M. Ludvtg, and P. Wiegreffe, J. Am. Chern. Soc .. 109, 7539 (1987).

19.)

D. J. Darensbourg and P. Wiegreffe, unpublished results.

20.)

D. J. Darensbourg and H. Pickner, Inorg, Chern, in press.

21.)

D. J. Darensbourg, R. L. Gray, and M. Pala, Organometallics, 1984, 3.. 1928.

22.)

D. C. Gross and P. C. Ford, Inorg. Chern.. 21, 1702 (1982).

23.)

D. J. Darensbourg, C. Ovalles, C. G. Bauch, Rev, Inorg, Chern 7, 315 (1985).

24.)

D. J. Darensbourg and C. Ovalles, Inom. Chern., 25, 1603 (1986).

25.)

V. L. Kuznetsov and AT. Bell, J. CataL 68, 374, (1980).

26.)

D. J. Darensbourg and D. J. Mangold, unpublished results.

27.)

J. G. Goodwin and C. Naccache, ,J Mol Cat, 14,259 (1982).

28.)

T. Bein and P. A. Jacobs, J. Chern. Soc .. Faraday Trans.. 79, 1819 (1983).

29.)

N. Herron G. D. Stucky, C. A. Tolman, Inorg. Chim. Acta. 100, 135 (1985).

30.)

D. J. Darensbourg and G. Gibson, unpublished results.

53

31.)

J. J. Verdonck, P. A. Jacobs, M. Genet, and G. Poncelet, J. Chern. Soc. Faraday I. 76, 403 (1980).

32.)

L.A. Pederson and J. H. Lunsford, J Catal, 61, 39 (1980).

33.)

B. Denise and R P. A. Sneeden, Chemtech. 108 (1982).

34.)

D. J. Darensbourg and Melissa Wagner, unpublished results.

54

REACTIVITY OF DINUCLEAR AND TETRANUCLEAR CLUSTERS OF MOLYBDENUM AND TUNGSTEN Malcolm H. Chisholm Department of Chemistry Indiana University Bloomington, IN 47405

ABSTRACT The reactivity of dinuclear and tetranuclear clusters of molybdenum and tungsten are reviewed with particular attention to Principles oxidative addition and reductive elimination sequences. pertaining to the design of a catalytic system are discussed in light of the above and specific attention is given to the hydrogenation and Recent studies of the isomerization of alkenes by W4 (H) 2 (OPri) 14 . reductive coupling and selective cross coupling of ketones/aldehydes to olefins are presented in their reactions with W2 (0R) 6 (py) 2 compounds. A two step mechanism is proposed based on the initial cleavage of the C-0 double bond to give an alkylidene bridged oxo compound followed by A pinacol coupling is ruled further reaction with aldehyde or ketone. Recent studies of the activation of carbvn monoxide by W2 (0R) 6 out. and W4 (0R) 12 clusters are presented with the latter providing the first example of the stoichiometric cleavage of C"'O to carbide and oxide ligands on a cluster. INTRODUCTION Dinuclear and polynuclear compounds containing metal-metal bonds represent a particularly attractive class of inorganic/organometallic The metal atoms provide a 1. compounds for studies of reactivity. The metal-metal bonds 2. template for the assembly of substrates . 1 serve as a reservoir for electrons and the presence of several metal atoms or a multiple bond of high order provide the opportunity for multielectron redox reactions that are not possible for mononuclear The presence of two or more metal atoms, especially in centers. 2 3. cases where the metal atoms are different, raises the possibility of The binding of organic 4. multisite activation of substrates. 3 fragments to cluster fragments often has a striking analogy to the organometallic chemistry of metal surfaces. 4 This, together with (3) above, raises the possibility that certain reactions will show an a reaction having a requirement for a specific "ensemble effect" number of metal atoms. 5

Metal-Metal Bonds and Clusters in Chemistry and Catalysts Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

55

Despite these advantages, the development of catalytic cycles based on cluster compounds has not been as fruitful as some of the pioneers in the field had hoped. Indeed present workers in cluster chemistry have to some extent inherited high mortgage payments because the field was oversold by the early researchers and the funding agencies during the 1970's. From this it should not be viewed that catalytic cycles based on clusters have not been discovered for the; have and these include olefin isomerization and hydrogenation, 6 hydroformylation, 8 • 10 hydro~enation of CO to ethylene glycol, 11 - 14 the water-gas shift reaction s- 18 and tertiary amine metathesis. 19 - 21 However, none have been taken to commercial production and few are unique to clusters or more efficient than catalysis by mononuclear systems. In addition, there is often the problem of establishing that cluster compounds are actually involved in the catalytic cycle. This problem is aggravated, in some cases, by relatively facile cluster degradation reactions. The uptake of substrates may lead to cleavage of M-M bonds and ultimately the formation of mononuclear species. Alternatively, elimination reactions may yield highly unsaturated species that condense further to give heterogeneous solutions. In these cases the cluster compounds may find new uses as catalyst precursors and may provide novel low temperature routes into alloys and other solid-state phases. 22 • 23 These topics are dealt with in detail in other talks at this symposium. What I wish to do is to focus on two types of reactions that require redox activity of a dinuclear center with a M-M multiple bond, namely reductive elimination and oxidative addition, and then to combine these with the "ensemble" effect to show that unique types of reactivity are possible for dinuclear and cluster compounds with M-M bonds. I define unique reactivity as a reaction pathway not possible for a mononuclear complex and not merely a formerly unknown reaction. Synthetic Routes to Dinuclear and Cluster Compounds of Molybdenum and Tungsten Two traditional routes to dinuclear and cluster compounds involve the use of either mononuclear precursors or polymeric materials with cluster subunits. An example of the former is the well known synthesis of Mo 2 (0 2 CR) 4 (M~M) compounds from the reaction between Mo(CO) and RCOOH in refluxing diglyme. 24 The detailed nature of this reaction is not known but it involves redox chemistry, carbonyl ligand eliminations, carboxylate bridge and metal-metal bond formation. An example of the second approach involves the heterogeneous reactions between the polymeric metal chlorides WC1 4 or MoC1 3 and LiNMe (4 or 3 equiv, respectively) in orfanic solvents such as THF/hexane {hat yield M2 (NMe 2 ) 6 (M=M) compounds. 2 In the case M = Mo, there is no oxidation state change but for M=W the reaction is more complex involving redox chemistry wherein either LiNMe 2 acts as a reducing agent or ligand redistributions occur effecting redox disproportionation: 3W(IV) -+ 2W(III) + W(VI). [The formation of some W(NMe ) in the reaction between WC1 4 and LiNMe 2 (4 equiv) is generally ~b~erved, consistent with redox disproportionation.] Starting with a dinuclear complex with a M-M multiple bond, it has been one of our goals to establish the fundamental rules for reductive elimination and oxidative addition reactions. In principle, these complementary reactions could form the basis for catalytic cycles of the type now well documented for the later transition elements. The d 6 -d8 relationship so commonly seen for octahedral ML6 and squareplanar ML4 complexes (M- Rh, Ir, Pd, Pt, Au) could be extended to the dinuclear chemistry of (M=M) 6+ and (M~M) 4 + where M = Mo and W.

56

Conversion of (M=M) 6 + to (MiM) 4+ With this in mind, my group tried to synthesize M-M quadruply bonded compounds from dialkyl M-M triply bonded compounds. The dialkyl compounds were synthesized in a straightforward metathetic procedure according to eq. 1. 26 (a)

(1)

(b)

The substitution of chloride by alkyl groups in eq. l(b) involving the use of either organolithium or organomagnesium reagents in hydrocarbon solvents (hexane, THF, ether), ~enerally at 0" to -78"C, has been extended to include aryl, 7 benzyl, 27 allyl, 28 cyclopentadienyl 29 and indenyl ligands. 29 The resultant 1,2diorganoderivatives M2 R2 (NMe 2 ) 4 can be purified by crystallization (hexane) or sublimation under vacuum. The initial strategy to induce reductive elimination was to convert the NMe 2 to 0 2 CNMe 2 ligands by the insertion of C0 2 into NMe 2 ligands. The latter reaction proceeds under mild conditions (T ~22"C) upon addition of C0 2 (1 atmos) to hydrocarbon solutions of M-NMe 2 containing compounds. 30 The approach proved successful for molybdenum. When the alkyl group R contained fi-hydrogen atoms the addition of C0 2 gave Mo 2 (0 2 CNMe 2 ) 4 with the elimination of alkane and alkene. 31 The alkane formed was derived from the formal transfer of the fi-H atom of one alkyl ligand to the a-carbon atom of the other as shown in eq. 2. Related studies showed that the reaction was an intramolecular reductive elimination and as such a dinuclear analogue of alkyl group disproportionation which is well documented in the reductive elimination reactions involving organoplatinum compounds. 32 22"C benzene

~

(2)

When R = benzyl, a similar reductive elimination occurred but bibenzyl was the major organic product. In the presence of a hydrogen atom donor, such as 1,4-cyclohexadiene, toluene and benzene were formed suggesting a homolytic Mo-C bond cleavage was operative. 33 For the compound Mo 2 ( CH 2 ) 4 (NMe 2 ) 4 , which contains a 1, 2dimetallacyclohex-1,2-ine central core, two equivalents of ethylene were liberated, eq. 3. 4

( 3)

Rather interestingly this strategy did not work in the synthesis of W2 (0 2 CNMe 2 ) 4 • The reactions between W2 R2 (NMe) 4 compounds and C0 2 are more complex. However, the addition of a carboxylic anhydride to 1, 2 -W2 R2 (NMe 2 ) 4 compounds did prove successful in providing a general synthetic route to W2 (0 2 CR') 4 compounds according to eq. 4 when the alkyl ligands R contained fi-hydrogen atoms. 35

57

22"C benzene

(4)

For R =benzyl, aryl, CH 2 SiMe 3 and CH 2 CMe 3 , the reaction between 1, 2-W 2 R2 (NMe 2 ) 4 compounds and carboxylic acid anhydrides gave W2 R2 (0 2 CR' ) 4 compounds whic~ 6 upon thermolysis or photolysis yielded W2 (0 2 CR') 4 compounds, eq. 5.

(i)

22"C benzene

( 5) (ii)

The ease of reductive elimination in reaction S(ii) followed the order R- benzyl> aryl > CH2 SiMe 3 > CH 2 CMe 3 . In related studies of the reactions between 1,2-Mo 2 R2 (NMe 2 ) 4 compounds and acid anhydrides R' COCOOR' the ease of reductive elimination was shown to be Mo > W and only in the case of R = CH 2 CMe 3 were compounds of formula Mo 2 R2 (0 2 CR') 4 isolated. 36 However, it seems that a common reaction pathway is operative for the two metals. Several important mechanistic questions remain unanswered concerning the detailed nature of the reaction pathway leading to alkane and alkene in reactions 2 and 4. For example, it is not known whether or not the ~-H atom transfer occurs by way of a metal-hydride intermediate, nor is the order of elimination of alkane and alkene known. After completing a designed synthesis of Mo~Mo bonds from (Mo=Mo) 6 + -containing compounds by alkyl group disproportionation, it came as a surprise, at least initially, to stumble upon an example in studies of the alcoholyses reactions of 1,2-Mo 2 R2 (NMe 2 ) 4 compounds. 37 The latter react with alcohols by initial replacement of NMe 2 ligands by OR' ligands but, depending upon the steric requirements of R and R', protolysis of the Mo-C(alkyl) bonds may occur leading to Mo 2 (0R') 6 compounds with the liberation of alkane, RH, in addition to amine. These reactions proceed most slowly for bulky combinations of R and R' , e.g. R = CH 2 CMe 3 and R' =But and Pri. Rather interestingly when R contains ~-hydrogen atoms, the formation of monoalkyl compounds Mo 2 R(OR') 5 is observed. Labelling studies, such as that shown in eq. 6, reveal that the elimination of alkane (one equiv) is not due to protolysis by the added alcohol. 37

o·c hexane

... (6)

In reactions involving .Mo 2 (Prn) 2 (NMe 2 ) 4 the initially formed monoalkyl complex was Mo 2 (Pr,)(OBut) 5 which, with time and in the presence of amine, isomerized to Mo 2 (Prn)(OBut) 5 • 37 Clearly a rather circuitous reaction sequence is implicated, one involving ~-hydrogen

58

atom transfer, reductive elimination of alkane and alkene followed by oxidative addition of ButOH and insertion of alkene into a metal Support for this general hydride derived from the R'OH hydrogen atom. scheme came from the alcoholysis reaction shown in eq. 7 where the structurally subsequently and isolated was alkoxide Mo~Mo Me 2C-CH 2 Presum~bly in reaction 7 the bulky alkene, characterized. does not compete effect~vely for access to the metal center.

oac

(7)

hexane

Mo 2 (0Pri) 4 (HOPri) 4 + BuiH + Me 2C-CH 2 The compound Mo 2 (OPri) 4 (HOPri) 4 is extremely labile and reacts with the donor ligands pyridine and PMe 3 to give Mo 2 (0Pri) 4L 4 compounds It also reacts with ethylene to give Mo 2Et(OPri) 5 (L - py, PMe 3 ). 38 thereby completing the proposed cycle for the formation of monoalkyl compounds in reactions of type 6. The aforementioned reaction that converts a Mo~+ center to a Mo~+ center under extremely mild conditions may be compared with the now classic dinuclear oxidative addition reaction shown in eq. 8. 39

(8)

Conversion of (M=M) 6 + to (M-M) 8+ Whereas reductive elimination from (M=M) 6 + centers occurs in the order M - Mo > W, the ease of oxidative addition is in the reverse order, M- W >Mo. For instance, whereas halogens and dialkylperoxides add to M2 (OR) 6 compounds, 40 mild reagents such as alcohols show redox Our first encounter with reactivity with M - W but not with M - Mo. this type of facile oxidative addition to the (W=W) 6+ center was during Addition of bulky studies of alcoholysis reactions of W2 (NMe 2 ) 6 • W2 (0But) 641 but the less sterically alcohols such as ButOH gave demanding alcohols with more acidic hydroxyl protons MeOH and EtOH gave tetranuclear compounds W4 (OR) 16 , where R - Me and Et, containNo intermediates were detected ing tungsten in oxidation state +4. 42 in alcoholyses reactions employing W2 (NMe 2 ) 6 and ROH where R - Et or Studies The reactions proceed very rapidly at room temperature. Me. of the reaction between W2 (NMe 2 ) 6 and the secondary alcohol PriOH revealed more insight into the complexities of the system. Addition of an excess of PriOH to hydrocarbon solutions W2 (NMe 2 ) 6 leads to a mixture of products as shown in eq. 9. 43

22°C

-----1·~ W4(H)2(0Pri)14 hexane

of

+

(9)

When reaction 9 was carried out at 0°C the initially formed cluster carbido the and W2 (0Pri) 6 (HNMe 2 ) 2 were products The latter compound is formed inca 5-10% yield W4 (C)(NMe)(OPr 1 ) 12 . and is apparently a product of a degradation reaction of a coordinated The mechanism leading to W4 (C)(NMe)(OPri) 12 is not NMe 2 ligand. 44 known but it is not unreasonable to speculate that the NMe 2 ligand

59

undergoes successively (a) ,8-H atom transfer to give a W2 (JL-CH2 -NMe) containing species, 45 (b) CH 2 -NMe bond cleavage to give CH 2 and NMe ligands and (c) cluster condensation with further C-H bond activation to yield the final product W4 (p. 4-C) (p.-NMe) (0Pri) 12 . In any event the major product at short reaction times is the Lewis base adduct W2 (0Pri) 6 (HNMe 2 ) 2 . At first we suspected that this might be an intermediate in the formation of W4 (C)(NMe)(OPr 1 ) 12 but this was soon ruled out. Furthermore, and to our great surprise, we found that starting with pure W2 (0Pri) 6 (HNMe 2 ) 2 dissolved in.hexane or toluene and adding Pr 1 0H failed to give yields of W4 (H) 2 (0Pr 1 ) 14 approaching those obtained in reactions starting with W2 (NMe 2 ) 6 , eq. 9, that were allowed to proceed for ~ 12 hours at room temperature. The direct addition of PriOH to W2 (0But) 6 leads to very rapid substitution of Buto by Pr 1 0 ligands but not to W4 (H) 2 (0Pr 1 ) 14 . . A black crystalline compound was obtained of empirical formula W(0Pr 1 ) 3 , based on elemental analysis, and approximate molecular formula W4 (OPri) 12 , based on cryoscopic molecular weight determinations. The samples, though crystalline, failed to yield suitable X-ray diffraction data for a molecular structure determination and the 1H NMR spectra were complicated and varied from sample to sample. Finally David Clark obtained a black crystalline sample from 1,2dimethoxyethane (dme) that gave satisfactory X-ray diffraction data and once again John Huffman came to our aid with the vital piece of information. The black crystals obtained from the reaction between W2 (0But) 6 and Pr 1 0H were, when crystallized from dme, a 1:1 mixture of W2 (0Pri) 6 and W4 (0Pri) 12 , eq. 10. 46 The unit cell contained two molecules of the dinuclear (M=M) 6+ -containing compound and two molecules of its dimeric 12-electron cluster, W4 (0Pri) 12 . A reevaluation of previous work showed that the black crystals obtained from hexane at~ -l5°C were, in fact, pure W4 (0Pri) 12 and in solution W4 (0Pr 1 ) 12 and W2 (0Pr 1 J 6 were in equilibrium. 47 (i)

(10)

(ii)

The addition of PriOH to hydrocarbon solutions of W2 (0Pri) 6 leads to very rapid alcohol exchange on the NMR time-scale - only one time averaged Pr 1 0H signal is observed. A slow exchange is observed for W4 (0Pri) 12 but !leither W2 (0Pri) 6 nor W4 (OPri) 12 re101ct with PriOH to give W4 (H) 2 (0Pr 1 ) 14 . However, if W2 (0Pr 1 ) 6 and Pr 1 0H are allowed to react in a hydrocarbon solvent in the presence of 6 equiv of HNMe 2 or 6 equiv of NEt 3 , then W4 (H) 2 (0Pr 1 ) 14 is formed. The dimethylamine liberated in the alcoholysis reaction 9 is therefore involved in the formation of W4 (H) 2 (0Pri) 14 . It appears that oxidative addition of Pr 1 0-H to the (WsW) 6 + is base promoted and addition of Na0Pr 1 in PriOH to W2 (0Pri) 6 brings about the same effect as shown in eq. 11. 43 22°C hexane The sodium ditungsten compound forms a diglyme adduct that is soluble in hexane and the molecular structure of W2 (H)(0Pri) 8 Na.diglyme reveals that the sodium ion is coordinated to two of the terminally bonded OPri oxygen atoms of a confacial bioctahedral moiety 0 3W(p.-H)(p.0)2W03 in addition to the three oxygen atoms of diglyme. The central NaW2 (H)0 8 skeleton is shown below.

60

0(39~ Addition of Me 2 NH/Cl- to a solution of NaW 2 (H) (OPri) 8 generates the tetratungsten dihydride according to eq . 12. 43

22°C hexane/THF

(12)

The tetratungstendihydrido compound W4(H) 2 (0Pr 1 ) 14 maintains its tetranuclear integrity in hydrocarbon solutions at low temperatures according to cryoscopic molecular weight determinations in benzene. 48 It can be sublimed in vacuo at £.!!, 80°C, 10- 4 Torr and shows a strong ion corresponding to [W2 (H) (OPri ) 7 ] + in the mass spectrometer. The molecule is fluxional on the NMR time-scale. There is a hydride resonance at £.!!, 6 8 ppm flanked by satellites due to coupling to two 1 J 1 aaw_ 1 H "' 100 Hz . equivalent 183 W nuclei : There is only one time averaged signal for the seven crystallographically distinct 0Pr 1 ligands in the temperature range -70° to +60°C at 300 MHz in toluened8. Evidently the alkoxide ligands scramble by an open and closing of bridges but, on the NMR time-scale, the hydride ligands are not scrambled over the four tungsten atoms . The addition of ·neutral donor ligands, L = pyridine and PMe 3 , causes a reversible reaction with the tetranuclear dihydride to give W~(~-H)(OPr 1 ) 7 L compounds, though no adduct has been isolated and fully characterized. The hydride ligand does not exchange with Pr 1 0D even in the presence of base suggesting that it is neither acidic nor basic and that a reductive elimination to W2 (0Pri) 6 (HOPri) is not operative. 48 The hydride does react with ethylene reversibly to give, by NMR spectroscopy, W4Et 2 (0Pri) 14 . Attempts to obtain the latter compound by crystallization yielded only W4 (H) 2 (0Pri) 14 . The dihydridotetratungsten complex is a catalyst for olefin isomerization, eq . 13, and, in the presence of both olefins and dihydrogen, olefin

a-olefins

W4 (H) 2 (OPri) 14

------------------------~.. ~

hexane/benzene, 22oc

internal olefins

(13)

61

hydrogenation occurs, though the details of latter reaction have not yet been investigated in any detail. It is sufficient to note that the system seems poised for catalysis based on reversible C-H, W-C and W-H bond forming reactions. Addition of w-Acid Ligands to M-M Multiple Bonds The fact that M-M bonds are generally weaker than M-L bonds renders dinuclear compounds with M-M multiple bonds susceptible to cleavage by w-acid ligands such as C=O, RNC and NO. See eqs. 14 49 and 15.50

...

22•c hexane

(14)

M = Mo, W, R = But or Pr 1 22•c hexane ...

(15)

Walton and his coworkers have studied extensively the cleavage of

M~+ units in their reactions with isocyanides, e.g. eq.

22·c MeOH, PF 6 -

...

16. 51

2Mo(CNR)~+

(16)

Schrock and coworkers were the first to observe a metathesis of M=M and C=C bonds in the cleavage reaction shown in eq. 17. 52 • 53

-----1·· o·c hexane

R- Me, Et, Pr 1

2(But0) 3W=CR

( 17)

In reactions with nitriles, RC=N, the W=W bond in W2 (0But) 6 is also cleaved to give (But0) 3W•N and (Buto) 3W•CR compounds. 53 Not all reactions between W2 (0R) 6 compounds and alkynes proceed in a manner akin to that of eq. 17; nor do all alkynes react with W2 (0But) 6 in this way. There are steric and electronic factors involved and in certain instances an equilibrium can be seen in solution between a dimetallatetrahedrane and the tungsten alkylidyne complex, eq. 18. 54 • 55

(18) Reaction 17 can be viewed as an oxidative cleavage of a W=W bond in which the alkyne is reduced by six electrons. The tungsten atom in (But0) 3 W=CR can be counted as W(6+) if the alkylidyne ligand is counted as a 3- ligand. The equilibrium 18 can then be considered as an internal redox reaction. 56 The recognition that, in a formal sense anyway, a (W~W) 6 + center could be oxidized to W~o+ or two W(6+) centers upon the addition of alkynes led us to investigate the reaction between ketones and W2 (0R) 6 compounds.

62

The Reductive Coupling and Selective Cross-Coupling of Ketones Aldehydes

and

In 1984, we reported that W2 (0Pr 1 ) 6 (py) 2 and acetone reacted in hydrocarbon solvents to give a novel tetranuclear oxo alkoxide, W4 02 (0Pri) 12 . 57 The organic molecule liberated in this reaction was tetramethylethylene and this suggested the stoichiometry shown in eq. 19. 22"C hexane

(19)

Our initial impression was that we had discovered a molecular model for the McMurray reagent. The latter involves the reduction of TiC1 3 with LiAlH 4 or Li/Na/K and serves as a useful reagent in organic synthesis for the formation of C-C double bonds. The reaction is believed to proceed via initial pinacolate formation. Cotton, Walton and their coworkers had observed the coupling of ketones to give pinacolate ligands at w:+ centers. 59 However, more recent studies of the reactions between W2 (0R) 6 compounds and ketones by Jeffrey Klang in this laboratory show that the reaction pathway in 19 is totally different. 60 First it should be stated that there are, as is usual for metal alkoxide chemistry, steric constraints. W2 (0But) 6 and acetone fail to react at 22"C in hydrocarbon solvents, as do W2 (0CH 2But) 6 and (But) 2C-O. Much of the recent work has focused on the use of W2 (0CH 2But) 6 (py) 2 as a starting ditungsten· alkoxide because the pyridine ligands dissociate reversibly in solution and the neopentoxide ligands allow considerable flexibility in terms of substrate access to the metal center. W2 (0CH 2But) 6 (py) 2 and acetone react in hydrocarbon solvents at o•c to give a compound of formula W2 (0CH 2But) 6 (0CMe 2)(py) that can be crystallized from hexane at -78•c in >80% yield. The NMR spectroscopic data for this compound indicate that the molecule lacks any element of symmetry. There are six different neopentoxide ligands each of which has diastereotopic methylene protons and the two methyl groups derived from acetone appear as two singlets of equal intensity. The ketonic carbon atom in the compound derived from Me 213 C=O is found as a resonance at 6 163.7 that is flanked by satellites of intensity 24% due to coupling to 183 W, I = 1/2, 14.5% nat. abund. The magnitude of the latter is typical of a W-Csp 2 bond coupling. 54 Collectively the data are consistent with a Jl-propylidene compound W2 (OCH 2But) 6 (0) (p.CMe2)(py) having one of two structures shown below.

63

Fig.

1.

A ball-and-stick drawing o~ the W2 (0CH 2But) 6 (JL-CMe 2 )(0)(py) molecule. W-W = 2. 705(1) A, W-O(oxo) = 1.684(4) A and WC(alkylidene) = 2.12(1) A (ave).

Though we have not as yet been able to obtain single crystals of W2 (0CH 2But) 6 (0)(CMe 2 )(py) suitable for a detailed single crystal x-ray study we have structurally characterized W2 (0CH 2 But) 4 (0 2 CCF 3 ) 2 (0)(CMe2)(py) which was formed by the addition of CF3 COOH to the former compound. The molecular structure found in the solid-state is shown in Figure 1 and verifies the cleavage of the ketonic C-0 bond. There is an octahedrally coordinated tungsten atom linked to a five coordinated tbp tungsten by a JL-OCH 2But and a JL-CMe 2 ligand. The oxo ligand occupies a terminal equatorial site of the trigonal bipyramid. The reaction between 9-fluorenone and W2 (0CH 2But) 6 (py) 2 in hexane at ooc leads to a similar 1:1 adduct which is formulated as a dit\.\ngsten oxo-JL-alkylidene. Addition of acetic acid to the latter liberates fluorene (identified by 1 H NMR spectroscopy and m.pt.), consistent with protonolysis of a fluorenylidene ligand. The initial reaction between a ketone and the ditungsten hexaalkoxide at ooc can be viewed as a four electron reduction of the ketone to alkylidene and oxo ligands and oxidation of the (W=W) 6 + center to (W-W) 10 +, eq. 20. 0°C hexane •

(20)

A further reaction between the ditungsten oxo-JL-propylidene ligand and acetone or other ketones or aldehydes occurs in hydrocarbon solvents at room temperature leading to the liberation of olefins and an as yet not fully characterized tungsten product presumably W2 (0) 2 (0CH 2 But) 6 . This stepwise reaction sequence allows for the selective reductive cross-coupling of ketones/aldehydes to olefins which was not previously possible. Some alkenes formed in this way are

64

listed in Table 1. In a typical reaction 300 to 500 mg of W2 (0CH 2 But) 6 (py) 2 was allowed to react with the ketone (3 equiv) in hexane at ca 22 •c for 24 h with stirring. Water was then added to destroy the tungsten alkoxide and the aqueous layer was extracted with ether and the organic product separated by column chromatography (Si0 2 , hexane/Et 2 0). The olefins were identified by NMR, mass spectrometry Table 1.

Alkenes Formed by Reductive Coupling or CrossCoupling of Ketones/Aldehydes by W2 (0CH 2 But) 6py2 .a

Aldehyde/Ketone

Olefin

51%

21%

o-CHO

3

0<

b 4

5

b

~H a

~b H

34%

44%

66%

6

36%

7

18%

Yields are unoptimized; b Olefins 4 - 7 were formed by cross-coupling reactions using W2 (0CH 2 But) 6 (0)(CMe 2 )(py). 65

and melting optimized. spectroscopy yield should

points. The isolated yields given in Table 1 are not From following the course of various reactions by NMR employing 13 C labelled ketonic carbon atoms the idealized be very much higher in most cases.

The reactions between W2 (OR) 6 compounds and ketones combine two known reactions in mononuclear chemistry. (1) The reaction between an early transition metal alkylidene and a ketone to give an alkene and a metal-oxo derivative. 61 (2) The cleavage of ketonic carbonyl bonds in the reactions between WC1 2 (PMe 2Ph) 4 to give W(6+) oxo-alkylidene complexes. 62 Activation of Carbon Monoxide:

Cleavage to Carbide and Oxide

The ability of W2 (0R) 6 compounds to reductively cleave c~c and c~N to alkylidyne and nitride ligands requires a formal six electron process. An obvious question arises: can this process be extended to c~o and N~N? Thus far we have not observed N2 activation. Probably N2 is too poor a ligand to coordinate to the W2 center and this provides a kinetic barrier. Thermodynamically there is every reason to believe that the reaction between W2 (0But.) 6 and N2 to give [ (But.0) 3 W~N]"" 63 is enthalpically favorable. However, carbon monoxide, which is both a better u-donor and a better 71"-acceptor than dinitrogen, binds to M2 (0R) 6 compounds. The addition of one equivalent of CO to M2 (0R) 6 compounds or their pyridine adducts, M2 (0R) 6 (py) 2 gives compounds of formula M2 (0R) 6 (~­ CO) where R =But and M2 (0R) 6 (py) 2 (~-CO), where R = Pr 1 and CH2 But, for both M - Mo and W. 50 •64 Members of each group of compounds have been fully characterized and it is instructive to compare the properties of the carbonyl adducts as a function of metal and coordination environment. Pertinent characterization data are given in Table 2. In all cases the addition of CO (one equiv) leads to a central M2 (~-CO) moiety having formally M-M and C-0 double bonds. The extremely low values of v(CO) reflect the extensive mixing of M-M ,.. and CO 1r* orbitals. The compounds are inorganic analogues of cyclopropenones.65 The values of v(CO) are lower forM= W than forM= Mo which reflects the orbital energetics of the M-M 71"-bonding orbitals. From photoelectron spectroscopy we know that the 1st ionization potential for M2 (0R) 6 compounds represents ionization from the M-M ,.. MO of the triple bond and that this IP occurs at lower energy by £g 0.5 eJ for M = W relative toM= Mo. 66 Put another way, the Wd,..-to-CO ,.. Table 2.

Compound

Bridging Carbonyl Adducts of Molybdenum and Tungsten Alkoxides. M-M

A

M-C

A

C-0

A

v(CO)cm- 1

513C ppm

Mo 2 (0But.) 6 (CO)

2.498(1)

2.02(1)

1.21(2)

1670

273

W2 (0But.) 6 (CO)

2.526(1)

2.00(1)

1.25(1)

1590

291

Mo 2 (0Pr1 ) 6 (py) 2 (CO)

2.486(2)

2.06(1)

1.19(1)

1655

331

W2 (0Pr 1 ) 6 (py) 2 (CO)

2.499(3)

2.06(1)

1.22(4)

1555

[W2 (0Pr 1 ) 6 (C0)] 2

2.657(1)

1. 95(1)

1.35(1)

1272

66

305

bonding is favored in the W2 (11-CO) compounds relative to Mo 2 (11-CO) compounds because the orbital energy separation is smaller for tungsten. The extremely low values of v(CO) for the bridging carbonyl ligands suggests that they should show nucleophilic behavior and, consistent with the ionic resonance forms drawn below, we find that dimerization of two M2 (11-CO) units forM~ W is possible, eq. 21. 67

o-

0

1

~ /"-.. M=M

c

' /- M + M22•c

~ W4 (C0) 2 (0R) 12 + ButOH -,-----1... hexane

(21)

Upon formation of the tetranuclear carbonyl compound the W-W bond distance increases from 2.52 to 2.66 A, the C-0 distance increases from 1.22 to 1.35 A but the W-C distance decreases from 2.05 to 1.95 A. The carbonyl oxygen to tungsten distance of 1. 97 A is also worthy of attention; it is too long to be cbnsidered a simple dative bond as is often seen in M (11-CO)-M' interactions, where M' ~a Lewis acidic metal center. 68 It i; more like a terminal alkoxide 0-W distance where a a bond is supplemented by ~-bonding, Op~-to-Wd~. 69 The bonding in these 11-CO compounds has been the subject of a theoretical treatment and it is sufficient to say that the calculations The CO ~* support the qualitative description presented above. orbitals are populated and the filled CO ~ orbitals are drained of electron density by Op~-to-Wd~ bonding. 70 The reaction sequence shown in eq. 22 provides a stepwise reduction of M-M and C-0 bond order from 3 to 2 to 1, and a four electron reduction of CO.

(i)

M=M

+

g /""M=M 0

c..o

(22)

(ii)

-+

The previously described reduction of ketones to oxo-alkylidenes compounds led us to think that the reaction between by W (OR) W (OR) 6 (J1-~0) and W2 (0R) 6 compounds might lead to cleavage of the C-0 b~nd of the carbonyl ligand to generate carbide and oxide ligands, eq.

23.44

67

+

M=M

-+

(23)

There is, however , a problem in selecting the appropriate attendant alkoxide ligands for the reaction shown in 23. The W2 (0R) 6 (~-CO) and W2 (0R) 6 compounds must be coordinately unsaturated. The ~-CO functionality of one molecule must be capable of being a ligand to W2 (0R) 6 • Furthermore, the reaction between W 2 (0R) 6 (~-CO) and W2 (0R) 6 must be preferred with respect to the dimerizations that yield W 4 (0R) 12 (~-C0) 2 and W4 (0R) 12 compounds , reactions 22(ii) and lO(ii), respectively. In reality all of the above are competitive as was seen in the reaction shown in eq. 24 . 22"C hexane

(24)

The use of 13 C labelled carbon monoxide allows the reaction shown in eq. 24 to be easily monitored by 13 C NMR spectroscopy. Evidence for the formation of compounds of formula W4 ( 13 C)(O)(OR) 12 is obtained from 13 C spectra that show a carbon signal at li ca 360 ppm flanked by satellites due to the various isotopomers having 183 W nuclei, I = 1/2, 14.5% nat. abundance. Three types of spectra are seen for compounds formulated as W4 (C) (0) (OR) 12 depending on the nature of the R group. See Figure 2 . However, in each it •is reasonable to invoke a common butterfly W4 (~-C) moiety as was found in the structurally characterized compound W4 (C)(NMe)(OPr 1 ) 12 44 and is similarly seen in

·~t\?l

l; c.·'w./

r l '?l f' i·. o::::::._~0 ·w ,./

o

j~'\??'w wtz:J-w·/

0 .·

366.5

365.5 ppm

Fig .

68

2.

13 C

364 5

349

347

348

346

ppm

signals of the carbido carbon in, from left to right, W4 (1 3 C)(O)(OR) 12 where R = CH 2 But, CH 2 -cy-Bu and Pr 1 , respectively. Spectra were ob.tained at 125.76 MHz, 22 •c from toluene-d8 solutions.

carbonyl supported tetranuclear carbido clusters, e.g. Fe 4 (J.1 4 C)(C0)13.71 In this geometry the carbido ligand is strongly bonded to the wingtip metal atoms but more weakly bonded to the two backbone For a certain W4 (JJ- 4 -C) moiety lacking metal atoms of the butterfly. any element of symmetry there are two large, W-wingtip-C, and two For a W4 (C)(O)(OR) 12 small, W-backbone-C, values of 1 Jtasw_13c· molecule having virtual C2v symmetry only two values of 1 Jtasw_13c will be seen, one large and one small, but the relative intensities of the two satellites will be twice that observed for a W4 (C)(O)(OR) 12 molecule lacking any element of symmetry. It was not possible to separate W4 (C)(O)(OR) 12 compounds from the The implication was, however, other compounds formed in reaction 24. So why not that it took two W=W centers to cleave carbon monoxide. start with a W4 (OR) 12 cluster, a 12 electron cluster formed from the coupling of two W2 (0R) 6 molecules, and take advantage of the multisite activation possibilities of a W4 unit? We have _now examined the reactions between W4 (OCH 2R) 12 compounds (R =But, Pr 1 , cy-C 4H7 , cy-C 5H9 , cy-C 6 H11 ) 72 and carbon monoxide (l equiv) in hydrocarbon solvents at room temperature and in toluene-d 8 in Care must be taken to limit the the temperature range -78"C to 22"C. addition of CO to l equiv. since the W4 (OCH 2R) 12 compounds can react with more than one equivalent of CO and this leads to competitive reactivity that detracts from the desired reaction, namely that shown in eq. 25. 73

22"C hexane

•

(25)

The compounds formulated as tetratungsten carbido-oxo clusters can be obtained as black crystalline products in ~ 40-60% yield based on the stoichiometry shown in 25. From following the reactions by 13 C NMR spectroscopy >70% of the added 13 CO ends up as the W4 (JJ.-C)(O)(OCH 2R) 12 We have as yet not been able to obtain a suitable single compounds. crystal for an X-ray diffraction study. Typically, the crystals showed diffraction to only small angles as was found for W4 (OCH 2R) 12 This is probably a result of hexagonal close-packed compounds. molecules with a molecular disorder.

How can we be sure that this carbonyl C-0 bond has been cleaved? Could the 13 C NMR data correspond to a W4 (JJ- 4 -CO) containing compound of To address this problem Charles the type characterized by Shriver? 74 Hammond 75 employed further 13 C NMR spectroscopic studies using a 60:40 Using high field and high resolution 13 C NMR tsctGo: tsctso mixture. spectroscopy there is a readily measurable isotope chemical shift The 13 C resonances for the mixture of 13 C16 0: 13 C18 0 employed effect. The chemical shift separation in these studies are shown in Figure 3. The compounds W2 (0But) 6 (JJ.-CO), W4 (JJ.-C0) 2 (DCH 2But) 12 is 0.050 ppm. and W4 (C) ( 0) ( OCH 2R) 12 were prepared from the 60:40 mixture of tsctGQ;Isctso and their 13 C spectra recorded in the region of interest. The presence of the C-0 bond is readily apparent in the spectra of W2 (0But) 6 (J.1-CO) and W4 (JJ.-C0) 2 (0CH 2But) 12 but for the W4 (C)(O)(OCH 2R) 12 compounds the 13 C spectra appeared identical to those obtained earlier. See Figures 2 and 4. We conclude that in reaction 25 the carbonyl C-0 bond has been cleaved and that the "ensemble requirement" for the reductive cleavage of CO by [W(OR) 3 ]n compounds is four.

69

1 ~C( 18 0)

= 0.048

ppm

~~-T-r,-,~~~-r•-r,-~~-T-r,-.-~~-r~

184.550

Fig.

70

3.

184.500

PPH

13 C signals of a 60:40 mixture of 13 C16 0: 13 C18 0 dissolved in toluene-d 8 at 22•c. The spectrum was recorded at 125 MHz.

0

II

/c""

W=W

290.5

291.0

Fig.

4.

PPM

290.0

313.40 313.30 313.20

PPM

366.0

365.5

PPM

365.0

13 C NMR spectra of samples prepared using a ca 60:40 mixture of 13 CO and 13 C18 0 recorded in toluene-d 8 , 22"C at 125.76 (a) W2 (0-t-Bu) 6 (JL-CO), (b) the central resonances (no MHz. satellites due to 1J1a 3w_1 3c are shown) of; [W2 (JL-CO)(OCH 2 -tBu)6]2 and (c) W4 (C) (O) (OCH 2 -c-Bu) 12 . The spectrum shown in (b) can be simulated as the sum of two singlets arising from [ 18 xW 2 (JL- 13 C16 0)] , 36% and [ 18 xW2 (JL- 13 C18 0)] 2 , 16% and an AB quartet for 18 jW 4 ( 13 C16 0)( 13 C18 0) with 3J 13 c_ 13 c ~ 3.1 Hz and 1 ~C ( 18 0) ~ 0.030 ppm.

CONCLUDING REMARKS From the previous description of recent work in this laboratory it is evident that dinuclear compounds with M-M multiple bonds are capable of showing reactivity that complements that of mononuclear chemistry. Such is seen in the case of oxidative addition and reductive elimination reactions from dinuclear molybdenum and tungsten centers. In addition the M-M multiple bond provides for multi-electron redox reactions and in this regard leads to reactions that are not possible in mononuclear. The role of the inorganic template effect in multisite activation, coupled with multi-electron redox is well illustrated by the reductive cleavage of carbon monoxide to carbide and oxide in the reactions between ~7 4 (OCH 2R) 12 compounds and C=O. ACKNOWLEDGMENTS I thank the National Science Foundation and the Department of Energy, Office of Basic Sciences, Chemistry Division, for financial support of various aspects of this work. Also I am grateful to my many talented co-workers whose names appear in the references. REFERENCES

1.

This is merely an extension of mononuclear chemistry where adjacent coordination sites must meet this role.

71

2. 3.

4. 5.

6. 7. 8. 9. 10. 11.

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

23. 24. 25. 26. 27.

72

It is no accident that in nature the Fe 4 S4 cluster in ferredoxin serves as an electron transfer reservoir. In the limiting case this can be taken to be a dinuclear earlylate transition metal containing complex and a considerable expenditure of research effort has been recently made in this area for the activation of molecules such as C=O: see C. P. Casey, R. E. Palermo, and A. L. Rheingold, J. Am. Chem. Soc. 108:549 (1986); C. P. Casey, R. E. Palermo, R. F. Jordan, and A. L. Rheingold, J. Am. Chem. Soc. 107:4597 (1985); C. P. Casey, R. F. Jordan, and A. L. Rheingold, J. Am. Chem. Soc. 3:504 (1984). J. T. Yates and M. R. Albert, in: "The Surface Scientists Guide to Organometallic Chemistry," ACS Publishers, Washington, D.C. (1987). Sachtler coined the term "ensemble requirement" to explain results obtained from alloy catalysts having an inactive component. The ensemble requirement of a catalytic reaction is the number of contiguous metal atoms needed for that specific reaction to occur: W. M. H. Sachtler, p. 434, Chemtech, July (1983). The term ensemble effect has also extended to certain carbonyl cluster reactions: D. F. Shriver and M. J. Sailor, Ace. Chem. Res. 21:374 (1988). Y. Doi, K. Koshizuka, and T. Keii, Inorg. Chem. 21:2732 (1982). G. SUss-Fink, J. Organomet. Chem. 193:C20 (1980). G. SUss-Fink and J. Reiner, J. Mol. Catal. 16:231 (1982). J. A. Smiejla, J. E. Gozum, and W. L. Gladfelter, Organometallics 5:2154 (1986). G. SAs-Fink, Angew. Chem. Intl. Ed. Engl. 21:73 (1982). R. L. Pruett and W. W. Walker, Union Carbide Corp. U.S. Patents 3 833 634 (1974), 3 957 857 (1976); J. L. Vidal, Z. C. Mester, and W. Walker, Union Carbide Corp. U.S. Patent 4 115 428 (1978). J. L. Vidal and R. C. Schoening, Inorg. Chem. 21:438 (1982). G. C. Demitras and E. L. Muetterties, J. Am. Chem. Soc. 99:2796 (1977). H.-K. Wang, H. W. Choi, and E. L. Muetterties, Inorg. Chem. 20:2661 (1981). J. C. Bricker, C. C. Nagel, and S. G. Shore, J. Am. Chem. Soc. 104:1444 (1982). J. C. Bricker, C. C. Nagel, A. A. Bhattacharayya, and S. G. Shore, J. Am. Chem. Soc. 107:377 (1985). M. W. Payne, D. L. Leussing, and S. G. Shore, J. Am. Chem. Soc. 109:617 (1987). P. C. Ford, Ace. Chem. Res. 14:31 (1981). R. D. Adams, H. Kim, and S. Wang, J. Am. Chem. Soc. 107:6107 (1985). R. D. Adams, J. E. Babin, and M. Tasi, Inorg. Chem. 25:514 (1986). See also Chapters by R. D. Adams and W. L. Gladfelter in this volume. For example see syngas and HDS catalysts derived from Cp~Mo 2 Fe 2 S 2 (C0) 8 and Cp~Mo 2 Co 2 S 3 : M. D. Curtis, J. E. Penner-Hahm, J. Schwank, 0. Baralt, D. J. McCabe, L. Thompson, and G. Waldo, Polyhedron 7:2411 (1988). For a novel low temperature route to MoW alloys see A. K. Cheetham, Nature 288:469 (1980). A. B. Brignole and F. A. Cotton, Inorg. Syn. 13:81 (1972). M. H. Chisholm, D. A. Haitko, and C. A. Murillo, Inorg. Syn. 21:51 (1982). M. H. Chisholm, D. A. Haitko, and J. C. Huffman, J. Am. Chem. Soc. 103:4046 (1981). M. J. Chetcuti, M. H. Chisholm, K. F0 lting, D. A. Haitko, J. C. Huffman, and J. Janos, J. Am. Chem. Soc. 105:1163 (1983).

28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

50. 51. 52. 53. 54. 55.

M.

H. Chisholm, M. J. Hampden-Smith, J. C. Huffman, and K. G. Moodley, J. Am. Chem. Soc. 110:4070 (1988). M. H. Chisholm, M. J. Hampden-Smith, K. A. Stahl, J. C. Huffman, J. D. Martin, and K. G. Moodley, Polyhedron 7:1991 (1988). M. H. Chisholm and M. W. Extine, J. Am. Chem. Soc. 99:782 (1977); idem, ibid, 99:792 (1977). M. J. Chetcuti, M. H. Chisholm, K. Folting, D. A. Haitko, and J. C. Huffman, J. Am. Chem. Soc. 104:2138 (1982). J. K. Kochi in "Organometallic Mechanisms and Catalysis," Academic Press, Chpt. 11 and references cited therein (1978). M. J. Chetcuti, M. H. Chisholm, K. Folting, J. C. Huffman, and J. Janos, J. Am. Chem. Soc. 104:4684 (1982). M. J. Chetcuti, M. H. Chisholm, H. T. Chiu, and J. C. Huffman, Polyhedron 4:1213 (1985); H. T. Chiu, Indiana University, Ph.D. Thesis (1986). M. H. Chisholm, H. T. Chiu, and J. C. Huffman, Polyhedron 3:475 (1984). M. H. Chisholm, D. L. Clark, J. C. Huffman, and W. G. Van Der Sluys, J. Am. Chem. Soc. 109:6817 (1987). M. H. Chisholm, and R. J. Tatz, Organometallics 5:1590 (1986). M. H. Chisholm, K. Folting, J. C. Huffman, and R. J. Tatz, J. Am. Chem. Soc. 106:1153 (1984). F. A. Cotton and R. A. Walton in "Multiple Bonds Between Metal Atoms," Wiley Publishers, Chpt. 3, Sect. 3 .1. 5, pp. 99-105 and references therein (1982). M. H. Chisholm, C. C. Kirkpatrick, and J. C. Huffman, Inorg. Chem. 20:871 (1981). M. Akiyama, M. H. Chisholm, F. A. Cotton, M. W. Extine, D. A. Haitko, D. Little, and P. E. Fanwick, Inorg. Chem. 18:2266 (1979). M. H. Chisholm, J. C. Huffman, C. C. Kirkpatrick, and J. Leonelli, J. Am. Chem. Soc. 103:6093 (1981). M. H. Chisholm, J. C. Huffman, and C. A. Smith, J. Am. Chem. Soc. 108:222 (1986). M. H. Chisholm, D. L. Clark, J. C. Huffman, and C. A. Smith, Organometallics 6:1280 (1987). K. J. Ahmed, M. H. Chisholm, K. Folting, and J. C. Huffman, J. Am. Chem. Soc. 108:989 (1986). M. H. Chisholm, D. L. Clark, K. Folting, J. C. Huffman, and M. J. Hampden-Smith, J. Am. Chem. Soc. 109:7750 (1987). M. H. Chisholm, D. L. Clark, and M. J. Hampden-Smith, J. Am. Chem. Soc. 110:574 (1989). M. Akiyama, M. H. Chisholm, F. A. Cotton, M. W. Extine, D. A. Haitko, D. Little, and J. Leonelli, J. Am. Chem. Soc. 103:779 (1981). (a) M - Mo: M. H. Chisholm, F. A. Cotton, M. W. Extine, and R. L. Kelly, J. Am. Chem. Soc. 100:3354 (1978). (b) M- W: M. H. Chisholm, F. A. Cotton, M. W. Extine, and R. L. Kelly, Inorg. Chem. 18:116 (1979). M. H. Chisholm, F. A. Cotton, M. W. Extine, and R. L. Kelly, J. Am. Chem. Soc. 101:7645 (1979). R. A. Walton, ACS Symp. Ser. 155:207 (1981) and references therein. R. R. Schrock, M. L. Listemann, and L. G. Sturgeoff, J. Am. Chem. Soc. 104:4291 (1982). R. R. Schrock and M. L. Listemann, Organometallics 4:74 (1985). M. H. Chisholm, D. M. Hoffman, and J. C. Huffman, J. Am. Chem. Soc. 106:6794 (1984). M. H. Chisholm, B. K. Conroy, B. W. Eichhorn, K. Folting, D. M. Hoffman, J. C. Huffman, and N. S. Marchant, Polyhedron 6:783 (1987).

73

56. 57. 58.

59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75.

74

M. A. Harmer, T. R. Halbert, W. -H. Pan, C. L. Coyle, S. A. Cohen, and E. I. Stiefel, Polyhedron, 5:341 (1986). T. P. Blatchford, M. H. Chisholm, K. Folting, and J. C. Huffman, J. Chern. Soc., Chern. Commun. 1295 (1984). J. E. McMurry, Ace. Chern. Res. 16:405 (1982). A similar reactivity is found in a reagent prepared by the reduction of WC1 6 with LiBun: K. B. Sharpless, M. A. Umbreit, M. T. Nieh, and T. C. Flood, J. Am. Chern. Soc. 94:6538 (1972). More recently a variety of active metals of the actinides have been employed as reagent for similar reductive couplings: B. E. Kahn and R. T. Riecke, Chern. Rev. 88:733 (1988). F. A. Cotton, R. A. Walton, D. DeMarco, and L. R. Falvello, J. Am. Chern. Soc. 104:7375 (1982). M. H. Chisholm and J. A. Klang, J. Am. Chern. Soc. llO:xxx (1989). R. R. Schrock, J. Am. Chern. Soc. 98:5399 (1976). T. C. Bryan and J. M. Meyer, J. Am. Chern. Soc. 109:7213 (1987). M. H. Chisholm, D. M. Hoffman, and J. C. Huffman, Inorg. Chern. 22:2903 (1983). M. H. Chisholm, J. C. Huffman, J. Leonelli, and I. P. Rothwell, J. Am. Chern. Soc. 104:7030 (1982). R. C. Benson, W. H. Flygare, M. Oda, and R. Breslow, J. Am. Chern. Soc. 95:2772 (1973). E. M. Kober and D. L. Lichtenberger, J. Am. Chem. Soc. 107:7199 (1985). M. H. Chisholm, D. M. Hoffman, and J. C. Huffman, Organometallics 4:986 (1985). See also F. A. Cotton and W. Schwotzer, J. Am. Chem. Soc. 105:4955 (1983). D. F. Shriver and C. P. Horwitz, Adv. Organometal. Chem. 23:219 (1984). M. H. Chisholm, Polyhedron 2:681 (1983). P. J. Blower, M. H. Chisholm, D. L. Clark, and B. W. Eichhorn, Organometallics 5:2125 (1986). J. S. Bradley, G. B. Ansell, and E. W. Hill, J. Am. Chem. Soc. 101:7417 (1979). M. H. Chisholm, K.· Felting, C. E. Hammond, J. C. Huffman, and M. J. Hampden-Smith, J. Am. Chem. Soc. 110:3314 (1988). M. H. Chisholm, C. E. Hammond, and M. J. Hampden-Smith, J. Am. Chem. Soc., submitted. E.g., as in Fe 4 (C0) 13 -: M. Manassero, M. Sansoni, and G. Longoni, J. Chem. Soc. Chem. Commun. 919 (1976). See Ref. 5 for reactivi.ty studies. For a review for 13 C18 0 isotope shifts see: P. E. Hensen, Ann. Rep. NHR Spectrosc. 15:105 (1983).

CLUSTERS AND THEIR IMPLICATIONS FOR CATALYIS

Richard D. Adams Department of Chemistry University of South Carolina Columbia, SC 29208

INTRODUCTION The ability of metal cluster complexes to produce facile multicenter transformations of small molecules has been the most exciting and important aspects of their chemistry. 1 Multicenter transformations can be divided into two categories: (1)• Multicenter Activation through multicenter coordination, as illustrated by bridging ligands, e.g., A~ and (2) Multicenter Reaction in which a ligand is coordinated to one or more metal atoms but undergoes a reaction at a vacant site on a neighboring metal atom, e.g., B. The potential to exploit these transformations for the development of new catalysts has been one of the principal reasons for the great interest in the synthesis and study of these compounds. 2-9 Multicenter coordination can produce significant changes in the structure, bonding and reactivity of a ligand. One of the most spectacular examples of ligand transformations induced by multicenter coordination is the intramolecular cleavage of the carbon-carbon multiple bond of a alkyne ligand to produce two alkylidyne ligand groupings, Eqs.(ll 10 and (2) 11 Another example H R

R

'

/

,(~\-/)

'c

u/

l

H

c/ 'u

DI

M---M

M

A

Metal-Metal Bonds and Clusters in Chemistry and CatalysiS Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

B

75

demonstrates the intramolecular cleavage of the C-O multiple bond of a dihapto triply-bridging benzoyl ligand to yield an alkylidyne ligand and an oxo ligand, Eq.(3). 12 This reaction could have implications for the metal surface-induced cleavage reactions involved in Fischer-Tropsch reactions. 13

R '

R

' Os--Os I \ "

' /\"t' / c--c

I

-l~~:}Y\-

llO"C

,,;' ...... 'Y . . . .

-CO

........

R

-~~~'l\' ,c,l /c' ,' W

H

Cp

'

hv

Os 6(C0) 11 + PhC- CPh

\

o--c

co

R

Cp

\

/

"

''?f~'fJ '-- c, / 'I __ Os-

.' . . I, '

-

-Os

_os~ v i s ,

I

c

( 2) Ph

I Ph

,...cu,tol

~~W~p\ /1\. ·

'os-1

-2

(I)

"" I ". /

'\ .,....

I\

I'

-Os---os

llO"C -2

co

(3)

The ability of mononuclear metal complexes to activate the alkenyl C-H bonds in olefins is very limited; 14 however, this process is facile in metal cluster complexes. The activation of alkenyl CH bonds by os 3 (C0) 12 has been studied in great detai1. 15 The reaction sequence begins by a ligand substitution of alkene for CO to form the complex os 3 (co) 11 !R(H)C=C(H)R]. Loss of co, presumably from one of the neighboring metal atoms, produces a vacant site (e.g. B) that can react with one of the CH bonds of the coordinated alkene ligand to yield the complex os 3 (C0) 10 [pC(R)=C(H)R](p-H), 1 containing a a-n coordinated bridging alkenyl ligand and a bridging hydride ligand, Figure 1. 15 b 76

,1/ Os

R 1

,\ /I"-. l_.c~H Os Os~ \\ R /1 /I ~c~ H

us•c j-co

\/ l ,\/ k~ 1\..-a Os os- ' /I !\" H -os-c

R=H; -CO

1

\

R:alkyl

\co

o._

...._I_.

,\os:----..... //\"'-~ / ___ o._ /'\,H,/" a"' 2

Figure 1.

c-c

'a

3

Multicenter activation of alkenes by triosmium clusters.

A co elimination from the third metal atom can produce a second CH activation in either of two ways. An a-CH activation, R = H, has been shown to yield the complex os 3 (C0) 9 [p 3-C=CH 2 J(p-H) 2 , 2 containing a triply bridging vinylidene ligand. Alternatively, a ~-CH activation R = H leads to the formation of the triply bridged alkyne complexes os 3 (co) 9 (p 3-RC=CR)(p-H) 2 , 3. One of the most revealing examples of a multicenter activation of the second type involves the cleavage of the phosphorus-carbon bond to one of the phenyl groups of the bridging diphenylphosphinidene ligand in the complex Ru 3 (co) 9 (pPPh2)(p-H), 4. 16 Complex 4 is ligand deficient by the amount of one CO ligand and literally contains a vacant coordination site on one ruthenium atom that is "partially" filled by a sideways interaction with one of the phosphorus-carbon a-bonds. When heated to 80°C, the phosphorus-carbon bond was cleaved and phenyl group was combined with the hydride ligand to yield c 6 a 6 , Eq.(4). The tetraruthenium cluster product Ru 4 (C0) 13 (p 3-PPh), 5 was formed by the addition of an adventitious mononuclear ruthenium carbonyl fragment to the triruthenium remnant from 4. 77

~pj)

0

f /~I,

so•c

Ru-

-c,n,

:R~<. iu~ I

/ \

4

I

)t::_ 'I

P,

I/ /Ru:..'\ / Ru~ \ -Ru' /I'-

(4)

5

Reactions of Diaminomethanes With Osmium Cluster Complexes Our research has been focused on the activation and transformation of tertiary amines and amine-containing hydrocarbyl ligands by osmium carbonyl cluster complexes. 17 We have used bis-dialkylaminomethanes, (e.g. CH 2 (NMe 2 ) 2 ), to act as reagents for introducing iminium, H2 CNR 2+, ligands to cluster complexes. One of these complexes is os 3 (COJ 10 (p-H 2 CNMe 2 )(p-H), 6 obtained from the reaction of os 3 (COJ 10 (p-HJ 2 with CH 2 (NMe 2 J 2 . 18 An ORTEP diagram of the molecular structure of 6 is shown in Figure 2. The molecule consists of a triangular cluster of three osmi1 atoms with an N,N dimethyliminium ligand bridging an edge of the

Figure 2.

78

An ORTEP diagram of Os 3 (C0) 10 [p-H 2CNMe 2 ](p-H), 6. Reprinted with permission from Organometallics 1988, 7, 963. Copyright (1988) American Chemical Society.

cluster. The carbon atom is bonded to one metal atom and the nitrogen atom is bonded to the other. Interestingly, the C-N vector is not parallel to the Os-Os bond, but it forms an angle that brings the CH 2 group closer to the axial CO ligand on Os(l). While the origin of this twist may be as simple as the differences in steric interactions between the NMe 2 and CH 2 group with the axial CO ligand on Os(l), the shift of the CH 2 group toward Os(l) is believed to favor the transformation that occurs when atom Os(l) is decarbonylated. When heated to 97°C, compound 6 is transformed to two products. 18 One of these is os 3 (C0) 9 [~ 3 -C(H)NMe 2 J(~-H), 7 formed by the loss of CO and the activation of one of the CH bonds on the CH 2 group. The second product is os 3 (C0) 10 (p-C=NMe 2 )(p-H), 8 formed by the activation of both CH bonds on the CH 2 group and the elimination of one formula equivalent of H2 . ORTEP diagrams of the molecular structures of 7 and 8 are shown in Figures 3 and 4, respectively. In compound 8, the iminium ligand was transformed into a triply bridging dimethylaminocarbene ligand. The carbon atom bridges two of the osmium atoms while the nitrogen atom is coordinated to the third osmium atom. The formation of 7 is believed to occur by the loss of a CO ligand from the CO-rich Os(C0) 4 group of the cluster and is followed by the activation of one of the CH bonds of the CH 2 group at the resultant vacant site, Eq. ( 5).

' v.····"··,~/ HH

MeMe

'le--N\/

Me H

Me

\/

'_.-N

l~n~// ' , _ /o'~s/"- --"1\,. \----~05-

/,, Os

6

-os-

(5)

H~

/i' 7

Compound 7 cannot be transformed to 8 at 97°C even in the presence of a CO atmosphere. Accordingly, it is believed that 8 is formed by a competing transformation that does not involve the loss of a CO ligand. Compound 8 could be formed by a cleavage of

79

Figure 3.

An ORTEP diagram of os 3 (C0) 9 [p 3-HCNMe 2 J(p-H) 2 , 7. Reprinted from Organometallics 1988, 7, 963. Copyright (1988) American Chemical Society.

Figure 4.

An ORTEP diagram of os 3 (C0) 10 [p-CNMe 2 ](p-H) 2 , 8. Reprinted with permission from Organometallic 1988, 7, 963. Copyright (1988) American Chemical Society.

80

the Os-N bond in 7 to yield a vacant site to permit cleavage of one of the CH bonds of the CH 2 group and formation of a C-bridged carbene ligand intermediate. This intermediate could be induced to eliminate H2 by coordination of the nitrogen atom of the amino group, see Scheme I. The mechanism of the activation of the second CH bond of the CH 2 group is less clear, but could occur at a vacant site on the third metal atom formed by cleavage of the Os-N bond, and be followed by a shift of the dimethylaminocarbyne ligand from a triply-bridging to an edge-bridging position. These latter transformations would require appropriate repositioning of some of the carbonyl ligands.

6

8

Scheme I The reaction of os 3 (co) 10 (p-SPh)(p-H) with cH 2 (NMe 2 ) 2 yielded the complex os 3 (C0) 9 [C(H)NMe 2 J(p-SPh)(p-H), 9 that contains a terminally coordinated secondary dimethylaminocarbene ligand, see Figure s. 19 When solutions of 9 are irradiated, it is decarbonylated presumably at the CO-rich Os(C0) 4 group and two products Os 3 (C0) 8 (p-CNMe 2 )(p-SPh)(p-H)§' 10 and Os 3 (C0) 8 [p-C(H)NMe 2 ](p 3sc6H4)(p-H)2, 11 are formed. 1 Compound 10, which contains a bridging dimethylaminocarbyne ligand, was apparently formed by an 81

Figure 5.

An ORTEP diagram of os 3 (CO)g[C(H)NMe 2 ](p-SPh)(p-H), 9. Reprinted with permission form J. Am. Chern. Soc. 1987, 109, 1414. Copyright (1987) American Chemical Society.

a-activation of the carbene CH bond at the vacant site on the neighboring osmium atom, see route a, Scheme II. Compound 11 contains a triply bridging sc 6 H4 ligand formed by an ortho-CH the phenyl ring of the benzenethiolato decarbonylated osmium atom route b. The carbene ligand in 9 was not transformed in the formation of activation on ligand at the

11.

Compound 11 can be transformed to 10 thermally at 69°C, but it is not believed to be an intermediate in the formation of 10 by photolysis. This is supported by an additional study that showed that Os 3 (C0) 10 !C(H)NMe 2 J(p-SC 6 F 5 )(p-H), 12 could be converted to the analog of 10, os 3 (COJ 9 (p-CNMe 2 J(p-SC 6 F 5 )(p-HJ 2 , The replacement of fluorine atoms for the hydrogen atoms

13. 19

on the phenyl group of 9 would prevent the formation of a fluorinated analog of 11 (none was observed) and also 13, if ring-metallation is a prerequisite to the formation of 13. When heated to 125°C, compound 10 is transformed to an isomer os 3 (co) 8 [c(Ph)NMe 2 J(p 3 -s)(p-H) 2 , 14 that contains a terminally-coordinated phenyl(dimethylamino) carbene ligand, see Figure 6. Compound 14 was formed by a shift of the phenyl group from the sulfur atom to the carbyne carbon atom. A crossover experiment confirmed that the shift occurred by an intramolecular 82

9

Scheme II

Reprinted with permission from J. Am. Chern. Soc. 1987, 109, 1414. Copyright (1987) American Chemical Society.

Figure 6.

An ORTEP diagram of Os 3 (C0) 9 [C(Ph)NMe 2 J(p-S)(pH)2' 14. Reprinted with permission from J. Am. Chern. Soc., 1987, 109, 1414. Copyright (1987) American Chemical Society.

83

A multicenter activation and shift process was proposed, 1 ~ see Scheme III. The cleavage of the carbon-sulfur bond is believe< to occur by an oxidative addition to the third osmium atom, and process.

may resemble the phosphorus-carbon bond cleavage process observe' by Carty for 4, Eq.(4). 16 Since 10 does not contain any vacant coordination sites, and since the formation of 14 does not require a ligand elimination, it is proposed that the s-c oxidative addition is promoted by the cleavage of one metalmetal bond. The unobserved intermediate c would contain a tripl: bridging sulfide ligand and a sigma-bonded phenyl group, see Scheme III. The shift of the phenyl group to the carbon atom of the carbyne ligand could be driven by a reformation of the cleaved metal-metal bond. Activation of Tertiary Amines Shapley et al. have reported that the reaction os 3 (C0) 10 (NCMe) 2 with NEt 3 yields the product Os 3 (C0) 10 [pC(H)C(H)NEt2J(p-H), 15. 2 Compound 15 contains a C(H)C(H)NEt 2 ligand formed by the cleavage of three hydrogen atoms from one o: the ethyl groups of the NEt 3 molecule. This ligand is coordinated across an edge of the cluster through the ~-carbon

°

atom alone. In an effort to induce further transformation of th1 C(H)C(H)NEt 2 ligand, 15 was irradiated in the hope of decarbonylating the Os(C0) 4 group of the complex. Decarbonylation was achieved and activation of the final CH bond on the ~-carbon atom occurred to yield the complex os 3 (C0) 10 [p 3 CC(H)NEt2](p-H)2, 16 that contained a triply-bridging

()

"'/VI

:s-Os-...........___ /

12s•c

/H?,oi

H\v !' Os--C /\ '\-R I

R

10

c Scheme III

84

14

C • H Activation in NEt3 '

\

HI

'1/,\/ os........_ /H

,..., I

NEt 2

/ II Os-e-e

-.. . . . . II

'u

Os

/I'

Et 2N

' e-e\// , H

I

',.....os~ _.._.. /,.....u'

I" __

{)s-

os

u-- /\' 16

17

Scheme IV

diethylaminovinylidene, CC(H)NEt 2 , ligand, see Scheme Iv. 21 When heated, 16 was transformed to an isomer os 3 (C0) 9 [p 3-uc 2NEt 2 J(pH)2, 17 by a 1,2 shift of the hydrogen atom from the a-carbon atom of the diethylaminovinylidene ligand to the a-carbon atom. This shift may have involved metal activation at some point, but details concerning the mechanism of this rearrangement have not yet been established. A drawing of the molecular structure of 17 is shown in Figure 7. The molecule contains a HC 2NEt 2 ligand that could be described as an ynamine (or aminoacetylene). It is a triply bridging ligand, but differs from all known examples of acetylene cluster complexes in that the amine-substituted carbon atom is coordinated to only one metal atom. 22 Other interesting structural features are the planar nitrogen atom, the short C-N bond distance of 1.33(2)A and the spectroscopic (NMR) inequivalence of the ethyl groups. All of these features are similar to those of aminocarbene ligands, and accordingly, we have chosen to describe this ligand as dimetalliomethyl(diethylamino)carbene ligand. In further support of the carbene

85

Figure 7.

An ORTEP diagram of Os 3 (C0) 9 [p 3-HC 2NEt 2 J(pH)2, 17. Reprinted with permission of Organometallics 1988, 7, 2241. Copyright (1988) American Chemical Society.

formulation, we have found 17 will react with secondary amines (e.g., Pr 2NH) in a carbene-like exchange reaction to yield the new ynamine ligand complex os 3 (C0) 9 {p 3-Hc 2NPr 2 ){p-H) 2 , 18, yield=74%. Laine et al. have reported that os 3 (co) 12 will serve as a catalyst precursor for the tertiary amine metathesis reaction, Eq.(6). 23 Their studies showed thatCH bond activation (6)

processes occur in conjunction with the C-N bond cleavage proces that leads to the exchange of the R groups. Recognizing that that 17 was formed from NEt 3 by CH activation processes, and tha the NR 2 group could be cleaved from the "ynamine" ligand, 17 was tested for its potential to perform tertiary amine metathesis. It was found that 17 will react NPr 3n at 97°C in a stoichiometri reaction to yield 18, 37% and NEt 2 Prn, and at 145°C, solutions o 17 will produce tertiary amine metathesis catalytically at rates comparable to those observed by Laine using os 3 (C0) 12 . Studies to establish the mechanism of this remarkable reaction are currently in progress.

86

In a related study, we found that the sulfur-containing osmium cluster os 3 (co) 10 (p 3-s) reacts with NMe 3 at 125°C to yield the new cluster complex os 3 (co) 8 [c(H)NMe 2 J(p 3-s)(p-H) 2 , 19 that contains a terminally coordinated secondary dimethylaminocarbene ligand. 24 It is structurally analogous to 14. The carbene ligand was apparently formed by the activation of two CH bonds on one of the methyl groups of an NMe 3 molecule. The hydrogen atoms were transferred to the metal atoms to become bridging hydride ligands. It was found that 19 will react both with secondary amines and with tertiary amines to give exchange of the amino group of the carbene ligand and form the new complexes n Os 3 (C0) 8 [C(H)NR 2 J(p 3-S)(p-H) 2 , R=Et, Pr . The reactions with tertiary amines also yield NMe 2 R formed by transfer of one alkyl group to the NMe 2 group of the carbene ligand. At 145°C, solutions of 19 will produce tertiary amine metathesis catalytically, but this catalysis is complicated by the presence of the anion [os 3 (co) 9 (p 3 -S)(p-H)]-, 20, a decomposition product derived from 19 by loss of the carbene ligand, loss of a proton and the addition of one CO ligand. Our independent investigations of 20, prepared by the reaction of tertiary amine with os 3 (co) 9 (p 3-s( )p-H) 2 have shown that it is also an active catalyst for tertiary amine metathesis. Studies of catalysis using 19 and 20 conducted in co 3oo solvent have shown deuterium incorporation into the alkyl groups of the mixed amine products and imply that metal-induced CH activation processes accompany the alkyl exchange reaction. Further studies are in progress.

Acknowledgement These studies were supported by the u.s. Department of Energy under Grant No. DEFG84ER13296.

References 1) 2) 3) 4) 5)

Adams, R. D.; Horvath, I. T. Prog. Inorg. Chern. 1985, 33, 127. Muetterties, E. L.; Krause, M. J. Angew. Chern. Int. Ed. Engl. 1983, 22, 135. Muetterties, E. L. Catal. Rev. Sci. Engl. 1981, 23, 69. Muetterties, E. L. Pure Appl. Chern. 1978, 50, 941. Laine, R. M. J. Mol. Catal. 1982, 14, 137.

87

6) 7) 8)

Ugo, R.; Psara, P. J. Mol. Catal. 1983, 20, 53. Ford, P. c. Accts. Chern. Res. 1981, 14, 31. Mark6, L.; Vizi-Orosz, A. in "Metal Clusters in Catalysis", Gates, B. C.; Guczi, L.; KnOzinger, eds. Elsevier, Amsterdam, 1987, Ch.S. 9) SUss-Fink, G. Polyhedron 1988, 7, 2341. 10) Park, J. T.; Shapley, J. R.; Churchill, M. R.; Bueno, c. J. Am. Chem. Soc. 1983, 105, 6182. 11) Fernandez, J. M.; Johnson, B. F. G.; Lewis, J.; Raithby, P. R. Acta Cryst. 1978, B34, 3086. 12) Shapley, J. R.; Strickland, D. S.; St. George, G. M.; Ziller, J. W.; Beanan, L. R. J. Am. Chern. Soc. 1984, 106, 1144. 13) a) Ponec, v. Cat. Rev. Sci. Eng. 1978, 18, 151. b) Herrmann, w. A. Angew Chern. Int. ·Engl. 1982, 21, 117. c) Henrici-Olive, G.; olive, s. Angew. Chern. Int. Engl. 1976, 15, 136. d) Muetterties, E. L. Chern. Rev. 1979, 79, 479. Bergman, R. G-i Seidler, P. F •; Wenzel, T. T. J. Am. 14) a) Chern. Soc. 1985, 107, 4358. b) Stoutland, P. 0.; Bergman, R. G. J. Am. Chern. Soc. 1985, 107, 4581. c) Silvestre, J.; Calhorda, M. J.; Hoffmann, R.; Stoutland, P. 0.; Bergman, R. G. Organometallics 1986, 5, 1841. d) Boncella, J. M.; Green, M. L. H. J. Organomet. Chern. 1987, 325, 217. 15) a) Johnson, B. F. G.; Lewis, J. Inorg. Chim. Acta 1979, 109, 271. b) Deeming, A. J. in "Transition Metal Clusters", Johnson, B. F. G., ed., Wiley & Sons, Chichester, 1980, Ch. 6. 16) MacLaughlin, S. A.; Carty, A. J.; Taylor, N. J. Can. J. Chern. 1982, 60, 87. 17) Adams, R. D.; Babin, J. E.; Kim, H. s. Polyhedron 1988, 7, 967. 18) Adams, R. D.; Babin, J. E. Organometallics 1988, 7, 1963. 19) Adams, R. D.; Babin, J. E.; Kim, H. s. J. Am. Chern. Soc. 1987, 109, 1414. 20) Shapley, J. R.; Tachikawa, M.; Churchill, M. R., Lashewycz, R. A. J. Organomet. Chern. 1978, 162, C39. 21) Adams, R. D.; Tanner, J. T. Organometallics 1988, 7, 2241.

88

22)

a)

Raithby, P. R.; Rosales, M. J. Adv. Inorg. Radiochem. 1985, 29, 169.

b)

Sappa, E.; Tiripicchio, A.; Braunstein, P. Chern. Rev. 1983, 83, 203.

23)

a) b)

Shvo, Y.; Laine, R. D. J. Chern. Soc., Chern. Commun. 1980, 753. Wilson, R. B. Jr.; Laine, R. M. J. Am. Chern. Soc. 1985, 107, 361.

24)

Adams, R. D.; Kim, H. S.; Wang,

s.

J. Am. Chern. Soc. 1985,

107, 6107.

89

METAL CLUSTERS IN THE SOLID STATE

Robert E. McCarley Department of Chemistry and Ames Laboratory, U.S.D.O.E. Iowa State University Ames, Iowa 50011

INTRODUCTION In solid state compounds metal clusters and metal-metal bonded arrays occur in both binary and ternary or quaternary phases.

Among transition

element compounds, cluster and metal-metal bonded species occur in binary halides (Nb, Ta, Mo, W, Re),l-B oxides (Nb, Mo, W, Tc, Re),l,7, 9 chalcogenides (Zr, Nb, Ta, Mo, Re) 7 • 10 - 13 and pnictides (Zr, Hf, Nb, Ta, Mo, W, Re).3,7,10

Ternary and quaternary phases containing a broad

representation of metal ions An+ (or Bn+), as in~(~~) or ~Bn(~~), or two different anions ~~Yz, as in oxide halides and sulfide halides, 14 are even more abundant.

Most recently the J. Corbett and A. Simon groups

have demonstrated a prolific chemistry of octahedral cluster species having either a nonmetallic or metallic element strongly bonded in the center of the unit.

The "interstitial" element provides electrons in

stabilized cluster orbitals resulting from strong interaction of the nonmetal p-orbitals (Be, B, C, N, Al, Si) 15 or metal d-orbitals (Fe, Ru) 16 with cluster orbitals of the same symmetry. Electron poor elements 7 such as Sc, Y, Ln (lanthanoids) and Zr can thereby be induced to form stable clusters, e.g., cs 3zr 6c1 16 c, Y6 I 10Ru, Gd10 c1 18 cc 2 ) 2 and Sc(Sc 6 I 12 (B, C)). As discussed previously1 the structures and properties of the solid state cluster compounds are controlled by a number of factors such as relative van der Waals and bond radii of X and M, atom ratios Nx/NM, and electron/metal ratio. favored in a

It is readily seen that M-M bonding should be

compound~

when the radius of X is small, the X/M ratio n

Metal-Metal Bonds and Clusters in Chemzstry and Catalysts Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

91

is small, and the number of electrons in valence orbitals not involved in M-X bonding is roughly equal to the number of orbitals.

In general,

discrete cluster units are promoted by larger, and condensed cluster units are encouraged by smaller X-anions, e.g. compare

~Mo 6 s 8

and

which

~Mo 4 o 6

have discrete octahedral clusters and infinite chains of trans edge-fused octahedral cluster units, respectively. In the chain structure it is essential that the van der Waals radius of oxygen be comparable to the metallic single bond radius 17 of Mo since the M-M and 0-0 spacings must be the same along the chain.

If the

required anion spacing is too large to permit good M-M bond formation along the chain, then break-down of the chain structure would be expected.

METAL OXIDE SYSTEMS Some General Relations In this paper we will focus on the metal oxide structures and emphasize the important recent developments which indicate that clusters of almost any size can be realized.

By proper control of the variables

mentioned above it appears that the clever investigator could choose conditions to prepare clusters of predetermined size and geometry in oxide systems. The ternary oxides of niobium 7 and molybdenum1 have been especially fruitful in providing examples of this behavior.

In contrast,

essentially no examples of clusters in oxide systems are known for the congeners tantalum and tungsten.

This is an important and striking

observation because it indicates that even though the niobium and molybdenum oxide compounds are often thermally stable at very high temperatures they nonetheless may be close to the thermodynamic line separating stability and instability.

For the reduced oxide systems

instability with respect to disproportionation would be the principal problem.

Evidently the Nb and Mo compounds are on the stable side, and

the Ta and W compounds on the unstable side with respect to disproportionation.

The Gibbs free energy changes for these

disproportionation reactions may differ by only a few calories, with for Ta and W, and ~G">O for Nb and Mo.

~G"
Cheetham and his coworkerslB

discuss this problem with reference to their lack of success in the preparation of Zn 2w3o8 , the tungsten analogue of the well-known zn 2Mo 3o8 structure.

92

Table 1.

Structure Types with Infinite Chains of Condensed Octahedral Cluster Units Known Examples X -

1, M- Na, K, In (I)

13.0

X

0.9, M- Sn(II)

13.8

X -

0.62, M

Ba(II)

13.25

X -

0.75, M

Pb(II)

13.5

x - 1.5, M - Mn(II) MM'Mo ob 4 7

M - Zn, Fe

14.5 14.5-14.75

M' - Sc, Ti, Fe, Zn, Al M - Li, Zn

14.5-15.0

M - Y, Lnc

14.0

x- 5.45, M- Ca a Metal cluster electrons per (Mo 2Mo 412 ) cluster unit. b These compounds have more complicated formulas than represented here, e.g. sc 0 _75 zn1 . 25 Mo4o 7 . c Ln- members of the lanthanide series. d Estimated from bond distance-bond order sums.

Compounds With Infinite Chains of Condensed Cluster Units The best known and most abundant examples of M-M bonded oxide systems are the several structure types involving infinite chains of trans edgeshared octahedral metal cluster units.

These are listed in Table I, and

views of the construction of the infinite chain and oxide framework are shown in Figure 1.

Basically in each structure type the molybdenum oxide

chains are interlinked in a different way through Mo-0-Mo bonding to create the particular oxide framework.

Within each framework the ternary

metal ions are located in 0-coordinated sites of geometry best suited for the particular metal ion.

For example, the metal ions in the

~Mo 4 o 6

structure are located in sites in tunnels formed by four surrounding

93

a

b

Fig. 1.

The structur e of InMo 4 o6 . (a) A projectio n of the structur e as viewed along the tetragon al c-axis; (b) the structur e of an individu al chain showing the trans edge-sha red condense d octahedr al cluster units. Mo-Mo bonds are shown by heavy lines.

94

~[Mo 4 o 6 ] chains, but the geometry of the coordination varies from cubic 8coordinate (M-Na, K, Ba), 19 • 20 to highly distorted cubic 8-coordinate (Pb), 21 to square pyramidal 4 coordinate (In), 22 to square planar (Sn). 23

Only for M-Na, K and In are all of these ternary metal sites filled to provide the stoichiometry MxMo 4o6 with x-1. When M-Ba, Pb and Sn, all divalent metal ions, vacancies occur in the ternary metal sites with resulting compositions x-0.62, 0.75 and 0.9, respectively. In the case of the Ba 20 and Pb 21 compounds the vacancies are ordered in super cells having 8 times and 4 times the dimension of NaMo 4o6 in the unique direction (c-axis).

Both size of the ternary metal ions and total

electrons transferred to the Mo 4o6 chains appear to be involved in this nonstoichiometry.

Cation-cation repulsions between the large divalent

ions in the fully loaded structure and instabilities resulting from transfer of 2 electronsfMo 4o6 unit favor lower x-values in the observed order. Instabilities associated with transfer of too many electrons, viz. more than 13e/Mo 4 repeat unit, show up as marked distortions of the Mo-Mo bonding within the infinite chains. 9 Thus, in the 13-electron cases NaMo 4o6 and InMo 4o6 the chains are highly symmetric as shown in Figure 1, with a repeat· distance of only ca. 2.86A in the chain direction.

In all

other cases with greater metal cluster electron count (MCE) the bonding becomes distorted by shortening and lengthening certain bonds.

Most often

pairwise distortions occur such that Mo-Mo bonds between apical atoms (3.0 ~

d(Mo-Mo) ~ 2.6A) and between atoms on the shared edges of the octahedral

cluster units (d(Mo-Mo) < 2.75A) alternate with stretched or nonbonded distances in adjacent positions along the chain.

The alternating short

and long distances are accompanied by a doubling of the repeat distance along the chain from about 2.86A to 5.72A.

These effects are illustrated in Fig. 2 for the chains found in Gd4Mo 4o11 . 24 Although extensive studies of the electrical resistivity of these

compounds has not been made, it appears that the compounds with doubled repeat distance along the chains will be low band-gap semiconductors. example resistivity measurements on InMo 4o621 and Sn0 . 9Mo 4 o6 23 show typical metallic behavior, while ZnMo 8o10 25 shows the resistivity-

For

temperature dependence of a semiconductor with Egap - O.OSeV over the range T < 200K. of a metal.

Above 200K the latter compound shows conductivity typical

Thus chains with certain MCE counts may undergo a Peierls-

like distortion at lower temperatures as should be expected for a material with essentially one-dimensional character. 95

Fig. 2. Two views of the condensed octahedral cluster chains as found in Gd4Mo 4 o11 . The top view shows the pairwise mating of the Mo atoms along the chain direction and the bottom view shows the alternating long and short bonds on the shared edges.

Only Mo

atoms are shown in these views.

Oligomeric Clusters As Segments of The Condensed Octahedral Cluster Chains The discovery of the compound In 11Mo 40 o62 and the elucidation of its fascinating structure 26 brought with it the realization that it is possible to dissect the chains of the type shown in Fig. 1 and thereby obtain discrete oligomeric cluster units.

In the wonderful example of

this compound, not one, but two oligomeric clusters were displayed in the same structure, and as a bonus two different oligomeric I~+ chains were also found!

Both kinds of oligomeric units are represented in the 7->< Mo 22 oaformulation (Ins7+) (In 6B+>< Mo 18 o28 34 ). The linear cation chains can

be understood as constructed from n-2 internal In+ atoms and two terminal In 2+ atoms to give the oligomeric general.formula In~n+ 2 )+. Presumably the I~n+ 2 )+ cations are selected to stabilize this structure because they almost perfectly fill the spaces between the anionic cluster units and provide just the needed MCE count by electron transfer to the anions.

A

representation of this structure illustrating only the Mo 18 , Mo 22 , Ins and In 6 units is shown in Fig. 3.

96

Fig. 3. The structure of In11Mo 40 o62 with oxygen atoms omitted.

Molybdenum atoms are shown as filled

circles and indium atoms as open circles.

Note the

presence of the In 6 chains between the Mo 22 clusters and In5 chains between the Mo 18 cluster units.

A general formula for the oligomeric series of Mo 4n+ 2o6n+4 may be derived from this structure. For the cluster anions in In11Mo 40 o62 we have members of the oligomeric series with n-4 or 5.

A discussion of the

possible electronic requirements (MCE) for these particular cluster units has been given by both the Simon group 7 and by Wheeler and Hoffman. 27

The

existence of In 11Mo 40 o62 makes it possible to predict that other members of the oligomeric series should be found. However, because of some uncertainty about the exact electronic requirements and compatibility of cations for specific members of the series, some difficulty could be expected in predicting specific compounds accurately. Remarkably, within the last year two papers have appeared which report the synthesis and structure of two additional members of this oligomeric series. The first of these, 28 BaMo 6o10 , represents not only a member with n-1, but also the first ternary oxide having individual octahedral clusters, of the Mo 6o12 type, as the essential structural unit. Based on other known clusters of the M6x12 type we might expect that the MCE of the cluster unit in BaMo 6o10 would be 14 to 16. 7 The observed MCE of 18 can be rationalized on the basis that additional electrons are needed to form the intercluster Mo-Mo bonds of low bond order, as shown in Fig. 4. By using the bond distances given by Wang, Wang and Lii, 28 bond order sums 9 can be used to estimate that ca. 14.5 electrGns participate in intracluster Mo-Mo bonding, and £A. 3.5 electrons compose the intercluster Mo-Mo bonds.

As illustrated in Fig. 4, the intercluster bonds knit the

97

Fig. 4.

Representation of Mo-Mo bonding in and between octahedral cluster units of BaMo 6o10 . Only Mo atoms are illustrated. Double lines represent intracluster bonds and single lines represent intercluster bonds.

Closed circles at one level

and open circles at level above or below.

Mo 6o10 units together to form infinite chains which run parallel to the orthorhombic b-axis.

Because of the intercluster Mo-Mo bonding we might

expect that this compound will exhibit low resistivity in the b direction, perhaps even metallic character. The second compound reported in 1988 and belonging to this oligomeric series is La 2Mo 10o16 . 29 In this case the edge-shared bioctahedral cluster unit represents the member with n-2. The MCE count of 34 for this cluster unit seems a little high (3.4e/Mo) when compared to all other members of this series (3.0 to 3.2e/Mo).

Molecular orbital calculations for this

cluster unit would be useful in developing a better understanding of this problem. As in the case of the related oligomeric molybdenum sulfide or selenide cluster anions Mo 6 nY~~!~)-(Y-S or Se), where all members with 1 ~

n ~ 5 are known, 13 we might expect the presently missing member of the Mo 4n+ 2o6n+4 series with n-3 to be found in future work.

It is also

conceivable that compounds having members with n > 5 may also be prepared if cations suitable for filling the spaces between the oligomeric anions, and for providing the necessary charge balance, can be devised.

98

Fig. 5.

3- cluster anion found in Structure of the Mo 8o14 NdMo 8o14 . Heavy filled lines represent Mo-Mo

bonds and unfilled lines represent Mo-O bonds . The Mo 8 cluster core is shown with 24 ligating oxygen atoms derived from sharing of oxygen atoms between cluster units in the crystal lattice.

A New Oligomeric Cluster Unit In this laboratory we have recently prepared the new compound NdMo 8o14 30 which contains a novel octanuclear cluster unit Mo 8o24 . This interesting cluster unit is illustrated in Fig . 5, which shows all oxygen atoms ligating the Mo 8 cluster core. Many of these oxygen atoms are shared between cluster units to interconnect the units and form a framework in which the Nd 3+ ions are embedded in sites with coordination number 12.

The connectivity of 0 atoms in the units is indicated in the

formula Nd((Mo 8o6o6;2)06/206/3) ·

99

Although the Mo 8 cluster unit may be viewed in a number of ways, e.g. octahedron plus two face capping atoms or two rhomboidal cluster units placed one on top of the other, the most fruitful view, we believe, is to consider it as two Mo 4 butterfly cluster units coupled together front to back. The butterfly units can be thought of as the repeat units in the infinite chains like those shown in Fig. 1 for the trans edge-shared mode of condensation of octahedral clusters.

If two such butterfly repeat

units can be condensed to form a discrete cluster unit, we might expect that it also should be possible to couple 3 or more units together in the same way.

Thus we have the basis for a new oligomeric series (Mo 4 )n

formed from the coupled butterfly units. NdMo 8o14 is therefore a member with n-2 of a possible oligomeric series with the general formula ~M 0 4n°6n+2"

The Mo-Mo bond distances in the Mo 8 cluster range from 2.66 to 2.84A with the average of 2.726A.

From the formula we deduce that there should

be 23 electrons available for Mo-Mo bonding. sum for all Mo-Mo bonds in the cluster unit,

From the Pauling bond order ~n-

12.1, we calculate an

average bond order n (ave.) - 12.1/18- 0.67 for the 18 Mo-Mo intracluster bonds.

This compares favorably with n (ave.) - 23/36- 0.64 calculated

for 23 electrons distributed over 18 bonds.

This comparison indicates all

23 cluster electrons are indeed involved in Mo-Mo bonding.

However it

does not appear that the odd number of cluster electrons gives rise to a magnetic moment residing on the cluster anions.

magnetic susceptibility

measurements indicate that the magnetic moment of NdMo 8o14 , 3.29~B· is derived solely from the Nd3+, with an expected moment of 3.62~B calculated from the ground state 4 1 912 term. We conclude therefore that pairing of the odd electrons must take place by coupling between cluster units.

The

nearest intercluster Mo-Mo distance of 3.059(2)A, observed between a cluster and each of two neighbors, signals electron delocalization and pairing by band formation as the most likely mode of unpaired electron coupling. Continued work in this laboratory emphasizes the search for new members of this oligomeric series with n > 2. forming other compounds

~Mo 8 o 14 ,

with M

Also the possibility of

divalent metal, is being

explored in order to establish if a variable electron concentration can be used to derive changes in the electrical properties.

100

REFERENCES 1. J.D. Corbett and R.E. McCarley, chapter in "Crystal Chemstry and Properties of Materials with Quasi-One-Dimensional Structures", J. Rouxel (ed.), D. Reidel Publishing Co., Dordrecht (1986). 2. R.E. McCarley, Phil. Trans.

~·

3. A. Simon, Ang:ew. Chern. Int. ,M.

Soc. Lond., A308, 141 (1982). ~.

ZQ, 1 (1981).

4. J.D. Corbett, Ace. Chern. Res. 14, 239 (1981). 5. C. Perrin, S. Ihmaine and M. Sergent, 6. A. Perrin and M. Sergent,

New~.

New~.

Chern. !l, 321 (1988).

Chern. 12, 337 (1988).

7. A. Simon, Ang:ew. Chern. Int. Ed. Engl. 1J.., 159 (1988). B. A. Perrin, C. Perrin and M. Sergent,

~.

137, 241

Less-Common~.

(1988). 9. R.E. McCarley, Polyhedron 2. 51 (1986). 10. H.F. Franzen, Prog:. Solid State Chern. 12, 1 (1978). 11. K. Yvon, Curr . .:Ism. Mater. Sci. }., 53 (1979). 12. J.D. Corbett,

~.

Solid State Chern. 39, 56 (1981).

13. R. Chevrel, P. Gougeon, M. Potel and M. Sergent,

~.

Solid State

Chern. 57, 25 (1985). 14. V.E. Fedorov, A.V. Mishchenko and V.P. Fedin, Russian Chemical Reviews 54, 408 (1985). 15. R.P. Ziebarth and J.D. Corbett,

~.

Am. Chern. Soc. 111, 3272 (1989)

and references therein. 16. T. Hughbanks, G. Rosenthal and J.D.

Corbett,~.

Am. Chern. Soc.

108, 1927 (1986). 17. L. Pauling, "The Nature of the Chemical Bond", 3rd Ed., Cornell University Press, Ithaca, NY (1960). 18. A.K. Cheetham, S.J. Hibble and H.R. Wakerley, Inorg:. Chern. 28, 1203 (1989). 19. C.C. Torardi and R.E. McCarley,

~.

Am. Chern. Soc. 101, 3963

~.

Less-Common Met. 116, 169

(1979). 20.

C.C. Torardi and R.E. McCarley, (1986).

21. K.-H. Lii and R.E. McCarley, unpublished research. 22. R.E. McCarley, K.·H. Lii, P.A. Edwards and L.F. Brough,

~.

Solid

State Chern. 57, 17 (1985) 23. B.A. Aufdembrink and R.E. McCarley, unpublished research. 24. P. Gougeon and R.E. McCarley, to be published.

101

25. K.-H. Lii, R.E. McCarley, S. Kim and R.A. Jacobson, J. Solid State Chern. 64, 347 (1986). 26.

H. Mattausch, A. Simon, E.-M. Peters, Inorg. Chern. 25, 3428 (1986).

27.

R.A. Wheeler and R. Hoffman, J. Am. Chern. Soc. 110, 7315 (1988).

28.

S.L. Wang, C.C. Wang and K.-H. Lii,

29.

S.J. Hibble, A.K. Cheetham, A.R.L. Bogle, H.R. Wakerley and D.E.

30.

P. Gougeon, C.D. Carlson and R.E. McCarley, to be published.

J.

Solid State Chern.

11.

407

(1988). Cox, J. Am. Chern. Soc. 110, 3295 (1988).

102

NATURE OF BIMETALLIC CLUSTERS

J.H. Sinfelt Corporate Research Science Laboratories Exxon Research and Engineering Company Annandale, N.J. 08801 INTRODUCTION Bimetallic clusters, as the name implies, are entities composed of atoms of two different metallic elements. They are useful in catalytic materials and were envisioned originally for this kind of application (1-Q). Research in this area was initiated by the author in the early 1960s. Thus, the time frame of interest for this review is similar to that considered in the first session of this symposium, which was concerned with developments in the area of metal clusters during the twenty-five year period beginning with the discovery (l) of the quadruple metal-metal bond. The bimetallic clusters of interest in catalysts generally have sizes smaller than about lOOA and are commonly in the size range of 10-50 A. In some catalysts, the clusters are so small that virtually every metal atom is a surface atom (Q). Before bimetallic clusters were investigated, monometallic clusters of similar sizes were well known as components of catalysts (~-11), although the term crystallite rather than cluster was generally used in referring to the metal entity involved. Such clusters differ from the clusters ordinarily considered by inorganic chemists in the sense that they are not parts of inorganic compounds. However, with molecules chemisorbed on them, even this difference becomes blurred. In a typical catalyst the clusters are dispersed throughout porous particles of a refractory material known as a carrier or support. The carrier is responsible for the small sizes of the metal clusters and for retarding processes leading to cluster growth. Materials such as silica or alumina, which can be prepared readily with surface areas in the range of 100-300 m2/g, are commonly used as carriers. In one simple method of preparation of catalysts containing monometallic clusters, granules of a carrier are wetted with an aqueous solution of some compound of the metal. The water imparted with the solution is removed in a drying step, and the carrier then contains a metal precursor species deposited on its surface, i.e., on the surface associated with the pores in the interior of the granules. The dried material is then exposed to a stream of hydrogen at a temperature high enough to accomplish reduction of the metal precursor. The reduction leads to the formation of the metal clusters (12). Metal-Metal Bonds and Clusters in Chemistfy and Catalysis Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

103

If compounds of two different metals are dissolved simultaneously in the original solution used to wet the carrier, there is the possibility that bimetallic clusters will be formed by the procedure just described. Indeed, the possibility that the procedure will yield bimetallic clusters is very high if the metal precursors present on the carrier are both easily reduced. Thus, bimetallic clusters of two Group VIII metals, of a Group VIII metal and a Group IB metal, or of two Group IB metals, are formed very readily on conventional carriers such as silica or alumina (2). The original research on bimetallic clusters had its roots in the physical chemistry of alloys, or solid solutions, of metallic elements, with a strong influence of concepts related to surface phenomena. If two metals A and B have certain properties (13), they will form solid solutions over the whole range of compositions from pure A to pure B. A bimetallic cluster of A and B may then be regarded simply as a small aggregate consisting of an alloy or solid solution of atoms of the two metals (2). If alloys of a given pair of metals have interesting catalytic properties, the dispersion of such alloys in the form of bimetallic clusters is a desirable feature since the high ratio of surface atoms to total atoms in the clusters generally means that the catalytic activity per unit mass of metal will be high. BACKGROUND ON METAL ALLOYS AND RELATED SYSTEMS Prior to the era of bimetallic clusters, metal alloys had been of interest in catalysis for fundamental investigations of the influence of the so-called "electronic factor" in surface reactions on metals (14-19). One type of alloy which received a great deal of attention comprised a combination of a metal from Group VIII of the periodic table with another from Group IB, e.g., Ni-Cu and Pd-Au. In terms of the energy band theory of electrons in metals (20), transition metals such as those of Group VIII possess d-bands whose states are not completely occupied by electrons. By contrast, the d-bands of nontransition metals such as those of Group IB are completely filled. The fact that Group VIII metals are generally much more active catalysts than the Group IB metals for a number of important reactions, e.g., the hydrogenation, dehydrogenation, and hydrogenolysis reactions of hydrocarbons, has been attributed to their incompletely filled d-bands. According to an early view of the electronic structure of a binary alloy composed of metals from groups VIII and IB, there was a single d-band which became increasingly populated with electrons as the amount of the Group IB metal increased (20). By investigating the dependence of catalytic activity on alloy composition, catalytic chemists reasoned that the influence of electronic structure could be deduced. However, the approach was not very fruitful in elucidating the electronic factor in the simple manner envisioned. It was based on the premise that the catalytic activity of a metal surface is determined by the electronic structure of the crystal as a whole. Today, this premise is generally not accepted. It has been supplanted largely by the view that the catalytic activity is determined by localized properties of surface metal atoms (~). Although studies with metal alloys as catalysts have not provided a simple way to probe the electronic factor, they have been very rewarding in other respects. For example, they have demonstrated that effects of alloying on the catalytic activity of a metal can depend markedly on the kind of reaction being catalyzed. Thus, for hydrogenolysis reactions of hydrocarbons, which involve the scission of carbon-carbon bonds, the

104

catalytic activity of a binary alloy compr~s~ng a combination of metals from groups VIII and IB is markedly lower than the activity of the Group VIII metal alone (21-26). In contrast, the catalytic activity of a Group VIII metal for reactions of carbon-hydrogen bonds, i.e., hydrogenation and dehydrogenation reactions, is affected relatively little when it is alloyed with a Group IB metal. In fact, the activity for such reactions may actually increase (22, 23). The discovery of this selectivity phenomenon stimulated much interest in bimetallic catalyst systems in the early 1970s. While many studies of the catalytic properties of alloys were originally undertaken with the hope of elucidating the electronic factor in metal catalysis, the results of such studies have been instrumental in reviving a point of view that structural features of the surface can play a significant role in determining catalytic activity (23, 25). Over half a century ago, the Russian chemist Balandin (27) had suggested that arrays of metal atoms called multiplets were the required sites for certain catalytic reactions. Today such reactions are said to be structure-sensitive (29), and the term "ensemble" (25, 30) is widely used in place of multiplet in referring to the arrays of metal atoms. If an alloy consists of a catalytically active metal component (e.g., a Group VIII metal) in combination with an inactive one (say a Group IB metal), and a large array of metal atoms of the active component is required to accommodate the chemisorbed intermediate for a reaction, the random interspersion of inactive metal atoms among the active ones will greatly decrease the availability of the required arrays and lower catalytic activity markedly (23, 25). It has often been suggested that the sites required for hydrogenolysis reactions contain larger numbers of active metal atoms than do the sites for hydrogenation and dehydrogenation reactions, and that this is a major factor in the selective inhibition of the hydrogenolysis activity of a Group VIII metal when an inactive Group IB metal is combined with it (1, 31). A complicating factor in the interpretation of the results of catalytic investigations on metal alloys is the fact that the composition of the surface of an alloy may be very different from the composition in the bulk. If accumulation of one of the components in the surface serves to lower the surface energy of the system, the surface will then be enriched in that particular component. This principle was first enunciated by Gibbs more than a century ago. Frequently, the component concentrating in the surface is the one with the lower heat of sublimation. For binary metal alloys consisting of a Group VIII metal and a IB metal, the latter has the lower heat of sublimation and would therefore be expected to concentrate in the surface. Support for this contention is provided by studies of such systems as nickel-copper (23, 32, 33), nickel-gold (34), and palladium-silver (35). A factor which must not be overlooked in discussions of the surface compositions of alloys is the effect of the gaseous atmosphere in contact with the alloy surface. The conclusions which have just been drawn for Group VIII-Group IB systems are for an inert atmosphere or vacuum, and do not consider the effect of chemisorbed gases. If the gas interacts very strongly and selectively with the Group VIII metal, for example, the Group IB metal will not be the predominant component in the surface. Thus, in an oxygen atmosphere, the surface of the nickel-gold system is rich in nickel rather than gold (34). Similarly, in an atmosphere of carbon monoxide, the surface of a palladium-silver alloy is rich in palladium, whereas silver would normally be the predominant surface component (35). In attempting to generalize these comments on the effect of a particular chemisorbed gas on the surface composition of an alloy, we state simply

105

that the component of the alloy with the greater affinity for the gas may be drawn to the surface (36). In the course of research on alloys as catalysts, it was discovered that combinations of metallic elements of interest need not be limited to those pairs which actually form alloys, i.e., solid solutions, in the bulk. The ruthenium-copper system provides a good example (Z, 24). Although ruthenium and copper are virtually completely immiscible in the bulk (37), bimetallic ruthenium-copper aggregates can be prepared in which the effect of copper on the catalytic behavior of ruthenium is similar to its effect on the behavior of nickel in nickel-copper alloys. In the ruthenium-copper system the two components exhibit significant interaction at an interface, despite the fact that they do not form solid solutions in the bulk. In a typical ruthenium-copper aggregate, the ruthenium forms an inner core, while the copper is present at the surface (1, 24). The copper is somewhat like a chemisorbed layer on the ruthenium. The bonding of copper to ruthenium at the interface is sufficiently strong to cause the copper to spread over the ruthenium surface in preference to forming separate aggregates of copper. Since the ability to form bulk alloys is not a necessary condition for a system of two metallic elements to be of interest as a catalyst, it is misleading to use the term alloy in referring to the bimetallic entities present in catalysts in general. A term such as bimetallic aggregate is therefore preferred in this context. For the particular case in which the bimetallic aggregates are present in a highly dispersed form on the surface of a carrier, the term bimetallic cluster has been adopted (1, Z), as discussed in the introduction. When the bimetallic clusters are so highly dispersed on a carrier that the dimensions approach those of molecular species, there is a reasonable analogy with the mixed-metal entities present in certain metal cluster compounds synthesized and characterized by the inorganic chemist. STRUCTURAL STUDIES OF SELECTED BIMETALLIC CLUSTERS In the early stages of the research on bimetallic clusters, methods of obtaining structural information were limited to chemical probes such as measurements of chemisorption isotherms and reaction rates (Z). At a later stage of the research the situation changed markedly, primarily as a result of some advances in the field of x-ray absorption spectroscopy during the 1970s. The advances were concerned with methods of analysis of extended x-ray absorption fine structure (EXAFS) data (38-40), and with improvements in obtaining the data with the use of synchrotron radiation (41). Studies applying EXAFS to bimetallic clusters were initiated in the middle 1970s (42), and since that time have been conducted with many different systems (43-50). When x-rays are absorbed by matter other than monatomic gases, a plot of absorption coefficient vs. x-ray energy exhibits oscillations on the high-energy side of an absorption edge. The oscillations constitute the extended x-ray absorption fine structure (EXAFS), as illustrated in Figure 1 for the K absorption edge of ruthenium in a catalyst consisting of ruthenium-copper clusters dispersed on silica (43). Electrons ejected from the K shells of ruthenium atoms on exposure to x-rays of sufficient energy (>22.1 Kev) are scattered by neighboring ruthenium and copper atoms. Interference between waves associated with the ejected electrons and waves associated with the backscattered electrons give rise to the oscillations in the absorption coefficient.

106

Ru-Cu

22

22.5

23

23.5

ENERGY, KeV

Fig. 1.

X-ray absorption spectrum of ruthenium-copper clusters at lOOK in the region of the K absorption edge of ruthenium (.2.) .

A photoelectron ejected from an atom as a result of x-ray absorption is characterized by a wave vector K given by the equation K ~ (2mE)l/2j b

(1)

where m is the mass of the electron, b is Planck's constant divided by In the treatment 2~. and E is the kinetic energy of the photoelectron. of EXAFS data, the absorption coefficient in the region of the EXAFS is divided into two parts. One part is independent of the environment of the absorber atoms and is identical to the absorption coefficient for the free atom. The other part is the oscillating part which constitutes EXAFS. Division of the latter part by the former normalizes the EXAFS oscillations. The normalized oscillations are represented by the quantity X(K), in which K is the photoelectron wave vector. The determination of X(K) from experimental EXAFS data has been discussed in detail elsewhere (39, 43). A plot of the function K·X(K) vs K is shown in the upper left-hand section of Figure 2 for the ruthenium EXAFS data on ruthenium-copper clusters presented in Figure 1. A Fourier transform of K·X(K) yields a radial structure function, which is shown in the upper right-hand section of Figure 2. The abscissa is the distance R from the absorber atom. The prominent peak is associated with scattering of photoelectrons by nearest

107

0.4

2o-.------.,

Ru-cu

0.2

""><

""

"'

0.0

0

10

X

-0.2

:; C<:

0

LJ..

Vl

-0.4

z

<[

C<: 1--

LJ..

2o-.-------.

0

0.2

w =>

0

1-

~

.....

><

z

l:)

0.0

<[

:;

""

10

v

-0.2 2

4

b

8

10

K, _A-1 Fig. 2.

12

14

1b

0

0

2

.

4

b

R, A

Normalized EXAFS data (ruthenium K absorption edge) for ruthenium-copper at lOOK, with associated Fourier transform, filtered transform, and inverse of the filtered transform (2). The inverse transform isolates the EXAFS due to the nearest neighbor metal atoms.

neighbor metal atoms about ruthenium absorber atoms. It is centered at a value of R which is not a true interatomic distance because of phase shifts (38, 40). In the analysis of EXAFS data, it is useful to invert the Fourier transform over a limited range of R. This procedure determines the contribution to EXAFS arising from shells of atoms within that range of R. In Figure 2, for example, the region of the Fourier transform for values of R between 1.7 and 3.1A is isolated in the lower right-hand section (2). An inverse transform of this region, which brackets the primary peak, yields the function shown in the lower left-hand section of Figure 2. This function represents the contribution to the EXAFS due to backscattering of electrons by nearest neighbor metal atoms (ruthenium and copper). When EXAFS data are obtained on bimetallic clusters, there are two EXAFS functions to consider, one for each component of the clusters (43, 44). If the analysis is limited to contributions of nearest neighbor backscattering atoms, each of the functions consists of the sum of two terms. The two terms represent the backscattering contributions of nearest neighbor atoms of the two metal elements comprising the clusters. Details of the quantitative analysis of EXAFS data on

108

bimetallic clusters, including the use of data on appropriate reference materials, can be found in our early papers (43, 44). Results of a number of EXAFS investigations have been reported from our laboratories for bimetallic clusters in which copper is one of the components. Of the systems included in these investigations, only two (Au-Cu and Pt-Cu) provide examples in which the components are known to form solid solutions over the whole range of compositions in the bulk (48, 49). Moreover, Au-Cu and Pt-Cu bulk alloys exhibit the well-known phenomenon of ordering (51, 52). For Au-Cu alloys with a 1:1 atomic ratio of Cu to Au, the ordering is such that the nearest neighbor atoms about a gold atom favor copper by a factor of two; about a copper atom the nearest neighbor atoms are correspondingly in favor of gold by the same factor. However, such ordering is not evident from the EXAFS studies on the bimetallic clusters (49). Thus, there may be an effect of aggregate size on ordering phenomena, with the consequence that bimetallic aggregates in the size range characteristic of bimetallic clusters have ordering properties very different from those of macroscopic alloy crystals. Bimetallic clusters of copper with the metallic elements ruthenium, rhodium, and silver of the second long period of the periodic table, and with rhenium, osmium, and iridium from the third long period, provide examples in which the components exhibit limited miscibility in the bulk. Results of EXAFS experiments on silica-supported Ru-Cu clusters with a diameter of approximately 30 A and a 1:1 atomic !atio of copper to ruthenium are in excellent accord with a model in which copper is present on the surface of the ruthenium (43). The copper atoms are coordinated extensively to ruthenium atoms in addition to other copper atoms, so that the environment of the copper atoms in Ru-Cu clusters is very different from that in pure copper clusters. The difference is clearly seen in measurements of the EXAFS associated with the K absorption edge of copper in the two types of clusters (Figure 3). The EXAFS functions in Figure 3 differ in both shape and magnitude, which is indicative of the different environment of the copper in the two types of clusters. These functions represent the contributions of nearest neighbor metal atoms. They were obtained by inversion of Fourier transforms of EXAFS data over ranges of distances chosen to isolate these contributions. The ruthenium-copper system represents an extreme case in view of the very limited miscibility of ruthenium with copper. A system which is less extreme in this respect is the rhodium-copper system, since the components both possess the face centered cubic structure and are slightly miscible at conditions of interest in catalysis (37). When one compares EXAFS results on rhodium-copper (46) and ruthenium-copper (43) catalysts in which the Cu/Rh and CufRu atomic ratios are both equal to one, there are differences which can be related to the differences in miscibilty of copper with ruthenium and rhodium. The degree to which copper concentrates at the surface appears to be lower for the rhodium-copper clusters than for the ruthenium-copper clusters, as evidenced by the fact that rhodium exhibits a greater tendency than ruthenium to be coordinated to copper atoms in such clusters. The rhodium-copper clusters presumably contain some of the copper atoms in the interior of the clusters.

On moving further right in the periodic table from rhodium to silver, one again observes only limited miscibility of this metal with copper (53). Measurements of the EXAFS associated with the K absorption edges of silver and copper in a catalyst containing silver-copper clusters on silica reveal extensive segregation of the components (49). 109

0.50

Cu

0.0

-

>l ..... ':-

"'

-0.50 0.4 Cu-Ru

0.2 0.0 -0.2

-0 .44---.--,.......,......,..--.-"'T"'...,.....T""""""'~~--.-...,.....-j 8 10 12 14 16 6 4 2 K,

Fig. 3.

A-1

Comparison of the copper EXAFS (K absorption edge) of copper and ruthenium-copper clusters at lOOK. The EXAFS shown is the part due to nearest neighbor metal atoms (~).

The results are similar to those obtained for ruthenium-copper and rhodium-copper clusters in this respect, but are different in another. For the silver-copper clusters, the EXAFS results indicate that the surface consists predominantly of silver rather than copper, since the silver has an average coordination number significantly lower than that of the copper. The conclusion is in agreement with that reported in studies of bulk alloys of silver and copper by other investigators (54). Thus, for bimetallic clusters of copper with the metals ruthenium, rhodium, and silver of the second long period of the periodic table, there is extensive segregation of the components from each other in all cases, as would be expected on the basis of the limited miscibilities of the components. In all cases, the component with the lower cohesive energy density (i.e., lower ratio of heat of sublimation to molar volume) appears to concentrate at the surface, at least in an atmosphere of hydrogen or inert gas. Bimetallic clusters of copper with the metals rhenium, osmium, and iridium of the third long period of the periodic table provide another interesting series in which the components exhibit very limited miscibility in the bulk (48). The bimetallic systems Os-Cu and Re-Cu may be regarded as completely immiscible (37, 55). The Ir-Cu system is a less extreme case, but the miscibility is still very limited (56). Despite the limited bulk miscibility of the components, bimetallic clusters are again observed. These systems are similar to Ru-Cu, but the clusters actually investigated were generally smaller than the Ru-Cu clusters already discussed. The diameters were 15A or lower, so that surface atoms constituted more than 70% of the total metal atoms present

110

in the clusters. Consequently, with a 1:1 atomic ratio of copper to the other component, the surface could not have consisted solely of copper. The Re-Cu clusters presented an extreme case in which the average coordination number of the atoms was only four. Such a low value implies that the clusters were extremely small, and one can readily visualize structures in which all of the atoms are surface atoms. CONCLUDING REMARKS The brief discussion of bimetallic systems with copper as a common component illustrates some of the structural features encountered with bimetallic clusters in general. Copper-containing systems were chosen for discussion because of the extensive EXAFS data available on them and also because they represent a reasonably complete series of bimetallic clusters in which one of the components has been changed in a systematic manner. From the standpoint of catalysis, bimetallic clusters of copper with metallic elements from Group VIII have the feature that the copper selectively inhibits the activity of the Group VIII metal for hydrogenolysis reactions of saturated hydrocarbons, thus improving the selectivity for reactions such as aromatization and isomerization (J). Other types of bimetallic clusters, such as those consisting of two metallic elements from Group VIII, or of rhenium and a Group VIII meta~. have also been investigated extensively with the use of EXAFS (45, 47, 50). For purposes of brevity, these types of systems have not been discussed in this short review, although their structural features are just as interesting as those found for copper-containing bimetallic clusters. Moreover, selected systems from this category are technologically very important, primarily in the catalytic reforming of petroleum fractions for the production of aromatic hydrocarbons for inclusion in automotive fuels (~). REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

Sinfelt, J.H. Chern. and Eng. News 1972 (July 3 issue), 50, 18. Sinfelt, J.H. J. Catalysis 1973, 29, 308. Sinfelt, J.H., Ace. Chern. Res. 1977, 10, 15. Sinfelt, J.H. Bimetallic Catalysts: Discoveries. Concepts. and Applications; Yiley: New York, 1983. Sinfelt, J.H. Ace. Chern. Res. 1987, 20, 134. Sinfelt, J.H. U.S. Patent 3.953.368; 1976. Cotton, F.A. Inorg. Chern. 1965, ~. 334. Spenadel, L.; Boudart, M. J. Phys. Chern. 1960, 64, 204. Sinfelt, J.H.; Yates, D.J.C. J. Catalysis 1967, ~. 82. Yates, D.J.C.; Sinfelt, J.H. J. Catalysis 1967, ~. 348. Yilson, G.R.; Hall, Y.K. J. Catalysis 1970, 17, 190. 641. Sinfelt, J.H. Annual Review of Materials Science 1972, Hildebrand, J.H.; Scott, R.L. The Solubility of Nonelectrolytes, 3rd ed.; Reinhold: New York, 1950; p.335. Schwab, G.M. Disc. Faraday Soc. 1950, ~. 166. Dowden, D.A.; Reynolds, P. Disc. Faraday Soc. 1950, ~. 184. Couper, A.; Eley, D.D. Disc. Faraday Soc. 1950, ~. 172. Best, R.J.; Russell, W.W. J. Amer. Chern. Soc. 1954, 76, 838. Hall, ~.K.; Emmett, P.H. J. Phys. Chern. 1959, 63, 1102. Hall, W.K.; Emmett, P.H. J. Pbys. Chern. 1958, 62, 816. Mott, N.F.; Jones, H. Theory of the Properties of Metals and Alloys; Oxford University Press: Oxford, 1936. Sinfelt, J.H.; Barnett, A.E.; Dembinski, G.W. U.S. Patent 3,442.973; 1969.

z.

111

22. Sinfelt, J.H.; Barnett, A.E.; Carter, J.L. U.S. Patent 3,617,518; 1971. 23. Sinfelt, J.H.; Carter, J.L.; Yates, D.J.C. J. Catalysis 1972, 24, 283. 24. Sinfelt, J.H.; Lam, Y.L.; Cusumano, J.A.; Barnett, A.E. J. Catalysis 1976, 42, 227. 25. Ponec, V.; Sachtler, W.M.H. J. Catalysis 1972, 24, 250. 26. Beelen, J.M.; Ponec, V.; Sachtler, W.M.H. J. Catalysis 1973, 28, 376. 27. Balandin, A.A. Z. Physik Chern. 1929, B2, 289. 28. Balandin, A.A. Z. Physik. Chern. 1929, B3, 167. 29. Boudart, M. Chern. Technol. 1974, ~. 748. 30. Dowden, D.A. Proceedings of the Fifth International Congress on Catalysis, Vol. 1; North-Holland: Amsterdam, 1973, p. 621-631. 31. Sachtler, W.M.H.; Van Santen, R.A. Advances in Catalysis 1977, 26, 69. 32. van der Plank, P.; Sachtler, W.M.H. J. Catalysis 1967, l. 300. 33. Cadenhead, D.A.; Wagner, N.J. J. Phys. Chern. 1968, 72, 2775. 34. Williams, F.L.; Boudart, M. J. Catalysis 1973, 30, 438. 35. Bouwman, R.; Lippits, G.J.M.; Sachtler, W.M.H. J. Catalysis 1972, 25, 350. 36. Sinfelt, J.H. Progress in Solid-State Chemistry 1975, 10(2), 55. 37. Hansen, M. Constitution of Binary Alloys, 2nd. ed.; McGraw-Hill: New York, 1958; pp. 607, 619, 620. 38. Sayers, D.E.; Lytle, F.W.; Stern, E. Phys. Rev. Lett. 1971, 27, 1204. 39. Lytle, F.W.; Sayers, D.; Stern, E. Phys. Rev. B: Solid State 1975, 11, 4825. 40. Stern, E.; Sayers D.; Lytle, F.W. Phys. Rev. B: Solid State 975, 11, 4836. 41. Kincaid, B.M.; Eisenberger, P. Phys. Rev. Lett. 1975, 34, 1361. 42. Lytle, F.W.; Via, G.H.; Sinfelt, J.H. Am. Chern. Soc .. Div. Petr. Chern. Preprints 1976, 21(2), 366. 43. Sinfelt, J.H.; Via, G.H.; Lytle, F.W. J. Chern. Phys. 1980, 72, 4832. 44. Sinfelt, J.H.; Via, G.H.; Lytle, F.W.; Greegor, R.B. J. Chern. Phys. 1981, 75, 5527. 45. Sinfelt, J.H.; Via, G.H.; Lytle, F.W. J Chern. Phys. 1982, 76, 2779. 46. Meitzner, G.; Via, G.H.; Lytle, F.W.; Sinfelt, J.H. J. Chern. Phys. 1983, 78, 882. 47. Meitzner, G.; Via, G.H.; Lytle, F.W.; Sinfelt, J.H. J. Chern. Phys, 983, 78, 2533. 48. Meitzner, G.; Via, G.H.; Lytle, F.W.; Sinfelt, J.H. J. Chern. Phys, 1985, 83, 353. 49. Meitzner, G.; Via, G.H.; Lytle, F.W.; Sinfelt, J.H. J. Chern. Phys, 1985, 83, 4793. 50. Meitzner, G.; Via, G.H.; Lytle, F.W.; Sinfelt, J.H. J. Chern. Phys. 1987, 87, 6354. 51. Cullity, B.D. Elements of X-ray Diffraction; Addison-Wesley: Reading, Massachusetts, 1956; pp. 370-372. 52. Schneider, A.; Esch, U. Z. Elektrochem 1944, 50, 290. 53. Stockdale, D. J. Inst. Metals 1930, 43, 193. 54. Betz, F.; Arias, M.; Braun, P. Nucl. Instrum. Methods 1980, 170, 347. • 55. Knock, B.; Star, W.M.; van Rongen, H.J.M.; van den Berg, G.J. Physica (Amsterdam) 1964, 30, 1124. 56. Raube, E.; Roschel, E. Z. Metallkd. 1969, 60, 142.

112

THERMOCHEMICAL ASPECTS OF ORGANOTRANSITION METAL CHEMISTRY.

INSIGHTS

PROVIDED BY METAL-LIGAND BOND ENTHALPIES Michel R. Gagne, Steven P. Nolan, Afif M. Seyam, David Stern, and Tobin J. Marks Department of Chemistry Northwestern University Evanston, IL 60208-3113 INTRODUCTION The past several decades have witnessed enormous growth in what we know about the synthesis, reactivity, reaction mechanisms, molecular structures, and electronic structures of organometallic molecules. Curiously, however, we know far less about the thermodynamics of most organometallic transformations and, in particular, about the exact strengths of metal-ligand bonds. In principle, such information offers a better understanding of metal-ligand bonding, a better insight into the course(s) of known reactions, and a valuable advantage in designing new transformations.l-5 For organometallic chemistry involving actinides, lanthanides, and early transition elements, our thermochemical/calorimetric investigations were motivated by growing evidence that deviations in reactivity patterns from those of middle and late transition elements could not be explained by kinetic factors alone. In the present chapter, we briefly review some of our recent results in this area, focusing upon how ancillary ligands affect metal-ligand bond enthalpies, metal-ligand bonding patterns as a function of position in the Periodic Table, and the thermochemically-assisted design of a new catalytic reaction pattern: organolanthanide-catalyzed hydroamination. DEFINITIONS AND METHODOLOGY The metal-ligand bond disruption enthalpy can be defined as shown in eq.(l) for the homolytic, adiabatic process shown in eq.(2).2,6 Assuming D(LnM-R) = 6H~(LnM) + 6H~(R•) - 6H~(LnM-R)

(1)

Ln = ancillary ligands LnM-R

- - > LnM + R•

(2)

the reactions are rapid, clean, and quantitative, relative metal-ligand bond disruption enthalpies can be readily obtained by protonolytic (eqs.(3),(4)) or halogenolytic (eqs.(5),(6)) titration calorimetry. The LnM-R + HX

- - > LnM-X + RH + 6Hrx

Metal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

(3)

113

(4)

AHrx - D(LnM-R) + D(H-X) - D(LnM-X) - D(R-H) LnM-R + X2

- - > LnM-X + RX + AHrx

(5)

(6)

AHrx - D(LnM-R) + D(X2) - D(LnM-X) - D(R-X)

advantages of titration calorimetry include the ability to monitor reaction stoichiometries in situ as an additional check on the analysis chemistry as well as the possibility of sequentially cleaving and analyzing a series of metal-ligand bonds at the same metal center. For the present research, instrumentation has been developed which allows the study of extremely air-sensitive compounds under vigorously anaerobic conditions.7 For a given LnM array and X ligand, it can be seen that the approach of eqs.(3)-(6) yields relative D(LnM-R) values which are anchoredlO to the corresponding D(LnM-X) parameters. While such information is adequate for analyzing numerous LnM-R ~ LnM-R' transformations, it would also be desirable to obtain absolute values of D(LnM-R) to understand both M and Ln effects. An approximate approach has been to equate D(LnM-X) values to the corresponding Dl(MXm) parameters where M is in the same formal oxidation state.6,8-10 While this transferability approximation seems reasonable, it would be of great interest to quantify the effects of the ancillary ligand array, Ln· A more rigorous approach to determining absolute D(LnM-R) values is possible when LnM/LnM-R pairs are available as shown in eqs.(7)-(ll).ll-13 Such studies have recently been carried out

- - > LnM-X + RX

.6Hrx(5)

(7)

---> LnM + 1/2 x2

.6Hrx(8)

(8)

---> 112 x 2

1/2 D(X-X)

(9)

R• + X•

D(R-X)

(10)

LnM + R•

D(LnM-R)

(11)

for (Me3SiC5H4 )3U-Rll and Cp2Ln-R12,13 complexes (Cp Eu, Yb).

'15-Me5C5; Ln- Sm,

LnM-R + x 2 LnM-X X•

R-X ---> LnM-R --->

,

ANCILLARY LIGAND EFFECTS AND TRANSFERABILITY Absolute metal-halogen bond disruption enthalpies have now been measured for a number of organoactinide and organolanthanide complexes using the approach of eqs.(7)-(ll). The numerical results are compiled in Table 1, where they are also compared to literature data for the corresponding homoleptic halides.l4-16 These data are illustrated graphically in Figures 1 and 2. From the invariance of D(Cp2Sm-OR)n in complexes where n = 1 and 212 as well as of D(Cp2Sm-I) in Cp2Sm(THF)I and in (Cp2Smi)n12 it can be argued that any association of (Cp2LnX)n complexes does not greatly affect derived D(Cp2Ln-X) values. These results indicate that Me3SiC5H4/Cp and halide ancillary ligands have remarkably similar influences on lanthanide-halide and actinide-halide bond disruption enthalpies. Thus, these quantities should have significant transferability. BOND ENTHALPY PATTERNS AS A FUNCTION OF POSITION IN THE PERIODIC TABLE While it is expected a priori that key metal-ligand bonding parameters may vary with the position of the metal in the Periodic Table, such patterns have not been investigated in depth. In Figures 3-5 are s~own comp~rative plots of D(M-X) bond enthalpy data from Cp2ZrX2, Cp2ThX2, Cp2SmX, and (Me3SiC5H4)3U-X systems for a wide range of X 114

Table 1.

Bond Disruption Enthalpy Data for Organo-f-Element Complexes and the Corresponding Homoleptic Halides in Kcal/Mol

LnM-X

D(M-X)

MXm

Dl(MXm)

Cp2SmCl

97.1(3.0)a

SmCl3

102(5)d

Cp2SmBr

83.6(1.5)a

SmBr3

86(5)d

Cp2Smi

69.4(2.4)a

Sml3

68(5)d

Cp2Eui

57.1(2.0)b

Eul3

65(5)d

Cp2Ybi

61.2(1.5)b

Ybl3

60(5)d

(Me 3SiC5H4)3UI

62.4(1.4)C

UI4

66(8)e

aReference 12 bReference 13 cReference 11 dReferences 14,15 eReference 16

•

0

Cl

Fig. 1.

Cp',SmX

x,smx

Br

Comparison of measured D(Cp2Sm-halogen) values to the first bond dissociation energies (refs. 14,15) of the corresponding samarium trihalides.

115

0 Cp' 2 ln-l

0

s

60

'-..

«i

"

~

,....,.

40

I

e p

20

0

Eu

Sm

Fig. 2.

Yb

Comparison of measured D(Cp2Ln-I) values to the first bond dissociation energies (refs. 14,15) of the corresponding lanthanide triiodides.

135

CIA

o'su

•

125

0

s

'til

" xI .!<

;:::

cp

115

Br aS"Pr

105

•

H 95

NMe 2

•

I

•

•

85

R

75 40

50

60

70

80

90

100

D(Sm-X), kcaJ/mol

Fig.3,

116

Comparison of measured D(Cp2Sm-X) data (ref.l2) to those of the corresponding Cp2Th(X)-X series (refs. 8,12 using an approximate thermodynamic anchor). For Sm, R ~ CH(SiMe3)2 while forTh, R ~ Me. The X= H and snpr data for Th are averages of D[Cp2Th(X)-S] and D(Cp2Th(OtBu)-X] data, and the X= halide data are from D1(ThX4) (ref. 16).

100

0 8 'a;

80

60

C)

.:.:

>
40

~

0

20

D(Sm-X), kcal/mol

Fig. 4.

Comparison of measured D(Cp2Sm-X) and D(Cp2Sm-L) data (ref. 12) to those of the corresponding D((Me3SiCsH4)3U-X] and D((Me3SiCsH4)3U-L] data of ref. 11.

Cl•

115

0 8 'a; 0

105 95

."'

>
... N q

85 75 65 40

50

60

70

80

90

100

D{Sm-X), kcal/mol

Fig. 5.

Comparison of measured D(Cp2Sm-X) data (ref. 12) to those of the corresponding Cp2Zr(X)-X series (ref. 10, anchored to D1(ZrCl4); average of D(Cp2Zr(X)-X] and D[Cp2Zr(Cl,I)-X]). For Sm, R= CH(SiMe3)2 while for Zr, R = CH3.

117

ligands. It can be seen that bonding patterns in these archetypical early transition element, actinide, and lanthanide systems, having rather different steric environments about the metal centers, are strikingly similar. Two gauges are readily available to help judge whether the above similarities in metal-ligand bonding patterns extend to middle and late transition metals. The first approach is to compare key D(M-X) - D(M-X') parameters as a function of position in the Periodic Table.ll Shown in Figures 6 and 7 are comparisons of the parameters D(M-H)-D(M-CH3)11-22 and D(M-I)-D(M-CH3)17-22, respectively. The former quantity is pivotal in determining the course of p-hydride elimination and C-H functionalization reactions while the latter plays an important role in a variety of oxidative addition/reductive elimination and metal-halogen interchange processes. It can be seen that D(M-H)-D(M-CH3) rises almost monotonically as the Periodic Table is traversed to the right while D(M-I)-D(M-CH3) falls almost monotonically. Clearly, these bonding patterns vary greatly and, to an appreciable extent, systematically with position in the Periodic Table. While portions of the above trends have been explained on the basis of promotion energies,23 bond polarity effects 24 and orbital repulsion effects,25 an equally appealing and complementary qualitative explanation can be derived from electronegativity24,26-29 considerations. In particular, the perturbation theory-based formalism of Matcha30 (eq.(l2))

30

Pal'tX.

25

0

E :::::.. 0

_,.u

.

20

.-::. :I:

u

I

15

.......

10

3Cl :I:

I

::IE

c

5

0

Sm

Fig. 6.

118

Th

u

Zr,Hf

Mo,W' Mo'

Mn

Ir

Ir'

Pt

Comparison of D(M-H)-D(M-CH3) for various transition metal complexes. ,Key: Cp2SmX (ref. 12,,where CH3 z CH(SiMe3)2 is assumed); Cp2Th(OR)X (refs.8,9); Cp2U(OR)X (ref.9); average Cp2MX2 (ref.lO); Cp2MX2 (ref.l7); CpMo(C0)3X (ref.l8); Mn(CO)sX (ref.l9); Ir(PMe3)2(Cl)(I)(CO)X (ref.20); Cpir(PMe3)X2 (ref.21); Pt(PEt3)2X2 (ref.22, adjusted assuming D(Pt-Me) - D(Pt-Et) z 5 kcal/mol).

(12) represents a considerable improvement over the original Pauling formulation and should be applicable. Here the x's are Pauling electronegativities. Starting from eq.(l2), it is possible to derive expressions for D(M-H)-D(M-CH3) (e.g., eq.(l3)) and D(M-I)-D(M-CH3).31 Figures 8 and 9 D(M-H)-D(M-Me) = l/2[D(Hz)-D(Mez)] + 105(e-O.Zl 9 (XM-XMe) 2 (13)

-e-0.219(XM-XH)2)

show theoretical curves generated for such parameters as a function of metal electronegativity. Although the present simplistic treatment ignores other important effects such as nonbonded repulsions, it can be seen that the overall patterns are in qualitatively good agreement with the trends in Figures 6 and 7, recognizing that transition metal electronegativity generally increases toward the right of the Periodic Table.26-29 Another gauge of metal-ligand bonding patterns involves the linearity of D(H-X) versus D(M-X) plots. For metal complexes of Ru and Pt, such plots are reported to be linear with a slope of 1 for a fairly wide range of X ligands.32 As can be seen in eq.(l4) (derived from eq.(l2)), if XM ~ XH• which is reasonable for middle and late transition

25

cp".smx 0

20

E

Cp' 1Th(OR)X

::::-0

Cp',U(OR)X

0

.Y.

,...::.

Cp' 1 ZrX, Cp 1loloX, Cp',H!X, Cp,WX,

15

~

:I:

u

I

:::;

0

loln(CO),X

10

Cp'lrPX,

-::::1

~ 0

5

0

Sm

Fig. 7.

Th

U

Zr,Hf

Mo,W Mo'

Mn

Ir

Ir'

Pt

Comparison of D(M-I)-D(M-CH3) for various transition metal complexes. Key: CpzSmX (ref.l2, where CH3 ~ CH(SiMe3)2 is assumed); CpzThXz (ref.9) assuming D(Th-I) ~ D1(Thi4); CpzUXz (ref.ll); average CpzMXz (ref.lO); average CpzMXz (ref.l7); CpMo(C0)3X (ref.lB); Mn(CO)sX (ref.l9); Ir(PMe3)(Cl)(I)(CO)X (ref.20); Cpir(PMe3)X2 (ref.21); Pt(PPh3)z(I)X (ref.22).

119

0

12

E

~ c u

8

.Yo

..

,...... :I:

4

' '

(J

I

:::::!l

a I'. I

a:::::!l Fig. 8.

0 XcH,

= 2~~--- -

-4

From Matcha Formalism

-8

L,,-~-r,-r,-ro~~-r~-r,-r,-r

1.20

1.45

1.70

xt.A

1.95

2.20

Correlation between D(M-H) - D(M-CH3) and XM calculated using eq.(l3) and metal Pauling electronegativities.

0

E

20

......... 0

u

16

..-.....,

12

~

I

(J

I

:::::!:

8

......... Q

From Matcha Formalism

4 I :::::!: ......... Q

0 1.20

1.45

1.70

1.95

2.20

xt.A Fig. 9.

120

Correlation between D(M-1) - D(M-CH3) and XM calcuated using an expression analogous to eq.(l3) (footnote 31) and metal Pauling electronegativities. XcH 3 is taken to be 2.3.

D(M-X) - D(H-X) + l/2[D(M 2 )-D(H 2 )] + 105(e-0· 219 (XH-XX) 2 -e

-0.219(xwxx)2 )

(14)

elements, the relationship reduces to a linear one for constant M. In contrast, for cases of more electropositive M, eq.(l4) predicts considerable deviation from linearity with D(LnM-X) values enhanced most for the more electronegative ligands. Figure 10 illustrates such a plot for Cp2SmX compounds.12 It is evident that the linear D(H-X) vs. D(M-X) relationship breaks down and, from numerical results presented elsewhere,lO that the deviations are roughly wh~t would be ex~ected from eq.(l4). A similar situation obtains for Cp2ZrX210 and Cp2Th(OR)X8,lO data.

DESIGNING NEW REACTIONS Metal-ligand bond disruption enthalpy data can provide invaluable assistance in designing new transformations, especially at metal centers where kinetic impediments are anticipated to be small. Thus, the remarkable facility with which early lanthanide-alkyl bonds undergo olefin insertion within bis(pentamethylcyclopentadienyl)metal coordination spheres33-36 (e.g., eq.(lS); Nt (1 atm ethylene; 25°C) ~ 1800 s-1 when X= (15) primary alkyl or hydride and Ln = La34) suggests that, in this environment, thermodynamically feasible but normally unobserved olefin insertion processes involving other metal-X bonds may also be facile. Coupling to proton transfer processes (e.g., eq.(l6)) would then provide a catalytic

130

0 120

s

'-..

<0 110

.

H

u

~

xI

ES Q

100

CH(TMS) 2

* NMe * 1r-C3Hs

90

2

BO PEt,

70 30

40

50

60

70

80

90

100

D(Sm-X), kcal/mol

Fig. 10. Correlations between D(H-X) values and the corresponding D(Cp2Sm-X) values (ref. 12). Ligands of the same general type are denoted by separate symbols and the lines represent leastsquares fits to the data points.

121

Table 2.

Entry

1

Product and Turnover Frequency Data for OrganolanthanideCatalyzed Hydroaminations/Cyclizations. Substrate

H 2N~

N1(hr- 1)

Pwduct

z.

Catalyst

1,1

'6

1

13(25°C)

1a

140(60°C)

la

125(25°C)

la

1,1

2

H2N~

:!.

'+

.8.

75(80°C)

lb

<1(80°C)

lc

.2.

84(25°C)

la

lJl

5(60°C)

la

H

3

4

5

NH2

~

H2N~

!

s.

--(:}

l)

():)--- l l

la

~I

(16) cycle for HX addition to olefins. For lanthanide amides (X~ NR2), (eq.(lS)) is estimated to be approximately thermoneutrall2,37 while eq.(l6) should be both rapid38 and exothermic.l2,39 As the first embodiment of this strategy, we have recently disclosed the facile and regiespecific organolanthanide-catalyzed hydroamination/cyclization of N-unprotected amino olefins,40 heretofore difficult homogeneous transformations proceeding (e.g., with Pd+2 catalysts) via distinctly different mechanistic pathways.41-44 The catalytic reaction of (Cp2LaH)2 (1a)34 with a variety of dry, degassed amino olefins (typically in 20-100-fold stoichiometric excess) proceeds to completion in hydrocarbon solvents as shown in Table 2. Several features of this hydroamination reaction are especially noteworthy. These include the formation of a six-membered heterocycle (5~10), the cyclization of internal amines (4~9, 6~11), and the rapidity of the ggm-dimethyl transformation (3~8). The latter observation strongly argues that ring formation is turnover-limiting. The present reactions are found to be ~ 99% regiospecific by NMR spectroscopy, with the exception of 6~11, where ca. 10% of the product is another species, the identity of which is currently under investigation.

122

Preliminary mechanistic data, in addition to the aforementioned ggmdimethyl effect, are in accord with the scenario in Figure 11 where olefin insertionjcyclization (1, eq.(lS)) is, i~ most cases, turnover-limiting. Kinetically, the hydroaminations in Table 2, entries 1, 2, and 4, are first-order in organolanthanide and zero-order in amino olefin over the entire course of the reactions (i.e., Nt is independent of olefin and lanthanide concentration), implying that proton transfer (ii, eq.(l6)) is the rapid step (as expected36). Also in agreement with this picture is the relative ordering of catalyst activities for 3~8: (Cp2LaH)2 (la) > [Me2Si(Me4Cs)2LuHl2 (lb) > (cp 2LuH)2 (lc) (Table 2)--identical to the p!eviously reported ordering for catalytic propylene oligomerization activity.33-36 The outcome of the isotopic labelling experiment 3-d2 ~ 8-d2 (eq.(l7)) further supports the proposed mechanism, revealing exactly

D

la (17)

the atomtransposition pattern expected in Figure 11. The observation that hydroaminationjcyclization rates are decreased in THF also supports a turnover-limiting ki process. Such effects are common in Cp2Ln-centered olefin transformations33-36 and reflect competition for the empty coordination site within the Cp2LnX coordination sphere, which is necessary for the insertion process. CONCLUSIONS This brief account has emphasized the value of metal-ligand bond enthalpy information for better understanding metal-ligand bonding and for

H

-{J

Cp'zl.aU

H

Fig. 11. Proposed mechanism of organolanthanide-catalyzed hydroaminationj cyclization of amino olefins.

123

aiding in the design of new types of transformations. While the emphasis has been on lanthanide, actinide, and early transition element chemistry, the broad generality of the approFches discussed should be evident. ACKNOWLEDGMENTS We thank the National Science Foundation for generous support under grant CHE8800813. D.S. thanks Rh8ne-Poulenc for a graduate fellowship. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

124

T. J. Marks, ed., " Metal-Ligand Bonding Energetics in Organotransition Metal Compounds", Polyhedron Symposium-in-Print, (1988). 1. G. Pilcher and H. A. Skinner, in "The Chemistry of the Metal-Carbon Bond," F. R. Harley and S. Patai, eds., Wiley, New York, (1982) pp. 43-90. J. A. Connor, Top. Curr. Chern. 71:110 (1977). J. Halpern, Ace. Chern. Res. 15:238 (1982). J. U. Mondal and D. M. Blake, Coord. Chern. Rev. 47:204 (1983). J. W. Bruno, T. J. Marks, and L. R. Morss, J. Am. Chern. Soc. 105:6824 (1983). L. E. Schock and T. J. Marks, manuscript in preparation. D. C. Sonnenberger, L. R. Morss, and T. J. Marks, Organometallics 4:352 (1985). J. W. Bruno, H. A. Stecher, L. R. Morss, D. C. Sonnenberger, and T. J. Marks, J. Am. Chern. Soc. 108:7275 (1986). L. E. Schock and T. J. Marks, J. Am. Chern. Soc. 110:7701 (1988). L. E. Schock, A. M. Seyam, M. Sabat, and T. J. Marks, in ref. 1, pp 1517-1530. S. P. Nolan, D. Stern, and T. J. Marks, J. Am. Chern. Soc., in press. S. P. Nolan, D. Stern, and T. J. Marks, Abstracts, 196th ACS National Meeting, Los Angeles, CA, Sept. 25-30, 1988, !NOR 378. J. E. Huheey, "Inorganic Chemistry", 2nd ed., Harper and Row, New York (1978), pp. 824-850. D. D. Wagman, W. H. Evans, V. B. Parker, L. Halow, S. M. Bailey, R. H. Schumm, and K. L. Churney, Natl. Bur. Stand. Tech. Note (U.S.), 1971, No. 34-35. Calculated from thermodynamic data given by: L. R. Morss, in "The Chern is try of the Actinide Elements" , 2nd ed. , J . J . Katz , G. T. Seaborg, L. R. Morss, eds., Chapman and Hall, London (1986), Chapt. 17. A. R. Dias, M. S. Salema, and J. A. Martinho-Simoes, J. Organomet. Chern. 222:69 (1981). S. P. Nolan, R. Lopez de la Vega, S. L. Mukerjee, A. A. Gonzalez, K. Zhang, and C. D. Hoff, ref. 1, pp. 1491-1498. J. A. Connor, M. T. Zarafani-Moattar, J. Bickerton, M. L. El Saied, S. Suradi, E. Caron, A. Al-Takhin, and H. A. Skinner, Organometallics 1:1166 (1982), and references therein. J. U. Mondal and D. M. Blake, Coord. Chern. Rev. 47:205 (1982). P. 0. Stoutland, R. G. Bergman, S. P. Nolan, and C. D. Hoff, ref. 1, pp. 1429-1440. R. L. Brainard and G. M. Whitesides, Organometallics 4:1550 (1985). J. B. Schilling, W. A. Goddard, III, and J. L. Beauchamp, J. Am. Chern. Soc. 109:5573 (1987). J. W. Bruno, M. R. Duttera, C. M. Fendrick, G. M. Smith, and T. J. Marks, Inorg. Chim. Acta 94:271 (1984). T. Ziegler, V. Tschinke, L. Versluis, and E. J. Baerends, in ref. 1, pp. 1625-1657. L. Pauling, "The Nature of the Chemical Bond," 3rd ed., Cornell University Press, Ithaca, New York, (1960), Chapter 3.

27. 28. 29. 30.

31.

32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44.

J. E. Huheey, "Inorganic Chemistry," 3rd ed., Harper and Row, New York (1983) pp. 144-160. B. E. Douglas, D. H. McDaniel, and J. J. Alexander, "Concepts and Models of Inorganic Chemistry," 2nd ed., Wiley, New York (1983) Chapt. 24. J. Mullay, Struct. Bond (Berlin) 66:1 (1987), and references therein. R. L. Matcha, J. Am. Chern. Soc. 105:4859 (1983). We utilize the formulation having an arithmetic mean expression for D(A-B) since this leads more straightforwardly to useful relationships such as eq.(l2). Both arithmetic and geometric mean approaches have been employed in the Pauling formulation,26-28 with the latter preferred in cases of sign ambiguities. Because D(I2)-D(Me2) is negative (a common artifact when the arithmetic mean is used in eq.(l2)),28 we arbitrarily set it equal to zero in the expression analogous to eq.(l3). This in no way affects the XM dependence of the function. H. E. Bryndza, L. K. Fong, R. A. Paciello, W. Tam, and J. E. Bercaw, J. Am. Chern. Soc. 109:1444 (1987). H. Mauermann and T. J. Marks, Organometallics 4:200 (1985). G. Jeske, H. Lauke, H. Mauermann, P. N. Swepston, H. Schumann, and T. J. Marks, J. Am. Chern. Soc. 107:8091 (1985). G. Jeske, L. E. Schock, H. Mauermann, P. N. Swepston, H. Schumann, and T. J. Marks, J. Am. Chern. Soc. 107:8103 (1985). G. Jeske, H. Lauke, H. Mauermann, H. Schumann, and T. J. Marks, ~ Am. Chern. Soc. 107:8111 (1985). Cp2LaNMe2 catalyzes ethylene polymerization (D. Hedden and T. J. Marks, unpublished results). P. J. Fagan, J. M. Manriquez, C. H. Vollmer, C. S. Day, V. Day, and T. J. Marks, J. Am. Chern. Soc. 103:2206 (1981). For the addition of NH3 to H2C=CH2, fiG 0 ~ -4 kcal/mol. M. R. Gagn~ and T. J. Marks, J. Am. Chern. Soc., in press. Y. Tamaru, M. Hojo, H. Higashima, and Z. Yoshida, J. Am. Chern. Soc. 110:3994 (1988), and references therein. J. P. Collman, L. G. Hegedus, J. R. Norton, and R. G. Finke, "Principles and Applications of Organotransition Metal Chemistry," University Science Books, Mill Valley, CA (1987), Chapts. 7.4, 17.1. A. L. Casalnuovo, J. C. Calabrese, and D. Milstein, J. Am. Chern. Soc. 110:6738 (1988). G. Pez, and J. E. Galle, Pure Appl. Chern. 57:1917 (1985).

125

METAL CLUSTERS AND SUPPORTED METAL CATALYSTS

B. c. Gates Center for Catalytic Science and Technology Dept. of Chemical Engineering University of Delaware Newark, Delaware 19716 INTRODUCTION One of the driving forces for the development of metal cluster chemistry has been the opportunities that these compounds offer in catalysis 1 . To an inorganic chemist, a metal cluster is a compound with two (or perhaps three) or more metal atoms2, 3, whereas to most researchers in classical heterogeneous catalysis, a metal cluster is a small, discrete group of metal atoms (referred to here as a metal aggregate) dispersed on a support, typically a porous metal oxide with a high internal surface area. Many industrial catalysts, ranging from those used in automobile exhaust converters to those used in petroleum naphtha reformers, consist of small metal aggregates dispersed on high-area supports or carriers. Understanding of the structures and properties of the metal aggregates is difficult because they are nonuniform in size, shape, interactions with the support, and (for many reactions) catalytic activity. The complexity of the structures of these materials has motivated the investigation of structurally simple supported catalysts, including molecular metal clusters, as models. Most of the attempts to prepare supported metal clusters have failed for the same reasons that most attempts to use metal clusters in homogeneous catalysis have failed: the clusters are fragile and easily converted into mononuclear metal complexes and/or into nonuniform metal aggregates. But recently some attempts have been successful in giving well-defined supported metals, and it is becoming clear that materials prepared from metal clusters on supports are of interest in providing new structures of highly dispersed metals on supports, some of which are entirely new types of catalysts. These developments are reviewed in the following paragraphs. In addition, investigations of supported metals derived from organometallic compounds (including metal clusters) are providing new insights into the structures of conventional supported metal catalysts; this is a secondary theme of the present review and is developed more fully elsewhere4. It is evident that metal cluster chemistry and metal catalysis are merging into common territory.

Metal-Meta/ Bonds and Clusters in Chemistry and Catalysis Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

127

SUPPORTED METALS DERIVED FROM METAL CLUSTERS: NEW STRUCTURES Structurally complex supported metals Most attempts at catalyst preparation from clusters of the group VIII metals have led to supported metal aggregates having little structural definition or resemblance to the precursors. For the most part, the preparations have not led to samples that are much different from those made conventionally from inorganic salt precursors. Nonetheless, an advantage of preparations from organometallics such as metal carbonyls is the ease of activation of the metals (by removal of ligands such as CO at low temperature) and the lack of residues such as chloride.ions on the support surface. Metal clusters may also be advantageous in the preparation of supported metals with alloy-like structures. There is a report of a commercially applied supported bimetallic catalyst derived from treatment of supported Rh particles with an organotin complex5, and there are reports of other supported bimetallics with interesting selectivities for oxygenated compounds in CO hydrogenation catalysis.6 Supported "molecular" catalysts Supported metal clusters. Supported metal catalysts can be made in a hierarchy of structures ranging from simple mononuclear metal complexes, to ensembles of these complexes, to metal clusters, to nonuniform aggregates or crystallites resulting from conventional preparations. One focus of the present review is structurally simple supported metal cluster catalysts, those which can be described as molecular analogues. There are now many examples of metal complexes and metal clusters anchored to polymers and metal oxides functionalized with ligands such as phosphines, and some are catalytically active for reactions such as olefin isomerization and hydrogenationl, but they are all unstable at temperatures exceeding about 100°C and will not be considered further here. There are also a number of metal complexes and metal clusters anchored directly to the surfaces of metal oxides. Some of the anchored mononuclear complexes of the transition metals (e.g., Cr, Mo, w, and Re) are quite stable, but almost all of the anchored metal clusters, exemplified by the well-described triosmium carbonyls [HOs3(COl1o{O-Si}] and [HOs3(COl1o{O-Al}], are limited in their stability and therefore in their potential catalytic applicationsl. (The curly brackets in this notation{} denote groups terminating the bulk metal oxide in these structures.) The focus here is on ( 1) the few known supported metal clusters that are robust and (2) simple, stable surface structures derived f~om metal clusters, especially those showing novel catalytic properties. Robust supported metal clusters have been prepared by reactions of metal cluster precursors, either in solution or in the gas phase, brought in contact with metal oxide supports 7 . Others have been prepared by reductive carbonylation of mononuclear metal complexes on support surfaces. Preparation chemistry of the latter type was demonstrated for neutral rhodium carbonyl clusters [RhG(CO)lGl on y-Al 2 o 3 8 and in zeolite y9, but these clusters lack stability at high temperatures. Extension of the chemistry to include robust clusters was made possible by the application of basic 128

supports, which allow the formation of anionic carbonyl clusters. 1 0 The surface organometallic chemistry of formation of Os and Ru clusters is analogous to the reductive carbonylation of halide complexes such as [PtC1612- in basic solution. For example, mononuclear carbonyl and halide complexes of Os on MgO have been converted into anionic metal clusters, often in high yields, by treatment with CO or CO + H211,12 (Fig. 1). Some of these clusters, especially those with encapsulated carbon atoms, namely, [OssC(CO)l41Z- and [Os1oC(COl2412-, are robust, existing on the surface even at temperatures of several hundred degrees Celsius, being stabilized by an atmosphere of CO or CO + H2 and the basic support ligands. The surface organoosmium chemistry occurring on the basic MgO surface (Fig. 1) closely parallels that observed in basic solutions. For example, the hydride carbonyl cluster [H30s4 (CO) 12l- is stable on the MgO surface at 300°C in the presence of equimolar CO + H2 at 10 atm, but when the H2 partial pressure is reduced, a cluster lacking hydride ligands, [Os1oC(COl241 2-, is formed. The chemistry of ruthenium carbonyl clusters on Mgo13,14 is more complicated; it is quite difficult to avoid formation of metal particles and mononuclear Ru complexes, even when robust metal clusters such as [RUGC(C0)161 2 - are present. The surface chemistry of rhodium carbonyl cluster anions is much less fully characterized 15, 1 6, but [Rh6 (CO) 161 has been converted into [Rh6(C0)1sl 2- and related species. The stability of a number of rhodium carbonyl cluster anions in basic solutions17 gives indications that interesting new discoveries may lie ahead. Encaged metal clusters. Metal aggregates in molecular-sieve zeolite cages are important catalysts, described in an extensive literature18. The molecular-scale cages of these supports may help preserve the dispersion of the metal and also offer the advantages of shape selectivity in catalysis. But the usefulness of zeolite-supported metals is limited by the tendency of the metals to migrate out of the pores and to sinter on the outer crystal surfaces of the zeolite. For example, small metal aggregates or clusters, e.g., of rutheniuml9 and of iron20 have been found to be highly selective catalysts for hydrogenation of CO, but their extreme lack of stability appears to severely limit their potential applications. It is evident that the surface organometallic chemistry illustrated in Fig. 1 provides a hint of how to stabilize metal carbonyl clusters in zeolites: the keys to stability of the supported metal clusters are the stability of the metal framework (suggesting metals such as Os), the basicity of the support, and the presence of stabilizing ligands such as carbonyl and hydride. Consistent with the surface organometallic chemistry shown in Fig. 1, zeolite-supported osmium carbonyl clusters have been synthesized in zeolite Y made basic by the incorporation of sodium clusters formed from sodium azide, as described by Fejes et al.2 1 and Martens et a1.22 Zeolite-supported osmium clusters were prepared by a shipin-the-bottle synthesis as the volatile [H20s (CO) 41 was introduced from the vapor phase into the molecular-scale pores of the support.23 Upon treatment with CO + H2, the mononuclear 129

w

0

200-250° C 1 atrn

CO or CO + H2

l

4-5 h

275°C, 1 atm

[Os 10 C(C0)24 ] 2 -/MgO

COCO+ or H2

x = 2 and 3

{0Mg} 2 0s (CO) x

A

150°C 2 h

/MgO

2-

C, 1 atm 4 h

[H 3 0s 4 (C0) 12 ]- /MgO

, 275°C, 11 atrn 12 h

CO + H 2 (1:3 molar ratio) 275°C, 10 atm > 2 days

CO + H2

[0s5C(C0 )14]

~

[H 3 Os 4 (C0) 12 ]-/MgO

275°C, 1 atm 4 h

co

Fig. 1. "Molec ular" organoo smium chemis try on the basic The subscr ipt 400 refers to the pretre atmen t temper ature MgO suppor t. of the MgO . l 0 , l l , l 2

Precurs or Prepare d from H2 0sC16+ Mg0400

2 h

275°C, 1 atrn

He

Precurs or Prepare d by Adsorpt ion of [Os 3 (COh2 l on MgO 400

metal carbonyl was converted into an osmium cluster, which is not yet fully characterized, but is thought from the infrared spectrum and the yellow color to possibly be [H30S4(C0)12l-, which just fits in the zeolite supercage and is too large to diffuse out through the apertures between the cages. This sample is stable in CO + H2 at temperatures in excess of 300°C, but when the CO:H2 ratio is decreased, the osmium migrates out of the pores and forms large particles, behaving more like metal aggregates and less like cluster anions. The stabilized metal cluster is a new class of catalyst and is discussed again below. Denuded metal clusters. The supported metal clusters mentioned above are all metal carbonyls, and to a catalytic chemist they seem to be "poisoned" metals. It is obvious that the catalytic opportunities would be greatly expanded if the ligands could be removed. But, as mentioned above, attempts to remove the carbonyl ligands from supported metal clusters have led to disruption of the metal frameworks and loss of structural uniqueness. Lamb et al.24 have now succeeded in removing the carbonyl ligands from a supported metal cluster without the loss of metal framework integrity. They worked with one of the most robust known clusters, [Os1oC (CO) 241 2 -supported on MgO (Fig. 1). Temperature-programmed decomposition showed that the decarbonylation in helium took place at about 300°C, and this temperature was therefore used for a vacuum treatment, followed by a treatment in hydrogen to remove carbonaceous fragments formed on the cluster surface by reaction of the carbonyl ligands. Characterization of the sample before, during, and after the treatments by extended X-ray absorption fine structure (EXAFS) spectroscopy gave evidence that the framework nuclearity had been maintained. The supported decaosmium is believed to be the first supported metal with a unique structure that has been denuded of its carbonyl ligands. It is probably not accurate, however, to refer to the cluster as denuded, since it was likely converted into a hydride cluster during the treatment--but the removal of the CO activated the cluster for hydrocarbon conversion catalysis, as discussed below. Ensembles of supported metal complexes. The process of cluster formation by reductive carbonylation can be reversed. Oxidative fragmentation converts metal clusters and aggregates into mononuclear metal complexes. Oxidative fragmentation of the alumina-supported triosmium clusters mentioned above has been suggested on the basis of electron microscopy to give ensembles of three mononuclear os (II) carbonyl complexes on the surface. 2 5 The bonding of these complexes to the surface has been characterized by EXAFS spectroscopy.26 A structural model is shown in Fig. 2. Each os2+ ion is coordinated to three oxygen ligands of the support, with an average metal-oxygen distance of 0.217 nm. A more thoroughly worked out example of this surface chemistry involves the oxidative fragmentation of MgO-supported [H2Re3(COl12l-, formed by deprotonation of [H3Re3(COl12l on the basic surface. 27,28 Again EXAFS analysis was done (in combination with infrared spectroscopy, temperature-programmed

131

Top View

Side View

Fig. 2. Structural model of osmium tricarbonyl complex on the (111) face of y-Al 2 0 3 The open circles represent oxygen, the full circles osmium, and the hatched circles CO.

decomposition, etc.) to determine the surface structure. The results showed that the fragmentation of the trirhenium clusters occurred to give ensembles, and the EXAFS data are in excellent agreement with a structural model consisting of three Re complexes on the MgO surface, with an average Re-Re near neighbor distance of 0.394 nm (Fig. 3). Again, the metal ion was coordinated to three oxygen ligands of the support, with an average Re-O distance of 0.215 nm.28 The broader importance of the oxidative fragmentation chemistry in metal catalysis is illustrated by the action of CO on a conventional, highly dispersed supported metal catalyst, Rh/y-Al 2 0 3 , prepared from a metal salt by impregnation followed by reduction. For years, researchers had debated the nature of this highly dispersed metal and the carbonyl species resulting

2.978

A

~

® MgO

(100l face

0

0 0

Re

10n

surface OH

0 atom

Mg atom

Fig. 3. Structural model of an ensemble of three Re carbonyl complexes on the (100) surface of MgO. The CO ligands (three per rhenium) are omitted for clarity.

132

from chemisorption of CO. What was not generally understood was that CO chemisorption, even at room temperature, can readily lead to the fragmentation and oxidation of very small supported rhodium aggregates. The noninnocence of CO as an adsorbate was demonstrated by Primet , 2 9 who monitored the reactions of CO with small alumina-supported rhodium aggregates using infrared spectroscopy. At temperatures above 100 K the Rh(I) geminal dicarbony1 species are formed at the expense of CO-covered Rh aggregates. These mononuclear species result from oxidative addition of surface hydroxyl groups to zerovalent metal centers. There appears to be a strong analogy to oxidative fragmentation of supported metal carbonyl clusters. The Rh(I) geminal dicarbonyl structures formed on the alumina surface as a result of fragmentation and oxidation of rhodium aggregates initiated by CO chemisorption have been characterized by infrared, X-ray photoelectron, and EXAFS spectroscopies30 and transmission electron microscopy.3 1 Each Rh atom is bonded on average to two CO ligands and three support oxygen atoms; a Rh oxidation state of +1 is consistent with the presence of one oxo and two hydroxo ligands. Robbins32 measured 13c NMR spectra and proposed that the Rh (I) (CO) 2 species remain clustered on the support after particle breakup; the process would then be analogous to the formation of ensembles by oxidative fragmentation of supported metal carbonyl clusters. Consistent with this view, transmission electron micrographs taken after exposure of highly dispersed Rh-on-alumina catalysts to CO still show 0.5-3.0 nm scattering centers.31 Summar~. Organometallic chemistry has been extended to surfaces, and a surface chemistry of metal carbonyl clusters is emerging. Most of the supported "molecular" metal clusters are fragile and limited in their potential catalytic applications, but the new materials include a small number of robust supported metal clusters, which are anions on basic supports, including MgO and basic zeolites. Stable surface species with novel, simple structures have also been formed by controlled fragmentation of metal clusters on supports to give ensembles of metal complexes having the same numbers of metal atoms as the cluster precursors.

SUPPORTED METALS DERIVED FROM METAL CLUSTERS: NEW CATALYSTS Supported "molecular" metal clusters The chemistry of osmium carbonyl anions on the basic MgO surface has been exploited to produce a family of surface-bound cluster anions that are stabilized by carbonyl and hydride ligands (Fig. 1) . The identities of these stabilizing ligands invite investigation of the catalytic properties of these species for CO hydrogenation. One of the MgO-supported metal clusters, [H30s4 (CO) 12l-, has been associated with the observed catalytic activity for this reaction, since catalysts made from different precursors, [H20s (CO) 4] 10 and [Os6 (CO) 1sl 2 -, 33 when brought to a temperature of 275°C in the presence of CO + H2, were converted largely into the supported tetraosmium cluster anion, as indicated by infrared spectroscopy.33 It is inferred that the tetranuclear cluster anion is a catalyst precursor, but the identity of the catalytically active species remains to be determined. There was no evidence of Os aggregates on the

133

catalyst surface, and experiments with Os aggregates formed from chloroosmic acid on alumina34 (which is not a strong enough base to stabilize the "molecular" clusters) were roughly two orders of magnitude more active than the MgO-supported osmium carbonyls (Table 1) . Neither this conventional catalyst nor the supported "molecular" catalyst had an unusual selectivity for the CO hydrogenation reaction (Table 1) . When the supported catalysts containing [H30S4(C0)1zl- were exposed to reactant gases with higher Hz: CO ratios, the tetranuclear clusters were converted into [OsloC(CO)z4lz-, and the activity was simultaneously lost. This result indicated that the decaosmium cluster anion is inactive for the CO hydrogenation reaction, and the suggestion was confirmed in experiments showing the lack of activity of a sample prepared by deposition of a salt of the decaosmium cluster anion on MgO. 33 The MgO-supported ruthenium carbonyls mentioned above have also been investigated as CO hydrogenation catalysts, 14 ,35 but there is spectroscopic evidence of aggregates of ruthenium metal on the catalyst surfaces in addition to ruthenium carbonyl clusters, and the identity of the catalytically active species is open to question.14,35 Encaged metal clusters The catalysts described in the preceding paragraphs are remarkable for their stability and the uniqueness of their structures, but not for their performance. An attempt was made to modify the selectivity of the supported osmium carbonyl catalyst by entrapping it in the molecular sieve support.Z3 The result was a catalyst having a high selectivity for formation of low-molecular-weight olefins from CO + Hz and a greater resistance to deactivation by metal sintering than has been observed for other metals in zeolites (Table 2) . Like the MgOsupported counterpart, it requires a sufficiently high ratio of CO to Hz in the reactor to prevent formation of Os aggregates, which sinter out of the zeolite pores and give an unselective catalyst. The encaged clusters are evidently a new kind of supported metal catalyst. Denuded metal clusters The activities of the supported "molecular" Os clusters mentioned in the preceding paragraphs are low in comparison with the activities of conventional supported Os aggregates, since the clusters are "poisoned" by the CO (and possibly H) ligands that are required to stabilize them.33 Catalysts that would be more useful because they would be more active could in prospect be formed by denuding a cluster of its ligands (except for the support surface) . Removal of these ligands without a change in nuclearity of the cluster would give another new kind of supported metal catalyst, unique because of its structural uniformity. This goal has been achieved after some years of effort in cluster, robust extraordinarily an with experiments [Os1oC (CO) Z4lz- supported on MgO, as described above. The original supported cluster anion was initially inactive as a catalyst for n-butane hydrogenolysis, but the activity slowly 134

(11

w

275 275 300

70C

<12d

y-Al203 MgO basic Y zeolite

H20sCl6

[H20s (CO) 4]

[H20s (CO) 4]

Os/y-Al203b

a

e

c d

b

Turnover frequency, s-1 Os aggregates on y-Al203 Determined by transmission electron microscopy Approximate diameter of zeolite supercage Only lower limit determined.

?

[H30S4(C0)12l-{Mg O}

T, °C

Avg. Os particle size,A

Catalyst support

Catalyst precursor

23 2 14 35 19 31 >0.04e

19

10

1 2 97

34 2 5 8

9

Ref.

74

1

70

104xTOFa

10

10

P,atm

Product distribution, mol% C1 C2 C3 C4 Cs

Performance of supported osmium catalysts for CO hydrogenation at low conversions.

Predominant form of catalyst

Table 1.

w en

LaY

NaY

Ru(NH3)6Cl3

[H20S (CO) 4] 300

252

254

T, °C

19

14

1

P,atm

Fe in toluene made from metal vapor.

NaY

Fe atomsa

a

Catalyst support

47

34

19

24

5

2

12

6

9

15

20

47

2

34

23

Product distribution, mol% C1 C2 C3 C4 C5

>480

-1

-1

Approximate catalyst lifetime, h

23

19

20

Ref.

Selectivities and stabilities of zeolite-supporte d CO hydrogenation catalysts.

Catalyst precursor

Table 2.

increased in a flow reactor, evidently as the CO ligands were displaced and reactant ligands gained access to the osmium. Alternatively, the catalyst was prepared as described above to remove the carbonyl ligands and form what was presumably an osmium hydride cluster on the support. 36 This was initially active for the hydrogenolysis reaction, confirming the role of CO ligands as catalyst poisons. The catalyst was approximately as active for n.-butane hydrogenolysis as a conventionally prepared supported osmium catalyst (made by aqueous impregnation with chloroosmic acid) . Most important, the nuclearity of 10 was retained during the catalytic reaction, as evidenced by EXAFS analysis of the used catalyst. Catalysts as uniform in structure as this offer unprecedented opportunities for fundamental understanding of structure sensitivity in metal catalysis. Dimers and pair sites The group of Iwasawa37 has pioneered in the synthesis and characterization of catalysts prepared from organometallic dimers designed to give pairs of atoms on metal oxide surfaces. The precursors include monomeric and dimeric complexes such as allyls of Cr, Mo, and Rh, and the supports are metal oxides. For example, Iwasawa et al.3B showed striking differences in the catalytic character between isolated silica-supported Mo complexes and pairs of Mo complexes for propylene oxidation. The pairs of Mo complexes on silica have a markedly higher selectivity for acrolein. Ensembles of mononuclear metal complexes Trinuclear ensembles of Re carbonyl complexes of MgO were prepared from [H3Re3 (CO) 12l, as described above. These are catalytically active for the structure-sensitive cyclopropane hydrogenolysis reaction, whereas a catalyst prepared from the mononuclear precursor [HRe(CO)s] (and presumably having isolated Re(I) centers on the surface) is inactive2B,39 (Table 3) . The two catalysts have nearly the same activity per Re atom for the structure-insensitive propylene hydrogenation reaction. These results demonstrate the need for neighboring metal centers for activation of the alkane and indicate a primitive design of surface catalytic sites allowing a discrimination between structure-sensitive and structure-insensitive reactions. Summary and prospects Metal cluster chemistry has stimulated active research in catalysis, all in about the preceding 10 years. The initial high hopes for efficient, subtle new catalysts have not been realized and likely will not be soon, if at all, because of the fragility of most metal clusters. For the same reason, almost all of the early attempts to translate metal cluster chemistry onto supports have been at best partially successful. But as researchers working with metal clusters on supports have learned how to control the chemistry, an extended family of "molecular" metal clusters on surfaces has emerged, and the

137

Table 3. MgO-supported rhenium catalysts: effects of surface site nuclearity on catalytic activity measured as turnover frequency, molecules/(Re atom•s) .28,29 CATALYST Catalytic reaction

Mononuclear Re complex on MgO, prepared from [HRe (CO) 5 ]

Propylene hydrogenation at 80°C Cyclopropane hydrogenolysis to give methane + ethane at 200°C

3.5 x 1o-2

Trinuclear Re ensemble on MgO prepared from [H 3 Re 3 (C0) 12 ] 2. 1 x 1o-2

4. 7

X

lQ-4

Re metal aggregates on MgO

-1o-1

-3

X

lQ-2

surface structures now include some robust metal cluster anions that are intriguing candidate catalysts, although not yet catalysts with remarkable activities or selectivities or catalysts that are stable except under restricted conditions (specifically, in the presence of CO + H2). Thus the outlook for useful new supported metal cluster catalysts is still not highly promising. On the other hand, metal clusters have been used to prepare new multinuclear surface sites that constitute new classes of catalysts and allow an unprecedented control over catalyst synthesis, adding a new legitimacy to the concept of catalyst design.40 There are almost untapped opportunities for preparation of new multicenter surface sites from metal cluster precursors, and there is every reason to believe that some may be highly stable structures. Even if the technological opportunities offered by metal clusters in catalysis have so far been disappointing, the scientific opportunities are grounds for optimism. The metal clusters are precursors used in organometallic syntheses to give precise ("molecular") surface structures. Exact determinations of these structures by techniques including EXAFS spectroscopy are now allowing determination of metal-support bonds in supported catalysts (placing the issues of metalsupport interactions on the firm foundation of chemical bonding and structure) and allowing preparation of supported metal clusters denuded of their ligands; families of these materials will allow quantitative characterization of the effects of structure sensitivity in metal catalysis for metal entities in the size range of greatest importance in technological catalysis.

138

We should imagine families of well-defined supported metal catalysts that can be characterized in a depth rivaling that attainable for molecular catalysts. From them may emerge advances in fundamental understanding of metal catalysis to rival those made possible some 15 years ago by the techniques of ultrahigh-vacuum surface science applied to single crystals of metal. ACKNOWLEDGMENTS This work was supported by grants from the National Science Foundation (CBT-8605699) and the Petroleum Research Fund, administered by the American Chemical Society. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

B. C. Gates, L. Guczi, and H. Kn6zinger, eds. "Metal Clusters in Catalysis," Elsevier, Amsterdam (1986). B. F. G. Johnson, ed., "Transition Metal Clusters," Wiley, Chichester (1980) . M. Moskovits, ed., "Metal Clusters," Wiley, New York (1986). B. C. Gates and H. H. Lamb, J. Mol. Catal., in press. J. P. Candy, 0. A. Ferretti, G. Mabilon, J. P. Bournonville, A. El Mansour, J. M. Basset, and G.Martino, J. Catal., .1.12. (1988) 210. M. Ichikawa, Polyhedron, 2 (1988) 2351. H. H. Lamb, B. C. Gates, and H. Kn6zinger, Angew. Chern. Int. Ed. Engl., £I (1988) 1127. J. M. Basset, A. Theolier, D. Commereuc, and Y. Chauvin, J. Organomet. Chern.,~ (1985) 147. G. Bergeret, P. Gallezot, P. Gelin, Y. Ben Taarit, F. Lefebvre, C. Naccache, and R. D. Shannon, J. Catal., l..Q.1 (1987) 279. H. H. Lamb and B. C. Gates, J. Am. Chern. Soc., 1na (1986) 821. H. H. Lamb, T. R. Krause, and B. C. Gates, J. Chern. Soc. Chern. Commun. (1986) 821. H. H. Lamb, A. S. Fung, P. A. Tooley, J. Puga, T. R. Krause, M. J. Kelley, and B. C. Gates, to be published. L. D'Ornelas, A. Theolier, A. Choplin, and J.-M. Basset, Inorg. Chern., 22 (1988) 1261. T. R. Krause, Ph.D. Thesis, University of Delaware, 1987. c. Dossi, R. Psaro, and R. Ugo, J. Organomet. Chern., ~ (1988) 259. P. Dufour, L. Huang, A. Choplin, R. Sanchez-Delgado, A. Theolier, and J.-M. Basset, J Organomet. Chern., ~ (1988) 243. J. L. Vidal and R. C. Schoening, Inorg. Chern., Zl (1982) 438. P. A. Jacobs in B. C. Gates, L. Guczi, and H. Kn6zinger, eds., "Metal Clusters in Catalysis," Elsevier, Amsterdam (1986) p. 357. H. H. Nijs, P. A. Jacobs, and J. V. Uytterhoeven, J. Chern. Soc. Chern. Commun. (1979) 180; 1095. L. F. Nazar, G. A. Ozin, F. Hugues, J. Godber, and D. Rancourt, J. Mol. Catal., Z1 (1983) 313.

139

P. Fejes, I. Kiricsi, I. Hannus, T. Tihanyi, and A. Kiss, in "Catalysis by Zeolites," B. Imelik et al., eds., Elsevier, Amsterdam (1980) p. 135. 22. L. R. M. Martens, P. J. Grobet, and P. A. Jacobs, Nature (London), ~ (1985) 568. 23. P. L. Zhou and B. C. Gates, J. Chern. Soc, Chern. Commun,, (1989) in press. 24. H. H. Lamb, M. Wolfer, and B. C. Gates, to be published. 25. J. Schwank, L. F. Allard, M. Deeba, and B. C. Gates, ~. ~., M (1983) 27. 26. F. B. M. Duivenvoorden, D. C. Koningsberger, Y. S. Uh, and B. C. Gates, J. Am· Chern. Soc., ~ (1986) 6254. 27. P. S. Kirlin, F. A. DeThomas, J. W. Bailey, H. S. Gold, C. Dybowski, and B. C. Gates, J. Phys Chern , ..9..Q. (1986) 4882. 28. P. S. Kirlin, F. B. M. van Zon, D. C. Koningsberger, and B. C. Gates, to be published. Chern Soc. Faraday Trans.I, 1.1 (1978) 2570. 29. M. Primet, J 30. H. F. J. van't Blik, J. B. A. D. van Zon, T. Huizinga, J. C. Vis, D. C. Koningsberger, and R. Prins, J. Am. Chern . .fulQ., 1.Q1. (1985) 3139. 31. D. J. C. Yates, L. L. Murrell, and E. B. Prestridge in J. Bourdon, ed., "Growth and Properties of Metal Clusters," Elsevier, Amsterdam (1980) p. 137. 32. J. L. Robbins, J. Phys. Chern., ..9..Q. (1986) 3381. 33. H. H. Lamb, T. R. Krause, and B. C. Gates, Proc 9th Int. Congr. Catal .• Calary. 1988, The Chemical Institute of Canada, Ottawa, 1988, Vol. 3, p. 1378. 34. E. 0. Odebunmi, B. A. Matrana, A. K. Datye, L. F. Allard, Jr., J. Schwank, W. H. Manogue, A. Hayman, J. H. Onuferko, H. Knozinger, and B. C. Gates, J. Catal,, ~ (1985) 370. 35. R. Pierantozzi, E. G. Valagene, A. F. Nordquist, and P. N. Dyer, J:, MQl, Ca:tal. , .2..l (1983) 189. 36. H. H. Lamb, M. Wolfer, and B. c. Gates, to be published. 37. Y. Iwasawa, AQy, !:;;;a:t;;a1., ll. (1987) 187. 38. Y. Iwasawa, N. Ito, H. Ishii, and H. Kuroda, J:, Ch~m. SQ~. Cbew CQIDID.l.lll. (1985) 827. 39. P. s. Kirlin and B. c. Gates, Nat.l.li:e ( LQilQQil) , .32..5. (1987) 38. 40. Y. Iwasawa and B. C. Gates, CHEMTECH, in press, March, 1989. 21.

140

MECHANISTIC FEATURES OF CARBONYL CLUSTER REARRANGEMENT

Brian F. G. Johnson,a Adrian Bott,a Robert E. Benfield,b Dario Braga,c Elisabeth A. Marseglia,d and Alison Rodgera a. University Chemical Laboratory, Lensfield Road, Cambridge, CB2 lEW, England b. Department of Chemistry, University of Kent, Canterbury, CT2 7NH, England c. Dipartimento di Chimica, 'G. Ciamician•, Universita di Bologna, via F. Selmi, 2, 40126, Bologna, Italy d. Cavendish Laboratory, Madingley Road, Cambridge, CB3 OHE, England

ABSTRACT A new scheme, which considers the sequential edge cleavage of ligand polyhedra, has been employed to explain the phenomenon of pseudorotation in simple homoleptic coordination complexes. The fluxional behaviour shown by cluster carbonyls, with special reference to Co2(COla, Fe3(COl12• and Co4(COl12• in the solid state can be explained in terms of small librations involving both the encapsulated metal polyhedron and the carbonyl envelope, and in solution by a combination of the same librational motions and concerted pseudorotation of the carbonyl polyhedron.

INTRODUCTION The stereochemical changes - fluxional and isomerisation - that many transition-metal carbonyl compounds exhibit have become amongst the most widely studied phenomena in inorganic chemistry and although the mechanisms of some may be conveniently explained in terms of reversible intermolecular carbonyl site exchange apart from simple ideas of localised rotations of terminal-bridge interconversion no simple, coherent approach applicable to all examples appears to have evolved. In this paper we wish to discuss such an approach which we believe offers a unified view of these fascinating phenomena. It is reasonable to claim that reactions in this category have become one of the most fascinating areas of study during the past decade. Although the phenomenon had been appreciated for some time the detailed examination of pseudorotation had to wait on the advent of a suitable analytical tool and nuclear magnetic resonance proved ideal for this purpose. There were also early indications that X-ray analysis would provide support and complementary data for molecular rearrangements along well defined pathways. In this case evidence derived for one molecular state (the solid) was initially employed to explain a phenomenon in another (solution) . More recently Magic-Angle Spinning (MAS) NMR

Metal-Meta/ Bonds and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

141

spectrometry has been used to make detailed studies of molecular motions in the solid, and in an earlier communication, we commented (1) on the way in which information derived by this method could be used in conjunction with information from the X-ray experiment to explain the motion in the solid. In this paper we shall examine possible complementary geometries for both mononuclear coordination compounds MLn and carbonyl clusters and offer a new, more convenient method of understanding these phenomena. MOLECULAR POLYHEDRA Before we consider some of the problems associated with coordination geometry it is essential that we recall some of the more important aspects of polyhedral geometries. The most symmetric of all polyhedra are the regular polyhedra of Plato. These are the tetrahedron (a), the octahedron (b), the hexahedron (or cube) (c), the icosahedron (d), and the dodecahedron (e) which are illustrated in Figure 1. These are the only convex polyhedra that are bounded by identical regular polygonal faces in which all vertices are equivalent. Of these polyhedra only three, viz., the tetrahedron, the octahedron, and the icosahedron, are fully triangulated and therefore stable. In contrast, the cube which has all square faces and twelve edges is unstable and will collapse (or contract) to the stable fully triangulated dodecahedron with eighteen edges. Similarly, the dodecahedron with all pentagonal faces and thirty edges is also unstable and will collapse to the stable fully triangulated 25-hedron with fifty-four edges. Polyhedral stability does not therefore necessarily correlate with high symmetry. Rather stability is associated with triangulation; the most stable polyhedra being those with all triangular faces. According to this view the most stable polyhedra for 4-to 12-hedra are those given in Figure 2. These are the polyhedra often observed in coordination chemistry and in the boranes and carboranes. We shall refer to them by their number of vertices, e.g. l4t, (5), (6}, etc. These polyhedra provide us with a prototypical array of geometric structures that reveals a useful and fascinating insight into chemical polyhedra. However, it is useful for our purposes to examine the relationship these polyhedra bear one to another. Consider the tetrahedron, [4J. Breakage of any one edge leads to a butterfly arrangement which, provided the extension is sufficiently large, corresponds to a trigonal bipyramid with one equatorial vertex removed we shall therefore refer to this as (S-1) . Hence, if we cap this butterfly we produce the trigonal bipyramid {SJ (see Figure 3) . Similarly, removal of one edge from the octahedron leads to the {7-1) polyhedron, and if we cap this polyhedral form the bicapped pentagonal bipyramid (7f (Figure 3) will result. In the same vein we may progress from {71 through (8-1J to (81, from (8} to (9-1} to (9), and so on until we reach (101 (see Figures 3-5) . If we consider these polyhedra to be constructed of equal spheres (or ligands) then progress from say the {41 to {5-1f to [SJ corresponds to an increase in size of the interstitial site. The site in the square based pyramid {6-11 being larger than that in the trigonal bipyramid (st . For coordination polyhedra the same progression would require a steady and systematic increase in the metal-ligand bond lengths. It is important to recognise that of the polyhedra given in Figure 2 only the tetrahedron, the octahedron and the icosahedron are such that their vertices are constrained to lie on the surface of a sphere. Putting this in chemical terms, in coordination polyhedra all M-L bond lengths are equal only in these three arrangements in Figure 2; all other

142

0~

c

b

e

d

Figure 1: The Platonic Solids

~

G/' 0;_\ f

The convex de~ahedra: Tetrahedron (4-hedron) Triangular dipyramid (6-hedron) Octahedron (8-hedron) Pentagonal dipyramid (1()-hedron) 12-lledron 14-lledron g 16-hedron h Icosahedron (20-hedron)

a b c d e f

g

h

Figure 2: The Stable Convex Deltahedra

Figure 3: Polyhedral Growth Sequence 1

143

polyhedra produce unequal bond lengths. For this reason alternative polyhedra are !Sf, {71, lBf, {91 and t10! in which all M-L bond lengths are equivalent have been sought. We shall not consider this aspect in detail but it is worth noting that a complete range of polyhedral forms in which the vertices lie on the surface of sphere have been derived. Thus, if we consider eight coordination the square antiprism {10-2) is an example of a polyhedron in which all M-L bonds are equal but which is not the most stable form. In coordination geometry we are frequently presented with the question of whether or not the coordination geometry approximates to one in which all L-L contacts are equal but the M-L bonds are not, or an alternative in which all M-L contacts are equal but longer. There is apparently no hard and fast answer to this question and examples of both occur. It should also be noted that there is a discontinuity in this polyhedral growth sequence (Figures 3-5) at twelve. According to the sequence developed above the polyhedron with twelve vertices corresponds to the omni-capped cuneane arrangement and not the icosahedron, although the two polyhedra are related by a simple edge transfer. The sequence of polyhedra for higher vertex numbers will be discussed in a subsequent paper (2). We shall now consider the effect of breaking two or more edges. Edge cleavage with retention of the maximum number of edges, will lead to the results shown in Figure 6 for the octahedron, and in Figure 7 for the triangulated dodecahedron. For {6f, the successive cleavage of appropriate edges leads first to (7-1f, then to rs-2), and finally to {9-3] - the trigonal prism:

£6J ~

{7-1}

~

{B-2)

~

(9-3}

The corresponding treatment of

rat _.

{9-11

~

110-21

~

(Bl

rn-3}

(Figure 7) leads to the sequence: ~

£12-4J

~

(13-51

~

l14-6J

In this sequence l9-1J is the bicapped trigonal prism, ~10-2) the square antiprism, and (14-6) the cube. There are no such simple names for the other members of this series. For (9t we find:

{91 ~ {10-1} ~

(11-21 ~ {.12-3} ~ [13-4~ ~ [14-5}

and for (101

flOf ~ {11-1J ~ {.12-2} ~ (13-3\ ~ {14-4} In each case the systematic and steady progression across the series leads to increasingly open polyhedra (i.e. with some square rather than all triangular faces) and a controlled increase in the M-L bond length. A decrease in stability will also follow the same progression. This will be important in our consideration of pseudorotation of coordination compounds.

PSEUDOROTATION The act of pseudorotation as it related to simple coordination compounds MLn requires that certain bond angles and bond distances in the parent compound change until the coordination shell of ligands, L, assumes its complementary geometry in what would constitute either the

144

Figure 4: Polyhedral Growth Sequence 2

Figure 5: Polyhedral Growth Sequence 3

{6}

{ 7-1} (1)

{8-2} (2)

{9-3} (3)

Figure 6: Multi-edge Cleavage for 6-coordination

145

®

___..

{1 0-2 }

{9-1 }

{8}

{ 13-5}

r:i1lJ ~

..-

'---...

{12-4}

-

\

{11-3}

@ { 14-6 }

Figure 7: Multi-edge Cleavage for 8-coordination

Figure 8: Berry Pseudorotation Mechanism for 5-coordination

146

transition state or the intermediate. The ease with which these processes can occur will depend upon the energy difference between the complementary geometries (the ground state or the transition state) which constitutes the barrier to the process. These in turn depend on the relative movements of atoms or groups of atoms within the parent. When the complementary geometries are energetically close, either because of the nature of the ligands or because of the inherent properties of certain coordination numbers, the barrier to pseudorotation will be small. In many cases the barrier will be increased by the need of the system to traverse an intermediate in which the ligand distribution or unfavourable bond angles and/or unfavourable bond distances lead to instability. The best examples of pseudorotation and stereochemical non-rigidity will be found for those complexes in which the barrier between complementary geometries is significantly small than the free energy of activation for bond fission. The problem usually is on deciding upon the appropriate, accessible, complementary geometry. For five coordination the complementary geometries - the trigonal bipyramid, (51, and the square pyramid, (6-lf, are energetically close to one another. Their relationship is clear to see. Expansion of one of the three equatorial edges of the trigonal bipyramid leads directly to the square pyramidal complementary geometry, and on further expansion along the same vector to the new trigonal bipyramid, (5J, (Figure B). This expansion corresponds to a low energy or soft vibrational mode of the trigonal bipyramid and does not demand an excessive extension of the M-L bonds. As a consequence, the tendency to undergo pseudorotation via the Berry twist mechanism is quite considerable as is observed for Fe(C0) 5 . This mechanism also provides an example of the simple, single diamond-square-diamond rearrangement process. In contrast, the octahedron is stereochemically extremely rigid. According to the Bailar Twist mechanism the most likely complementary geometry for pseudorotation is the trigonal prism and a trigonal twist of 60° about one of the octahedral threefold axes will convert an octahedron into a trigonal prism and vice versa. The barrier to this pseudorotation is generally regarded as being similar to or greater than the activation energy for bond fission. The high barrier to pseudorotation in octahedral systems is thus easy to understand. First, it requires a considerable movement of ligands L, second, there is a significant decrease in the number of L-L contacts, finally, and most importantly, the conversion to the trigonal prismatic intermediate requires a substantial increase in the M-L bond lengths. In the trigonal bipyramidal case we have an example of a relatively small movement of atoms or ligands leading to pseudorotation, and in the octahedron an example of a large movement leading to a relatively major reorganisation of the molecular geometry. Clearly, the extent of these motions is dependent on the magnitude of the M-L bond strength (and to a much less extent on the L-L interactions), but most importantly on the physical state of the sample. First and foremost the complementary geometry must, obviously, be available by a low energy process corresponding to a soft normal mode of the molecule. Secondly, and very importantly, the route to the proposed complementary geometry must not involve any unreasonable extension of the M-L bond length. Finally, the process should sustain as many L-L contacts as possible. It is important to recognise that although two polyhedra may be easily converted one to another the need to maintain reasonable M-L bond distances throughout the process is of paramount importance. It is also clear that, given the different constraints imposed on the molecule in either the solid or solution barriers will vary widely from phase to phase.

147

In the past it has proved to be relatively more difficult to derive satisfactory complementary geometries for coordination numbers other than 4, 5 and 6. For larger coordination numbers, e.g. seven or eight, the availability of several closely related polyhedral forms has usually been taken as sufficient to explain the generally observed low barriers to fluxionality and no general scheme appears to exist. In this paper we wish to present such a general scheme which we believe offers a convenient approach to these fascinating phenomena and permits a simple view of suitable complementary geometries. As we described above for five coordination the complementary geometries - the trigonal bipyramid (Sf and the square pyramid f6-lf are energetically close to one another, and so the essential, low energy pseudorotation process may be viewed as involving two steps:

(sl ~

{6-11 ~

{s! •

Of course we must not forget that there are three, equivalent, equatorial edges within the trigonal bipyramid and that the soft mode under consideration will lead to cleavage of any one of them. Hence, at any given moment three 'different' {6-1] polyhedra may be produced from a given sample of ML 5 in the ground state although of course for a homoleptic MLs complex they are indistinguishable. For the octahedron which is stereochemically extremely rigid, the Bailar Twist is usually considered to occur in a one step process as outlined in Figure 9. However, here we will choose, for convenience, to consider the process as occurring in three distinct steps according to Figure 6. Cleavage of one octahedral edge, which corresponds to a soft vibrational mode, generates intermediate (1), of two edges intermediate (2), and of three edges intermediate (3). Examination of these three species (1), (2) and (3) reveals that (1) corresponds to a nido-pentagonal bipyramid {7-lf, that (2) corresponds to an arachno-dodecahedron l8-2f; and finally (3) to a hypo-tricapped trigonal prism {9-34. Thus, the Bailar Twist mechanism could be viewed as a six-step process: [6f ...., (7-1) _. [B-2f -

{9-31 -+ [B-21 ' ...._ l7-1J' -

[6)'

The high barrier to pseudorotation in octahedral systems is thus more easily understood. It is interesting to observe that the formation of (7-lJ from (61 corresponds to a opening of one polyhedral edge and is closely related to the process (5f~(6-lf. Hence, the energies for the two processes might be expected to be similar. However, the intermediate (7-lJ does not collapse to a new t6l and ligand exchange does occur. It is also worth noting that any compound possessing an f7-ll geometry might be expected to show a lower barrier to interconversion since it has only two steps to the (9-31 complementary geometry. This is found to be the case for XeF6. In this case rearrangement presumably occurs via the reaction sequence: (7-lt- (B-2t-+ {9l -

fB-20 ' ... (7-lf

This illustrates what we believe to be a general phenomenon, viz. pseudorotation always proceeds via polyhedral expansion i.e. M-L stretching. Thus, for a compound with a square pyramidal (6-lf ground-state, pseudorotation is expected to take place via a (7-2J intermediate and for a compound with a {9-lf ground-state through a (10-21 intermediate, etc. For seven coordination there is no single-step suitable complementary

148

geometry available. Single edge cleavage, as illustrated in Figure 4 will lead to a nido-dodecahedron (8-1} which cannot serve as a complementary geometry since operation of the diamond-square-diamond process in this instance will not return us to the parent polyhedral form (7J . Instead the capped-octahedron will be produced. This may be regarded as unfavourable for a homoleptic ML7 complex since one M-L bond (viz. that involving a bond to the capping ligand) is required to undergo a relatively large extension. A further edge cleavage is required to give the more favourable arachno-tricapped prism, {9-2}, which can serve as an appropriate complementary geometry. Hence, ligand rearrangement in this case may be viewed as the five-step process:

l71 ---)

(8-1) ---) (9-2f ---) (8-11' ---)

171'

Again the edge cleavage corresponds to a soft vibrational mode, and it is important to remember that although we are choosing to consider the process as occurring in two independent steps it will in reality occur in one concerted motion. This may be visualised as taking place via the simultaneous opening of two edges (joining two vertices of connectivity 4 and 5, and there are ten such edges). For eight coordination we have a situation similar to that found for coordination number five. The parent polyhedron is the dodecahedron (8) . Single edge cleavage takes us to the bicapped trigonal prism or (9-11 which can clearly serve as a suitable complementary geometry since further extension along the same vector leads to a new dodecahedron (8}'. In this case therefore the rearrangement process may be viewed as:(81 ---) {9-1} ---) (8}

I

Coordination number nine is somewhat similar. Here again single edge cleavage leads us directly to a suitable complementary geometry, in this instance a nido-bicapped square antiprism - {10-ll (Figure 5) and pseudorotation may occur via the process: (91 ---) {10-1l ---) (91' The general scheme shown in Figure 10 for pseudorotation in coordination compounds thus emerges. It may be extended to higher (>9) or lower (>5) coordination numbers although care must be exercised when dealing with numbers > 11. This aspect will be the subject of a further publication (2) .

FLUXIONAL BEHAVIOUR AND ISOMERISATIONS IN CLUSTER CARBONYLS Rearrangement processes in metal carbonyl clusters are considerably more complicated than in simple coordination compounds. First, they involve the reversible intramolecular site exchange of CO ligands bonded to the metal cluster framework, secondly, to the structural reorganisation within the metal framework and thirdly to the rearrangement of the metal polytopal or polyhedral unit within the carbonyl polyhedron. These three processes are not independent. Until recently however, fluctional behaviour of cluster carbonyls was regarded as only involving physical movement of the CO-ligands about the metal cluster unit which, in general, was taken to be rigid. Here we shall be concerned with fluxionality in both the solution and solid phases. We shall demonstrate that the arguments presented above for the simpler mononuclear coordination compounds may be extended to cluster fluxionality - at least for samples in solution - but that other

149

Figure 9: Bailar Twist Mechanism for 6-coordinate Complexes

COMPLIMENTARY GEOMETRIES {4}

{5-1}

<6-2>

{5}

{6-1> {6}

{7-1>

{8-2}

{9-3}

{7}

{8-1}

{9-2}

<8>

<9-D {9)

Note:

{6-2}

{9} {9-1} {9-2}

<9-3}

square tricapped trigonal prism

= bicapped

trigonal prism

nonocapped trigonal prism

= trigonal

prism

Figure 10: General Scheme for Pseudorotation

150

<10-1}

important factors such as the relative motion of the metal cluster unit within the ligand polyhedron must also be considered. The structures of a large number of cluster carbonyls have been determined by X-ray analysis. These are normally taken to be the ground-state structure of the molecule. Within these structures the CO ligand may adopt one of several of bonding modes. It can act as a terminal group bonded to a single metal atom or function as a bridge spanning two or three metal atoms. These bridges are not always symmetrical and varying degrees of distortion have been recorded. However, one of the more fascinating features to emerge from these structures has been the occurrence of different structural types in apparently closely related compounds. For example, for the series of dimeric carbonyls of Co, Fe, and Mn, Co2(COls has a structure in the solid with two CO bridges, Fe2(C0) 9 a structure three CO bridges and Mn 2 (COl 1 o a structure with no bridges (see Figure 11). For the trimeric clusters Fe 3 (CO)l2 has a molecular structure of Cs symmetry involving two CO bridges spanning one Fe-Fe edge, whereas the structures of Ru3(COl 1 2 and os 3 (C0) 12 have D3 h symmetry with all CO groups terminal (Figure 12). Similarly co 4 (CO)l2 and Rh 4 (CO)l2 have molecular structures of C3v symmetry with both edge bridging and terminal groups but Ir 4 (CO)l2 has a Td structure and all CO groups terminal (Figure 13) . A less dramatic but nevertheless important structural difference is found for the two closely related hexametal carbonyls Rh6(CO)l6 and Ir6(CO)l6· These have very similar structures; each possessing an Oh-arrangement of metal atoms each with two terminally bound CO groups but differ in the arrangement of the four remaining carbonyls. For rhodium these occupy four tetrahedrally disposed face-bridging sites whereas for iridium these have shifted to four tetrahedrally disposed edge-bridging sites. It was observations of this kind which led to the development of the 'CO-polyhedral model'. A number of researchers working in the area had commented on the fact that the CO ligands in the binary carbonyls tended to occupy positions which to a fair approximation defined the vertices of regular or semi-regular polyhedra. In 1966 Dahl and Blount (3) observed a nido-icosahedron of CO groups in the anion [Fe3(C0) 11 HJ- and concluded that the missing vertex was occupied by the H-ligand. On the basis of this conclusion Dahl (4) was able to correctly deduce the molecular structure of Fe3(COJ 1 2 at a time when crystal disorder prevented a detailed analysis. In 1975, when many other molecular structures of the binary carbonyls had been established we (5) proposed that these CO-polyhedral geometries might be of more fundamental importance. Close examination of a wide range of CO-arrangements led to the conclusion that the CO polyhedron was determined by purely steric factors. In Fe 3 (C0) 12 , for example, the carbonyls define the minimum energy polyhedron, viz. the icosahedron. The bridged-structure of c 5 -symmetry then arises as the simple geometric consequence of placing a triangle within this icosahedron (Figure 14) . In contrast, on the assumption that the CO has a effective radius of 302pm (derived empirically), it was deduced that the Ru 3 and Os3 units are too large to be accommodated satisfactorily in the same polyhedron. Hence, the clusters Ru3(COl12 and Os3(COl12 possess the slightly less favourable anticuboctahedral packing of carbonyls in which there is a larger interstitial site (Figure 15) . So far we have been concerned only with structure in the solid state. In solution structures are sometimes different and alternative isomeric forms may exist. A particularly good example is Co2(CO)g. In the solid, this molecule has a well established structure with two bridging carbonyls. In solution, however, it has been shown, by IR spectroscopy, that several other isomeric forms, some without bridges also exist. Similar observations have been recorded for Fe3(C0) 1 2 which also exhibits

151

Figure 11: Molecular Structures of (a) Co2(CO)a (b) Fe 2 (CO)g and (c) Mn2(C0) 1 o

152

(b)

(a)

Figure 12: The Molecular Structures of (a) Fe3(COl12 and (b) Os3(COl12

Cube- Octahedron.

3: 6: (3)

Figure 13: The Molecular Structure of Ir4(C0) 1 2

153

an IR spectrum that is markedly solvent and temperature dependent These and other suggesting the existence of alternative· isomeric forms. similar observations have led to the view that for many carbonyl clusters interconvert through relatively minor deformations and it is clear that for a given carbonyl a number of structures exist with energies similar to that of the ground-state (solid-state) and which can be excited by thermal energy. As a consequence a classical (localised) bond description may not be appropriate since the ground-state structures does not sit in well defined potential in the potential energy surface The existence of these readily excited states governing nuclear motion. leads to the dynamic phenomenon of fluxionality - the subject of this paper. In this part of the paper we shall deal with the nature of these isomeric forms, and with fluxionality in both the solid and solution. Three carbonyl prototypes will be considered, viz. Co2(C0)9, Fe3(C0) 1 2 and Co4(C0)12• but similar arguments may be extended to all other systems. Co2(C0) 8 Structure and Isomers In the alternative polyhedral description of the structure of this molecule in the solid the CO arrangement may be considered to be the bicapped trigonal prism, (9-1}. For eight CO ligands we can also consider other likely, closely-related polyhedra viz. the dodecahedron {8), and the square antiprism (10-21. The relationship between these three polyhedra is clear to see. The parent or lowest energy polyhedron is the dodecahedron {81, single edge-cleavage takes us to the bicapped trigonal prism (9-1] and cleavage of a second appropriate edge brings us to the square antiprism (10-2). As we progress across the series: -) (10-21 the polyhedra become increasingly more open and -) (9-ll (81 the interstitial site more spacious (Figure 16) . We can thus immediately visualise three isomeric forms of Co2(CO)a corresponding to the insertion of the linear Co-Co unit into each of these three possible polyhedra, viz. lBl, (9-1}, and tl0-21 (or even more if we consider even more spacious polyhedral forms such as the cube {14-61. These are shown in Figure 16. However, other isomers also based on these same CO-polyhedra are also possible. The linear Co-Co unit may adopt at least two different orientations within each polyhedron as shown in Figure 16. For example, the two possibilities for the bicapped trigonal prism are shown in 2a and 2b. Form 2a corresponds to the observed geometry in the solid state and described above in which there are three terminal CO's per Co atom and two CO-bridges; the two bridging In contrast, 2b, CO's corresponding to the two polyhedral caps in l9-l). which be derived from 2a by a small rotation of the Co2 vector, corresponds to a structure contains only terminally bonded CO's, four per cobalt. A similar result is found in examples la and lb although the arrangement of carbonyls is (81, and in 3a and 3b where the CO-polyhedron is [10-21 . Fluxional Behaviour Most, if not all, eight coordinate structures studied in solution There is very little energy have been found to be highly fluxional. difference between the various possible polyhedra, viz. (8J, [9-ll or (10-21, and Co2(Co) 8 should show a fluctuating structure in solution and possibly a permanent distortion from C3v symmetry in the solid.

154

-

-

Figure 14: Molecular Structure of Fe3(COl12 According to the Polyhedral Model

M3(C0)12

Anti -Cube- Octahedron.

M = Ru or Os

Figure 15: Molecular Structure of M3(C0) 1 2 According to the Polyhedral Model

155

The nature of these fluctuating structures in the solid is also easily visualised. Consider form 2a. Rotation of the Co-Co vector about its two-fold axis - and this will coincide with the Co-Co vibration will initially lead to the formation of two asymmetric bridges and eventually to the non-bridged form 2b. Hence, to the apparent equilibration of bridging and terminal groups. In reality, there will be an oscillation or libration of the CO-Co unit about this two-fold axis and at no point do ligands actually migrate from one metal atom to the other at this stage of the fluxional process. However, this libration will also be expected to perturb the CO polyhedron about the same molecular c 2 axis and it would not seem unreasonable to associate this with the distortion of the bicapped trigonal prism {9-1J towards the square antiprism {10-2) . Eventually, as the temperature is increased, and the amplitude of the libration becomes sufficiently large a new isomer with the square antiprismatic geometry will be reached. At this stage intermolecular CO ligand migration could then occur via the sequence of pseudorotation processes for eight-coordination as outlined above, i.e.

leading to CO scrambling. Thus, we might expect the overall complementary geometries in this case to correspond to those shown in Figure 16 and which are formed by a combination of the libration of the Co2 vector and the pseudorotation of the CO polyhedron (as with a mononuclear species) . This sequence of motions can easily be seen to occur in solution and in the solid since it involves a relatively small movement of all atoms or groups. But in the solid, because of the restraints imposed by the lattice, conversion from the bicapped trigonal prism to the square antiprism is not expected to be a high energy process. It is also important to recognise that such a process is more easily allowed in the solid because of 'hole' in the {9-1J polyhedron, i.e. it is a low energy process requiring the minimum movement of the CO ligands. This will not always be the case. Fe 3 (C0) 12 Structure and Isomers X-ray studies have shown that Fe3(COl12 shows orientational disorder in the solid state. In the molecular structure two carbonyls are in asymmetric than the other two, all other carbonyls are terminally bonded. The molecule possesses C1 symmetry but is very close to c2 with a pseudo-two-fold axis passing through the middle of the bridged Fe-Fe vector and the opposite Fe atom. Random occupancy of one or other of two possible orientations in the crystal lattice results in the presence of an inversion centre within the unit cell statistically relating two crystallographic 'half molecules' in each of the two orientations. According to our polyhedral model the iron triangle occupies the interstitial site within an idealised but slightly flatened icosahedron, the optimum geometry for twelve CO ligands. In an argon matrix at 20K the carbonyl has an I.R. spectrum reasonably consistent with this geometry. There is a weak band at 2110 cm-1, weak and medium intensity bands at ca. 1870 and 1830 cm-1 and six medium to strong bands between 2015 cm- 1 and 2060 cm-1. Inn-hexane there is a weak band at 2103 cm-1, four bands of variable intensity in the region 2000 - 2050 cm-1 but only weak broad bands at ca. 1867 and 1838 cm-1. This change in IR was the first indication of the presence of different isomeric forms. Several possibilities exist. The simplest, according to the polyhedral mode, corresponds to a reorientation of the iron triangle within the 156

Ia

a

a

b

Bicapped trigonal prism 2

b Square antiprism 3

a

b

Figure 16: Isomeric Forms of Co2(COls According to the Polyhedral Model

[2]

[3]

Figure 17: Molecular Structure of Co4(COl12 According to the Polyhedral Model

157

icosahedron of CO groups. Rotation of this triangle by 15° about the molecular C2 axis which passes through the unique iron atom Fe(l) produces a second isomeric form of Fe3(COl12 with D3-symmetry (Figure 14) . This is also based on an icosahedron of CO groups but in contrast to the solid state structure all CO groups are terminally bonded and all Fe-Fe distances are the same (260 pm) . This rotation of the iron triangle will also cause the slightly flattened icosahedron of carbonyls to become regular. Other isomers are possible. The most obvious being based on the less stable anticuboctahedron of carbonyls with a pseudo-Os3(COl12 structure (see e.g. ref. 9). Other possibilities, e.g. one with a bicapped right pentagonal prismatic CO arrangement exist, but as mentioned above the relationships between twelve-vertex polyhedra are not so easily understood, and these will be discussed more fully in a subsequent publication. Fluxional Behayiour

Gansow et al (6) and Cotton (7) have shown conclusively that Fe3(COl12 undergoes rapid CO-equilibration in solution and Cotton and Troup have predicted and verified (8) that the activation energy for this equilibration process is <5 kcal mol-l, thus excluding the possibility that the predominant solution structure may be obtained by recording a 13c n.m.r. spectrum at the slow exchange limit. These measurements al~o confirm that an easily accessible complimentary geometry is available to this cluster. However, variable temperature MAS 13c n.m.r. solid measurements on Fe3(COl 1 2 show that at temperatures below -95°C, the observed spectrum is consistent with the crystal structure indicating two bridging and ten terminal carbonyls. At 24·C, there are six resonances of similar intensities (2:2:2:2:2), and none of the observe chemical shift values are consistent with bridging or semi-bridging carbonyls. A librational motion of the iron triangle of a few degrees about the molecular two-fold axis, as described above, will first convert the symmetry from C1 to C2, and is sufficient to eventually change the molecular symmetry from C2 to (pseudo) D3 in which all the CO ligands are terminally bound. Since those carbonyls that adopt a bridging mode in the C2 conformation will adopt a terminal bond in the D3 conformation, the room temperature spectrum is easily understood (9). Furthermore, at low temperatures (-95°C) the amplitude of this libration is considerably reduced, and hence, the MAS 13c spectrum observed is consistent with the crystal structure. Further, more extensive libration of the iron triangle about any one of the now equivalent molecular two-fold axes would lead eventually to the complete, but apparent equilibration of all CO ligands. However, it must be remembered that although the CO-polyhedron may undergo a slight perturbation towards say a cubeoctahedron during this librational motion at no stage will the CO ligands undergo intermolecular exchange. This behaviour is in contrast to that in solution where complete CO equilibration can be achieved by a combination of such librational modes and the pseudorotation of the CO-icosahedron possibly through a cubeoctahedral intermediate. Co 4 (C0) 12 Structure and Isomers The structure of Co4(COl12 has also been studied independently by different groups (10) and has been shown to be disordered in the solid state. Random occupancy of one or other of two possible molecular orientations in the crystal lattice results in the presence of a two-fold axis within the unit cell which passes through one basal Co atom and the

158

centre of the opposite triangular face. The molecule itself has only C1 symmetry with three carbonyl groups in edge-bridging positions around a tetrahedral face, the remaining carbonyls being terminally bound. As with Fe 3 (C0) 12 the CO-polyhedron is again approximately icosahedral, and the molecular symmetry is only slightly distorted from C3v (Figure 17). The IR spectrum at ambient temperature of both solid and solution samples is consistent with this symmetry and the same structure is thought to persist in both phases. Nevertheless, we find that the spectrum does change with temperature and evidence for a new isomeric form has been (11). Within the icosahedron of carbonyl ligands at least two other orientations of the tetrahedron are possible (Figure 17). (12). Rotation of the Co 4 tetrahedron about the unique C3 molecular-axis which passes through the apical Co atom and the centre of the basal co 3 face produces a new form (2) which is also based on the icosahedral of carbonyl groups but in contrast to isomer (1) all CO groups are equivalent and terminal. This isomer has Td symmetry. alternatively, rotation of the co 4 tetrahedron about any one of the three C3 molecular axes which pass through a basal Co atom and the opposite Co3 face produces a third isomeric form which also contains CO-bridges (3) (12). The high temperature IR indicates a change in those frequencies associated with the CO-bridging region and is consistent with D2 d symmetry. Hence, isomer (3) is the preferred. Fluxional Behaviour Definitive evidence for the structure of Co 4 (C0) 1 2 in solution has been difficult to obtain, but it is now generally accepted that the same c 3 v structure persists, at least at low temperatures. At higher temperatures the compound is fluxional. MAS 13c n.m.r. spectra for Co 4 (C0) 12 have been recorded over a wide temperature range and above 62°C only one broad signal can be seen indicating total equilibration of all carbonyls. Interconversion of the C3v structure and a higher energy complementary geometry of T symmetry produced by libration of the Co 4 tetrahedron about the molecular c 3 axis would provide a rationale of this behaviour (14). However, this does not tie in well with evidence of molecular motion coming from X-ray studies. As an alternative we suggest that a librational motion of the co 4 tetrahedron of a few degrees about one of the c 3 • axes would be sufficient to change the molecular symmetry from c 3 v to D2 d, in which although the icosahedral arrangement of CO ligands is preserved a new bonding arrangement is found. In agreement, the mean square displacement parameters reported for Co 4 (C0) 1 2 indicate a preferred librational motion of the co 4 tetrahedron about this pseudo-three-fold-axis. As with Fe3(C0) 12 additional motion is possible in solution. The postulated librational modes in themselves will bring about apparent CO equilibration without CO intermolecular exchange (1) will be accompanied by the additional possibility of pseudorotation of the CO-icosahedron possibly through a cubeoctahedral transition state or intermediate of Td symmetry similar to Ir4(COl12• and of the type we have described elsewhere (13). Other possible intermediates or transition states are possible and will be discussed in a future publication (2) .

CONCLUSIONS We have shown that the observed or postulated pseudorotation in simple coordination compounds may be understood in terms of simple polyhedral edge cleavage mechanism. The main advantage of this mechanism being the ease with which the complementary geometries for a range of coordination numbers may be derived and the wide range of energy barriers to rearrangement processes understood. Of particular importance is the need to limit the extent of M-L bond extension and sustain as many as

159

possible L-L contacts. We have also shown that the observed solid state n.m.r. data for cluster carbonyls can be rationalised in terms of low energy librations of the metal polyhedron within the CO-polyhedron, rather than the higher energy CO intermolecular migrations previously postulated. The m.s.d. parameters derived from X-ray data provide evidence for such librations; the thermal ellipsoids of the iron atoms in Fe 3 (co) 12 indicate motion of the iron triangle about its pseudo three-fold axis, and those of the cobalt atoms in Co4(COJ 1 2 indicate motion about the three c 3 axes of the cobalt tetrahedron. Other concerted motions of the metal atoms and the CO groups must also be involved in these dynamic processes in order to account for CO migrations in solution; these concerted motions must be consistent with the symmetry of the normal modes of vibration of the system and will involve amongst others pseudorotation of the CO-polyhedron in ways related to those postulated for simple coordination compounds. The formation of cluster isomers also follows naturally from these arguments. Similar arguments may be applied generally to cluster carbonyls but great care must be exercised in drawing comparisons between the solid and solution states especially when ionic compounds are under consideration.

REFERENCES 1. 2. 3. 4. 5.

6. 7. 8.

C. E. Anson, R. E. Benfield, A. W. Bott, B. F. G. Johnson, D. Braga, and E. A. Marseglia, J Chern Soc . Chern Coromun , 889 (1988). B. F. G. Johnson, to be published. J. F. Blount and L. F. Dahl, Inorg Chern , 1373 (1964) C. H. Wei and L. F. Dahl, J 8mer Chern Soc , 91:1351 (1969). Chern Commun , 211 (1976); R.E. B. F. G. Johnson, J Chern Soc Benfield and B. F. G. Johnson, J Chern Soc , Dalton Trans , 1554 (1978) . Affier Chern Soc., 0. A. Gansow, A. R. Burke, and W. D. Vernon, J 94:2550 (1972). F. A. Cotton and J. M. Troup, J Affier Chern Soc , 96:4155 (1974) and references therein. F. A. Cotton and J. M. Troup, "Symposium on Metal Carbonyl Chemistry", Ettal

9. 10. 11. 12. 13.

160

(1974).

B. F. G. Johnson, J Chern Soc . Chern Commun , 703 (1976). F. H. Carre, F. A. Cotton, and B. A. Frenz, Inorg Chern, 15:380 (1976) . C. E. Anson, R. E. Benfield, A. W. Bott, B. F. G. Johnson, D. Braga, and E. A. Marseglia, unpublished observations. R. E. Benfield and B. F. G. Johnson, ~J~~Cah~e~m~_ws~o~c~~D~a~l~t~olin~T~r~aun~s~, 1554 (1978); Royal Society Annual Meeting Liverpool (1980). R. E. Benfield and B. F. G. Johnson, ~J~~Cdh~e~m~~s~o~c~_JD~a~l~t~o~n~T~r~aun~s~, 1743 (1980)

SELECTIVE OXIDATION CHEMISTRY ON SOLUBLE OXIDES: A PROGRESS REPORT V. W. Dayab W. G. Klemperer*b, S. P. Lockledgeb, D. J. Main , F. S. Rosenbergb, R.-C. Wangb, and 0. M. Yaghib acrystalytics Company Lincoln, Nebraska 68501 Department of Chemistry University of Nebraska Lincoln, Nebraska 61801 bDepartment of Chemistry University of Illinois Urbana, Illinois 61801 INTRODUCTION The phrase "selective oxidation" invokes the world of industrial chemistry concerned with the partial oxidation of organic molecules. two grounds for this.

There are

The first is economic, since almost half ot the major

products of the usage of heterogeneous catalysis (see Table I) are produced by selective oxidation of organic materials.

The second ground for associ-

ating selective oxidation chemistry with the industrial community arises from the difficulty that academic chemists have encountered when attempting to understand heterogeneous selective oxidation processes on a molecular level and simulate them with homogeneous analogues.

The detailed mechanisms

of methanol dehydrogenation and ethylene epoxidation are unclear, and even the vaguest outlines of the C-H activation mechanism involved in xylene and butane oxidations are obscure. This article is a report on progress in the authors' laboratories toward performing selective organic oxidations homogeneously using soluble oxides in an effort to mimic and understand the corresponding heterogeneous processes.

Methanol oxidation will be treated first, since consider-

able success has been achieved using stoichiometric molybdate reagents. Next, the problem of oxygen transfer from dioxygen to olefins will be addressed in terms of an iridium system where limited progress has been made. *To whom correspondence should be addressed.

Metal-Metal Bands and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

161

Table I.

Major Products of the Usage of Heterogeneous Catalysisa

Reactant(s) Crude Oil so 2 , o 2

Product Hydrocarbon Fuels Sulfuric Acid Ammonia

N2, H2 NH 3 , o 2

Nitric Acid

CO, H2

Methanol

Global production, k I r 10 12 1.4x 10 11 9 X 1011 1

X

2.5

X

1010

1. 5 X 1010

1

X

1010

8

X

10 9

Polyethylene

6

X

Formaldehyde

5

X

Acrylonitrile

3

X

E._- Xylene, 02

Phthalic Anhydride

2

X

10 9 10 9

_g_-Butane, 02

Maleic Anhydride

4

X

10 8

C2H4, 02 Unsaturated Vegetable Oils, H2 C2H4 CH 30H, 02 c3H6 , NH 3 , 0 2

Ethylene Oxide Hydrogenated Vegetable Oils

10 9 10 9

a from I. M. Campbell, "Catalysis at Surfaces," Chapman and Hall: New York, 1988; p. 5. Finally, the nature of C-H activation over vanadium oxides will be considered, not in terms of achieving partial oxidation but in terms of synthesizing soluble polyoxovanadates that might have features normally associated only with solid vanadium oxide surfaces. METHANOL DEHYDROGENATION Formally speaking, the oxidation of methanol to formaldehyde is a dehydrogenation reaction, more specifically, an oxidative dehydrogenation:

(1) Considerable progress has been made toward understanding the mechanism of alcohol oxidation over Moo 3 by using temperature-programmed desorption with simultaneous microbalance and mass spectral detection. 1 • 2 Scheme I shows the reaction sequence for methanol oxidation derived from these studies, one of which involves low temperature dissociative chemisorption of methanol, methoxide oxidation at 150°-250°C, and catalyst reoxidation at still A detailed mechanism for the first two steps was proposed in 1977 by Edwards, Nicolaidis, Cutlip, and Bennett. 3 This mecha-

higher temperatures.

nism, shown in Scheme II, proposes dissociative adsorption of methanol

162

across a molybdenum-oxygen double bond followed by oxidation involving participation of a second molybdenum center.

Although this mechanism has re-

ceived no experimental support, its key features have been endorsed by Allison and Goddard 4 • 5 on the basis of general valence bond quantum mechanical computations.

These calculations support addition of methanol across

an Mo=O bond followed by an oxidation step that involves an adjacent molybdenum center.

Unfortunately, there is no experimental precedent for the

formation of stable surface methoxides of the type observed in methanol oxidation by addition across Mo=O bonds.

Instead, oxomolybdenum(VI) alkoxides

favor dioxomolybdenum structures that could be obtained by addition across molybdenum-oxygen single bonds. 6 Formation of this type of intermediate obviates the need for a dual-site oxidation process and allows the possibility of a one-site oxidation mechanism.

A mechanism employing these two altern-

ative steps is shown in Scheme III.

Scheme I

0

HO

0

I

I

OCH 3

\1

0

\1

+

II

Mo--0--Mo

Mo--0--Mo

HO

OCH3

HO

0

OCH

\1

II Mo--0--Mo

OH 2

I

Mo--0--Mo--

Scheme II

163

Mo

Mo

#~

0

0

0

~

+

#~

"'-../ 0

0

'\./ Mo #~

0

0

0

0~

-•------'~~

~

#~

0

+

Mo #~

OCH 3

"'-../ Mo

HO

OCH 3

. . . _ _...o........._ /

0

"'-..--~cH 2 Mo

#'\.OH

0

Scheme III

Several features of the mechanism shown in Scheme III are supported by reactions of soluble molybdates.

First, alchohols have been shown to react with the "molybdic anhydride" [(P 3o9 )Mo0 2 ] 2o4- by addition across a molybdenum-oxygen single bond to form "esters" [(P 3o9 )Mo0 2 (0R) 6 ] 2- having the molybdenum coordination environment proposed in Scheme III. etry of the ethoxide complex 6 is shown in a.

The geom-

a

This complex, when heated to at least 180°C in acetonitrile or nitrobenzene solution, releases acetaldehyde.

The type of oxidation mechanism

shown in Scheme III involving polar 1,2-addition of a C-H bond across an Mo=O bond is supported by the behavior of analogous cyclopentadienyl and pentamethylcyclopentadienyl complexes.

The cyclopentadienyl complex

(C 5H5 )Mo0 2 (0CH 2CH 3 ), b, liberates acetaldehyde at about 80°C in solution, supporting a 1,2-addition mechanism since the c5H5 ligand is far more

164

b

electron-donating than the P3o9

3-

ligand.

7

Formation of a coordinated

2 n-

aldehyde intermediate is consistent with the reluctance of the sterically crowded pentamethylcyclopentadienyl complex (C 5Me 5 )Mo0 2 (0CH 2CH 3 ), c, to generate acetaldehyde, even at 125°C in solution.

c

ETHYLENE EPOXIDATION The oxidation of ethylene to ethylene oxide by dioxygen is formally an oxygen transfer reaction:

(2)

The industrial process for producing ethylene oxide in this fashion utilizes silver supported on alumina as a catalyst and proceeds as indicated

165

Ag(ox.)

Ag(ox.)I0 2(ads.)

Ag(ox.)IO(ads.)

)

Scheme IV

in Scheme

rv. 8

Three features of the catalytic system are noteworthy.

First, the silver catalyst is not present in zero-valent form, but must be oxidized in order to achieve good selectivity.

Second, adsorbed oxygen

but not adsorbed dioxygen reacts with ethylene to form the product.

Final-

ly, the catalytic cycle cannot be extended to olefins having allylic hydrogens since the presence of allylic hydrogens leads to a great loss in selectivity. Several homogeneous catalytic systems have been developed that are capable of o2 epoxidation of olefins. 9 Mechanistic details of these processes are obscure, however, due in part to an inability to isolate reaction intermediates.

The iridium olefin complex d undergoes selective

o2

oxidation via intermediates that may be related to those involved in heterogeneous and homogeneous epoxidatione

Specifically, complex d reacts with

2-

d

o2

at ambient temperature according to equation (3) to form a single product, e, in >90% yield as determined by 31 P NMR. 10 A key intermediate (3)

in the reaction can also be observed by 31 P NMR, isolates in crystalline 166

2-

e

10 form, and identified as an isomer of e, the oxametallacyclobutane f:

f

Observation of this intermediate draws a clear analogy between the iridium phosphate system under discussion and the heterogeneous catalytic system of Scheme IV: if the cyclic intermediate f were to convert to an epoxide g instead of the allylic complex e, a homogeneous catalytic cycle

2-

g

might be developed.

This possibility is not realized, however, since

allylic hydrogens in f are sufficiently acidic to favor formation of e by proton transfer to the alkoxide oxygen.

XYLENE/BUTANE OXIDATION The oxidation of xylene to phthalic anhydride, equation (4), and butane to maleic anhydride, equation (5), are complex oxidations involving oxidative dehydrogenation and oxygen transfer steps. In both cases, vana"d . 11,12 an d date catalysts are emp 1 oye d , v20 5 on Ti 0 2 f or xy 1 ene ox1 at1on $-(V0) 2 P2 o7 for butane oxidation. 13 • 14

167

~

~

+ 3 02

+ 3 H2o

(4)

0

~

+ 7/2 02

+ 4 H2o

(5)

0

Little is known about catalyst-substrate interactions in these systems on a molecular level, but the nature of the reactive site for the initial step, C-H activation, may be surmised from structural data. 14

A

recurring motif in oxovanadate chemistry is square pyramidal coordination at penta- or tetravalent vanadium, with a very weak metal-ligand interaction at the vacant coordination site trans to a vanadium-oxygen multiple bond. 15

One version of how this type of weak association might be in-

volved in catalyst-substrate interactions can be observed in the recentlyreported

16

soluble inclusion complex [CH 3CN C: v 12 o32 4- )], h:

h

Here, an acetonitrile molecule is suspended in the center of a basket-like v12032

4-

cage of corner- and edge-sharing VOS square pyramids.

Efforts

are currently being made to determine whether less polar guest molecules can bind to the v 12 o32 vl2032

168

4-

cage and to investigate the ability of the

4- cage to undergo redox reaction chemistry.

ACKNOWLEDGMENT We acknowledge the National Science Foundation for support of this work.

Mr. Chris Frank provided invaluable technical assistance in the

synthesis of polyvanadate complexes, and Dr. Scott Wilson performed X-ray crystallographic structure determinations of the organomolybdate complexes b and c.

REFERENCES l.

W. E. Farneth, F. Ohuchi, R. H. Staley, U. Chowdhry, and A. W. Sleight, J. Phys. Chern. 89, 2493 (1985).

2.

W. E. Farneth, R. H. Staley, and A. W. Sleight, J. Am. Chern. Soc. 108,

3.

J. Edwards, J. Nicolaidis, M. B. Cutlip, and C. 0. Bennett, J. Catal.

4.

J. N. Allison and W. A. Goddard III, J. Catal. 92, 127 (1985).

5.

W. A. Goddard III, Science 227, 917 (1985).

6.

V. W. Day, W. G. Klemperer, C. Schwartz, and R.-C. Wang, in NATO Ad-

2327 (1986). 50, 24 (1977).

vanced Science Institute Series, Series C; Mathematical and Physical Sciences, Vol. 231: Surface Organometallic Chemistry: Molecular Approaches to Surface Catalysis; J. M. Basset, B. C. Gates, J. P. Candy, A. Chaplin, M. Leconte, F. Quignard, and C. Santini, eds.; Kluwer: 7.

Dordecht, 1988; p. 173. V. W. Day, W. G. Klemperer, F. S. Rosenberg, and R.-C. Wang, manuscript in preparation.

8.

(a) R. A. Van Santen and H. P. C. E. Kuipers, Adv. Catal. 35, 265 (1987); (b) J. T. Roberts and R. J. Madix, J. Am. Chern. Soc. 110, 8540 (1988).

9.

(a) J. T. Groves and R. Quinn, J. Am. Chem. Soc. 107, 5790 (1985); (b) C. L. Bailey and R. S. Drago, J. Chern. Soc., Chern. Comm. 179 (1987); (c) M. M. T. Khan and A. P. Rao, J. Mol. Cat. 39, 331 (1987); (d) R. A. Leising and K. J. Takeuchi, Inorg. Chern. 26, 4391 (1987); (e) J.-C. Marchon and R. Ramasseul, J. Chern. Soc., Chern. Comm. 298 (1988).

10. V. W. Day, W. G. Klemperer, S. P. Lockledge, and D. J. Main, manuscript in preparation. 11. M. S. Wainwright and N. R. Foster, Catal. Rev.-Sci. Eng. 19, 211 (1979). 12. R. Y. Saleh and I. W. Wachs, Appl. Catal. 31, 87 (1987). 13. G. Centi, F. Triffiro, J. R. Ebner, and V. M. Franchetti, Chern. Rev.

88, 55 (1988).

169

14. M. A. Pepera, J. L. Callahan, M. J. Desmond, E. C. Milberger, P. R. Blum, and N. J. Bremer, J. Am. Chern. Soc. 107, 4883 (1985). 15. For representative examples, see (a) S-(V0) 2P2o7 : Yu. E. Gorbunova and

s.

A. Linde, Sov. Phys. Dokl. (Engl. Transl.) 24, 139 (1979); (b) a-

VP05: B. Jordan and C. Calvo, Can. J. Chern. 51, 2621 (1973); (c) SVP05: R. Gopal and C. Calvo, J. Solid State Chern. 5, 432 (1972); (d)

v2o5 : R. Enjalbert and J. Galy, Acta Crystallogr. Sec. C C42, 1467

(1986); (e) K12v18o42 ·16H 20: G. K. Johnson and E. 0. Schlemper, J. Am. Chern. Soc. 100, 3645 (1978); (f) [NMe 4 J6 rv 15 o36 ]•Cl·4H2o: A. Muller, E. Krickemeyer, M. Penk, H.-J. Walberg, and H. Bogge, Angew Chern., Int.

Ed. Engl. 26, 1045 (1987); (g) K6 [As 6v15o42 (H 20)]·8H 20: A. Muller and J. Doring, Angew. Chern., Int. Ed. Engl. 27, 1721 (1988). 16. V. W. Day, W. G. Klemperer, and 0. M. Yaghi, submitted for publication.

170

THE STUDY OF CLUSTERS OF POLYLITHIUM ORGANIC COMPOUNDS AND STRUCTURAL STUDIES OF POLYLITHIUM ORGANIC COMPOUNDS Richard J. Lagow Department of Chemistry The University of Texas at Austin Austin, Texas 78712 Clusters of polylithium organic compounds, (C Li ) , are perhaps the n

m x

mQst interesting and unusual clusters in all of main group chemistry from a structural and reactivity point of view.

These unusual clusters, such

as (CLi 4 ) 3 and (CLi 2H2 ) 4 , were discovered by Richard Lagow and research associates in 1982 1 ' 2 thus opening this field of chemistry for years of fascinating structural investigations to follow.

Compounds from first

row elements in which the central element had a nearly empty 2p orbital have a very pronounced tendency to form polymeric or oligomeric clusters such as those observed and studied for the boron hydrides, (B H ) • This n mx fact has been well established. The 1976 Nobel Prize in Chemistry was awarded for structural studies of the boron hydrides and boron hydride clusters.

Polylithium organic compounds promise to offer an even more

fascinating story from a structure and bonding point of view since the lithium electron configuration has a completely empty 2p orbital (1s 2 2s 12p 0 ) in contrast to the well known electron configuration for boron hydride (1s 2 2s 22p 1 ). Boron hydrides are often called "electron deficient species" and if this is so, clusters of polylithium compounds have even more chances for multicenter bonding. The field of lithium organic compounds and its chemistry was opened around 1970 when Richard Lagow3 and associates at the University of Texas synthesized such species as tetralithiomethane, tetralithioethylene, hexalithioethane, trilithiomethane, hexalithiobenzene, 1,1-dilithiocyclopropane, bis(trimethylsilyl)dilithiomethane, etc. and Robert West 4 and associates at the University of Wisconsin synthesized such exotic species as c3Li 4 and c4Li 6 . Previously, Karl Ziegler had prepared the compound dilithiomethane, 5 but his report had suffered a credibility problem when others could not reproduce it and was not widely accepted. 1 Metal-Metal Bonds and Clusters in Chemistry and Catalysts Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

171

In 1981, the polymeric nature of polylithium organic compounds and their thermal stabilities and rearrangements was discussed by Lagow and co-workers. 2 Very fundamentally the thermal stability of most polylithium organic compounds is limited to about 225 •c where they undergo rearrangement and there exists no vapor pressure of polylithium organic compounds due to their high polymer nature, until temperatures approaching 800-1000 •c. A breakthrough in our laboratory first applied to dilithiomethane 1 was that polylithium compounds could be transported over distances as great as ten centimeters without substantial decomposition by the nonequilibrium method of very rapid vaporization (flash vaporization).

This technique was applied and adapted to a mass

spectrometric environment.

A tungsten conical basket is employed as a

vaporization source for a mass spectrometric probe (Figure 1) and is dusted in a very high quality dry box with an organolithium compound (powder). This is then transported in an ultra high vacuum valve system to the mass spectrometer (the probe also protrudes through the bottom of the dry box with high quality seals and connecting systems). 6 What Tungsten Wire Vacuum Feedthrouah

Prabe Body Copper Leads

Figure 1.

Probe tip of flash-vaporization apparatus.

happens here, of course, is that the filament is raised to a temperature of 1500

•c in less than a second and the polylithium compounds vaporize

so quickly that they do not undergo the type of Morrison-Lagow rearrangements to lithium acetylide and other more stable species previously noted. 2 This nonequilibrium method has produced, as we expected, extraordinarily interesting polymeric and oligomeric species for the lithium compounds in spite of the fact that our mass spectrometric facilities at the University of Texas leave much to be Polylithium compounds are three dimensional polymers and not, in their natural state, gas phase species. We had predicted 2 polylithium compounds would be polymeric in the gas phase. Observed were species

desired.

ranging in most cases from the monomer to the tetramer and often as high as the octamer with our type of flash vaporization mass spectroscopy. Most frequently the dimer or the trimer is the species of highest concentration and less often the monomer is the species of highest intensity.

Results are summarized for the lithiomethanes in Tables 1

through 4 (the reason there are many species in the trilithiomethane 172

Table 1.

Major Ions from the Flash Vaporization Mass Spectrum of CH 3Li

Positive Ion

m/e

(CH 3Li)n

(22)

(CH 3Li)nLi (CH 3 ) Li 1 n n-

(22)

Table 2.

+ 7 n (22) - 7 n

1-4 2-4

Major Ions from the Flash Vaporization Mass Spectrum of CH 2Li 2 m/e

(cH 2Li 2 )n

(28)

(CH2)nLi2n-1 (CH 2Li 2 )nLi

(28)

(CH 2Li 2 )nH (CLi2)nH2n-1 (C2Li2)n

Table 3.

n n (28) + n (28) + n (28) n (38) n

n 1-6 7

2-6

7

2-5

1

1-3

1

1-4 1-3

Major Ions from the Flash Vaporization Mass Spectrum of HCLi 3

Positive Ion

m/e

(CH3Li)n (CH3Li)nLi

(22)

(CH2Li 2 )n (CH 2Li 2 )nLi

(28)

(CHLi 3 )n (CHLi 3 )nLi (CLi 4 )n (CLi 4 )nLi (C2Li2)n (CH 3 ) Li 1 n n(CH2)nLi2n-1 CnLi4n-1

1-4

n

Positive Ion

CnHnLi3n-1

n

(22)

n

+7 n

n (28) n (34) n (34) n (40) n (40) n (38) n (22) n (28) n (34) n (40) n

n 1-4 1-4

+7

1-4 1-4

+7

1-4 1-4 1-4

+7

1-4 1-4

- 7

2-6

- 7

1-6

- 7

1-5

- 7

1-4

173

Table 4.

Major Ions from the Flash Vaporization Mass Spectrum of CLi 4

Positive Ion

m/e

(CLi 4 )n

(40)

(CLi2)nLi2n-1 (CLi 4 )nLi

(40)

Table 5.

1-4 - 7

1-4

7 n + (38)n (38)n - 7

1-3

(40)

(C2Li2)n (C 2Li)nLin_ 1

n

n

n

1-4 1-3

Selected Ions from the Flash Vaporization Mass Spectrum of c3Li 4 Prepared by the West Method

Positive Ion

m/e

CLi 2

26

CLi 3

33

c2Li c2Li 2 c2Li 3 c2Li 4 c3Li 2 c3Li 3 c3Li 4 c3Li 5

31 38 45 52 50 57 64 71 128 192

(C3Li4)2 (C3Li4)3

spectrum is that the sample was not highly pure and contained other species such as dilithiomethane). as high as n

=

6.

Note the dilithiomethane gives species

The Robert West compound,

c3Li 4 ,

gives interesting

species such as the trimer (C 3 Li 4 ) 3 (Table 5). It is also noteworthy along these lines that due particularly to the thermal rearrangement of solid methyllithium to dilithiomethane, 5 the mass spectrum of methyllithium was unattainable by standard mass spectrometric techniques. Using our technique it was possible 7 to compare the spectra, and establish for the first time what was always suspected--that the vapor of methyllithium should be tetrametric.

One

can now compare the spectra of methyllithium with the classic mass spectral work of Joe Berkowitz and Ted Brown 8 • 9 (see Tables 6 and 7) on ethyllithium and t-butyllithium.

Of course, here the secret is that methyl-

lithium is vaporized before it rearranges to dilithiomethane. 174

Positive Ions in the Flash-Heated High-Resolution Mass Spectrum of Halide-Free Methyllithium

Table 6.

a

Calcdb Mass

Measd Mass

Rel Error, ppm

Rel Intensityc

Me 3Li 4+

73.13442

73.13479

5.1

8

Me 2Li 3+

51.09495

51.09512

3.3

4

CH 2Li 3+

35.06365

35.06380

4.3

1.5

CHLi 3+

34.05583

34.05596

3.8

1

CLi 3+

33.04800

33.04815

4.5

<1

CH 3Li 2+

29.05548

29.05559

3.8

100

CH 2Li 2+

28.04765

28.04774

3.2

2

CHLi 2+

27.03983

27.03996

4.8

3

CLi 2+

26.03200

26.03206

2.3

1

Ion

~e =methyl.

bBased on C

12.000 ... , H = 1.0078246, and Li = 7.016001

cintensities varied ± 2-3% from spectrum to spectrum and were determined by comparison of sample line densities to standard Ar 2+/Ar+· amu.

spectra whose line densities were known relative to integrated ion current.

Comparison of the Positive Ion Mass Spectrum of Methyllithium with Those of Ethyllithium and tert-Butyllithiuma

Table 7.

R

CH b 3

not found

R5 Li 6 +

not found

R4Li 5+ (1.3) R3Li 4+ (47.5)

R3 Li 4+ (8) R2 Li 3+

(4)

RLi 2+ (100)

(24)

R2Li 3+ (15) RLi 2+ (100)

aRelative intensities in parentheses. noted.

bThis work.

ethyllithium vapor.

R3Li 4+ (43) R2 Li 3+ (4.5)

cReference 8.

RLi 2+ (100) 70-eV electrons used, except where

Intensities are for saturated

75-eV electrons used.

~eference 9.

Intensities

are for a mixture of 88% tert-Butyllithium and 12% tert-butoxylithium.

175

As previously stated, polylithium compounds vaporize in clusters with the dimer or trimer being more intense in the mass spectrum than the monomer.

In almost all cases one observes a "parent ion" for the monomer

as well as higher polymers.

It is thought that the exact distribution

and nature of the clusters observed depend on the mass spectrometric technique which is used to produce the clusters from the three dimensional solid lattice.

In this respect, the flash vaporization mass

spectrometric technique which was developed to see the polylithium compounds is viewed in our laboratory as an extreme method and was primarily invented to overcome lack of appropriate equipment at the University of Texas.

A new high resolution spectrometer with

multifaceted techniques has been obtained from the United States National Science Foundation and it is felt that when such techniques as field desorption, fast atom bombardment (FAB) and perhaps laser desorption are applied in this field that even higher molecular weight species will be generated.

In subsequent sections of this manuscript we will discuss

structures and predictions of structures of polylithium organic compounds.

While the structures of monomers appear to be of great

interest and importance, the structures of clusters and oligomers should be both extraordinarily interesting and unusual.

Collaboration is

underway with Professor Josef Michl and his research associates in which such extraordinarily interesting species are to be vaporized into Nobel gas matrices and their structures studied by vibrational electronic and a wide range of other spectral techniques.

An apparatus has been designed

where a steady stream of powder drops on a high temperature surface.

The

vapor produced is deflected into a matrix as it fragments into monomers, dimers:

lT

High

Vacuum

Polyllthium Power Feed

I

Environment

Low Temperature Matrix for Various Spectroscopic Investigations

I

~~~rature _)J # Surface

-+ " " ' (

l

Deposition of

vapor species

Already some interesting work has emerged on photoacoustic spectra of some of the dilithio compounds and simple polylithium compounds from the Michl collaboration.

What the most recent experiments showed was that

the apparatus had to be very highly refined to carefully exclude oxygen, water and other impurities but our experiments have been proven the

176

feasibility of such techniques.

A schematic for an apparatus designed to

flash vaporize polylithium clusters onto a cryostat is pictured below:

The structures of monomeric polylithium species are also of great interest. The stature of the field of polylithium organic compounds was given a very strong boost in 1974 when Paul Schleyer and John Pople entered this exciting area via ab initio calculations. Their bold and venturesome calculations have led to additional interest in this field throughout organic chemistry, inorganic chemistry, and structural Additionally, one can not help but conclude that if the striking predictions of Schleyer, Pople, and co-workers are reasonably accurate for gas phase monomers, the bonding and structure of

chemistry around the world.

polylithium compounds in higher molecular weight clusters and in the solid state must be even more intriguing. Among the most striking and potentially significant predictions are that certain polylithium 10 compounds will exhibit planar rather than tetrahedral carbon atoms, that others will have electron-defic ient vapor-phase monomers involving 11-16 . and that those olefins with 1,1-dilithio bridging lithium s1tes, substituted bonds will be skewed instead of planar. 17 The vapor species of dilithiomethan e, (CH 2Li 2 )n' are of particular interest in view of the possibility proposed by the work of Schleyer, 10 of the existence of fluctional cis-tetrahedra l, Pople, and co-workers and possibly trans gas-phase forms. Trilithiometha ne monomer is also of great interest in that regard. According to Schleyer, planar tetracoordinat e carbon should be stabilized preferentially (1) by delocalizing the lone pair, (2) by providing more electron density to

car~on

by sigma donation and (3) by

177

Q cu-Orlrlhromethane

p-OrbrtJI syslem m d•lrthtOmethane

traiU'·DIIIIhlomethlne

z

enforced reduction of the angle around the planar carbon atom by means of small rings. 10 • 18 Because it has pi acceptor and sigma donor character together with its small atomic radius, lithium may be particularly effective in stabilizing the planar arrangement.

Roald Hoffmann had

previously observed that lithium substitution could be very advantageous for stabilizing planar carbon in a treatise on planar carbon. 18 Ab initio molecular orbital calculations on dilithiomethane (CH 2Li 2 ) -10 19-20 monomer ' have revealed remarkable features: cis-trans isomers 1 and

with the cis isomer being the more stable species, and a low planar-tetrahedral inversion barrier that may result in rapid ~.

interconversion of cis planar and tetrahedral CH 2Lig(g). Schleyer and Pople have further proposed a "homoaromatic" stabilization 10 of the p orbitals giving rise to some TI-bonding between the cis and lithium atoms 3 and 4. Using a Gaussian 80 program, Streitwieser and co-workers 21 have explored the energy differences between ionic triplet structures which have significant Li-Li bonding in both planar and tetrahedral geometries and singlet models with tetrahedral and planar symmetry. An interesting model emerges as an aggregate of triplet methylene bonded to Li • 2 little difference in energy (-.8 Kcal/mol- 1 ) between planar and

Very

tetrahedral carbon sites is a feature of the Streitwieser model as well. Similarly planar structures have been predicted to be more stable for alkane structures, such as CLi 4 , CHLi 3 , ~. 10 LiH 2CCH 2Li, ~. 22 CF 2Li , 2 10 and cis-C H Li , 8. 10 However structure l will be extraordinarily 3 4 2 difficult to observe and certainly would be a challenging candidate for our matrix collaboration since lithium halide elimination would occur at

z,

temperatures on the other side of -80 178

•c.

H H-'

\/'

~

H--e

e--H

I

H

lr

Ll

H

lr

\I

e

l~\

lr

lr

lr

li

e

1\ F

lr

I, 2- Orlrthraethane

Tetr•lrthromethane

Orfluarcrdrlrthrornethane

5

I

7

H1 H I, 1-0rlrthrocyclopropane

•

Unsaturated organolithium compounds may show even greater anomalies in structure.

Hydrocarbon olefins have been generally found to be two-

dimensional, that is, planar.

However, the most stable conformation for

the lithium-substituted analog of ethylene is predicted to be a skewed olefin structure

2 with

a 5 kcal/mol rotation barrier holding the vapor

monomer in a skewed configuration with the lithium site orthogonal to the unsubstituted olefin sites. 17 This calculation has been conducted at higher levels by Fritz Schaefer 23 with the same conclusion. The skewed olefin effect, one which one can see we now have the molecules to probe, is one that we in the Lagow research group really expect to see.

The vacant p orbitals on

lithium have a very high probability of interacting with the electron rich

~-systems

and the distances are reasonable.

The most stable configuration of the tetralithioethylene species 10 has been predicted to be a planar structure with "electron-deficient" three-center bonding leading to a bridging lithium configuration. 12 This unsual bridging structure has also been postulated for the acetylene derivative c2Li 2 , ll• rendering it planar rather than linear. 12

..."'··...

H

li

··..

H/c

Lr---e

I, I - Oilrthroethylene

'

/

\

I/

\

'

\

I

\ /

I

e---li

'u'

Perlrthroethrlene 10

e

\

I

\

I

\

Lr

li

\

I

\

I

\ /

c

Dilithroacetvlene II

The idea of "hypervalent" lithium compounds arose out of the work of 24 25 Schleyer ' and is not beyond the realm of possibilities. The original 179

Schleyer proposal was that the molecule dilithium (Li 2 ) could act as a ligand and supply one or two sites on the CLi 4 monomer structure with an electron:

----l----z-1---------1----Li

Li

CLis Li

Li

Li

Certainly Li 2 is a minor species in the gas phase vapor of lithium.

An even more intriguing model has been recently proposed by Frank Weinhold of Wisconsin 26 in which (CLi 6 ) and (CLi 8 ) are treated.

In the

Weinhold proposal on occupancies of "natural atomic orbitals' (NAOs), a very simple model for bonding in CLin species is suggested in which formal charge can be stabilized by strong donor-acceptor interactions from the filled Ne-like C(IV) core to the enveloping cage of unfilled Li 2s orbitals, producing predictions of preferred geometries for donor-acceptor stabilized species.

These are highly symmetric and very

similar to close packing about a central spherical ionic core.

Under

such conditions the number of bonds is limited only by geometrical constraints of core and ligand radii.

Thus in the Weinhold projection

CLi 6 has Oh octahedral geometry and CLi 5 is as Schleyer predicted n3h. 11

Table 8.

Analytic Expressions for Ground-Sate Eigenvalues of Hiickel-Type Matrices for Some n-Orbital Cages of High Symmetrya

n

symmetry

2

D.,.,h

-s180

3

0 3h

- 28 120

4

Td

- 38 109.47

5

03h

6

oh

-(S180 + 48 90)

8

oh

-( 38 70.53 + 38 109.47 + 8 180)

lowest eigenvalue

2 -l/ 2 ( 8 180 + 28 120) [ 1 + {1 - 8 ( 8 180 8 120 - 38 90)/(S180 + 2S )2}1/2] 120

aEach eigenvalue is expressed in terms of overlap integrals S

8

between

orbitals separated by angle 8 (as measured from the cage center). 2s STOs at 2.0

l

For Li

from the cage center, the overlap integrals used to

construct Hucke! matrices for various n were s 60 = 0.7368, s 63 _43 = 0.7141, s 70 _53 = 0.6676, s 72 = 0.6581, s 90 = 0.5496, s 109 _47 = 0.4543, s 116 _57 = 0.4261, s 120 = 0.4136, s 144 = o.3487, s 180 = o.3143.

180

(XI

b

oh

oh

D3h

2.094

2.058

c

a

c

a

c

a

2.056

2.087

2,056

2.087

2.056

2.087

1.929

1.929

-81.63028d

-82.00344

-74.45440b

-74.59934

-82.104959

-82,483723

-74.885704

-75.000853

-75.001260

-67.396262

(6,05)e

6.93

(6.02)c

6.81

7.53

7.28

7.26

9.34

9.44

9.19

9.19

9.08 7.18

7.14

8.97

9.13

natural population on C DHa

7.05

7.09

6.81

-67.520562

-67.13154

Mulliken population on C DHa 3-21G

calcd total energy (au) DHa 3-21G

Present work.

assessed as being much more negative in the natural population analysis than in the Mulliken population analysis.

a

c ST0-3G basis set and geometry (1.979 a , 1.953 c ). 3-21G optimized geometry (2.022 a , 1.999 c ), f e d Note that the charge on carbon is ST0-3G basis set and geometry (2.063). 3-21G optimized geometry (2,094).

6

CLi 2+

CLi 6

5

CLi +

D3h

CLi 5-

Td

0 3h

2-

Td

Symmetry

C-Li bond length clt)

SCF Total Energies and Carbon Populations for Some Lithiated Carbon Species CLi , as Calculated with n 3-21G and Dunning-Hay (DH) Split Valence Basis Sets,f

CLi 5

CLi4

CLi 4

Species

Table 9.

Figure 2.

Comparison of 2p NAOs in neutral carbon atom (top) and the carbon atom of CLi 4 (bottom), showing the much greater diffuseness of the latter compared to "normal" carbon orbitals.

(The orbitals shown are the "pre-NAOs", preceding

interatomic orthogonalization, whose overlaps allow direct estimates of NAO Fock matrix elements and are more readily visualized.)

Contour intervals are 0.02 au with outermost

contour at 0.02 au. Carbon-lithium bonds of such a CLi 6 or CLi 8 variety, if they exist, would certainly have a lower bond strength than species such as CLi 4 , whether the Schleyer description or the Weinhold description proves best. We have a feeling that if they existed, they could not be seen (and indeed only CLi 5+ has been seen) in our current mass spectrometric work (flash vaporization) with saturated species such as (CLi 4 )n. Polylithium compounds synthesized via the lithium vapor approach before they are further purified are available in our laboratory and are the ideal materials for such a quest. Work in our laboratory has provided the most substantial support for these striking Schleyer-Pople-Weinhold predictions of species such as In particular the species CLi 5+ was first seen in our laboratory when we were studying the vapor species associated with CLi 5 , CLi 6 , and CLi 8 .

Table 10.

Pentacoordinate Carbocations CLi 5 _nHn+

Ion

Calculated Mass

Measured Mass

CLi5

47.08000

47.08019

(CLi 4 )n

CHLi 4+

41.07183

41.07196

CH 2Li 3+

35.06365

35.06380

CH 3Li 2+

29.05548

29.05559

(CHLi 3 )n (CH2Li 2 )n, (CH 3Li) 4 (CH 2Li 2 )n, (CH 3Li) 4

CH4Li+

23.04730

[sought in CH4 + Li reaction]

182

Source of Ion

tetralithiomethane 24 , later in our laboratory we observed a number of other five coordinate carbocations 27 (see Table 10). The structures of the first five coordinate carbocations have also been calculated 24 • 25 . Presently our mass spectrometric facilities do not permit us to determine whether these species come from the more likely ion molecule reactions in the vapor phase or the more interesting generation from neutral precursors such as (CLi 5 )n' (CLi 6 )n' or (CLi 8 )n. In the recently completed structure of dilithiomethane, it is possible to see how one might generate by certain decomposition and vaporization processes gas phase species such as CLi 6 and CLi 8 . Our laboratory in conjunction with those of Galen Stucky of the University of California at Santa Barbara and David Cox of Brookhaven National Laboratory have recently completed the neutron powder diffraction structure of dilithiomethane and the same collaborators have the synchrotron microcrystal structure in progress on the same compound along with the synchrotron microcrystal structures of other polylithium compounds soon to be studied.

A first approximation of the structure of

dilithiomethane, (CLi 2H2 )n' has the antifluoride structure shown below:

0

c

0 u

One can see in a perfect antifluoride structure that each carbon atom is surrounded by four lithium atoms and then the "hypervalent" clusters are bridged by two lithiums to the next carbon.

Thus a network polymer is

observed for dilithiomethane. Neutron diffraction data were measured at 5, 101 and 293 K on a triple-axis diffractometer at Brookhaven National Laboratory.

All lines

were indexed in a body-centered tetragonal system with lattice constants at the above temperatures: a=b=5.6216(6), c=10.865(2) @5K a=b=g=90°; a=b=5.631(1), c=10.901(1) @101K; a=b=5.833(1), c=10.938(1) @ 293K.

The structure was solved by a combination of (1) using Patterson techniques and estimated peak intensities, (2) packing models based on the relative intensities of the Bragg reflections and (3) comparisons with known 183

.....

o

4r\'r-er... 0

00

0

~

4 0

,...

eLi c

iooo.. 0

G)

to...13

Figure 3.

Lithium and carbon atom positions for CD 2Li 2 within the tetragonal unit cell.

binary structures of the form AB 2 • The fact that the c unit cell, axis is approximately 2 x a and the above considerations led to the refined structure shown in Figure 3 in the space group I4 1;a · The structure is a very distorted version of the antifluorite structure of Li 2o28 and Be 2c29 in which one of the cubic cell edges is doubled (Figur~ 3) . To a first approximation the carbon atoms are located at the face centered sites of the parent cubic antifluorite cell and the lithium atoms are near the (0.25,0 . 25,0 . 25) body diagonal position.

The shortest lithium- lithium atom distance in the refined

tetragonal structure is 2.26(2)

K which

can be compared with the Li-Li distance of 2. 314(1) ~in Li 20, 2. 383(6) Kin [LiC 6H11 ] 2 , 30 and 2.56(1) in CH 3Li 31 Both the lithium and carbon atoms were found to be in ordered sites. ~axis

Qc u D

Figure 4 . 184

CD 2 coordination in CD 2Li 2 •

A

Table 11. Space Group:

Structural Parameters CD 2Li 2 .

I4 1/a(2nd setting, 88, International Tables for Crystallography)

Lattice Constants (A):

a=b=5.6216(6), c=l0.865(2) @ 5 K a= b = g = 90 • a=b=5.631(1), c=10.901(1) @ lOlK a=b=5.833(1), c=10.938(1) @ 293K

Structural Coordinates (5 K): Atom

p

= 0.098, R = 0.077, c 2 p

y

X

C(l) D(1) D(2) Li(1)

wR

U(isotropic)x10

z

0.25 0.323(2) 0.044(2) 0.461(3)

0.0 -0.034(3) -0.019(6) 0.189(2)

= 1. 70. a

0.4267(8) 0.336(1) 0.406(1) 0.028(1)

0.00(3) 10.6(9) 6.1(6) 0. 00(7)

aStructure Refinement with GSAS by Bob Von Dreele of Los Alamos. Selected Bond Distances and Angles. Distance (A)

Atoms C(l)-D(l) C(1)-D(2) C(1)-Li(1)

Li( 1 )-D(l)

Li(l)-D(2)

Li(l) -Li( 1)

Atoms

1.09(1) 1.18(1) 2.63(1) 2.17(1) 2.73(1) 2.53(2) 2.36(2) 2.15(1) 1. 72(2) 2. 28( 1) 2.03(2) 2.16(1) 2.38(2) 1.99(1) 2.06(2) 2.26(2) 2.86(1) 2.78(1) 3.09(1)

D(l)-C(l)-D(l) Li(1)-C(1)-Li(1)

Angle 100.2(5) 153.6(7)

The arrangement of the lithium atoms about the carbon atom is shown in Figure 4.

A crystallographic

c2

axis passes through C(l) in the

vertical direction of the figure and only one of the two disorder orientations for the deuterium atoms is shown. 32 The closest carbon-

!]

is comparable to that in cyclohexyllithium30 [2.172(5) A] or in ethyllithium33 [2.19(1) A]. The shortest Li-D distances [1.72(3) to 2.17(2) XJ 34 are compatible, but shorter than Li-H lithium approach [2.17(2)

185

distances previously observed for organolithium compounds [e.g. 1.96(1) K for the shortest alpha carbon hydrogen atom in cyclohexyllithium30 • 35 ]. The puckering of the four atom lithium rings shown in Figure 4 optimizes the close approach of C(1)-Li(1) [2.17(2) ~] and Li(1)-D(1) [1.72(2) A]. The observed D(1)-C(1)-D(2) angle is 100(1) 0 with C-D distances of 1.09(1) A[C(1)-D(1)] and 1.18(2) A[C(1)-D(2)]. The local geometry about the carbon atom (Figure 4) suggests two types of C-D coordination if one considers closest neighbor interactions. The coordination of D(1) can be viewed as (I) below, while that for D(2) ~C-H···M I

;H· ..

-c-·····M \ II

is best represented by the semibridging model (II) with coordination to the C-D bond [C-Li, 2.17(2); Li-D, 1.99(3) A]. This configuration is also found in cyclohexyllithium30 • 35 [C-Li, 2.184(3); Li-H, 2.00 XJ. (III) and (IV) have been previously described for the CH 3 groups in LiB[CH 3 ] 4 36 [C-Li, 2.207(9); Li-H, 2.115(8) X (III) and e-Li, 2.359(11); Li-H, 2.231(10) A (IV)]. In the field of polylithium organic compounds there have been extraordinary predictions of unusual geometries and in particular "hypervalent" polylithium compounds. First ab initio calculations by Schleyer and Pople 25 • 37 and later Frank Weinhold 26 using a natural orbital approach have forecast stabilities of such species as CLi 5 , CLi 6 and even CLi 8 neutral species. To date the most supportive experimental studies have been the observation of CLi 5+ in the mass spectrum of tetralithiomethane, (CLi 4 )n, by Lagow and co-workers 24 and subsequently the observation of other five coordinate carbocations. 27 Schleyer and Pople have also predicted the geometry of these additional carbocations. 38

This dilithiomethane structure may be very relevant to

the predictions of Schleyer, Pople, and Weinhold.

We should note that

the lithium coordination number for carbon in this structure of dilithiomethane is 8.

AlLhough there are extraordinary features such as

the C-D-Li bonding and the partial ionic character of the polymeric network, it is possible to envision vaporization processes occurring which would create species a CLi 6 neutral or CLi 8 neutral species considering the observed C-Li geometry.

Although Lagow and co-workers

have observed only CLi 3H2+ from the flash vaporization mass spectrometric study of dilithiomethane, 27 milder mass spectrometric techniques currently under development at the University of Texas may well result in the observation of such "hypervalent" species.

Such experimental

evidence would be extraordinary and in that event it wou.1.d be of great 186

interest to study the origin of such mass spectrometrically observed species to determine whether their precursors were these neutral species. Another unusual and interesting feature of the structure is the extraordinary coordination number of the central carbon atom.

The

current dilithiomethane structure contains a carbon atom which is surrounded by eight lithium atoms, six of which are less than 2.4 A away. If one considers the additional bonds to the 2 deuterium atoms, all carbon atoms in the structure could be considered to be at least eight coordinate and possibly even ten coordinate.

Although coordination

numbers for carbon of up to six are known in a few transition metal compounds, a coordination number of eight is the highest coordination number of carbon yet reported as far as we can determine. Prior to the neutron diffraction structure obtained by the Stucky, Lagow and Cox research groups, the solid 13 c spectra of dilithiomethane was studied. The 13 c spectra of solid CH 2 ( 7Li) 2 , CH 2 ( 6 Li) 2 and CH 36 Li 1H- 13 c cross po 1 ar~zat~on · · · ang 1e sp~nn~ng · · · · d us~ng mag~c were o b ta~ne (CPMAS) techniques in collaboration with C.S. Yannoni. 39 The weak signal 7 13 C spectrum of CH 2 ( Li) 2 at 150 •c observed in the proton-decoupled shown in Figure Sa is due to broadening from dipolar coupling with the 7Li nucleus. This type of broadening was eliminated through use of 40 t h roug h mag~c . . 40 an d t hroug h . ang 1e sp~nn~ng . . d ecoup 1 ~ng tee hn~ques,

(a)

(b)

Figure 5.

Proton-decoupled solid-phase 13 c NMR spectra of dilithiomethane: (a) CH 2 ( 7Li) 2 , spectrum obtained at 150 •c; (b) CH 2 ( 6 Li) 2 , 6 Li-decoupled spectrum obtained at 25

•c.

Peak is 10.5 ppm downfield from Me 4Si. 187

(b)

(c)

Proton-decoupled solid-phase

13

C NMR spectra of methyllithium obtained at -150 •c: (a) CH 3 7Li; (b) CH 36 Li, no 6 Li decoupling; (c) CH 36Li, 6 Li

Figure 6.

decoupled.

Peak is 16 ppm upfield from Me 4Si.

utilization of materials containing the lithium-6 isotope. The room temperature 13 c NMR spectrum of CH 2 ( 6Li) 2 shown in Figure 5b consists of a single line, 80 Hz FWHM. The chemical shift is 10.5 ppm downfield from TMS.

The

1

H and

6

Li decoupled

13

6

C spectrum for CH 3 Li is shown in Figure 6. The spectrum consist of a single line with a chemical shift of 16 ppm upfield from TMS. The 13 c NMR spectrum of CH 37Li in various solvents is well documented. 41 The chemical shifts for CH 3 7Li range from 11 ppm (Et 3N) to 15 ppm (THF) upfield from TMS. The small upfield displacement of the 13 c chemical shift of methyllithium (relative to methane) was interpreted as evidence for predominant sp 3 hybridization of carbon in this species.

Furthermore, excess charge density on the carbon was

indicated to be small.

188

The downfield shift of CH 2Li 2 relative to methyllithium may be attributed to an increase in charge density at the methylene carbon with Electropositive substituents are known to function as effective o-donors. 18 The very small linewidth for CH 2 ( 6 Li) 2 compared with that observed for CH 2 ( 7Li) 2 shows that the stratagem of using CH 2 ( 6 Li) 2 and decoupling 6 Li works, and should prove useful for 13 c studied of other organolithium increasing lithium substitution.

materials. Further information about the structure of dilithiomethane and a great deal of information about the charge distribution in the compound were obtained in a collaborative study on ESCA spectrum with Mike Hall and co-workers at Texas A&M University. 42

(a) Methylilth•um

700

4000 Li Is

400

2

ot

!!

<:"'::J

a u

() 0

0

u;

<:

::J

~

(b) Diithiomethone

2500

1000

1500

600

291

287

283

58

279

54

50

Binding Energy (eV)

Figure 7.

C ls and Li ls photoelectron spectra of (a) methyllithium and (b) dilithiomethane. Contaminant hydrocarbon at 285.0 eV is used as a reference.

189

0.0

-:c

-1.0

en

"6

.Eu

-2.0

"'

.t:: (.)

..!!!

(.)

-3.0

-4.0

Calculated qc

Figure 8.

Fenske-Hall MO charges for C in

c3H4

(1),

CH 3Li (2), and CH 2Li 2 (3) plotted against observed C 1s chemical shift. Vertical bars represent uncertainty of ± 0.2 eV in binding energy.

The horizontal bar represents the

standard deviation in the average charge of three CH 2Li 2 model structures.

Table 12.

Compound

Binding Energies, Chemical Shifts, and Calculated Charges. C 1s BE, eV

C 1s shift, eV

Calcd qc

Li 1s BE, eV

Calcd qLi

C3H6 CH 3Li

285.0a

0.0

-0.03

282.6

-2.4

-1.02

54.0

+0.86

CH 2Li 2

280.9

-4.1

-1. 55b

53.9

+0.71b

(0.07)

(0.04)

b Average of monomer, dimer, and trimer structures. Standard deviation in parentheses. aUsed as reference.

190

In the synthesis area, a very important new development is a high yield general synthesis for gem 1,1-dilithio compounds.

Gem dilithio

compounds are very unusual new reagents and of interest from a structural and theoretical point of view. 10 • 11 In 1985, our research group 3j modified and refined a Karl Ziegler pyrolysis and thermal rearrangement of methyllithium generating a high yield synthesis for dilithiomethanes.

This synthesis made

dilithiomethane an accessible reagent and added polylithium chemical to the repertoire of conventional organic synthetic chemists. Recently we have established that the Kawa-Lagow-Ziegler synthesis is a very general and broadly applicable synthetic method. 3j .The method can be modified so that synthesis of higher 1,1-dilithio compounds such as 1,1-dilithioneopentane, (trimethysi1yl)dilithiomethane and bis(trimethylsilyl)dilithiomethane are possible in 88, 97 and 95% yield; 3j the synthesis of tetramethyl-1,1-dilithiocyclopropane in 90% yield; 3k and the synthesis of 9,9-dilithiofluorene. 43

(Figure 9.)

Many of

the compounds which have been prepared in the R.J. Lagow research laboratory by the new Kawa-Lagow-Ziegler synthesis combined with those prepared by the original Lagow lithium vapor-chlorocarbon replacement

Me3So

2

'

c

H

/

H/ 'L•

'-./ /c'-..

Mo3So ICH313 c 2

'

/

..

I"/

'-.Li

180"C 90%

•

M"4S•

Lo + IMo3 Sil:! CHz

Me3Si/ "LI

'/

>

.,.

u

c H/ ' L i

(CH 3 12 C

H

c

""-/ c

(CH3 13C 180"C 88%

/Lo

c H/ 'Lo

Mo3Si

H

c H/ '-u

(CH.Y~/

170"C 95%

Lo

ICH3lzC 2

..

H

Me3So 2

Mo3S•, 150"!; 96%

• (CH 314C

Li

(CH 3lzC

Ll

ICH3lzC/

• I'/ /c""-

(CHyzC

I

'-cHz

.~"

if"

Figure 9.

Synthesis of gem-dilithium compounds by the Kawa-Lagow-Ziegler Pyrolysis. 191

reaction has led to a number of interesting compounds which should play a key role in establishing the accuracy of the Paul Schleyer, John Pople and Fritz Schaefer forecasts:

Me Mel>

Li Li

Me

Me

These are all prime candidates for study in the Nobel gas matrix spectroscopy program joint between the Richard Lagow and Josef Michl research groups at the University of Texas.

These and many more

dilithiomethanes and 1,1-dilithio olefins may reveal the extraordinary features predicted over the last decade about the structures of monomers. The preliminary NMR data, some field desorption mass spectrometry data from our laboratory, and flash vaporization mass spectroscopy done in our laboratory coupled with an earlier prediction of quadruply bridged 11 are beginning to create for such dilithio lithium by Schleyer and Pople species a structural picture where there are two orthogonally sets of bridging lithium.

We are finding that very bulky groups prevent the

extensive oligomerization characteristic of the polylithium compounds and will probably lead to structures (we have both single crystal and microcrystal synchrotron studies in progress) where the lithium compounds exist in the crystal lattice as dimers:

R~.--~:7:--- R ::.:.....u~

R

R

Li

There is also in our laboratory under development, a new liquid phase "phantom" synthesis which will be accessible with the equipment of most organic laboratories and which promises to be an extraordinarily general synthesis for polylithium organic compounds. Acknowledgement We are grateful to the National Science Foundation (CHE-8521390) and Robert A. Welch Foundation (F-700) for support of this work.

192

Dedication This manuscript is dedicated to my close friend and former colleague, F. Albert Cotton.

I have had the highest respect for Al

Cotton as a person and as a scientist for many years.

Al has contributed

to my career in many ways and is a person from whom I have learned much. He encouraged me early in my career, when there were long odds against the RJL idea that a new form of matter such as polylithium organic compounds could exist when there were suggestions and reports in the literature that such compounds were unstable and impossible to prepare. Al Cotton is a mentor who supported and encouraged the early studies in this area.

F. Albert Cotton on an earlier occasion was also responsible

for selecting and purchasing the first bottle of outstanding wine ever consumed by RJL, starting an exciting new hobby. References 1. 2. 3.

4.

5. 6. 7. 8. 9. 10. 11. 12. 13.

J.A. Gurak, J.W. Chinn, Jr. and R.J. Lagow, ~Am. Chern. Soc. 104:2637 (1982). L.A. Shimp, J.A. Morrison, J.A. Gurak, J.W. Chinn, Jr., and R.J. Lagow, ~Am. Chern. Soc. 103:5951 (1981). (a) C. Chung and R.J. Lagow, ~Chern. Soc., Chern. Commun. 1970 (1972); (b) L.A. Shimp and R.J. Lagow, ~Am. Chern. Soc. 95:1343 (1973); (c) L.G. Sneddon and R.J. Lagow, J. Chern. Soc., Chern. Commun. 302 (1975); (d) J.A. Morrison, c.-chung, and R.J. Lagow, J. Am. Chern. Soc. 97:5015 (1975); (e) J.A. Morrison and R.J. Lagow, Inorg. Chern. 16:2972 (1977); (f) L.A. Shimp, C. Chung, and R.J. Lagow, Inorg. Chim. Acta 29:77 (1978); (g) K.M. Abraham and R.J. Lagow, Tetrahedron Lett. 3:211 (1979); (h) L.A. Shimp and R.J. Lagow, ~Am. Chern. Soc. 101:2214 (1979); (i) L.A. Shimp and R.J. Lagow, ~ Org. Chern. 44:2231 (1979); (j) H. Kawa, B.C. Manley, and R.J. Lagow, ~Am. Chern. Soc. 107:5313 (1985); (k) H. Kawa, B.C. Manley and R.J. Lagow, Polyhedron 19/20:2023 (1988). (a) R. West and P.C. Jones, J. Am. Chern. Soc. 91:6155 (1969); (b) G.A. Gornowicz and R. West, ~Am. Chern. Soc. 93:1720 (1971); (c) R. West, P.A. Carney, and I.C. Mim~ J. Am. Chern. Soc. 81:3788 (1965); (d) G.A. Gornowicz and R. West, J. Am. Chern. Soc. 93:1714 (1971); (e) W. Priester, R. West, and T.~ Chwang, J. Am. Chern. Soc. 98:8413 (1976). - - -- -K. Ziegler, K. Nagel, and M. Patheiger, ~ Anorg. Allgern. Chern. 282:345 (1955). F.J. Landro, J.A. Gurak, J.W. Chinn, Jr., and R.J. Lagow,~ Organomet. Chern. 249:1 (1983). J.W. Chinn, Jr. and R.J. Lagow, Organometallics 3:75 (1984). (a) J. Berkowitz, D.A. Bafus, and T.L. Brown, ~ Phys. Chern. 65:1380 (1961); (b) T.L. Brown, Ann. N.Y. Acad. Sci. 136:98 (1966). M.Y. Darensbourg, B.Y. Kimura~E. Hartwell, and T.L. Brown, ~Am. Chern. Soc. 92:1236 (1970). ~Collins, J.D. Dill, E.D. Jemmis, Y. Apeloig, P.v.R. Schleyer, and J.A. Pople, ~Am. Chern. Soc. 98:5419 (1976). E.D. Jemmis, P.v.R. Schleyer, and J.A. Pople, ~ Organomet. Chern. 154:327 (1978). Y. Apeloig, P.v.R. Schleyer, J.S. Brinkley, J.A. Pople, and W.A. Jorgensen, Tetrahedron Lett. 43:3923 (1976). E.D. Jemmis, D. Poppinger, P.v.R. Schleyer, and J.A. Pople, J. Am. Chern. Soc. 99:5796 (1977). -193

G. Rauscher, T. Clark, D. Poppinger, and P.v.R. Schleyer, Angew. Chern. 90:306 (1978). 15. E.D. Jemmis, J. Chandrasekhar, and P.v.R. Schleyer, ~Am. Chern. Soc. 101:2848 (1979). 16. A.J. Kos, D. Poppinger, P.v.R. Schleyer, and W. Thiel, Tetrahedron Lett. 21:2151 (1980). 17. Y. Apeloig, P.v.R. Schleyer, J.S. Binkley, and J.A. Pople, ~Am. Chern. Soc. 98:4332 (1976). 18. (a) R. Hoffmann, R.G. Alder, and C.F. Wilcox, Jr., J. Am. Chern. Soc. 92:49g2 (1970); (b) R. Hoffmann, Pure~ Chern. 28:1'ii"l(1971).-19. W.D. Laidig and H.F. Schaefer, ~Am. Chern. Soc. 100:5972 (1978). 20. E.W. Ni1ssen and A. Skancke, ~ Organomet. Chern. 116:251 (1976). 21. S.M. Bachrach and A. Streitwieser, Jr., J. Am. Chern. Soc. 106:5818 ------ (1984). 22. A.J. Kos, E.D. Jemmis, P.v.R. Schleyer, R. Gleiter, J.A. Pople, and U. Fischback, J. Am. Chern. Soc. 103:4996 (1981). 23. W.D. Laidig an~H~ Schaefer, ~Am. Chern. Soc. 101:7184 (1979). 24. J.W. Chinn, Jr., F.J. Landro, P.v.R. Schleyer, J.A. Pople, and R.J. Lagow, ~Am. Chern. Soc. 104:4275 (1982). 25. P.v.R. Schleyer, E.D. Jemmis, J. Chandrasekhar, E.-U. Wurthwein, A.J. Kos, B.T. Luke, and J.A. Pople, J. Am. Chern. Soc. 105:484 - - -- -(1983). 26. A.E. Reed and F. Weinhold, J. Am. Chern. Soc. 107:1919 (1985). 27. J.W. Chinn, Jr. and R.J. Lagow~.~Chern. Soc. 106:3694 (1984). 28. Swanson et al., Nat!. Bur. Std. Mon~25:1 (1962). 29. Stackelberg and Quatram, Z.~s~hern. Leipzig B27:50 (1934). 30. R. Zerger, W. Rhine, and G.D. Stucky, J. Am. Chern. Soc. 96:6048 - - -- -(1974). 31. (a) E. Weiss and A.C. Lucken, ~ Organomet. Chern. 2:197 (1964); (b) E. Weiss and G. Hencken, ~ Organomet. Chern. 21:265 (1970). 32. Refinement of this model was carried out with the soft constraint that D(1)-D(2) be separated by 1.78(25) A. This was required because of the close approach of D(1) to the twofold axis passing through the carbon atom. 33. E. Weiss, Acta Cryst. 16:681 (1963). 34. G.S. Smith, Q.C. Johnson, D.K. Smith, D.E. Cox, R.L. Snyder, R.-S. Zhou and A. Zalkin, Brookhaven Synchrotron Light Source Annual Report, 227 (1988). 35. E. Zintl and A. Harder, ~ Phys. Chern. 28:478 (1935). 36. (a) W. Rhine, D. Groves and G.D. Stucky, J. Am. Chern. Soc. 93:1553 (1971); (b) W. Rhine, S.W. Peterson and G:D.-siucky, J. Am. Chern. -- -Soc. 97:6401 (1975). 37. ~P.v.R. Schleyer, E.-U. Wurthwein, E. Kaufmann, T. Clark, and J.A. Pople, ~Am. Chern. Soc. 105:5930 (1983); (b) P.v.R. Schleyer, "New Horizons of Quantum Chemistry," P.-O. Lowdin and B. Pullman, eds., D. Reidel, New York (1983), pp. 95-109; (c) E.-U. Wurthwein, P.v.R. Schleyer, and J.A. Pople, J. Am. Chern. Soc. 106:6973 (1984); (d) P.v.R. Schleyer, E.-U. Wurthwein~nd J.A. Pople, J. Am. Chern. -- -Soc. 104:5839 (1982). 38. Private communication between RJL and P.v.R. Schleyer prior to the Lagow communication (ref. 24). 39. (a) J.A. Gurak, J.W. Chinn, Jr., C.S. Yannoni, H. Steinfink, and R.J. Lagow, Inorg. Chern. 23:3717 (1984); (b) C.S. Yannoni, J.A. Gurak, J.W. Chinn, Jr., R.A. Kendrick, and R.J. Lagow, Inorg. Chim. Acta 96:L75 (1985). 40. W.S. Veeman and E.M. Mengerand, ~ Magn. Reson. 46:257 (1982). 41. L.D. McKeever, R. Waack, M.A. Doran, and E.B. Baker, J. Am. Chern. -- -Soc. 91:1057 (1969). 42. J.W. Chinn, Jr., G.F. Meyers, M.B. Hall, and R.J. Lagow, ~Am. Chern. Soc. 107:1413 (1985). 43. H.P.S. Chauhan, H. Kawa, and R.J. Lagow, ~ Org. Chern. 51:1632 (1986). 14.

194

ORGANOMETALLIC CHEMICAL VAPOR DEPOSffiON OF GaAs AND RELATED SEMICONDUCTORS USING NOVEL ORGANOMETALLIC PRECURSORS Alan H. Cowley Department of Chemistry The University of Texas at Austin Austin, TX 78712

INTRODUCTION Gallium arsenide and related ternary materials are used extensively in the electronics industry (Table I). Other III- V semiconductors such as indium phosphide are employed in the telecommunications industry and current interest in indium antimonide stems from its potential use in infrared sensors. Broadly speaking, two approaches are available for the preparation of III- V materials, namely molecular beam epitaxy (MBE) and organometallic chemical vapor deposition (OMCVD). Of these approaches, the OMCVD method is more suited to the production of larger quantities of compound semiconductors. Typically, the OMCVD method is based on the reaction of Me 3Ga or Me 3In with PH3 or AsH3 at 600-

700

·c. e.g.l

M= Ga, In; E = P, As The conventional OMCVD method possesses a number of disadvantages which include problems associated with handling highly toxic materials such as AsH3, difficulties in controlling the stoichiometry of the product, and unwanted side reactions. It would also be desirable to lower the deposition temperatures to avoid the diffusion of impurties, spurious interfacial reactions, and film stresses. Exciting opportunities exist, therefore, for the synthetic chemist working closely with materials growth and evaluation experts to prepare novel, tailored precursors that produce the desired III- V materials at lower temperatures. In a very fruitful collaboration with my colleague, Professor Richard A. Jones, we have begun to develop a wide range of what we refer to as single-source precursor clusters. These clusters Metal-Meta/ Bonds and Clusters in Chemtstry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

195

Table I.

Some Uses of Gallium Arsenide and Related Materials

Field effect transistors Mixer diodes Transferred electron oscillators Hetero-junction bipolar transistors IMPA TT diodes High electron mobility transistors Light emitting diodes Photocathodes Laser diodes Quantum well laser diodes Integrated opto-electronics Waveguides Solar cells

Field effect transistors Transferred electron oscillators High electron mobility structures Light emitting diodes Laser diodes Quantum well laser diodes Integrated opto-electronics Avalanche photo-detectors

196

are of the general type RnMxEyRm'• where M and E are group III or group V elements, respectively, and Rn and Rm' are substituents that are capable of facile hydrocarbon elimination. Our precursors feature the additional advantages of ease of purification, short-term air stability, and relatively low toxicity. Synthetic Chemist:ty It is clear from the current review literature2 that, with some notable exceptions,3 little attention has been paid to the organometallic chemistry of gallium and indium. The initial stages of our program were therefore concerned with exploration of the range of possible III-V organometallic compounds. As illustrated in Figure 1, compounds of 1:1, 1:2, 1:3, and 1:4 stoichiometry have now been isolated and characterized. Our entry into this field was provided by the metathetical reaction of lithium phosphide or arsenide with the

r

appropriate metal trichloride which afforded compounds of 1:3 stoichiometry, 1.4

t-Bu2

/M~ E-t-Bu 2

t-Bu 2E

1 In the same year, Wells et al.5 described the synthesis ofGa[As(mesityl) 2] 3 via the reaction of (mesityl)2AsSiMe3 with GaCl3. Compounds of type 1 are useful reagents for the synthesis of mixed metal compounds and clusters with 1: 1 stoichiometry. For example, treatment of 1 with trialkylgallium or trialkylindium affords very high yields of 1:1 compounds of the type [R 2ME-t-Bu 2]z (2). 6 These compounds can also be made by treatment of GaCl3 with one equivalent of t-Bu2ELi (E=P,As) and two equivalents ofRLi(R =Me, n-Bu).4,6

2 MC13 + 2 t-Bu 2ELi + 4RLi

--~

2 197

Me

R '/

R-E

R

'\.I

E-R

I

I

Me-.M

M-Me

Me" " ' /

E /'

R

1:1

'

t-Bu

Me

' M'\./

'Me

"\.

/

t-Bu2sb....._

t-Bu Sb /

Cl

/'

Ga,

_;~Ja

"

Sb

/

Sb-t-Bu2

/ 't-Bu

t-Bu

R

1:2

M=Ga,ln; E=P,As,Sb 1:4

1:3

E-t-Bu2

~ t-Bu2E/

Figure 1.

198

~E-t-Bu2

Ill-Y Compounds of Various Stoichiometries

Alkane or arene elimination also represents a viable synthetic approach to Ill-V cluster compounds and indeed this method was used in pioneering studies in the 1960's by Coates and co-workers.3 We and others7 have since used the alkane or arene elimination method to prepare several 1: 1 compounds, a recent example of which is shown below. 8

At the present time, rather few compounds of 1:2 stoichiometry have been prepared. Wells et al.5.9 have synthesized gallium-arsenic compounds of the type [XGa{AsR2h1n (X=Cl,Br; R= mesityl, Me3SiCH2) either by salt or silyl halide elimination, and in our laboratories we have prepared an indium-antimony compound of the same stoichiometry (3) by silyl halide elimination.lO t-Bu

'Sb, /

CI., II,,

t-Bu

In

/

"

••• Sb-t-Bu2

In

t.''

/ " Sb/~ Cl

t-Bu2Sb

/"- t-Bu

t-Bu

3

Interestingly, the formation of 3 constitutes the first report of a compound with indium(lll)-antimony a-bonds. It is also appropriate to point out that it is not possible to prepare 3 by salt elimination because of the reduction of In(Ill) to ln(O) by the stibido anion, [t-Bu2Sbr. The silyl halide methodology can also be very useful when alkane elimination becomes sluggish due to the presence of bulky ligands. 5 Sterle effects also play a role in the stoichiometry and properties of the final product. As pointed out above, the reaction of MCl3 (M=Ga,In) with the bulky phosphide or arsenide anions, [t-Bu2EL results in 1:3 complexes, 1. 4,6 If the less bulky phosphide or arsenide reagents, Ph2ELi, are employed in the same stoichiometry, only intractable products can be obtained. However, when the reactant ratio is changed to 1:4 crystalline "ate" complexes can be isolated.11

199

THF

M

=

Ga, In; E

=

P, As

In addition to the foregoing, mention should also be made of the 3:1 stoichiometry

complex, [(THF)Br2Gal)As which was synthesized via the reaction of GaBr3 with (Me3Si)3As. 12 Structural Information By means of X-ray crystallography, it has been established that gallium-phosphorus4 and gallium-arsenic4, 13 dimers of 1:1 stoichiometry adopt diborane-like structures, 4. An

4 M=Ga; E=P, As; R=Me, n-Bu; R'=t-Bu

approximately tetrahedral geometry is evident at both theM and E centers and the M-E bond lengths are close to the sums of the respective single-bond covalent radii. The geometry of each M2E2 ring is planar and there is no indication of M ... MorE ...E cross-ring bonding. These essential structural features have since been found for several other such dimers and also for the 2:1 stoichiometry complexes [BrGa{As(CH2SiMe3)2hl9 and [Clln{Sb-t-Bu2hh.10 In the latter complexes the exocyclic R2'E groups adopt a mutually trans disposition. The weakly bonded dimer of 1:3 stoichiometry, [{(Me3SiCH2)2As}3Gah, also possesses structure 4 14. However, in this case the Ga2As2 ring is puckered. The trimers exhibit a wide range of conformations for the M 3E 3 rings. For example, [Cl2GaSb-t-Bu2l3 15 and [Br2GaAs(CH2SiMe3)2l39 adopt twist-boat and irregular boat

200

conformations respectively. The case of [Me2InAsMe2]3 is particularly interesting because of the presence of puckered and approximately planar conformations in the same asymmetric unit.15 Dynamic 1H NMR data for [Me 2InAsMe 2] 3 establish that the In-methyl groups remain equivalent to -80 ·c. Clearly, it will be of interest to acquire more structural information on the trimers. However, the available data imply that the various possible M3E3 ring conformations are close in energy. Finally, the "ate" anions, [M(EPh2)4L have been found to possess C1 skeletal structures. II These anions are therefore much less symmetrical than transition metal analogues which adopt D 2d skeletal geometries.l6 Deposition Studies At the present time, most of our deposition studies have focussed on 5 as the single-source precursor. I? Films of gallium arsenide have been grown in a cold-wall OMCVD reactor.

t-Bu Me ·.,,

"-/ As

/ 'Ga/ Me,;::?' "'-. t-Bu

' As

/

t-Bu

'Ga''

..• Me

~Me

/"'-. t-Bu 5

This reactor was designed and built in the laboratories of Professor John G. Ekerdt of the Department of Chemical Engineering at the University of Texas at Austin, and it is a pleasure to acknowledge this important collaboration. The reaction conditions are summarized in Table IT, together with a listing of the substrates employed and the substrate cleaning protocols. The use of the conditions outlined in Table IT resulted in typical fllm growth rates of 0. 7 to 1.0 Jlm per hour (as measured by Dektak). The gross compositions of the gallium arsenide films were established by X-ray photoelectron spectroscopy (XPS). Within experimental error, a Ga:As ratio of 1:1 was confirmed in each case by measurement of the intensities of the Ga3d and As3d signals at 18.8 and 40.9 eV, respectively. Carbon was not detected in the XPS experiments which implies that the carbon impurity levels are less than 1000 ppm. Secondary ion mass spectroscopy (SIMS) is a more sensitive technique for the detection of impurities. SIMS

201

Table II.

OMCVD Reaction Conditions

Reactor Pressure: 1x1o-4 torr

Substrate Temperature: 550 - 700 ·c Saturator Temperature: 130 ·c

Carrier Gas: H2, He Substrates: Quartz cleaned in trichloroethylene rinsed in methanol rinsed in deionized water dried with nitrogen Si (100) 3" off toward [011] cleaned in trichloroethylene rinsed in methanol rinsed in deionized water etched in 20% hydrofluoric acid rinsed in deionized water dried with nitrogen heated to 775

·c under hydrogen

GaAs (100), semi-insulating and Si doped cleaned in trichloroethylene rinsed in methanol rinsed in deionized water

rinsed in deionized water dried with nitrogen heated to growth temperature in vacuum

202

studies of our films revealed the carbon levels to be comparable to those found in electronic grade gallium arsenide. Apart from carbon, traces of Na, Al, Si, K, Cr, Fe, and In were also detected. Undoubtedly, these impurities were present in the starting materials or were introduced during the synthetic procedures. X-ray diffraction (XRD) measurements indicate that polycrystalline GaAs growth occurred on all three substrates. Interestingly, the GaAs films grown on semi-insulating GaAs(lOO) and Si-doped GaAs(lOO) exhibit low-temperature photoluminescence (PL) signals, thus implying the presence of crystalline domains. Preliminary thermal decomposition studies on 5 show that methane and isobutylene are the main hydrocarbons produced during layer growth. This, plus the solid state crystal structure of the compound which indicates a relatively short H ...Ga interaction of a tert-butyl-As hydrogen atom4 lead us to propose tentatively a decomposition pathway involving transfer of this H atom to the Ga and expulsion of CH4. This process, accompanied by a cleavage of the As-C bond producing isobutylene would then result in the clean removal of the hydrocarbon groups from the inner Ga2As2 core. Finally, attention is drawn to the very recent work of Bradley et al.18 on the use of the indium-phosphorus analogue of 5 for the growth of epitaxial indium phosphide using metal organic molecular beam epitaxy (MOMBE). Ac1mowled&ment I am most grateful to the organizers of the IUCCP Symposium for the opportunity to

help recognize Professor F. A. Cotton's seminal work on the quadruple metal-metal bond. Gratitude is also expressed to the National Science Foundation, the Robert A. Welch Foundation, and the Texas Advanced Technology Research Program for fmancial support. It is also a pleasure to thank Professors Richard A. Jones and John G. Ekerdt for their collaboration and the following co-workers for their dedication and hard work: B. Benac, R. Geerts, D. Giolando, P. Harris, D. Heaton, K. Kidd, M. Mardones, C. Nunn, J. Power, S. Schwab, and D. Westmoreland.

References

Alwl. Pbys. Lett. 12:156 (1974).

1.

H. M. Manasevit,

2.

D. G. Tuck, 1981, Chapter 7 in Volume 1 of "Comprehensive Organometallic Chemistry," G. Wilkinson, F. G. A. Stone, and E. W. Abel, eds., Pergamon Press Ltd.

3.

G. E. Coates and J. Graham, J. Chern. Soc. 233 (1963); 0. T. Beachley and G. E. Coates, J. Chern, Soc. 3241 (1965).

4.

A. M. Arif, B. L. Benac, A. H. Cowley, R. L Geerts, R. A. Jones, K. B. Kidd,

J. M. Power, and S. T. Schwab, J. Chern. Soc., Chern, Commun. 1543 (1963).

203

5.

C. G. Pitt, A. P. Purdy, K. T. Higa, and R. L. Wells, Or~anornetallics 5:1266 (1986); C. G. Pitt, K. T. Higa, A. T. McPhail, and R. L. Wells, Inorg. Chern. 25:2484 (1986).

6.

A.M. Arif, B. L. Benac, A. H. Cowley, R. A. Jones, K. B. Kidd, and C. M. Nunn,

7.

0. T. Beachley, J.P. Kopasz, H. Zhang, W. E. Hunter, and J. L. Atwood, L.

New J. Chern. 12:553 (1988). Organornet. Chern. 325:69 (1987). 8.

A. H. Cowley, R. A. Jones, M.A. Mardones, and C. M. Nunn, to be published.

9.

A. P. Purdy, R. L. Wells, A. T. McPhail, and C. G. Pitt, Organornetallics 6:2099 (1987).

10. A. R. Barron, A. H. Cowley, R. A. Jones, C. M. Nunn, and D. L. Westmoreland, Polyhedron 7:77 (1988). 11. C. J. Carrano, A. H. Cowley, D. M. Giolando, R. A. Jones, C. M. Nunn, J. M. Power,Inor~.

Chern. 27:2709 (1988).

12. R. L. Wells, S. Shafieezed, A. T. McPhail, and C. G. Pitt, J. Chern. Soc., Chern. Cornrnun. 1823 (1987). 13. R. L. Wells, A. P. Purdy, A. T. McPhail, and C. G. Pitt, J.

Or~anornet.

Chern.

308:281 (1986). 14. R. L. Wells, A. P. Purdy, K. T. Higa, A. T. McPhail, and C. G. Pitt, J. Organornet. Chern. 325:C7 (1987). 15. A. H. Cowley, R. A. Jones, K. B. Kidd, C. M. Nunn, and D. L. Westmoreland, J. Organomet. Chern. 341:C1 (1988). 16. M. H. Chisholm, F. A. Cotton, andM. W. Extine, Inorg. Chem 17:1329 (1978);

R. T. Baker, P. J. Krusic, T. H. Tulip, J. C. Calabrese, and S. S. Wreford, J. Am. Chern.Soc. 105:6763 (1983). 17. A. H. Cowley, B. L. Benac, J. G. Ekerdt, R. A. Jones, K. B. Kidd, J. Y. Lee, and J. E.

Miller, J. Am. Chern. Soc. 110:6248 (1988). 18. D. A. Andrews, G. J. Davies, D. C. Bradley, M. M. Fak:tor, D. M. Frigo, and E.A.D. White, Semicond. Sci. Techno!. 3:1053 (1988).

204

SURFACE CHEMISTRY OF MIXED-METAL SYSTEMS

D. W. Goodman Department of Chemistry Texas A&M University College Station, TX 77843-3255 INTRODUCTION Interest in bimetallic catalysts has risen steadily over the years because of the commercial success of these systems [1]. This success results from an enhanced ability to control the catalytic activity and selectivity by tailoring the catalyst composition [2,3]. A long-standing question regarding such bimetallic systems is the nature of the properties of the mixed metal system which give rise to its enhanced catalytic performance relative to either of its individual metal components. These enhanced properties (improved stability, selectivity andjor activity) can be accounted for by one or more of several possibilities. First, the addition of one metal to a second may lead to an electronic modification of either or both of the metal constituents. This electronic perturbation can result from direct bonding (charge transfer) or from a structural modification induced by one metal upon the other. Secondly, a metal additive can promote a particular step in the reaction sequence and, thus, act in parallel with the host metal. Thirdly, the additive metal can serve to block the availability of certain active sites, or ensembles, prerequisite for a particular reaction step. If this "poisoned" reaction step involves an undesirable reaction product, then the net effect is an enhanced overall selectivity. Further, the attenuation by this mechanism of a reaction step leading to undesirable surface contamination will promote catalyst activity and durability. The studies reviewed here are part of a continuing effort [7-10] to identify those properties of bimetallic systems which can be related to their superior catalytic properties. EXPERIMENTAL These studies were carried out utilizing the specialized Metal-Meta/ Bonds and Clusters in Chemistry and Catalysis Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

205

apparatus described in references [11,12]. This device consists of two distinct regions, a surface analysis chamber and a microcatalytic reactor. The custom built reactor, contiguous to the surface analysis chamber, employs a retraction bellows that supports the metal single crystal and allows translation of the catalyst in vacuo from the reactor to the surface analysis region. Both regions are of ultrahigh vacuum construction, bakeable, and capable of ultimate pressures of less than 2 X 10- 10 Torr. Auger spectroscopy (AES) is used to characterize the sample before and after reaction. A second chamber was equipped with Auger spectroscopy, low energy electron diffraction (LEED) and a mass spectrometer for temperature programmed desorption (TPD). RESULTS AND DISCUSSION A pivotal question to be addressed of bimetallic systems (and of surface impurities in general) is the relative importance of ensemble (steric or local) versus electronic (nonlocal or extended) effects in the modification of catalytic properties. In gathering information to address this question it has been advantageous to simplify the problem by utilizing models of a bimetallic catalyst such as the deposition of metals on single- crystal substrates in the clean environment familiar to surface science. Many such model systems have been studied but a particularly appealing combination is that of cu on Ru. cu is immiscible in Ru which facilitates coverage determinations by TPD [4] and circumvents the complication of determining the 3-d composition. The adsorption and growth of Cu films on the Ru(0001) surface have been studied [4-10,13-20] by work function function measurements, LEED, AES, and TPD. The results from recent studies [4-10] indicate that for submonolayer depositions at lOOK the cu grows in a highly dispersed mode, subsequently forming 2-d islands pseudomorphic to the Ru(0001) substrate upon annealing to 300K. Pseudomorphic growth of the copper indicates that the copper-copper bond distances are strained approximately 6% beyond the equilibrium bond distances found for bulk copper. A comparison of CO desorption from Ru [7] from multilayer cu (.10ML) on Ru and 1ML Cu on Ru is shown in Fig. 1. The TPD features of the lML Cu (peaks at 160 and 210K) on Ru are at temperatures intermediate between Ru and bulk cu. This suggests that the monolayer cu is electronically perturbed and that this perturbation manifests itself in the bonding of co. An increase in the desorption temperature relative to bulk Cu indicates a stabilization of the CO on the monolayer Cu suggesting a coupling of the CO through the Cu to the Ru. The magnitude of the CO stabilization implies that the electronic modification of the cu by the Ru is significant and should be observable with a band structure probe. Recent angular resolved photoemission studies [7] indeed show a unique interface state which is likely related to the altered co bonding on Cu films intimate to Ru. Figure 2 shows the results [7] of CO chemisorption on the CujRu(OOOl) system as a function of the Cu coverage. In each case the exposure corresponds to a saturation coverage of co. 206

co-cu

0

1 ML Cu/Au

A.u


200

100

300

400

100

Temperature (K)

Figure 1.

TPD Results for CO adsorbed to saturation levels on clean Ru(OOOl), on multilayer CU, and on a lML Cu covered Ru(OOOl). (Reproduced with permission from Ref. 7. Copyright 1986, J.E. Houston).

0

o..u
TEMPERA TURE-K

Figure 2.

TPD results corresponding to CO adsorbed to saturation levels on the clean Ru(OOOl) surface, and from this same surface containing various coverages of Cu. (Reproduced with permission from Ref. 7. Copyright 1986, J.E. Houston).

207

Most apparent in Figure 2 is a monotonic decrease upon addition of Cu of the CO structure identified with Ru (peaks at 400 and 480K) and an increase of the CO structure corresponding to Cu (peaks at 160 and 210K). The buildup of a third feature at -300K (indicated by the dashed line) is assigned to correspond to co desorbing from the edges of Cu islands. Integration of the 200, 275, and 300K peaks provides information regarding island sizes, that is, perimeter-to-island area ratios, at various cu coverages. For example, at ecu = 0.66 the average island size is estimated to be approximately 50A in diameter. This island size is consistent with an estimate of the 2-d island size corresponding to this coverage of 40-60A derived from the width of the LEED beam profiles [7]. Model studies of the CujRu(0001) catalyst have been carried out [8] for methanation and hydrogenolysis reactions. These data suggest that copper merely serves as an inactive diluent, blocking sites on a one-to-one basis. Similar results have been found in analogous studies [21] introducing silver onto a Rh(111) methanation catalyst. Sinfelt [22] has shown that copper in a CujRu catalyst is confined to the surface of ruthenium. Results from the model catalysts discussed here then should be relevant to those on the corresponding supported, bimetallic catalysts. Several such studies have been carried out investigating the addition of copper or other Group 18 metals on the rates of CO hydrogenation [23-25] and ethane hydrogenolysis [25] catalyzed by ruthenium. In general, these studies show a marked reduction in activity with addition of the Group 18 metal suggesting a more profound effect of the Group 18 metal on ruthenium than implied from the model studies. A critical parameter in the supported studies is the measurement of the active ruthenium surface using hydrogen chemisorption techniques. Haller and coworkers [26,27] have recently suggested that hydrogen spillover during chemisorption may occur from ruthenium to copper complicating the assessment of surface Ru atoms. Recent studies in our laboratory [5,6] have shown directly that spillover from Ru to Cu can take place and must be considered in the hydrogen chemisorption measurements. H2 spillover would lead to a significant overestimation of the number of active ruthenium metal sites and thus to significant error in calculating ruthenium specific activity. If this is indeed the case, the results obtained on the supported catalysts, corrected for the overestimation of surface ruthenium, could become more comparable with the model data reported here. Finally, the activation energies observed on supported catalysts in various laboratories are generally unchanged by the addition of Group 18 metal [26-28] in agreement with the model studies. These arguments suggest that Ru specific rates for methanation and ethane hydrogenolysis on supported CujRu catalysts approximate those values found for pure Ru. As a consequence, the rates for cyclohexane dehydrogenation reaction on supported CujRu, similarly corrected, must exceed those specific rates found for pure Ru. The uncorrected specific rates for cyclohexane dehydrogenation on the supported CujRu system remain essentially unchanged upon addition of cu to Ru [28]. An activity enhancement for cyclohexane dehydrogenation in the mixed CujRu system relative to pure Ru is most 208

surprising given that Cu is less active for this reaction than Ru. Figure 3 shows the effect of the addition of cu to Ru on the rate of cyclohexane dehydrogenation to benzene. The overall rate of this reaction is seen to increase by approximately an order of magnitude at a copper coverage of 3/4 of a monolayer. This translates to a Ru specific rate Above this coverage, the rate falls to an enhancement of -40. activity approximately equal to that of Cu-free Ru. The observation of non-zero rates at the higher Cu coverages is believed to be caused by three dimensional clustering of the Cu overlayers [29]. Similar data have been obtained for this reaction on epitaxial and alloyed AujPt(111) surfaces [30]. The rate enhancement observed for submonolayer cu deposits may relate to an enhanced activity of the strained cu film for this reaction due to its altered geometric [29] and electronic [9] properties. Alternatively, a mechanism whereby the two metals cooperatively catalyze different steps of the reaction may account for the activity promotion. For example, dissociative H2 adsorption on bulk Cu is unfavorable due to an In the activation barrier of approximately 5 kcaljmol [31]. combined CujRu system, Ru may function as an atomic hydrogen sourcejsink via spillover tojfrom neighboring Cu. A kinetically controlled spillover of H2 from Ru to Cu, discussed above, is consistent with an observed optimum reaction rate at an intermediate cu coverage. Finally, we note the differences between a Ru(0001) catalyst with or without added Cu with respect to attaining steady-state reaction rates. On the cu-free surface, an induction time of approximately 10 minutes is required to achieve steady state activity. During this time, production of benzene is quite low while the hydrogenolysis to lower alkanes, primarily methane, is significantly higher than at steady-state. During this induction time the carbon level (as determined by Auger spectroscopy) rises to a saturation value coincidental with the onset of steady state reaction. This behavior suggests that a carbonaceous layer on the metal surface effectively suppresses carbon-carbon bond scission, or hydrogenolysis, on the Ru surface. Cu addition leads to an enhanced rate of benzene production with little or no induction time. That is, the initial rate of cyclohexane hydrogenolysis, relative to the Further, Cu reduces the cu-free surface, is suppressed. relative carbon buildup on the surface during reaction. Thus, cu may play a similar role as the carbonaceous layer in suppressing cyclohexane hydrogenolysis whiie concurrently stabilizing those intermediates leading to the product benzene. In addition, copper may serve to weaken the chemisorption bond of benzene and thus limit self-poisoning by adsorbed product. This latter possibility has been proposed by Sachtler and Somorjai [30] to explain the role of Au in Au/Pt(111) catalysts for this reaction. A weakening of benzene chemisorption satisfactorily accounts for our observation that the reaction changes from zero order in cyclohexane on Ru(0001) to approximately first order upon the addition of cu. A second bimetallic system which has been thoroughly 209

studied is nickel adsorbed onto tungsten [32,33]. On both W(llO) and W(lOO), Ni is adsorbed layer-by-layer. Annealing Ni layers with coverages less than 1.3 ML to 1200K produces little change in the Ni(848eV)/W(179eV) AES ratio. However, for Ni coverages above 1.3ML, a 1200K anneal results in a very slow increase in this AES ratio with coverage, indicating either alloy or 3-dimensional island formation. co adsorption [32] as a function of Ni coverage on W(llO) has been investigated. As the Ni coverage is increased from 0.3 to l.OML, adsorption on the W(llO) substrate decreases, as evidenced by a reduction in the co features between 225 and 350K, while a maximum for lML Ni compares with 430K for the co TPD peak maximum for Ni(lll) [35]. Increasing the Ni coverage above l.OML results in a broadening of the 380K CO TPD peak and in the development of a shoulder feature, suggestive of bulk Ni co desorption, at -430K. co chemisorption on the Ni/W{lOO) surface is similar to co adsorption on Ni/W(llO). As the Ni coverage is increased from 0.3 to l.OML, decreasing intensity in the TPD features associated with W(lOO) are evident near 300K. At a Ni coverage of lML, the CO TPD peak maximum is reduced by approximately 50K from the corresponding peak maximum on Ni{lOO) [35]. For coverages greater than lML, a clear shoulder at 420- 450K is observed, indicating that second and successive Ni layers have chemisorptive properties very similar to bulk Ni. Thus the W substrates clearly alter the chemisorptive properties of the first Ni layer, but have only slight effects on the second and subsequent layers. That CO chemisorption is perturbed on strained-layer Ni is not surprising in view of CO chemisorption behavior on other metal overlayer systems. For example, on CujRu it has been proposed that charge transfer from Cu to Ru results in decreased occupancy of the cu 4s level. This electronic modification makes Cu more "nickel-like," and results in an increase in the binding energy for co. Similarly cu;w [36] also exhibits charge transfer to the substrate and a increase in CO binding strength to Cu. In another case where the CO binding energy increases, Ni/Ru [37], an increase in the density of states is observed close to the Fermi level. The increased electron density may result in increased metal-CO backbonding, which in turn would increase the binding energy of co. In contrast to the above examples, co on Ni/W is less strongly bound to the Ni monolayer than to bulk Ni. One explanation for this effect is that the charge transfer observed from Ni to W results in a shift of the Ni d levels, relevant to CO bonding, to higher binding energies (i.e. farther from the Fermi level). Indeed, such an effect has been observed in the case of Ni/Nb(llO) and Pd/Ta(llO) [38). Similarly, results on other group VIII metal - W systems [3940] have shown a decrease in the CO binding strength. The catalytic activity of strained layer Ni on W(llO) for methanation and ethane hydrogenolysis has been studied as a function of Ni coverage [41]. The activity per Ni atom site for methanation, a structure insensitive reaction, is independent of the Ni coverage (Fig. 4) and similar to the 210

10.0

0

•.

!

8.0

c

-.• 0

u

.

Gl

~

c-;

8.0

4.0

a:

2.0

0.0

o.o

1.2 Copper coverage (monolayars)

Figure 3.

Relative rate of reaction versus surface Cu coverage on Ru(OOOl) for cyclohexane dehydroH2 / genation to benzene. PT ; 101 Torr. (Reproduced cyclohexane ; 100. T ; 650 K. with permission from Ref. 10. Copyright 1987, D.W. Goodman).

>

()

~ 10_,

::I

a

Ill

...a:

a::

Ill

>

0

z

a:

::I

..... 10 2 Ill

z c

:z: .....

Ill

:1!

• .:I IlL NI/W ( 1 101 .I IlL NI/W (1101 A 1.0 IlL NI/W 11101 0 .4 IlL NI/W ( 1001 . 7 IlL NIIW ( 1001 • + .!I IlL NI/W ( 1001

0

1.4

1.6

1.8

n

2.0

2.2

2.4

1000/T (K- 1)

Figure 4.

Arrhenius plot for CH4 synthesis over several different Ni coverages on W(l10) and W(lOO) at a total reactant pressure of 120 Torr, H2/CO ; (Reproduced with permission from Ref. 41. 4/1. Copyright 1987, D.W. Goodman).

211

-•

'

>-

CJ 10 1

z Ill

:I

0

Ill

a:: ~ a:: Ill > Hi 2 0

z

a::

:I

1-

Ill

z

c

...

:z: Hi 3

Ill

::1

NIIW(1101

e

o

•

10- 4 1.5

Figure 5.

Table 1. Adsorbate

Cu cu Ni Ni Ni Pd Pd Pd Fe Fe

212

0.1 ML Nl 0.4 ML Nl 1.2 ML Nl

2.1

1.9

1.7

Arrhenius plots of the rates of ethane hydrogenolysis versus Ni coverage on W(110) at a total pressure of 100 Torr, H2/C2H6 = 100. (Reproduced with permission from Ref. 41. Copyright 1987, D.W. Goodman).

Comparison of Strained-Metal Overlayer Systems

Substrate

Ru ( 0001) W(llO) W(llO) W(100) Ru (0001) W(llO) W(100) Ta (110) W(llO) W(100)

Atom Density Mismatch/ML

6% 20 21 42 15 10 35 18

9 35

Pseudomorphicj Change Epitaxial in co Layers Desorption T 1/2 1/1 1/1 1/1 1/1 1/1 2/2

50K 80 -50 -50 50 -200 -170 1/1

1/2 2/2

-230 -50 -60

activity found for bulk Ni. The activation energy for this reaction is lower on the strained metal overlayer, however, very likely reflecting the lower binding strength of CO on the bimetallic system. In contrast, ethane hydrogenolysis, which is a structure sensitive reaction over bulk Ni, displayed marked structural effects on the Ni/W system [41]. We have observed, as shown in Figure 5, that the specific rate, or rate per surface metal atom, but not the activation energy, is a strong function of metal coverage on the Ni/W(110) surface, suggesting that the critical reaction step involves the need for a single, sterically unhindered Ni atom. On the Ni/W(100) surface the specific reaction rate was independent of Ni coverage. In addition, the rate on bulk Ni(100), Ni/W(110) in the limit of zero coverage, and Ni/W(100) were all equal, as were the activation energies. This implies that on Ni/W(100) the Ni atom geometry is sufficiently open to allow unhindered access to each Ni atom. Apparently on the Ni/W(110) surface only island edges and individual atoms display activity similar to the Ni(100) surface: the island interiors, in contrast, exhibit behavior similar to Ni(111) which has a much lower specific rate and higher activation energy. As the Ni coverage is reduced, the number of active, Ni(100)-like atoms increases, leading to an increase in the specific rate. The activation energy, however, remains unchanged. We have studied several other metal overlayers on W(110), W(100), and Ru(0001) substrates [42]. Table 1 lists properties of the metal overlayers, and the effect of the substrate on co chemisorption. In general only the first monolayer grows pseudomorphically, though more than one monolayer may be stable before three dimensional islands are formed (e. g. Cu/Ru grows two stable layers). The binding strength of co is always altered from the bulk metal, though the magnitude of the effect is seemingly more dependent on the metal overlayer, than on the degree of strain induced by the substrate (represented as the atom density mismatch). As with Ni/W and Cu/Ru, the effect on co binding energy extends primarily to only the first monolayer: subsequent layers exhibit behavior close to the bulk metal. ACKNOWLEDGEMENT We acknowledge with pleasure the support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences. REFERENCES 1.

T. B. Grimley and M. Torrini, J. Phys. Chern., 87, 4378

2.

J. H. Sinfelt, "Bimetallic Catalysts:

(1973).

3. G. 4. J.

5. D. 6.

D.

Discoveries, Concepts, and Applications", John Wiley & Sons, NY, NY, 1983. M. Schwab, Disc. Faraday Soc. 8, 166 (1950). T. Yates, Jr., c. H. F. Peden, and D. W. Goodman, J. catal., 94, 576 (1985). w. Goodman, J. T. Yates, Jr., and c. H. F. Peden, Surf. sci., 164, 417 (1985). w. Goodman and c. H. F. Peden, J. Catal., 95, 321 (1985). 213

7. J. E. Houston, c. H. F. Peden, D. s. Blair, and D. W. Goodman, Surf. sci., 167, 427 (1986). 8. J. E. Houston, c. H. F. Peden, P. J. Feibelman, and D. W. Goodman, I&EC Fundamentals, 25, 58 (1986). 9. J. E. Houston, C. H. F. Peden, P. J. Feibelman, and D. R. Hamann, Phys. Rev. Lett., 56, 375 (1986). 1o.c. H. F. Peden and D. w. Goodman, J. catal., 104, 347 (1987). 11.D. w. Goodman, R. D. Kelley, T. E. Madey, and J. T. Yates, Jr., J. Catal. 64, 479 (1980). 12.D. w. Goodman, Ann. Rev. Phys. Chern., 37 (1986) 425; D. W. Goodman, Accts. Chern. Res., 17, 194 (1984); D. W. Goodman, J. Vac. Sci. Tech., 20, 522 (1982). 13.H. Shimizu, K. Christmann and G. Ertl, J. Catal., 61, 412 (1980). 14.J. c. Vickerman, K. Christmann and G. Ertl, J. Catal., 71, 175 (1981). 15.S. K. Shi, H. I. Lee and J. M. White, Surf. Sci., 102, 56 (1981). 16.L. Richter, s. D. Bader and M. B. Brodsky, J. Vac. Sci. Techn., 18, 578 (1981). 17.J. c. Vickerman and K. Christmann, surf. sci., 120, 1 (1982). 18.J. c. Vickerman, K. Christmann, G. Ertl, P. Heiman, F. J. Hirnpsel, and D. E. Eastman, Surf. Sci., 134, 367 (1983). 19.S. D. Bader and L. Richter, J. Vac. Sci. Techno!., A1, 1185 (1983). 20.C. Park, E. Bauer, And H. Poppa, Surf. Sci., submitted for publication. 21. D. w. Goodman in "Heterogeneous Catalysis" (Proceedings of IUCCP Conference), Texas A&M University, 1984. 22.J. H. Sinfelt, G. H. Via, and F. W. Lytle, Catal. Rev.-sci. Eng., 26, 81 (1984). 23.A. K. Datye and J. Schwank, J. Catal., 93, 256 (1985). 24.G. c. Bond and B. D. Turnham, J. Catal., 45, 128 (1976). 25.L. J. M. Luyten, M. v. Eck, J. v. Grondelle, and J. H. c. v. Hooff, J. Phys. Chern., 82, 2000 (1978). 26.A. J. Rouco, G. L. Haller, J. A. Oliver, and c. Kemball, J. catal., 84, 297 (1983). 27.G. L. Haller, D. E. Resasco, and J. Wang, J. Catal., 84, 477 (1983). 28.J. H. Sinfelt, J. Catal., 29, 308 (1973). 29.Sachtler, J.W.A. and Somorjai, G.A., J. Catal. 89, 35 (1984). 30.Balooch, M., Cardillo, M.J., Miller, D.R., and Stickney, R.E., Surf. Sci. 50, 263 (1975). 31.P. J. Berlowitz and D.W. Goodman, Surf. Sci. 187, 463 (1987). 32.M. Balooch, M.J. Cardillo, D.R. Miller, R.E. Stickney, Surf. Sci. 50, 263 (1975). 33.J. Kolaczkiewicz and E. Bauer, Surf. Sci. 144, 495 (1984). 34.K. Christmann, 0. Schober, G. Ertl, J. Chern. Phys. 60, (1974). 35.D. w. Goodman, J.T. Yates, Jr., T.E. Madey, Surf. Sci. 93, L135 (1980). 36.I. Harnedeh and R. Gomer, Surf. Sci. 154, 168 (1985). 37.J. E. Houston, J.M. White, P.J. Berlowitz, D. w. Goodman, surface sci., in press. 38.M. w. Ruckman, M. Strongin, X. Pan, ibid. 39.D. Prigge, W. Schlenk, E. Bauer, Surf. Sci. 123, L698 (1982). 40.R. w. Judd, M.A. Reichelt, R.M. Lambert (to be submitted for publication). 41.C. M. Greenlief, P. J. Berkowitz, D. W. Goodman, J. Phys. Chern. 91, 6669, (1987). 42.P. J. Berlowitz, c. H. F. Peden, and D. w. Goodman, Mat. Res. Soc. Syrnp. Proc., 83, 161 (1987). 214

ORGANOMETALLIC CHEMICAL VAPOR DEPOSITION OF ALUMINUM NITRIDE AND ALUMINUM METAL

David C. Boyd*,

Richard T. Haasch*, Kwok-Lun Ho#, Jen-Wei Hwang*,

Roland K. Schulze*, John F. Evans*, Wayne L. Gladfelter*, and Klavs F. Jensen# Departments of #Chemical Engineering and Materials Science and *Chemistry University of Minnesota, Minneapolis, Minnesota

55455

INTRODUCTION Research Goals and Approach.

Thin films of solids are important because

of a particular physical property, such as electrical or thermal conductivity, mechanical strength, or refractive index.

These properties are determined by

both the atomic (crystal) structure of the material as well as the microstructure (i. e. grain size, pore volume, etc.).

The fundamental question which forms the

foundation of our research program in materials is:

What is the relationship

between the chemical reaction mechanism of the film forming step and the atomic and microstructural features of the film? Because we can systematically change the mechanism of a reaction by modifying the molecular structure of the precursor, we may develop an additional parameter upon which the physical properties of the film would depend. Establishing this relationship between film structure and deposition mechanism requires that we know both of these.

Therefore our research

involves elucidating the reaction mechanism and at the same time cataloguing the changes in film structure as reaction conditions are modified.

The studies

of the deposition mechanisms are conducted using the same guidelines and some of the same experimental methods which are common to solution-based mechanistic studies.

As listed below, however, several additional techniques

common to the study of surfaces are also employed.

Metal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

215

analysis of ill products.

1.

Qualitative and quantitative

2.

Kinetic studies of the film growth rate.

3.

Isotopic labelling studies.

4.

Temperature programmed reaction spectroscopy (TPRS)

5.

Surface sensitive spectroscopic techniques (XPS)

Materials of Interest.

Aluminum nitride has the wurtzite crystal

structure and several physical properties that make it especially interesting, both as a bulk material and as a thin film.

As a large band gap (6.2 eV) III-V

compound 1, which is very hard, resistant to chemical attack and high melting (2400°C)2, it has useful properties for coatings.

Its piezoelectric nature makes

it potentially important for surface acoustic wave devices, and its high thermal conductivity may be exploited in applications for packaging electronic microcircuits3.

The nitrogen sources in these previous studies have included

ammonia4,S (the most common source), hydrazine6 and alkylaluminum amido compounds 7,8.

We describe here studies on the use of the azide9,10 group as the

reactive nitrogen source and compare these results to precursors which contain direct N-C bonds. Metallic aluminum films are used primarily for their high conductivity and reflectivity.

Organoaluminum precursors, especially triisobutylaluminum

(TIBA), have received a great deal of attention for their ability to generate high quality AI films.ll-1 7

Typical deposition conditions for the low pressure CVD of

AI on Si or other substrates using TIBA include substrate temperatures of 260°C, TIBA temperature of 4S°C, with argon as the carrier gas at pressures up to 1 torr. These conditions result in deposition rates of up to 0.15 JJ.m/min.l4 The films produced by such methods have low resistivities (2.8 - 3.5 JJ.C-cm) and other properties which compare well with AI films prepared by evaporation.l6 A disadvantage observed in the use of TIBA is the rough surface morphology which leads to poor reflective properties.ll

We have developed a new process

for the CVD of Al that gives highly reflective films at temperatures less than 100°C at rates that are two orders of magnitude faster than TIBA 18.

EXPERIMENTAL Two reactors were used to study the CVD of AlN and Al.

The "survey"

reactor had a hot-wall quartz tube which was usually operated without carrier

216

gas and had a base pressure of 5 x w-5 torr.

A removable liquid Nz trap placed

between the quartz tube and the pump allowed the trapping and quantitation of all volatile products except Nz, Hz, and CH4.

A schematic of the second reactor

in which all growth rate measurements for AlN depositions were performed is shown in Figure 1.

Heating Unit

Figure 1

Pure hydrogen, produced by a Matheson Pd-alloy purifier, was used as the carrier gas for the [EtzAlN3]3, and was also used to purge the viewport on the reactor chamber.

The reactor chamber was a six-way stainless steel cross. The susceptor was a Mo plate heated radiatively by a Mo wire carrying a current of a few amperes.

The growth rates were measured in situ with a 6328

laser interferometry

assembly 19.

A He-Ne

Exhaust gases and unreacted precursor were

cracked by passing them through a high-temperature furnace and filtering before venting to a hood. Si(lOO) substrates were prepared by degreasing with trichloroethylene and methanol, oxidizing with nitric acid, and etching with HF. Typical growth conditions for this system were as follows: substrate temp., 400 - 800°C; [EtzAlN3l3 temp., 40 - 60°C; gas manifold pres., 3 - 5 torr; reactor pres., 3 - 5 torr; azide/Hz flow rate, 30 - 60 seem; makeup Hz flow rate, 5 - 20 seem. The analytical data for the AlN films, obtained using X-ray photoelectron spectroscopy (XPS), are summarized in Table 1, and they emphasize that the films produced in the two different reactors have similar compositions.

217

RESULTS AND DISCUSSION Aluminum

Nitride

Summarized below are the results of several studies on AIN film growth. Most of the work involves the use of dialkylaluminum azides as the precursor. These, along with the dialkylaluminum amides, have been found to be the most successful of the organoaluminum precursors to AIN.

At the end of this section

we have included a comparison of films preduced with precursors containing a N-C bond. Gas Phase Products.

The analysis of the gases produced from the reaction

of [Et2AIN3l3 ( T(precursor) = 40°C, T(reactor) = 500°C) was obtained using gas chromatography and reproducibly showed 70% C2H4, 30% C2H6, and a very small amount of butane and butenes (C4/C2 = 0.03).

Based on the mass of the

precursor consumed during the deposition and the total pressure of the C2H4, C2H6, and C4 products trapped, 89% of the precursor ethyl groups were accounted for in this analysis. Similar results were obtained when the furnace temperature was increased to 650°C (C2H4 = 76%, C2H6 = 24%, C4/C2 = 0.02, 95% recovery); or when the deposition was conducted under 1 torr of H2 carrier gas (C2H4

= 65%,

C2H6 = 35%, C4/C2 = 0.01). For comparison, analysis of the gaseous

products from CVD of [Et2AINH2h20 under similar conditions, indicates the formation of C2H6 was significantly favored over C2H4 (C2H4 = 25%, C2H6 = 75%, C4/C2 = 0.02, 81% recovery).

Table 1.

Atomic Compositions of Thin Films Using XPS after Ar+ sputtering.

Prc~:J.mi!Ha

[Et2AIN3]3

Rcil!:l!H Tr;:mg (°C) quartz 500

~AI

P'r;Q

~N

~c

43

45

10

2

45

10

1

[Et2AIN3l3

metal

480

44

[Me2AIN3l3

quartz

44

11

2

quartz

500 550

43

[Et2Al(NH2)l3

44

41

7

8

[Me2Al(NMe2)h

quartz

700

42

19

38

1

[Me2Al(Azir)]3

quartz

500

34

18

33

15

[Et2Al(NH-t-Bu)]2

quartz

700

44

36

15

4

Et3AINH2(t-Bu)

quartz

500

30

2

5

63

a) azir = NCH2CH2; Me= CH3; Et = CH2CH3; t-Bu = C(CH3)3

218

These results are consistent with a operative in cleaving the Al-Et bond.

P-hydrogen

elimination

mechanism

An alternative mechanism, homolytic Al-

Et bond cleavage, would form ethyl radicals which would couple or disproportionate to yield C4H 10 or C2H4/C2H6. respectively.

The gas phase

reactivity of ethyl radicals has been studied thoroughly by Lalonde and Price21, from which they concluded that radical coupling is favored over disproportionation by a factor of 10.

The analogous ratio under the

experimental conditions employed in the current CVD studies is 0.03 indicating that liberation of gaseous ethyl radicals is not the predominant mechanism for Al-Et bond cleavage.

The dramatic change in the C2H4 to CzH6 ratio that occurs

upon changing the N-source to an NH2 group is consistent with a change in the Al-Et bond cleavage path.

Presumably, the availability of the weakly acidic

hydrogens of the NH2 group allows the facile protonolysis of the Al-C bond20. Rates and Mechanism of FUm Growth Usjng fEt2A!N.llJ: the dependence of the growth rate on substrate temperature.

Figure 2 shows In the low

temperature regime ( < 525°C), the growth rate increases with temperature with an apparent activation energy of 26.4 kcal/mol.

At higher temperatures, the

growth rate shows a weaker dependence on temperature (apparent activation barrier of 5.2 kcal/mol).

This activation barrier is greater than that (-2

kcal/mol) corresponding only to mass transfer limited growth.

Therefore, the

change in the activation energy at high temperatures most likely represents a change in the chemical mechanism combined with mass transfer effects. Behavior such as this has been observed in other CVD studies22.

Additional data

is being collected to evaluate the relative importance of surface reactions versus homgeneous (gas phase) reactions. These results suggest elementary steps for several phases of the deposition process. Scheme 1 shows one of the possible scenarios for the mechanism of the reaction.

It must be emphasized that the relative timing of

these events remains unknown.

Further studies using temperature

programmed reaction spectroscopy are planned to more completely elucidate the mechanistic details. Nitrogen Sources Containing N-C Bonds. Both azide (N3) and amide (NHz) are comparably effective as sources for nitrogen in the deposition of AlN.

It

was of interest to explore how much the variation in structure of the N-source would influence the AlN deposition, and we were particularly interested in determining the effect of incorporating a N-C bond into the precursor.

The

219

~

I 'H

-N2

Et 2 AIN 3 (g) _____. Et 2 AJN 3 (s) _____.

1-l:!C

'AIN (s) I

Et

Scheme 1.

800

700

600

500

10

l.81J. m/hr 1.4

Ea=5.23kcal/mol

1.0

9 Ill

... ...... -<

Ea=26.4kcal/mol

I

~

.d

~

"'

8

Azide!H 2 H 2 Makeup Precursor Temp Reactor Pressure

50 seem 7.5 seem 50°C 3 torr

0.2

7+---~--~--T---~--~--T---T-~T-~

0.9

1.1

1.3 1000/T(K-1)

Figure 2

220

0.6

1.5

precursors studied included [Me2Al(NMe2)l2 23, [Me2Al(Azir)]3 (Azir = aziridine)24, Et3AlNH2(t-Bu)25, and {Et2Al[NH(t-Bu)] )225. All of these structures contain direct Al-N bonds, and with the exception of Et3AlNH2(t-Bu), all exist as cyclic four- or six-membered rings.

The structures differ from the

azide and amide precursors primarily due to the presence of the N-C bond.

The

substantial impact made by the presence of the N-C bond is highlighted in Table 1.

All of these films contained large concentrations of carbon, and one

contained only small amounts of nitrogen.

In the case where the nitrogen

source is the t-butylamine-triethylaluminum donor acceptor complex, the small concentrations of N indicate that a facile Al-N cleavage is operative, simple dissociation of the donor-acceptor bond.

e. g.

It should be noted that

[Me2Al(NMe2)]2, which also gave films with low N values, was the most stable of all the precursors examined.

Even at oven temperatures of 450°C, this

precursor would pass through the hot tube unchanged.

The lower N content of

the film may be the result of an Al-N bond cleavage process involving hydrogcn elimination of the NMe2 ligand.

fi-

The intermediate case of [Et2Al(NH-

t-Bu)]2 is interesting because it contains no P-hydrogens on the amido ligand. It is also related to [Mc2Ga(As-t-Bu2)]2 which was recently reported to give GaAs films26.

The results show that most of the nitrogen is incorporated into

the film; unfortunately, the carbon content is also high.

The compound

containing the 3-mcmber, highly strained aziridine ring, [Me2Al(Azir)]3, gave results similar to those found for [Me2Al(NMe2)]2.

Aluminum A series of stable, volatile donor-acceptor complexes of alane (AlH3) have been known for many years.27 They can be readily synthesized in one step from LiAlH4 27,28, and although they arc air sensitive (primarily due to moisture), they are much less sensitive than the trialkylaluminums.

The

structures of these compounds are shown in eq. 1, and among the known donors are Me3N, Et3N, Me3P, MezS, and THF27. Trimethylamine is unique among these donors in its ability to form a complex with two donors, eq. 1.

NMe 3

I H.;' I

H.. ~'''Al--H

(1)

NMe 3

221

As these complexes contain no AI-C bonds, the elegant studies29 on the surface reactivity of TIBA would suggest they might eliminate H2 at lower temperatures to form AI films (assuming the relatively weak donor-acceptor bond is also In fact, from the early studies describing these compounds, it

readily cleaved).

was known that they decomposed to form AI at temperatures >100°C.27

Patents

describing the use of amine-alane complexes for the vapor phase30,31 and electroless solution AI plating32 have appeared, as has a recent report of the laser-induced deposition of AI using these compounds.3 3 Deposition Conditions and Film Properties.

In a typical deposition the

precursor vessel was opened to the reactor system for four minutes during which time the inside of the quartz tube and the substrates were coated with an AI film.

The extent of the film down the length of the furnace and the At

appearance of the film were dependent on the specific reaction conditions.

the exit of the furnace a small amount of a crystalline deposit of precursor (or (Me3N)AIH3) was observed.

Unreacted precursor was also found in the liquid

nitrogen cooled trap placed between the reactor and the diffusion pump.

As the

trap was located between the capacitance manometer and the reactor, the pressures measured resulted from the H2 expelled during the deposition. This appeared to be a sensitive measure of the reproducibility of a given set of reaction conditions. specific example.

The behavior usually observed is best described with a

With the precursor vessel at 25 °C and the furnace at 180 °C,

the pressure would stabilize at a constant value of approximately 0.2 torr.

Both

lower furnace and precursor vessel temperatures lowered the observed pressures. The rate of AI deposition was determined by masking part of the wafers prior to the deposition and measuring the step height created for a given deposition time.

With a precursor temperature of 25 °C and the reaction tube at

In depositions where the reactor 180 °C, the deposition rate was 0.9 ~m/min. and its contents were treated with TiCI4 prior to the alane precursor, a more even distribution of the AI layer was observed.

Characterization of the AI Films.

Figure 3 shows the results of the Auger

electron spectral profile as a function of sputtering time.

The top layers of the

film exhibit the usual oxide coating, and the carbon and nitrogen content is

222

also adsorbed from the atmosphere.

All of these elements decrease rapidly to

within the detection limits of the method as the sputtering proceeds. In the films where TiCl4 was used to pretreat the surface, no Ti or Cl was detected in the AI films or at the interface with the silicon wafer.

This is consistent with

the suggestion made regarding the TiCl4 promoted deposition of aluminum from TIBA that less than a monolayer of TiCl4 actually adsorbs during the pretreatment. 1 4 X-ray diffraction shows the formation of polycrystalline AI films on the surface of Si(lOO) or glass slides (Figure 4a). Thicker films deposited without any TiCl4 pretreatment gave nearly the expected34 intensity distributions, while an increasing deviation from the polycrystalline AI pattern was observed Figure 4b, however, shows the striking effect caused by pretreating the surface with TiCl4 and growing the films at 100 °C. The films

for thinner films.

show nearly complete preference for growing with the (111) face parallel to While Figure 4b was taken from a film grown on Si(lOO), similar

the surface.

patterns are obtained on simple glass slides. Figure 5 clearly illustrates this same orientation preference on TiCl4 treated polyimide films. The microstructure of the films was examined using scanning electron microscopy (SEM), and, once again, a dramatic effect of pretreating the film with TiCI4 was observed. The grain size observed averages 2 - 3 JJ.m in films deposited at higher temperatures (180 °C) without TiCI4 pretreatment.

In the

regions near the edge of the AI created by the mask, small crystallites of AI were visible on the Si.

Scanning towards the AI film reveals an increase in

crystallite size, but not a corresponding increase in the number of small crystallites.

This suggests that the rate of crystallite growth is greater than the

rate of nucleation,

which undoubtedly contributes to

the surface

roughness of

Those films grown on Si(lOO) wafers or glass slides that were pretreated with TiCI4 exhibited a much higher number of crystallites in this the final film.

same near-edge region.

On those films grown on pretreated surfaces at 180 °C

the grains averaged 1 JJ.m, whereas the films grown on pretreated surfaces at 100 °C exhibited an average grain size of 0.15 JJ.m. The resistivities of the films were evaluated using a four-point probe and were in the range of 2.5 - 4.5 JJ.n-cm.

The low resistivities observed are

223

40.0 35.0 ];o 'ii r::

30.0

.!

.5 .>1.

ilD.. ,g

25.0 20.0

.>1.

c

15.0

II

D..

10.0 5.0 0.0

-,E:;::::~~:;:::::;:=;==::=;=;=:=:~~~;:::::::;~:::::;::::;:=:,.~ 0.0

2.0

4.0

6.0

8.0

1 0.0

12.0

14.0

1 6.0

18.0

20.0

sputter time (min.)

Figure 3.

Auger sputter profile of an AI film on Si(lOO).

virtually identical to those reported for AI films grown using TIBA, and they are near that of bulk AI.ll - 11 The surface roughness of many of the films meant that their specular reflectivity was very low.

Typically, the total reflectivity was greater than 90%

at a wavelength of 550 nm, but the specular contribution to this value was less than 5%.

For the films grown at 100°C on TiCl4 treated surfaces, the relative

contributions of the diffuse and specular components to the total reflectivity were reversed.

Similar reflectivities are also obtained from AI deposited on

treated polyimide films. The adhesive property of films to their substrates is very important, but it is a property that is difficult to quantify.

The qualitative test used in this

study to evaluate the adhesive nature of these AI films was the Scotch tape test. All of the films grown on Si(lOO), glass, or polyimide at 180 °C or above remained intact as the tape was peeled away from the AI film.

Some of the films

grown at the lower temperatures on Si(IOO) were partially removed as the tape was peeled away.

These films were also the ones grown on the pretreated

surface, and the reason for the lower adhesion is not apparent. the AI films to polyimide was always good.

224

The adhesion of

5.00

a 4.00

,..... ., "0

.,

c:

0

:I

0

3.00

.r::.

~

.,"u

'!l

2.00

c:

:I

0

u

1.00

0.00 30.0

Ll 50.0

\_

~

k

70.0

90.0

two theto

5.00,-------,-------------------------~----------------~

b 4.00

,..... ., "0

c:

.,0

:I

0

3.00

.r::.

~ u

.,"

'

!l c:

2.00

:I

0

u

, .00

)

~

h

0.004-------~--~~--r--------.--~~~~~----~n-----~ 50.0 70.0 30.0 90.0

two theta

Figure 4. X-ray diffraction of thin films deposited; a) on Si(lOO) at 280 °C without any TiCl4 pretreatment, and b) on Si(lOO) at 100 °C following treatment of the substrate with TiCl4.

225

4.50 4.00

.,

..._ "C

3.50

.,c I:

3.00

2,

"0

2.50

u

.,"

....... !l

""u0

.L.OO

1.50 , .00 0.50

...._....._

0.00 30.0

II.

50.0

70.0

90.0

two theta

Figure 5.

X-ray diffraction of an Al film on polyimide.

Chemistry of the Deposition.

(Me3N)2AlH3 is an effective precursor for

the formation of thin films of high purity, low resistivity aluminum in a low pressure chemical vapor deposition reactor.

The microstructure of the film is

very dependent on the reaction conditions, especially the temperature and the pretreatment of the surface. In the following discussion we will comment on a possible mechanism for the deposition and offer some suggestions on the role of TiCl4. We will then compare the deposition reactions of (Me3N)2AIH3 with TIBA. Studies of eq. 1 in the gas phase suggest35 that by 80 °C the equilibrium lies mostly to the left making the predominant species in the gas phase (Me3N)AlH3. Although less well-documented, higher temperatures were reported to cause dissociation of the second Me3 N releasing AIH3.

Alane itself is

known to form an intractable polymeric solid at room temperature.27

In the

above study35 precipitation of AI metal or of solid alane did not occur as evidenced by the reported reversibility.

This earlier study differs from that

described here in that it was conducted in a sealed vessel where volatile products were not continuously removed. Studies of the thermal decomposition of liquid (Me3N)2AIH3 and (Me3N)AlH3 have also suggested that dissociation of the Me3N precedes hydrogen loss.36

Based on these studies the following

sequence is proposed for thin film growth on aluminum surfaces.

226

(Me3N)2A1H3(g)

(Me3N)AIH3(g) + Me3N(g)

(2)

(Me3N)A1H3(g)

AlH3(g) + Me3N(g)

(3)

AIH3(g) + 2 Al(s) ~ 3 AIH(s)

~ 3/2 H2

3 AIH(s)

(4)

+ 3 Al(s)

(5)

Eqs. 2 and 3 are based on the previous studies, whereas eq. 4 is based on the observed surface chemistry of TIBA29.

It was found that upon adsorption of

TIBA on AI surfaces, the three isobutyl groups behaved identically.

That

coupled with the fact that molecular TIBA desorption was never observed suggested that upon adsorption the alkyl groups became equivalent by migration to adjacent surface aluminum atoms.

The similarity between alkyl

and hydride ligands is the basis for proposing the step shown in eq. 4. Finally, it is known that aluminum surfaces do not dissociatively adsorb H2 (the reverse of eq. 5) at low pressures.

Studies of eq. 5 have been accomplished by reacting

atomic hydrogen with aluminum surfaces at low temperatures. Temperature programmed desorption studies showed that H2 desorbs from AI surfaces around 60 oc29,37.

We note that this is between the lowest temperature where AI films

were observed (100 °C; lower temperatures were not examined) and room temperature where no AI films formed.

While it is tempting to suggest that eq.

5 is the rate limiting step, we point out that the temperature range required for deposition is also the range where eqs. 2 and 3 become significant.

It is

important to note that our data does not rule out other possible steps, such as direct adsorption of (Me3N)2AIH3 or (Me3N)AlH3.

Kinetic studies of the

deposition are planned which should help delineate the mechanism. The mechanism outlined in eqs. 2 - 5 highlights the difference between growth of the AI on AI and initial nucleation of the AI crystallites; the latter being the slower process. Some consideration of the role of TiCl4 can be addressed in this light. Alane, a powerful reducing agent, would undoubtedly react rapidly with TiCl4. Evidence of this can be observed in the trap placed at the exit of our CVD reactor.

Even at very low temperatures (-100 °C) dark green

to blue coloration is observed, and as the trap is further warmed towards room temperature, a very exothermic reaction takes place.

While no spectral data

regarding the products is yet available, we would anticipate Ti-H [or perhaps aluminohydride,

Ti(A1H4)] complexes to form initially.

Reductive elimination

of H2 from the Ti would be expected to be facile, thus providing a route to metallic AI.

It should be kept in mind that in our procedures both the Si(lOO)

and glass surfaces are coated with hydroxyl groups which offer a route (the first step is shown in eq. 6) for binding the Ti to the surface.

227

Si-OH + TiCl4 ~ HCI + Si-O-TiCl3

(6)

Another factor which could contribute to the surface roughness is gas phase nucleation of particles of AI or (AIH3)n.

The degree to which this is

important for the film growth could addressed by examining the kinetics of the deposition. There is a striking similarity in physical properties and microstructure between the aluminum films grown from (Me3N)2AIH3 and those grown from TIBA.

The difference in using these two precursors comes in the processing

parameters. growth

Although there is an emerging understanding of the surface

mechanism29, detailed kinetic studies of film growth under typical CVD

conditions are not available for TIBA.

The conditions reported by Green and

coworkersl5 are typical and involve heating the precursor to 45°C and the substrate to 220 to 300 °C.

With

deposition pressures in the range of 0.2 to 0.5

torr, growth rates of 0.02 to 0.08 11m/min. were observed. While the pressures in our reactor were not measured, the H2 pressure after the trap was in the range of 0.2 torr for a reactor temperature of 180 °C.

We would expect the

pressure in the reactor to be somewhat higher, but less than the equilibrium vapor pressure of (Me3N)2AIH3, which is 1.8 torr at 25 °C.35 The growth rate observed, 0.9 11m/min., is over an order of magnitude greater than that found for TIBA. The other difference comes in the threshold temperature for aluminum deposition, which is below 100 oc for (Me3N)2AIH3, and around 180 °C for TIBA. The high rates and low temperatures of deposition should make (Me3N)2AIH3 especially attractive for growing aluminum on temperature sensitive

substrates.

COMMENTS

Using several techniques, we have studied the deposition of thin films of AIN and AI metal from novel organometallic precursors.

While we can suggest

reasonable reaction mechanisms for the depositions, much work remains to provide experimental support for these ideas.

Because most surface sensitive

methods are easily applied to metals, we anticipate the extent of our knowledge will advance further and faster for AI deposition studies than with the studies of the insulating AIN films.

The long term nature of this program becomes

clear when considering that only after establishing the mechanism of at least two different precursors for the same solid film can we address the relationship between mechanism and film structure.

228

Current work in our laboratories also

involves developing precursors for the deposition of other main group and transition metal nitrides and selected metals.

ACKNOWLEDGEMENTS This research was supported by a grant from the Materials Chemistry Initiative of the National Science Foundation (CHE-8711821) and by the Center for Interfacial Engineering at the University of Minnesota.

REFERENCES

1.

Yim, W. M.; Stofko, E. J.; Zanzucchi, P.J.; Pankove, J. I.; Ettenberg, M; Gilbert, S. L. J. Appl. Phys. 1973, 44, 292.

2.

Weast, R. C. (ed) "Handbook of Chemistry and Physics, 51st Ed." 1970, The Chemical Rubber Company: Cleveland.

3.

Baker, R. T.; Bolt, J. D.; Chowdhry, U.; Klabunde, U; Reddy, G. S.; Roe, D. C.; Staley, R. H.; Tebbe, F. N.; Vega, A. J. Mat. Res. Soc. Symp. Proc. 1988, 121, 471.

4.

Manesevit, H. M.; Erdmann, F. M.; Simpson, W. I.

J. Electrochem. Soc. 1971,

118, 1864. 5.

Sayyah, K.; Chung, B. C.; Gershenzon, M. J. Crystal Growth 1986, 77, 424.

6.

Gaskill, D. K.; Bottka, N.; and Lin, M. C. J. Crystal Growth 1986, 77, 418.

7.

Interrante, L. V.; Carpenter,Jr., L.; Whitmarsh, C.; Lee, W.; Garbaukas, M.; Slack, G. A. Mat. Res. Soc. Symp. Proc. 1986, 73, 359.

8.

Interrante, L.V.; Lee,W.; McConnell, M.; Lewis, N.; Hall, E. J. Electrochem.

Soc., 1989, 136, 472. 9.

Schulze, R. K.; Mantell, D. R.; Gladfelter, W. L.; Evans, J. F. J. Vac. Sci.

Technol., 1988, A6, 2162. 10.

Boyd, D. C.; Raasch, R. T.; Mantell, D. R.; Schulze, R. K.; Evans, J. F.; Gladfelter, W. L. Chemistry of Materials, 1989,1, 119.

11.

Levy, R. A.; Green, M. L. J. Electrochem. Soc. 1987,134, 37C.

12.

Pierson, H. 0. Thin Solid Films 1977,45, 257.

13.

Malazgirt, A.; Evans, J. W. Me tall. Trans. 1980, 11 B, 225.

14.

Cooke, M. J.; Heinecke, R. A.; Stem, R. C.; Maes, J. W. Solid State Technol.

15.

Green, M. L.; Levy, R. A.; Nuzzo, R. G.; Coleman, E. Thin Solid Films 1984, 114,

1982, 25(12), 62. 367. 16.

Levy, R. A.; Green, M. L.; Gallagher, P. K.; Ali, Y. S. J. Electrochem. Soc. 1984,

131, 2175.

229

17.

Kobayashi, T.; Sekiguchi, A.; Hosokawa, N.; Asamaki, T.

lap. J. Appl. Phys.

1988, 27, L1775. 18.

Gladfelter, W. L.; Boyd, D. C.; Jensen, K. F. Chemistry of Materials, in press.

19.

Sugawara, K.; Yoshimi, T.; Okuyama, H.; Shirasu, T. J. Electrochem. Soc. 1974,

121' 1233. 20.

Sauls, F. C.; Interrante, L.V., private communication.

21.

Lalonde A. C.; Price, S. J. W. Can. J. Chern .. 1971, 49, 3367.

22.

Jasinski, J. M; Meyerson, B. S.; Scott, B. A. Ann. Rev. Phys. Chern .. 1987, 38, 109.

23.

Davidson, N.; Brown, H. C. J. Am. Chern. Soc. 1942, 64, 316.

24.

Atwood J. L.; Stucky, G. D. J. Am. Chern. Soc. 1970, 92, 285.

25.

Gosling, K.;

26.

Cowley, A. H.; Benac, B. L.; Ekerdt, J. G.; Jones, R. A.; Kidd, K. B.; Lee, J. Y.;

Smith, J. D.; Wharmby, D. H. W. J. Chern. Soc. 1969, (A), 1738.

Miller, J. E. J. Am. Chern. Soc. 1988, 110, 6248. 27.

Wiberg, E.; Amberger, E. Hydrides of the Elements of Main Groups 1-IV; Elsevier: Amsterdam, 1971; Chapter 5.

28.

Kovar, R. A.; Callaway, J. 0. Inorg. Syn. 1977, 17, 36.

29.

Bent, B. E.; Nuzzo, R. G.; Dubois, L. H. J. Am. Chern. Soc. 1989, 1 I I, 1634.

30.

Whaley, T. P.; Norman, V. U. S. Patent No. 3 206 326, (1965).

31.

Carley, D. R.; Dunn, J. H. U. S. Patent No. 3 375 129, (1968). U. S. Patent No. 3 462 288, (1969).

32.

Schmidt, D. L.; Hellmann, R.

33.

Baum, T. H.; Larson, C. E.; Jackson, R. L. Abstracts of Papers, Fall Meeting,

34.

Powder Diffraction File, International Center for Diffraction Data,

35.

Heitsch, C. W. Nature, 1962, 195, 995.

Boston, MA; Materials Research Society:

Pittsburgh, PA, 1988; 84.12.

Swarthmore, PA, Card #4-0787. 36.

Nechiporenko, G. N.; Petukhova, L. B.; Rozenberg, A. S.

Ser. 37.

230

Khimicheskaya 1975, 1697.

Paul, J.; Hoffmann, F. M.

Surf. Sci. 1988,194, 419

Izv. Akad. Nauk SSSR,

SOLID STATE CARBON-13 NMR OF METAL CARBONYLS

Brian E. Hanson Department of Chemistry Virginia Polytechnic Institute and State University Blacksburg, Virginia 24061-0212

INTRODUCTION The first carbon-13 NMR spectrum of a metal carbonyl was reported in 1958 for iron pentacarbonyl by Cotton and Waugh. [1] The result, a single resonance in the carbonyl region, has been shown to be correct by several subsequent investigations. [2] In fact the single resonance for Fe(C0) 5 persists to -160 °C.[3] The structure of iron pentacarbonyl in solution is trigonal bipyramidal thus the carbon-13 NMR spectrum for this complex should contain 2 signals of relative intensity 2 to 3. Of the possible explanations for the single resonance it is now accepted that the trigonal bipyramidal structure undergoes a reorientation which is fast on the NMR time scale and exchanges equatorial and axial carbonyl groups. The most likely mechanism for the exchange of carbonyls is the well known Berry psuedorotation. [4] From the very first carbon-13 NMR study of metal carbonyls then it was apparent that not only are metal carbonyls likely to be dynamic in solution but also that NMR spectroscopy would be an invaluable tool for the investigation of mechanisms which describe the dynamic behavior of metal carbonyls. The phenomenon described by the formal breaking and reforming of bonds while no net chemical change occurs, is referred to as fluxionality. [5] Fluxionality has proven to be common for metal carbonyls, and particularly for metal carbonyl clusters, in solution. [6] The first Magic Angle Spinning, MAS, carbon-13 NMR of a binary metal carbonyl was recorded for Fe3(COl12 and suggested that certain types of fluxional behavior may also be anticipated for metal carbonyls in the solid state. [7] Although molecular motions are common in the solid state it appears that fluxionality is limited to a few structures of a special type and thus is the exception rather than the rule. Although the phenomenon is relatively rare an appropriate description of the dynamic behavior of these exceptional cases is important to the understanding of the structure and bonding in metal carbonyls. In this contribution the results from solid state NMR studies of metal carbonyls is reviewed with a special emphasis on molecules which exhibit dynamic behavior in the solid state.

MAGIC ANGLE SPINNING NMR SPECTROSCOPY The technique of MAS NMR is becoming more commonly practised in inorganic and organometallic chemistry. A brief outline is presented

Metal-Meta/ Bonds and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

231

here of the special problems associated with studying metal carbonyls in the solid state. Spin Lattice Relaxation Times. Binary metal carbonyls pose a difficult problem for study by solid state carbon-13 NMR. This is due to the fact that most metal carbonyls have extremely long carbon spin lattice relaxation times, T1 •s. When a spectrum is recorded using a simple carbon 90 degree pulse experiment, pulse delays should be approximately 5 T1 •s to prevent saturation of the signal. This is prohibitively long for many metal carbonyls which have carbon T1's of up to 30 minutes in the solid at room temperature. [8] In chloroform solution, the 13c T1 's for Mo(C0)6 at 15 MHz and 50 MHz are 180 and 20 seconds, respectively. The strong inverse field dependence on T1 indicates that the relaxation in Mo(C0)6 is dominated by the chemical shift anisotropy (CSA) mechanism. In the solid state Mo(C0) 6 has a T1 greater than 8 minutes at room temperature due to slow molecular correlation times. A large contribution from the CSA mechanism can be anticipated for metal carbonyls since chemical shift anisotropies of up to 430 ppm have been observed for these compounds. In this respect metal carbonyls are unique since the CSA relaxation mechanism typically does not contribute significantly to carbon T 1 •s for simple organic compounds. Long carbon T1's in solution may be shortened by the addition of a paramagnetic relaxation agent. [9] In general this technique is not very effective for crystalline solids however in special cases where the unpaired electron density is sufficiently high, carbon nuclei can be dynamically polarized by the electrons. [10] Relaxation by spin rotation, as well as chemical shift anisotropy, has been shown to be important for Ni(C0) 4 and Fe(Co) 5 at low field. [11-13] Dicobaltoctacarbonyl, Co2(COl8 and tetracobaltdodecacarbonyl, co 4 (C0) 1 2, are special cases. In addition to the large CSA of the carbonyl ligand, cobalt-59 has a large quadrupolar coupling constant and a magnetogyric ratio very close to that of carbon. Thus, relaxation by the dipole-dipole mechanism can be significant for cobalt carbonyls. The reported T1's for these two compounds are anomalously short, 0.16 and 0.5 s respectively for Co2(CO)e and Co4(COl12· Triirondodecacarbonyl also has a short spin lattice relaxation time, 80 s at 30 °C. [16] Each of these compounds have been proposed to be fluxional in the solid state [14-16] and thus should have shorter molecular correlation times, tc, than static metal carbonyls in the solid. This contributes to the shorter T1's. Other organometallic complexes which appear to demonstrate dynamic behavior in the solid state include C04-xRhx (CO) 12r x = 1 or 2, [17] (!J.-1]8-C8H8l Fe2 (CO) Sr [18] (!J.-1]8-C8Hs) 2Ru3 (CO) 8r [18] (1]4-CsH8) Fe (CO) 3r [19] (1]6-C6H5CH3)M(CO) 3r M = Mo,Cr, [20] and pi bonded cyclopentadienyl rings. [21] Compounds with pi bonded polyene rings demonstrate motion corresponding to the relative rotation of the ring to the metal carbonyl fragment. For binary metal carbonyls long carbon T1's in the solid state are overcome by enrichment in 13co to enhance the signal and by long pulse delays to prevent saturation. In substituted metal carbonyls the cross polarization experiment[22] can be employed to overcome the long T1 times. Chemical Shift Anisotropy. In the solid state the chemical shift depends not only on the isotropic shielding, , at the nucleus, but also on the orientation of the nucleus with respect to the applied magnetic field. The anisotropic contribution to the shielding is given by equation 1 for an axially symmetric site in which two of the principal axes of the shielding tensor are identical, cr11 = cr22 = cr~

232

and a 33 = ~1- Most of the tensors of carbonyl groups in metal carbonyls have been shown to have either axial or near axial symmetry. [23]

(1) The angle $ in equation 1 is defined as the angle between 0"33 and B0 , the applied magnetic field (see Figure 1). The time average of the chemical shift anisotropy, aa(t), vanishes when the sample is rapidly spun at an angle of 54.7° with respect to B 0 • This is true even though (Details all possible orientations of spins occur in a powdered sample. of the magic angle spinning geometry are more completely discussed by Yannoni [22] in a review of the magic angle spinning experiment) . Thus, since the observed shielding is simply the sum of the isotropic and anisotropic components (equation 2), the result of magic angle spinning is the observation of the isotropic chemical shift. G=

+

(2)

Ga

m""~~

<2\ 54.7°

~

sample

0 Fig. 1.

Experimental geometry for the removal of chemical shift anisotropy through magic angle spinning. B0 represents Rapid spinning at the angle the applied magnetic field. 54.7° causes the time average of the CSA to go to 0.

The isotropic shift is, of course, the shift observed in solution and there is good agreement between chemical shifts obtained on powders in the magic angle spinning experiment with solution data. However, crystal effects sometimes cause splitting of lines which are degenerate This has been demonstrated for several metal carbonyls in solution. including (1] 5 -c5H5)Fe2 (CO) 4, [24] Ru3 (CO) 12, [25,26] and Os3(COl12· [26,27] Elimination of the chemical shift anisotropy is important for studying metal carbonyls since !\a, [(0"11 +0"22)/2-0"33], may be as large as 430 ppm for these molecules. [22,26] For effective removal of the CSA, the spinning rate must be comparable to !\a , in Hz, to avoid complication from spinning sidebands. At an observation frequency of 15 MHz for carbon, spinning rates of 2 to 2.5 KHz are sufficient to eliminate most spinning sideband problems. However the spinning requirements become more severe as the observation frequency increases. When the spinning rate is slow compared to the CSA line narrowing still occurs however a progression of sidebands will be observed. The trade off in loss of intensity in the centerbands and the possibility of overlapping sidebands is compensated by the ability, in many cases, to

233

extract the chemical shift parameters [28] . This has been accomplished Figure 2 illustrates the situation when a for many metal carbonyls. [26] MAS spectrum is recorded in the slow spinning regime. The spectrum shown is of Rh6(COl16 and was reported by Oldfield et al. [26] For comparison the breadline NMR of the same molecule as reported by Gleeson and Vaughan [22] is shown also.

(b)

(a)

.zoo Fig.2

0

zoo

400

400

zoo

0

Magic angle (b) (a) The breadline NMR for RhG(CO)l6· [22] The centerbands in the MAS spinning NMR for Rh6(C0)16. [26] spectrum are labeled 3i and the sidebands are seen to map out the chemical shift anisotropy of the powder pattern. The breadline spectrum is plotted with positive chemical shifts to the right of 0 ppm on the x axis. (downfield)

REARRANGEMENTS OF PENTACOORDINATE COMPLEXES IN THE SOLID STATE The solution behavior of pentacoordinate metals is now well established from NMR investigations of many transition metal phosphine Although many mechanisms have been proposed to complexes. [29-31] describe the dynamic behavior of

pentacoordinated complexes in solution

2 are generally accepted. First, a simple Berry psuedorotation and its permutationally equivalent rotations have been shown to be consistent with the simultaneous exchange of the 2 axial groups with 2 of the equatorial groups in complexes such as [Rh(P(OMel3lsl+. [3,29,30] Second, for complexes containing a hydride ligand, a hydride tunnelling The basis for this comes, in part, mechanism has been proposed. [29] from the recognition that the structure of many complexes of the type The Berry HML 4 can be described as a hydride capped tetrahedron. [29] psuedorotation and the hydride tunnelling mechanisms are shown schematically in Figure 3.

Iron pentacarbonyl has been studied by Iron pentacarbonyl. Evaluation of breadline NMR at temperatures from 4.2 to 213 K. [12] the spin lattice relaxation times over part of this temperature range and calculation of the expected line shapes led to the conclusion that a molecular rearrangement occurs in solid Fe(C0) 5 . Two possible mechanisms for the motion were considered, a psudedorotation , as described above, and a rotation about the molecular 3-fold axis of the trigonal bipyramidal molecule. The authors concluded that a Berry The exchange psuedorotation better described the experimental data. [12] frequency for the motion was measured to be 2.4 x 104 s-1 at -60 °C.

234

'"'-k

--

--

.

-A

(b)

Fig. 3. (a) Representation of the Berry psuedorotation for the exchange of axial and equatorial ligands in a trigonal bipyramid. (b) The hydride tunnelling mechanism moves the hydride ligand from face to face in a capped tetrahedral structure.

From the reported variation of the rate with temperature an activation energy of 0.5 kcal mol-l can be calculated. Although this is similar to what is estimated for axial equatorial exchange of carbonyls in liquid Fe(CO)s the exchange frequency is much slower than the corresponding rate of molecular reorientation in liquid Fe(CO)s at -20 °C, i.e. 1.1 xlolO s-1. [11,13) The rate estimated in the solid state at -60 °C is sufficiently fast to exchange axial and equatorial carbonyl resonances on the NMR time scale provided the isotropic chemical shifts of the axial and equatorial carbonyl groups are within 1000 Hz. This is a reasonable expectation since the range of chemical shifts observed for terminal carbonyl ligands in neutral iron carbonyl derivatives is ca 208 to 220 ppm. [32) Figure 4a summarizes the previous breadline NMR data for Fe(CO)s. [12) We have recently obtained MAS carbon-13 NMR for iron pentacarbonyl in the temperature range -26 to -118 °C. [33) Some of these spectra are shown in Figure 4b. In Figure 4b, at temperatures of -38 °C and lower, 2 signals at 216 and 208.1 ppm are observed. The relative integrated intensities, including spinning sidebands, are 2:3 respectively for the 2 signals consistent with a trigonal bipyramidal structure. This represents the first time that distinct axial and equatorial signals have been observed in a metal pentacarbonyl. The melting point of Fe(CO)s is approximately -20 °C. All MAS spectra recorded between the temperatures -20 and -30 °C show 3 signals in the carbonyl region. These occur at 216, 208.1, and 211.6 ppm. The last of these corresponds to the isotropic shift of iron pentacarbonyl. [2b) The presence of this signal however is not indicative of rapid axial-equatorial exchange in the solid state. Rather the simultaneous observation of 3 signals indicates the presence of both a liquid and a solid phase in the sample at these temperatures. Further evidence of this is seen in the first set of spinning sidebands; the sidebands corresponding to the solid phase are observed but a sideband representing the average of axial and equatorial signals is not present.

235

(a)

(b) ..........,,...,~/YO"-..~~

~~Aw~ -so•c

60 Fig 4.

40

20

0 KHz

215.0

208.1

ppm

(a) On the left are the breadline NMR spectra of Fe(CO)s. [12] The estimated exchange frequencies are shown. (b) On the right are the MAS NMR spectra for Fe(C0) 5 from -118 to -26 °C. [33] The resonances at 216.0 and 208.1 ppm are assigned to the axial and equatorial carbonyls respectively.

Clearly it can be concluded from these results that axialequatorial exchange of carbonyls is not rapid on the NMR time scale in solid Fe(C0) 5 at temperatures of -26 °C and lower. This is in contrast to the conclusion reached from breadline NMR. [12] We propose therefore that the motion indicated by breadline NMR in iron pentacarbonyl is best represented by a rotation about the 3-fold axis and not a psuedorotation. Further examination of the MAS NMR results for Fe(C0) 5 shows that the linewidths of both peaks diminish as the temperature is lowered. Although in principle a plot of the log of the linewidth vs 1/T will yield an activation energy for the process responsible for the line narrowing, in the present case the data is not sufficiently good to put a great deal of faith in the result . (From 5 data points an activation energy of 0.6 kcal mol-l is calculated however the correlation coefficient from the least squares fit is only 0.7.) JNEL~l fHFe!COl~lThe room temperature MAS NMR spectrum for the anion in [NEt 4 J [HFe(C0) 4 J shows a single line in the carbonyl region. [34] This is contrary to what is expected from the structure of the anion and suggests a dynamic process in the solid state. We have recently recorded the MAS NMR spectra for this compound as a function of temperature and demonstrate that it is indeed fluxional in the solid state. [35] Some of these spectra are shown in Figure 5.

As noted above the structure observed for compounds of the type HML 4 is best described as a face capped tetrahedron. The structure of the hydridotetracarbonylferrate has been reported as the (Ph3P)2N+, PPN+, salt; [36] the anion is represented in Figure 6. From the bond distances and angles in the anion the structure can be described as

236

either a distorted trigonal bipyramid or as a face capped tetrahedron. Although the structure of the anion as the tetraethylamonium salt has not yet been reported it is reasonable to expect that its geometry is similar to that observed with PPN+ as the cation. At temperatures of -60 °C and lower 2 carbonyl resonances are seen at 228 and 220 ppm. The peak at 235 ppm appears to be independent of the other two and is most likely due to an impurity. The peaks at 228 and 220 ppm are in a 1 to 3 ratio consistent with 1 axial and 3 equatorial carbonyls as expected. Coalescence for the 2 site exchange occurs at approximately -40 °C . After coalescence the signal broadens extensively at temperatures near -13 °C . This phenomenon is described by Rothwell and Waugh, and is a consequence of magic angle spinning with high power proton decoupling.[37] On top of the broad line observed at -13 °C the peak at 235 ppm persists as a fairly sharp signal. At 30 °C the peak due to the anion is seen at 221.6 ppm which is close to the weighted average position predicted from the chemical shifts seen at 105 °C . The impurity gives a small peak at 233.1 ppm. Over the temperature range studied the signal due to the impurity broadens and resharpens indicating that it is dynamic also. The linewidths from 4 spectra recorded at 0 °C and higher were fitted to the Arrhenius equation to obtain an activation energy for the process responsible for the exchange. This was calculated to be 7.0 ±0.7 kcal mol-l. The fact that Fe(C0)5 does not undergo axial-equatorial exchange in the solid state leads us to the the conclusion that a psuedorotation or similiar process such as a twist mechanism is unlikely in the solid state for (HFe(C0)4]-. A hydride tunnelling mechanism appears to be more likely since it requires minimal movement of the carbonyl ligands. [35]

10"

228 Fig. 5. Fig. 6.

220

ppm

Magic angle spinning NMR spectra of [HFe(C0)4]-.[35] Representation of the solid state structure of [HFe (CO) 4 ]- determined as the PPN salt. [36]

237

DINUCLEAR METAL CARBONYLS The simple binary metal carbonyl dimers Mn2(COl1o, [38] have each been investigated by 13c MAS and co 2 (COl8 [14] Fe 2 cco) 9 , [24] NMR. Of these the cobalt carbonyl dimer is unique in showing dynamic The structure of dicobaltoctacarbonyl behavior in the solid state. [14] is shown in Figure 7a [39] and the l3c MAS NMR spectrum at 40 and -40 °C are shown in Figure 7b. The low temperature spectrum shown in Figure 7b was recorded under the TOSS pulse technique [42] which eliminates spinning sidebands from Two signals are clearly seen, at 234 and 182 ppm, the spectrum. [14] which are consistent with the bridged structure in the solid state. The activation energy for the exchange process was estimated to be 11.7± 0.6 kcal mol-l. [14] It has previously been proposed that the dynamic behavior of the cobalt dimer can be interpreted by the concerted motion of the 2 cobalt atoms and the carbonyl framework. [40] This was proposed on the basis that the ligand polyhedron defined by the bridged structure observed in the solid state is similar to the polyhedron defined by one of the all The problem terminal structures proposed to exist in solution. [40,41] with this mechanism is that the carbonyls do not all become equivalent even though all have access to a "terminal" environment. Thus the high temperature spectrum should consist of more than one signal. The spectrum however is fairly sharp and very symmetrical at observation frequencies of 15, 22.6 and 50 MHz at temperatures of 20 °C and higher. This suggests that either all 8 carbonyls become equivalent on the NMR time scale or that there is a surprising amount of accidental degeneracy of signals in the high temperature spectra. Here we propose an alternative mechanism for the complete exchange of bridging and terminal carbonyls in solid Co2(C0)8· Although direct proof of this mechanism may be difficult to obtain there is indirect evidence in support of it.

(a) (b) Cel;!(COla

----~~------T-~C_)_ TOSS

-40 500

from TMS)

Representation of the molecular structure in Fig. 7. (a) (b) The MAS NMR spectra at 40 and crystalline Co2 (CO) 8. [39] The low temperature spectrum was -40 °C for Co2CC0)8· [14] recorded using the TOSS technique.

238

In brief, a homolytic cleavage of the Co-Co bond in solid dicobaltoctacarbonyl would lead to 2 Co(C0)4 radicals in the solid state. These would exist as caged radical pairs which could recombine to give the bridged structure. Prior to recombination the tetrahedral radicals could rotate about one of four, 3-fold axes to exchange carbonyl positions. Such a rotation is well established for adamantane which also has a tetrahedral structure. [37] Rotation in solid Ni(C0) 4 is relatively slow as determined by NMR experiments. [12] However experiments on solid nickel carbonyl must be performed below its melting point, ie -25 °C; [12] the rate of rotation may be expected to be faster at room temperature in Co(C0) 4 . Recombination of the radical pair would lead to complete interchange of carbonyl positions. An experiment of relevance to the existence and fate of cobalt tetracarbonyl radicals in polycrystalline dicobaltoctacarbonyl was reported in 1965 by Keller and Wawersik. [43] They reported that samples of freshly sublimed Co2(COls trapped on a cold finger at -196 °C show the presence of Co(C0)4 radicals in the EPR spectrum at -196 °C. When the sample was warmed to room temperature and recooled to liquid nitrogen temperatures the EPR signal due to the tetracarbonyl cobalt radical disappeared. Thus in the gas phase there 'must be an equilibrium between dimeric dicobaltoctacarbonyl and cobalt tetracarbonyl. This is shown in equation 3 below. 2 Co(C0) 4

(3)

The Co-Co bond enthalpy in dicobaltoctacarbonyl is estimated to be 20 kcal mol-l, [44] thus the dissociation of the dimer in the gas phase must be entropy driven to generate a significant concentration of the mononuclear radicals. From the EPR experiments it was estimated that the concentration of cobalt radicals in the solid state is ca 3%. If this represents the equilibrium concentration of radicals in the gas phase then the free energy change for the reaction shown in equation 3 is calculated to be ca. 2 kcal mol-l. The fact that the EPR signal is quenched upon warming the sample suggests that there is sufficient mobility to allow radical centers to recombine to form Co2(CO)s. A plausible mechanism for the movement of radical centers is the disruption of a radical pair by an adjacent isolated Co(C0) 4 radical to generate a new Co-Co bond and a new radical. Ultimately the cobalt radicals would be able to combine as they come into contact with one another.

CARBONYL CLUSTERS WITH TWELVE LIGAND VERTICES The solution dynamics of clusters with 12 ligand vertices has been discussed in length in the literature. [6c,6d,45] Exchange of groups between different sites in the clusters may be interpreted either as a polytopal rearrangement of the entire ligand polyhedron or by a local exchange of ligands at a single metal center or between 2 or more metals. In some cases both the polytopal and local mechanisms are equivalent. In others the dynamic behavior can only be described by a local mechanism. [6] With the idea of a polytopal rearrangement of 12 vertices came the idea that there may be a relative motion of the ligand polyhedron around the central metal core of a cluster. Thus it may be possible that the ligand polyhedron remain essentially fixed in space while the metal core moves with respect to the ligand polyhedron. In 1976 this idea lead Johnson to propose a novel mechanism for the intramolecular exchange of the 12 carbonyl ligands in Fe3(COl12 in which the ground state Czv structure achieves a D3 configuration by rotation of the Fe 3 triangle

239

about the 2-fold rotation axis. [45] Evidence for motion of the metal core in triirondodecacarbonyl is now available from MAS 13c NMR spectroscopy. There is additional evidence that suggests dynamic behavior in the solid state for Co4-xRhx(CO)l2• x = 0,1 or 2.[17] These are best described by a motion of the metal core with respect to the carbonyl polyhedron. Triirondoctecacarbonyl. The correct determination of the solid state structure of Fe3(COl12 represents a milestone in the chemistry of metal carbonyls. [46,47] Figure 8 shows a representation of the molecular structure. In the crystal the molecule lies on an inversion center. Since the molecule itself does not contain an inversion center the crystal lattice is disordered. The X-ray structure shows the space average of two orientations related by the inversion center. These are projected onto one another in Figure 8. Figure 9 shows the 13c MAS NMR spectra of Fe3(COl 1 2 at 27 and -80 °C .[48] These were recorded at 75 MHz. Figure 10 shows the correlation of the resonances at -80 with the resonances at 27 °C . The high temperature limiting spectrum at low field shows 6 sharp resonances of equal intensity. [7, 49] At high field a higher temperature would be required to obtain the same degree of line narrowing. Within the constraints of the crystal structure there is only one way in which to obtain a 6 line spectrum with the observed chemical shifts for solid Fe3(COl 1 2. This is represented in Figure 11. The mechanism suggests that the time average structure represented by the exchange of the 2 orientations of the iron triangle is observed at room temperature in the NMR spectrum.

Fig 8. The 2 orientations are shown projected onto the plane of the iron triangle. The molecule sits on a crystallographic inversion center which requires half occupancy of the 2 orientations.

It is important to remember that all 12 carbonyl ligands of one orientation of the Fe3(COl 1 2 molecule are crystallographically unique. The symmetry related carbonyls reside in a separate molecule of different orientation. In principle then 12 resonances should be obtained in solid Fe3(COl12· The observation of 6 lines at room temperature implies either a dynamic process or an accidental degeneracy of 6 pairs of 2 carbonyl groups since the 6 resonances have equal

240

240

200

220

Fig. 9. The high field (7 tesla, 75 MHz) MAS 13 c NMR spectra for solid Fe 3 (COJ 12 at 27 and -80 •c.

238.0 236.1

--··.

··.:::~.._-:.,_

..

obs. ··~

·~:::~J==

225.2 224.1

----::.::::::=··· ----··_::_.-_.._··-~:::::::::=··---- 213.2 210.4 _ _ _ _ _ _ 1~----······-------~---215.6

203.9 201.3 199.2

!calc I

(225.8) (224.1)

••••

(213 0)

209.6

(210.4)

202.2 ==~~=-~~:l-:.::::~::::t==== __ .............. . 201.4

(202.1)

-so •c

(201.5)

27 •c

Fig. 10. Proposed correlation of NMR signals from the spectra shown in Figure 9 at -80 and 27 •c.

241

intensity. As mentioned above a rotation of the iron triangle by 60° within the plane of the triangle satisfies the symmetry requirement of generating a time averaged inversion center and making 6 pairs of 2 carbonyl groups equivalent on the NMR timescale. Furthermore the 2 bridging carbonyls labeled 1 and 2 in Figure 11 are exchanged with the terminal carbony1s labeled 3 and 4. The correlation of the bridging carbonyls with terminal carbonyls is illustrated in Figure 10. Thus the peaks at 238.0 and 236.1 ppm have a relative integrated intensity of 2 compared to 10 for the terminal carbonyls at 199 to 215.6 ppm in the spectrum at -80 °C. At room temperature the peaks at 225.2 and 224.1 ppm have a relative intensity of 4 compared to 8 for the terminal peaks at 210 to 213 ppm.

Fig. 11. Schematic representation of the proposed mechanism to describe the observed MAS NMR spectrum of Fe3(COl12· Recently it has been suggested that a 2 fold rotation of the iron triangle about the axis containing the unique iron atom and bisecting the Fe-Fe bond bridged by CO ligands can explain the dynamic behavior revealed by the MAS NMR spectra. [50] A motion of this type was originally considered to be inconsistent with the NMR data. [7] Although it is clear from the crystallographic data [46,50] that a rocking motion about this axis takes place the motion does not exchange different

carbonyl groups.

This is true even if a structure of "psuedo D3"

symmetry is obtained in which all carbonyls are terminal. Referring to Figure 11 such a rocking motion never makes the two bridging carbonyls, 1 and 2, equivalent. Nor do the bridging carbonyls become equivalent with any other CO group in the molecule. This must be so since there is no crystallographically imposed 2 fold rotation axis in the molecule and, although the symmetry may approximate D3, the molecule cannot have true D3 symmetry in the crystal lattice. It is possible however that CO's 1 and 2 may experience a "terminal" environment if the "psuedo D3" structure can be achieved. Thus the resonances associated with the 2 carbonyl groups, 1 and 2, may shift toward a terminal position as the the "D3" structure is populated but never will the relative intensities of these peaks be greater than 2 compared to 10 for the terminal carbonyls. The low temperature spectrum obtained (Figure 9) is consistent with a low energy oscillation of the iron triangle but an explanation of the NMR spectra obtained at room temperature and higher requires an additional motion of the iron atoms. An argument against the in plane rotation of the iron triangle is that very close Fe-e contacts develop as the Fe3 group moves [50] . This is true provided the carbonyls remain rigidly fixed in space. However as noted previously [7,16,40] it is necessary for the carbonyl polyhedron to expand and for the positions of the carbonyl groups to move since the carbonyl positions in the symmetry related molecules do not overlap perfectly. The motion of the carbonyl groups must be correlated with the motion of the iron atoms. Also the molecules in the

242

215

168

ppm

Fig. 12. Variable temperature MAS NMR for co 4 (COl 12 in the solid state. The asterisked peak is due to delrin.

s5•c 69•c

61•c

Js•c

25•c

500

400

300

200

100

0

ppm

Fig. 13. Variable temperature MAS NMR for co 2 Rh 2 (COl 12 in the solid state.

243

crystal are staggered such that there is space for some expansion of the polyhedron in the plane of the irons. It should be noted that although the polyhedron defined by the carbonyl groups in solid Fe3(C0) 1 2 is approximated by an icosahedron, the icosahedron is flattened in such a way that the short dimension is perpendicular to the iron triangle. The related clusters Ru3(COl12 and Os3(C0)12 have also been investigated by magic angle spinning NMR in the solid state. [26-28] Neither molecule exhibits a fluxional process in the solid state. Line narrowing of the carbonyl signals in Ru3(COl12 as a function of temperature has been attributed to motion of CO at specific bonding sites in the cluster. [26] M~~lZ Clusters. Magic angle spinning NMR data is available for the clusters Co4-xRhx(C0)12 where x = 0, 1, 2, or 4. [15,17,51] Data for Co4(COl12 and Co2Rh2(COl12 are shown in Figures 12 and 13 respectively. The structure of tetracobaltdodecacarbonyl, like triirondodecacarbonyl, is based on an icosahedral arrangement of carbonyl ligands which surrounds the metal core. [52-54] Also, like Fe3(COl12• the tetracobalt cluster is disordered in the solid state. The molecule occupies a site of 2-fold rotational symmetry in the crystal which requires a rotational disorder in the solid. The 2 orientations observed in the crystal are represented in Figure 14. The structure of Co2Rh2(COl12 has not been determined crystallographically, however since both co 4 (C0) 1 2 and Rh4(COl12 have structures based on an icosahedral array of carbonyls it is reasonable to expect that the bimetallic cluster has a similar structure.

The variable temperature spectra were obtained at 22.6 MHz and are clearly complicated by 59co-13c dipolar and quadrupolar interactions. This is a serious problem for co 4 (C0) 12 . As noted previously this leads

Fig 14.

244

Schematic representation of the 2 orientations of the co 4 tetrahedron within the carbonyl icosahedron in Co 4 (C0) 12 . The view is down the 2-fold rotation axis that relates the 2 orientations.

to apparently anomalous chemical shifts for the carbonyl carbons. [15] Recently a room temperature spectrum for Co4(COl12 at 68 MHz has been The obtained and shown to be consistent with the crystal structure. [51] fact that Co 4 (C0) 12 is dynamic in the solid state is inferred from the behavior of the spinning sidebands in Figure 12. The spinning rate is constant at 2600 Hz at each temperature. However while at room temperature the third order sidebands are observed at 75 °C only the first order sidebands are seen. This suggest a dynamic process which reduces the chemical shift anisotropy takes place at temperatures above The line shape changes over this temperature range are 24 °C. completely reversible which suggests that the sample did not decompose at the high temperatures. The resolution is not sufficiently good to say that all 12 carbonyl ligands become truly equivalent. The spectra for Co2Rh2(COl12• are also consistent with a dynamic process in the solid state. Again all line shape changes are reversible which argues against sample decomposition. At elevated temperature loss of sideband intensity and broadening of the signals suggests a process which not only reduces the chemical shift anisotropy but also exchanges some chemical environments. It was previously suggested that rotation of the cobalt tetrahedron in Co 4 (CO)l2 within the ligand icosahedron could account for some of the apparent exchange of carbonyl environments in this molecule. However as noted above in the discussions of the dynamic behavior of Co2(COls and Fe3(COl12 all carbonyl environments will not become equivalent. The symmetry related environments in the disordered Co4(COl12 can be exchanged via a rotation of Co4 tetrahedron about the crystallographic 2-fold rotation axis. Given the width of the high temperature limiting NMR spectrum for Co4(COl 1 2 it is impossible to say if all 12 carbonyls have become truly equivalent on the NMR time scale. When the data for Co4(COl12 and Co2Rh2(COl12 are taken together it is apparent that some exchange of carbonyl environments takes place. This is most likely due, in part, to rotation of the metal polyhedron with respect to the ligand polyhedron. Figure 15 compares the room temperature spectra for the series of Co4-xRhx(C0)12 clusters. As cobalt is introduced into the cluster the chemical shifts as obtained at 22.6 MHz are shifted anomalously upfield. As noted above this is most likely due to the coupling with the quadrupolar 59co nucleus. Of these clusters only Rh4(COJ 1 2, appears to be completely static in the solid state. This may be due to the increased size of the metal core within the ligand polyhedron. The related metal carbonyl, Ir4(COl12• has a molecular structure based on a cubooctahedral arrangement of the carbonyl ligands. [6c] Results from MAS NMR on this molecule suggests that it is static in the solid state. [26] CONCLUSIONS Although in a few cases, notably Co4(COl12 and Fe3(COl12, fluxional behavior in the solid state may be viewed mechanistically as the interchange of different orientations in a disordered structure, it should not be deduced that disorder in the solid state is either the cause or result of the fluxional process. Clearly there are examples of non-disordered solids, eg Co2(COls, which are fluxional in the solid and there are disordered solids which are static.

245

400

200

0

ppm

Fig. 15. Room temperature MAS NMR spectra for the clusters Co4(COl12• Co3Rh(COl12• Co2Rh2ICOl12• and Rh4(CO)l2·

REFERENCES 1. 2. 3. 4. 5.

6.

7. B. 9. 10. 11. 12. 13.

246

Cotton, F. A.; Danti, A. ; Waugh, J. S.; Fessenden, R. W.; J. Chern Phys. ; 1958, 29, 1427. (a) Bramley, R.; Figgis, B. N.; Nyholm, R. S.; Trans. Faraday Soc., 1962, 58, 1893; (b) Mann, B. E.; J. C. S. Chern. Soc.; 1971, 1173. Meakin, P.; Jesson, J. P.; J. Am. Chern. Soc.; 1973, 95, 1344. Berry, R. S.; J. Chern. Phys., 1960, 32, 933. (a) Cotton, F. A., Chapter 10 in "Dynamic Nuclear Magnetic Resonance Spectroscopy", L. M. Jackman, and F. A. Cotton, eds., Academic Press, New York, 1975; (b) Cotton, F. A.; J. Organomet. Chern.; 1975, 100, 29. (a) Cotton, F. A.; Hanson, B. E.; in "Molecular Rearrangments," P. Mayo ed., Academic Press, 1980; (b) Aime, S.; Milone, L.; Progress in NMR Spectroscopy, 1977, 11, 319; (c) Benfield, R. E.; Johnson, B. F. G.; in "Transition Metal Clusters" B. F. G. Johnson, ed. Wiley Interscience, New York, 1980.; (d) Johnson, B. F. G.; Benfield, R. E.; J. C. S. Dalton Trans., 1978, 1554. Dorn, H. C., Hanson, B. E., Motell, E., Inorg. Chim. Acta., 1981, 54, L71. Duncan, T. M.; Yates, J. T., Jr.; Vaughan, R. W.; J. Chern. Phys., 1980, 73, 975. Cotton, F. A.; Hunter, D. L.; White, A. J.; Inorg. Chern.; 1975, 14, 703; and references therein. Kendrick, R. D.; Maresch, G. G.; Yannoni, C. S.; submitted J. Mag Res. Spiess, H. W., Mahnke, Ber. Bunsen Gesell.; 1972, 76, 990. Spiess, H. w., Groesca, R.; Haeberlen, U.; Chern. Phys.; 1974, 6, 226. Sheline, R. K., Mahnke, H., Angew. Chern. Int. Ed. Engl.; 1975, 14, 314.

14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30 31. 32. 33. 34. 35. 36 37. 38. 39. 40. 41. 42 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.

Hanson, B. E., Sullivan, M. J., Davis, R. J., J. Am. Chern. Soc., 1984, 106, 251. Hanson, B. E., Lisic, E. C., Inorg. Chern., 1986, 27, 716. Hanson, B. E., Lisic, E. C., Petty, J. T., Iannaconne, G.; Inorg. Chern., 1986, 25, 4082. Lisic, E. C., Ph.D. Dissertation, VPI&SU, 1986. (a) Fyfe, C. A.; Lyerla, J. R.; Yannoni, C. S.; J. Am. Chern. Soc.; 1979, 101, 1351; (b) Lyerla, J. R.; Yannoni, C. S.; Fyfe, C. A.; Ace. Chern. Res., 1982, 15, 208 Campbell, A. J.; Cottrell, C. E.; Fyfe, C. A.; Jeffrey, K. R.; Inorg. Chern., 1976, 15, 2690. Wagner, G. w., Hanson, B. E., Inorg. Chern.; 1987, 26, 2019. Benn, R.; Grundey, H.; Nolte, R.; Erker, G.; Organornetallics, 1988, 7, 777. Gleeson, J. W., Vaughan, R. W., J. Chern. Phys, 1983, 78, 5384. Yannoni, C. S., Ace. Chern. Res., 1982, 15, 208 Dorn, H. C., Hanson, B. E., Motell, E., J. Organornet. Chern, 1982, 224, 181. Airne, S.; Botta, M.; Gobetto, R.; Osella, D.; Inorg. Chirn. Acta, 1988, 146, 151. Walter, T. H., Reven, L., Oldfield, E., J. Phys. Chern., in press. Hasselbring, L., Lamb, H., Dybowski, C.; Gates, B.C.; Rheingold,A.; Inorg.Chirn. Acta, 1987, 127, 149; Herzfeld, J., Berger, A. E., J. Chern. Phys., 1980, 73, 6021. Meakin, P.; Muetterties, E. L.; Jesson, J. P.; J. Am. Chern. Soc.; 1972, 94, Meakin, P.; Jesson, J. P.; Tebbe, F. N.; Muetterties, E. L.; J. Am. Chern. Soc.; 1971, 93, 1797. Shapley, J. R.; Osborn, J. A.; Ace. Chern. Res., 1972, 6, Wilkinson, J. R.; Todd, L. J.; J. Organornet. Chern.; 1974,-77, 1 Hanson, B. E.; submitted J. Am. Chern. Soc. Whitmire, K. H.; personal communication. Whitmire, K. H.; Hanson, B. E.; submitted J. Am. Chern. Soc. Smith, M. B.; Bau, R.; J. Am. Chern. Soc.; 1973, 95 Rothwell, W. P.; Waugh, J. S.; J. Chern. Phys.; 1981, 74, 2721. G. Wagner, personal communication. (a) Leung, P. C.; Coppens, P.; Acta Cryst.; 1983, B39, 535; (b) Sumner, G. G.; Klug, H. P.; Alexander, L. E.; Acta Cryst.; 1964, 17, 732. Hanson, B. E.; chapter 2 in "Advances in Dynamic Stereochemistry," M. Gielen, ed. Freund, London, 1985. Johnson, B. F. G.; J. C. S. Chern. Commun.; 1976, 211. Dixon, W. T.; J. Chern. Phys.; 1982, 77, 1800. Keller, H. J.; Wawersik, H.; z Naturforsch. B., 1965, 20, 938. Connor, J. A.; Topics in Current Chemistry; 1977, 71, 71. Johnson, B. F. G.; J. c. s. Chern. Commun., 1976, 703. Wei, C. H.; Dahl; L. F.; J. Am. Chern. Soc.; 1969, 91, 1351. Cotton, F. A.; Troup, J.; J. Am. Chern. Soc., 1974, 96, 4155. Hanson, B. E.; submitted Inorg. Chern. Hanson, B. E.; Lisic; E. C.; Petty, J. T.; Iannaconne, G. A.; Inorg. Chern.; 1986, 25, 4062 Anson, C. E.; Benfield, R. E. Bott, A. W.; Johnson, B. F. G.; Braga, D.; Marseglia, E. A.; J. C. S. Chern. Cornrnun., 1988, 889. Airne, S.; Botta, M.; Gobetto, R.; Hanson, B. E.; Inorg. Chern.; in press. Wei, C. H.; Dahl, L.; J. Am. Chern. Soc., 1966, 88, 1821. Wei, C. H.; Inorg. Chern., 1969, 8, 2384. Carre, F. H. Cotton, F. A.; Frenz, B. A.; Inorg. Chern., 1976, 15, 381.

247

SURFACE CHEMISTRY OF METAL AND SEMICONDUCTOR CLUSTERS

R. E. Smalley Rice Quantum Institute and Department of Chemistry Rice University Houston, Texas 77251

INTRODUCTION In recent years a rather remarkable evolution of molecular beam experiments has taken place. This evolution is remarkable not so much in the

high

intricacy

level and

of

technology

involved,

sophistication

this

or

the

technical

increasing level evolution

has

of

been

proceeding fairly continuously ever since the first crude experiments several decades ago.

What is remarkable is that a qualitative change is

taking place in the very subject matter of these experiments. Rather than being limited to the study of intrinsically atomic or molecular phenomena, the newest molecular beam machines are quite at ease with quite large assemblies of atoms, at times extending to species where the atom count runs into the hundreds.

Although these species are still

molecular in the sense that they contain a definite number of atoms bound in a particular way, the physics and chemistry of these species is often quite distinct from the small molecule realm.

In many respects these

objects are nearly macroscopic. For example, consider even a rather small cluster molecule,

one

composed of 10 atoms of nickel.

Techniques have now been worked out which permit beams of such clusters to be prepared quite routinely 1 . They can be cooled to near absolute zero,

and studied either as a neutral, or a cold ion (either positive or negative) 2 • 3 . They can be extracted from the beam and trapped in an inert matrix, or injected into

Metal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

249

a superconducting magnet and trapped in an electromagnetic field for many minutes without "touching" anything4 . Since nickel is a fairly refractory metal, one expects the nickel atoms in Ni10 to be bound to each other rather strongly.

There will be

an inside and an outside to such a molecule, and this outside surface of the Ni10 molecule should be

the dominant player

chemical process one might wish to study.

in any physical or

In this respect Ni10 is no

different form most medium sized molecules studied in traditional organic and inorganic chemistry.

But in almost all other respects, NilO• is very

different. The lOth cluster of nickel is, after all, just a microscopic piece of the bulk metal.

Using the new cluster beam techniques, the

surface

chemistry of species like Ni10 is now beginning to be studied, and this chemistry is turning out to be far more like the catalytic dissociative chemisorption of bulk nickel surfaces than the chemistry of an isolated nickel atom or organo-nickel complex 5 . Most small molecules studied by conventional molecular beam techniques exist in a single well-defined ground electronic state with the first excited electronic state several eV higher in energy.

In sharp contrast, Ni10 is estimated to have a vast

number of excited electronic states within 1 eV of the ground state. It is instructive to estimate just how vast this number of lowlying states can be for even this rather small cluster. The current feeling about the bonding in transition metal clusters on this far right hand

side

of

the

periodic

table

is

that

they

are

predominantly by interaction of the 4s atomic orbitals.

held

together

In the case of

nickel, each atom comes to the cluster with one 4s electron and nine 3d electrons.

The 4s electrons become involved in the bonding, while the

3d9 holes remain primarily intact around each atom.

For each such atom

there are (including spin) 10 possible states of the 3d9 hole, so the total number of electronic states of Ni10 for a given type of 4s bonding is 1010 ! For small molecules the paradigm for physical science is that of the atom: find the individual eigenstates and fit their energy patterns to that of a simple Hamiltonian. successful

While this has been a fine formula for

science in small molecules, molecule with 10 10 electronic states.

250

it

is

clearly useless

for

a

Some other approach will be

necessary to understand these Ni10 molecules, and obvious place to start is in the concepts of solid state physics and surface science. Instead of talking about individual quantum states and wavefunctions, it is more useful to consider bands of states, and the variations in spatial extent and nodal properties with energy. The Molecular Surface Science Approach This electronic state "catastrophe" in Ni10 is just one example of the microscopic --> macroscopic transition that is making a qualitative change in the subject matter of modern cluster beam experiments.

Other

examples include the question of what is best view of the bonding in the molecule: is it like balls held together with sticks (covalent), or a set of

well-screened

(metallic).

ion

cores

bound

together

by

a

sea

of

electrons

Or questions of dynamics: how does one handle the many

interconnected electronic degrees of freedom, or the concept of strain release during fragmentation of a large cluster framework, etc. Along with these new conceptional worries comes a whole new set of reasons to be doing such experiments in the first place.

Since these

clusters act in many ways like a bulk surface, they should serve as excellent "molecular models" of such surfaces 6 . In fact, to many scientists currently involved in this cluster beam field, this molecular model aspect is the principal appeal. As is the case with all simple models, these cluster models should not be expected to be perfect replicas of the real thing.

They will be

useful as models only to the extent that some of the essential new physics and chemistry that we need to learn about real surfaces will also hold sway in the small cluster.

If so, it may be that many of the new

insights will be gained first on these small clusters.

The hope of

course is that many of the powerful techniques we normally associate with small

molecule

science

can

be

extended

to

these

cluster

models.

Particularly intriguing is the notion that it may soon be possible to calculate the physical and chemical properties of these clusters by high level quantum chemical techniques, and to test the effectiveness of the approximations used by direct comparison with experimental measures of exactly the same clusters in the laboratory. Cluster models of surfaces are most likely to be of direct use when one

concentrates

on

the

microscopic

details

of

surface

chemistry,

251

especially when this chemistry occurs on the surface of what is primarily a covalently bonded material.

silicon,

research has been carbon, paper summarizes emphasizing

one

a

few the

of

So a natural early topic of concentrated and gallium arsenide.

promising new

these

along

of the recent developments most

This

ion

techniques:

short lines,

cyclotron

resonance (ICR) of the cluster surface chemistry while the particles are trapped at high vacuum in the field of a superconducting magnet.

A SUPERSONIC CLUSTER BEAM ICR APPARATUS Figure 1 shows a schematic of the supersonic cluster beam FT-ICR Extensive discussion of the design

apparatus recently developed at Rice.

details of this machine have appeared elsewhere 4 ' 7 , early

examples

its

of

to

application

the

along with several

chemisorption of H2 on transition metal clusters 8 .

dissociative

of

study

however,

Here,

the

supersonic cluster beam was arranged to produce clusters of silicon and germanium.

Schematic of the supersonic cluster beam FT-ICR

Fig. l.

apparatus.

These

semiconductor

clusters

were

produced

by

pulsed

laser

vaporization (2nd harmonic of a Nd:YAG) of a disc of the corresponding semiconductor

in

a

originally developed

pulsed in our

supersonic group

for

nozzle. the

study

Using of

an

arrangement

gallium

arsenide

clusters 9 , this disc was slowly rotated and translated so that the laser

252

etched the surface in a spiral pattern, smoothly removing material from the surface at a rate of roughly 1000 mono layers per pass. nozzle the hot atoms

In such a

and small molecules originally produced in the

laser-driven plasma are entrained in a fast flow of helium carrier gas at nearly 1 atm pressure, and the resultant mixture is allowed to freely expand into a vacuum, producing an intense supersonic jet which is then skimmed to

form a well-collimated beam.

Using a second laser

(ArF

excimer at 1930 angstroms) to irradiate the clusters just as they begin the supersonic expansion it is possible to generate a dense plasma of positively and negatively charged clusters which then receive the full cooling of the supersonic expansion. Similar techniques have been in use in our group for a few years now in a wide range of experiments with cluster ion beams in the size range

from

2

to

were

several

hundred

originally

atoms

developed

per

cluster.

Although

the

the spectral and 10 photofragmentation study of the positive cluster ions , the negatively techniques

for

charged species "l:lave turned out to be particularly useful since they permit the detailed study of the UPS patterns for these clusters as a function of cluster size 11 . Here, however, as shown in Figure 1 the clusters were directed by means of a pulsed extraction field and a pair of einzel lenses down along the central axis of a superconducting magnet operating in the persistence mode at a peak field of 6 Tesla. Ordinarily,

injection of ions into such a strong magnetic field

would be extremely difficult owing to the so-called "magnetic mirror" effect.

Ions passing into regions of increasing magnetic field begin to

turn in response to the v x B component of the Lorentz force.

The effect

is to cause the ion trajectories to cross the magnetic field lines with increasing steepness, which in turn increases the Lorentz force further. As a result most incoming trajectories actually never make it into the Fortunately, as described in an earlier paper center of the magnet. from this group 7 • 8 , there is a set of initial trajectories that do succeed in cleanly transiting the fringing field of the magnet.

It is

those which lie within a narrow cone converging to a point some 20 em short of the center of the magnet.

These trajectories are such that they

are nearly tangent to the local magnetic field lines in that region of the fringing field where the Lorentz force is first strong enough to substantially affect the ions motion.

There the trajectory begins a slow

spiral about the local field line and follows it smoothly into the center of the magnet.

253

As shown in the figure, the forward momentum of the cluster ion is slowed in two successive deceleration steps as it approaches the ICR trap.

This trap is composed of a 15 em long, 4.8 em diameter cylinder

divided lengthwise into 4 sectors.

One opposing set of two of these

sectors were used to provide RF excitation in order to coherently pump the cyclotron motion of the clusters.

The other set of two sectors were

connected to a very sensitive differential amplifier so that the weak image currents generated by the circling cluster ions could be detected, digitized, and the resultant time-dependent waveform submitted to a fast Fourier transform. along

the

Drift of the cluster ions out of the trap by motion

magnetic

field

lines was

prevented by

a

small

repelling

potential on two electrodes on either end of the cylinder. Due to the crude initial pulsed extraction of the cluster ions from the supersonic beam, cluster

ion

packet

a as

considerable energy spread is present in the it

enters

the

magnet.

As

a

result

it

is

inefficient to decelerate the cluster ions to less than 5-10 eV as they approach the ICR trap.

This and the fact that often 10-100 pulses of the

supersonic cluster beam apparatus are necessary to "fill" the ICR cell forces us to use a thermalizing gas in the cell during injection, and to pulse the entrance electrode to the cell down momentarily as each cluster packet

arrives.

The

experiments

discussed below used

thermalizing gas held at 1 x 10- 5 torr

in the

neon as

the

ICR trap during the

injection of (typically) 100 cluster ion pulses at 10 pulses per second. After an additional 5 seconds to insure the clusters were effectively thermalized,

this neon gas was allowed to pump away prior to the RF

excitation pulse and subsequent detection of the cyclotron resonance. Once injected and thermalized, cluster ions may be stored in this trap at high vacuum for many minutes. As illustrated in Figure 2 the resultant FT-ICR mass spectrum of clusters trapped in this way can be of superb mass resolution, excellent signal to noise.

with

This spectrum is due to the very special

cluster of carbon, C6o+, which is now widely believed to have the form of a hollow aromatic network with a symmetric bonding pattern identical to that of a modern soccer ball 12 . The smaller peaks to higher mass are due to 13 c isotopic variants of the molecule. They appear roughly as

254

expected in accord with the 1.1% normal terrestrial isotopic abundance of

13c. Silicon clusters, on the other hand,

show no tendency to form a

specially stable 60 atom species. Instead the cluster ion distribution from the source is generally found to be a slow monotonically decreasing of cluster size 13 •14 . Figure 3 shows a section of this

function

distribution in the 44-54 atom size range as detected in the FT-ICR apparatus.

The fine structure here is due to the several isotopes of

silicon in natural abundance.

1.4

1.0

Fig.

2.

1.6 KILOHERTZ

1.8

2.0

2.2

FT-ICR mass spectrum of C6o+ at high resolution in a magnetic trap at 6 Tesla.

The horizontal frequency axis

is relative to a 125 kHz heterodyne carrier. The lower 13 12 13 frequency ICR peaks are due to c c59+ and c2 12 c 5s+ respectively.

255

47 50 !)4

53

52

49

45 46

51

44

I

ollro.,j ~

~ ~

1500

lw.ll

~

....... ~

1400

_I

1lOO MASS

Figure 3.

~

~--

(AMU)

FT-ICR of positive silicon clusters.

Unlike carbon where one isotope is dominant, most elements in the periodic table have at least two, and often many more, isotopes in major abundance.

As seen here with silicon these

isotopes begin to cause

somewhat of a problem when one is trying to identify the composition of a cluster by mass alone -- even with the sensational mass resolution of FT!CR.

A "SWIFT" TECHNIQUE One of the

great virtues

of the

ICR environment is

cluster mass has its own cyclotron frequency,

that each

so in principle it would

appear possible to selectively excite each cluster at resonance until its cyclotron radius became large enough to hit one of the side electrodes, thereby sweeping it from the trap.

If this were cleanly done for all but

a single isotopic form of the cluster, one could eliminate the isotopic

256

Ge 11

Get2

IIll,

,,111111 880

880

Fig.4.

860

860

,IIlii 840

820

MASS

(AMU)

MASS

820 (AMU)

840

IIIlLI 800

780

800

780

FT-ICR of positive germanium clusters in 11-12 atom size The boxed in region of the top panel received

range.

uniform RF excitation by the SWIFT technique in order to sweep all but one isotopic variant of the 11 and 12 atom clusters.

The bottom panel shows the FT-ICR spectrum of

the contents of the ICR trap after this SWIFT ejection.

mass confusion.

For example, the top panel in Figure 4 shows the FT-ICR

mass spectrum of germanium clusters in the 11-12 atom mass range. The broad spread of multiple peaks seen here is due to the 5 major isotopes of germanium in the 70-76 amu mass range.

In order to eject all

but a single isotopic variant of these clusters, one would ideally like to arrange the RF excitation such that a uniform RF power was given to

257

all masses shown in the blocked off area to the top panel of this figure. The

excitation

of

as

RF

the

to

desired has

one

and

well

leave

will

cluster

this

the

that

is

course,

of

problem,

exposed

to its

cluster

single mass worry

highly

motion

cyclotron

is

off-resonant

that

excited.

A beautiful loss

the

of

Transform" coworkers

this

solution to

desired

mass

technique

(SWIFT)

at Ohio State 15 .

is

the

first

problem of off-resonant excitation Inverse

Fourier

suggested by Alan Marshal

and his

"Stored

Here one

Waveform the

calculates

discrete

inverse

Fourier transform of the RF power spectrum necessary eject the unwanted stores

masses,

this waveform in a

and then uses

large memory,

it to

generate an analog time-dependent excitation voltage that is sent to the excitation

of

plates

the

cell.

ICR

complicated RF waveform which

is

excitation of the desired cluster.

The

guaranteed

result

an

is

minimize

to

extremely

off-resonant

The bottom panel of Figure 4 shows

the result of this SWIFT sweeping of all but a single isotopic variant of the 11 and 12 atom germanium clusters from the ICR trap.

RESULTS FOR THE CHEMISORPTION OF AMMONIA ON SILICON Using this FT-ICR device it is now possible to begin the study of cluster surface chemistry for a wide variety of materials.

One of the

most interesting examples of this sort of approach is shown in Figure 5. Here in the top panel one can see the positively charged clusters of silicon in the 44p54 atom size range as they are detected in the ICR trap after

injection,

thermalization,

and

SWIFT

ejection

most

of

of

the

isotopic congestion. In the bottom panel of figure 5 one can see the effect on this cluster distribution of a 5 second exposure to 4 x 10- 7 torr NH3 at near room temperature.

A typical silicon cluster ion will experience roughly

70 collisions with NH3 during this exposure. clusters

such as

Note that some silicon

Si44+ and Si46 + have chemisorbed NH3

so efficiently

under these conditions that very little of the bare cluster remains.

In sharp contrast, under the same conditions other clusters such as Si45 + show little evidence of any reaction whatsoever.

Preliminary

indications of such sharp reactivity variations were published previously

258

48

51

49

52

45

47

1400

1~00

_{

1$00

Fig.

5.

1300

.J....

.......-~o
1400 MASS

44

46

(AMU)

'

J

j

. .I.

I

1300

Surface chemistry probe of silicon clusters in the 44-54 The top panel shows the (SWIFTed) atom size range. contents panel

of the

shows

IGR cell before

the

distributi on of a

result

on

this

The bottom

reaction. bare

5 second exposure

to

silicon l

x

cluster

l0- 6

torr

pressure of ammonia at 300K.

259

by our group 16 .

Now that more extensive data has been accumulated it is

clear that these variations are quite real and reproducible. A broad summary of the reactivities of the silicon clusters in the 3-65 atom size range, all measured as a percent fractional depletion of the bare cluster signal is shown in figure 6. clusters

are

quite

relatively inert,

reactive,

there

are

a

Note that even though most few

special

namely those at 11,25,39, 45,

currently in the process

ones

that

and 64 atoms.

are

We are

of measuring the absolute reaction rates

for

these surface chemisorption reactions.

Such dramatic dependence of reactivity upon cluster size has one of the hallmarks of the new cluster studies.

been

It is a particularly

striking effect in dissociative chemisorption reactions of species like

100 1

/1

80

...,.,.,

"0

60

J

~~

c.., 0

""

\ I

40

20

0

10

20

30 # Si

so

40 atoms

60

70

Fig. 6. Measured fractional depletion of bare silicon clusters due to chemisorption of NH3 at 2 x 10- 7 torr during a reaction exposure cluster ion).

260

(roughly

35

NH3

collisions

for

5 s each

H2 and N2 on small transition metal clusters. to find it here. here

for

very

But it is quite surprising

First, sharp variations in reactivity are being seen large

clusters.

In

all

previous

cases

the

measured

reactivity variations largely died out before the cluster size reached 30 atoms, yet here even at 64 atoms special behavior is evident.

Second, it

is far easier to produce amorphous silicon by rapid quenching than it is for typical metals.

It would not have been surprising to have found

these silicon clusters to be essentially amorphous, with each cluster size having many geometrical isomers represented in the ICR trap, each with

a

distinct

happening.

reaction

chemistry.

Yet

that

clearly

cannot

be

The inertness of clusters like Si45+ indicates that there is

a special structure for this cluster size and that somehow the 45-atom clusters have been able

to anneal

sufficiently to

settle

into

this

structure. Much work remains to be done to quantify these reaction rates, test for the presence of multiple isomeric forms for some of the more reactive clusters,

measure

the

rates

of subsequent reactions

after the

first

chemisorption event, explore other reactants, study the surface chemistry of the negative silicon clusters, etc.

But already there is a clear

challenge from this early data to theorists interested in understanding silicon surface chemistry. The experimental fact that certain silicon cluster sizes are special is a clue as to how silicon restructures. Somehow these special clusters like Si39+ and Si45+ have found a way to restructure

all

around

the

chemisorption site remains.

surface

so

that

not

a

single

active

Yet when even a single atom is added or

removed, such a perfect restructuring no longer appears possible. This is just the sort of intriguing experimental result one hopes to find with this new cluster approach to surface science.

Although

silicon clusters in this size range still constitute a terrific challenge to

electronic structure

theorist,

it

is not a ridiculous

challenge.

Already extensive ab initio calculations have been published for silicon clusters and their chemistry in the 2-10 atom size rangel7-1 8 . Interestingly, local spin density theory19 and even an elementary tight binding hamiltonian20 appear to agree quite well with high level ab initio theory on the structures of the small silicon clusters.

So it is

261

detailed

quite

that

conceivable

and

calculations

predictive

be

will

available over the next few years for silicon clusters throughout the 2As yet there is no compelling explanation for the

100 atom size range.

new reactivity data based on such fundamental calculations, although some interesting first steps along this road have been published recently2 1 • 22 examining in some detail a structural suggestion originally put forward by J. C. Phillips2 3 .

suggested by Kaxiras 24 as a

symmetry has be cluster.

strikingly beautiful model with full tetrahedral

A

the Si45+

candidate for

Interestingly the surface of this structure is characterized by

the same restructured bonding motifs as those found on the (111) face of bulk silicon.

CONCLUSION This example of silicon is only one of quite a number that are As

beginning to come out of the new generation of cluster beam studies.

mentioned a number of times above, transition metal surface chemistry has been

another

experiments

area similar

that

discovered

of to

there

those is

a

above

discussed

ICR

cluster

silicon,

for

we

have for

reactivity

of

variation

dramatic

in

example,

For

activity.

great

dissociative chemisorption of H2 as a function of cluster size 4 • 8 . reactivity sensitivity to size appears at least to first

This

order to be

roughly independent of net charge on the cluster, but highly responsive to

in

changes

cluster

geometry.

being

found

in

either

19-th

The

cluster

of

niobium

is

It appears to have roughly equal likelihood of

particularly interesting. a

very

reactive

successfully on nearly every collision, reactive by over 5 orders

form

that

chemisorbs

H2

and another form which is less

of magnitude.

On listening

to

the

oral

presentation of this paper at the conference, Karl Kharas realized that this may be

explained by

two

highly symmetrical

available to metal clusters of 19 atoms 25 .

packing arrangements

His proposal for the inert

form is a simple octahedron composed of tight packing of rows of 1,4,9,4., and 1 atoms.

This cluster has eight identical faces, each of which is a

section of the (111) surface plane of the bulk metal.

Such smooth faces

on transition metal surfaces are often found to be fairly unreactive. Kharas's

proposal

for

the

reactive

form

is

composed

of

two

interpenetrating icosahedra made up of rows of 1,5,1,5,1,5, and 1 atoms with the central pentagonal ring staggered with respect to the outer two. The surface of cluster as a concave "belt" of surface sites around its

262

middle,

providing a

number

of

reasonable

locations

for

dissociative

addition of H2. The

emerging picture

is

then that

some of

these

small,

naked

clusters studied in the molecular beam may have beautifully simple and perfect structures.

While as yet we do not have a simple direct probe of

the structure, in the fullness of time it should be possible for us to develop experimental proof of one or another hypothesis for the structure of such special clusters as C60• Si45+, and Nb19+.

Such nanocrystalline

pieces of the bulk may then serve as wonderful testing grounds for the development of a deep, fundamental understanding of at least a part of surface science.

ACKNOWLEDGEMENT The author would like to acknowledge simulating discussions with Karl Kharas and Efthimios Kaxiras on the likely structure and surface chemistry of some of the clusters discussed,

and for communication of

their fine papers on this subject in advance of publication. Cluster research in the authors laboratory is supported by the Office of Naval Research (gallium

(semiconductor clusters),

arsenide

clusters),

the

the US Army Research Office

Department

of

Energy,

Division

of

Chemical Sciences (bare metal clu;ters), the National Science Foundation (carbon clusters and chemisorbed cluster chemistry), and the Robert A. Welch Foundation.

REFERENCES 1. a) T. G. Dietz, M. A. Duncan, D. E. Powers, and R. E. Smalley, J. Chern. Phys. 74 6511 (1981). b) V. E. Bondybey and J. H. English, J. Chern. Phys.76 2165 (1982). 2. L. S. Zheng, P. J. Brucat, C. L. Pettiette, S. Yang, and R. E. Smalley, J. Chern. Phys. ~ 4273 (1985). 3. L. A. Bloomfield, M. E. Geusic, R. R. Freeman, and W. L. Brown, Chern. Phys. Lett. 121 33 (1985). 4. J. M. Alford, F. D. Weiss, R. T. Laaksonen, and R. E. Smalley, J. Phys. Chern. 90, 4480 (1986). 5. M. D. Morse, M. E. Geusic, J. R. Heath, and R. E. Smalley, J. Chern. Phys. 83 2293 (1985).

263

6. R. E. Smalley in Comparison of Ab Initio Quantum Chemistry with Experiment: State of the Art edited by R. J. Bartlett (Reidel, New York, 1985) pp 53-65. 7. J. M. Alford, P. E. Williams, D. J. Trevor, and R. E. Smalley, Int. J. Mass. Spectrom. Ion Phys. 72, 33 (1986). 8. J. L. Elkind, F. D. Weiss, J. M. Alford, R. T. Laaksonen, and R. E. Smalley, J. Chern. Phys. 88, 5215 (1988). 9. S. C. O'Brien, Y. Liu, Q. Zhang, J. R. Heath, F. K. Tittle, R. F. Curl, and R. E. Smalley, J. Chern. Phys. 84, 4074 (1986). 10. P. J. Brucat, C. L. Pettiette, L. S. Zheng, M. J. Craycraft, and R. E. Smalley, J. Chern. Phys. 85, 4747 (1986). 11. K. J. Taylor, C. L. Pettiette, M. J. Craycraft, 0. Chesnovsky, and R. E. Smalley, Chern. Phys. Lett. 152, 347 (1988). 12. R. F. Curl and R. E. Smalley, Science 242, 1017 (1988). 13. L. A. Bloomfield, R. R. Freeman, and W. L. Brown, Phys. Rev. Lett. 54 2246 (1985). 14. Q. L. Zhang, Y. Liu, R. F. Curl, F. K. Tittle, and R. E. Smalley, J. Chern. Phys. 88, 1670 (1988). 15. A. G. Marshall, T. C. L. Wang, and T. L. Ricca, J. Am. Chern. Soc. 107, 7893 (1985). 16. J. L. Elkind, J. M. Alford, F. D. Weiss, R. T. Laaksonen, and R. E. Smalley, J. Chern. Phys. 87, 2397 (1987). 17. K. Raghavachari, J. Chern. Phys.

84~

5672 (1986).

18. K. Raghavachari, and C. M. Rohlfing, J. Chern. Phys. 89, 2219 (1988). 19. P. Ballone, W. Andreoni, R. Car, and M. Parrinello, Phys. Rev. Lett. 60, 271 (1988). 20. D. Tomanek and M.A. Schluter, Phys. Rev. B 36, 1208 (1987). 21. D. A. Jelski, Z. C. Wu, and T. F. George, Chern. Phys. Lett. 150, 447 (1988). 22. J. R. Chelikowsky, Phys. Rev. Lett. 60, 2669 (1988). 23. J. C. Phillips, J. Chern. Phys. 88, 2090 (1988). 24. E. Kaxiras, Chern. Phys. Lett., in press. 25. K. C. C. Kharas, Chern. Phys. Lett., in press.

264

THE ELECTRONIC STRUCI'URE OF METAL DIMERS AND METAL CLUSTERS: 1HE EIGHTEEN-ELECTRON RULE vs. SKELETAL ELECTRON-PAIR COUNTING

Michael B. Hall Department of Chemistry Texas A&M University College Station, Texas 77843-3255

INTRODUCTION The 18-electron or effective atomic number rule (EAN)l has been extremely valuable to organometallic chemists in directing their thinking toward new compounds and "explanations" of the structure and bonding in known compounds. In metal dimers one often postulates the degree of metal-to-metal bonding by enforcing the 18-electron rule. As chemists made larger and larger clusters, there came the realization that the 18-electron rule was difficult, if not impossible to enforce. This problem lead to the formulation of skeletal electron pair counting (SEPC)2 and cluster valence electron counting (CVEC). 3 A strong connection between the bonding in main group clusters such as the boranes and that in organometallic or low oxidation state transition metal clusters can be made through the isolobal principle.4,5 Again SEPC and the isolobal principle have lead to suggestions for new compounds and new synthetics routes, as well as "explanations" for the structure of existing compounds. Without any doubt these qualitative electron and orbital counting procedures have contributed substantially, and will continue to contribute, to the development of the chemistry of organometallic compounds, especially dimers and dusters. However, in spite of their usefulness in organizing our thinking and providing useful relationships, it is not obvious that any of these electron counting principles leads to an accurate picture of the bonding. They clearly do a good job of rationalizing and predicting the molecular structure, but how well do they do in rationalizing and predicting the electronic structure at a deeper level? We will examine this question by discussing in some detail the bonding of the clusters Os3(COl9X2 (X=S, Sel and the dimers (CpM}2(J.1-E0)2 (M=Co, Rh, Ir and E=C, N). Previously, we have published the photoelectron spectra and approximate molecular orbital calculations of these compounds.6,7 RESULTS AND DISCUSSION

These clusters may be viewed as square based pyramids with the unique Os forming the apex and the X's and other two Os's forming the base, 1.

Merai-Meta/ Bonds and Clusters in Chemistry and CatalysiS Edited by J.P. Fackler, Jr. Plenum Press, New York, 1990

265

0

'c

l

/0

c

";/ o,~/~,//o

x-,s- c-o

o-c-r c

c

I

I

0

0

1

One can think of 1 as an electron precise duster which satisfies the EAN rule at all atoms (Seat the main group atoms and 18e- at the metal atoms). To do so one can treat the X's as x2-, the apical Os as QsO (dB), and the basal Os's as Os2+(d6). The principle resonance structure would then appear as 2

/f\

as2• -..- as~as2•

'\J/ ~-

2

In this structure the bonds have been drawn as dative bonds (arrows) to coincide with the original oxidation state choice. However, we do not wish to imply that these bonds are necessarily dative or that the initial oxidation state is an accurate reflection of the final charges. In our view an equally accurate (or equally inaccurate) representation of the electronic structure would have all covalent bonds, 3. One should view 2 and 3 as resonance structures; the "true" total wavefunction being a mixture of these and other related resonance forms.

3

266

The key feature in all these forms is the existence of eight 2-center, 2-electron (2c/2e) bonds. To make these bonds an Os center (the apical Os in 2 and 3) has used one of its 5d t2g-like electron pairs (one of the three pairs stabilized by backbonding to the carbonyls) in the Os-Os bonding. One can also describe this view in terms of hybrid orbitals, again our description should not be taken too literally as any hybrid orbital description is a gross approximation. Here x2- can be viewed as "sp3" with three of the hybrids used in X-Os bonds and one remaining as a lone pair. The basal Os's are "d2sp3" with three hybrids used for bonding to CO, and two used for Os-X bonds, and one used for each Os-Os bond. The seven coordinate apical Os is "d3sp3" with three hybrids for CO bonding, two for Os-X bonds, and two for Os-Os bonds. On all Os's the unused Sd orbitals are occupied; thus, the basal Os's have d6 cores and the apical Os has a d4 core. Regardless of oxidation state choice, there are exactly the correct number of hybrids and electrons remaining to form eight 2c/2e bonds. An alternative view of this cluster is provided by skeletal electron pair counting (SEPC). Here, we use the isolobal principle and make an analogy to the electron deficient boron hybrides. For our Os3(CO)gX2 clusters the Os(C0)3 fragments are isolobal to BH fragments, where each Os(C0)3 contribute 3 orbitals and 2 electrons to cluster bonding. The remaining 6 d electrons on each Os are not involved in the cluster bonding but are on! y involved in bonding to the carbonyls. Each X contributes 3 orbitals and 4 electrons. Thus, the Os3(CO)gX2 clusters can be thought of as 5 vertex polyhedra with 7 skeletal electron pairs. By analogy to the boron hydrides, they are nido clusters (n+2 pairs) and are derived from an octahedron with one vertex missing. The SEPC approach succeeds in predicting the observed geometry. In small, lower symmetry, clusters one can often arrive at a localized bonding picture by following the prescription of Lipscomb.B For the boron hydride analog, BsHg, the square base of the pyramid is bonded by 3c/2e B-H-B bonds and the capping BH group is bonded to the four basal borons by two 2c/2e bonds and one 3c/2e bond. A number of different arrangements are possible. In our lower symmetry system one may choose from either 4 or 5.

5

4

Thus, one could have the 3c/2e bond holding the capping Os to the other Os (4) or to the X groups (5). A third possibility also exists in which the 3c/2e bond is a central rather than an open bond (6). There would be four equivalent resonance forms like 6.

6

The principle differences between the 18-electron description and the SEPC description is the number of electron pairs involved in cluster bonding. In enforcing the lBe- rule we must involve one of the t2g-like pairs, which SEPC and the isolobal analogy suggest are involved

267

principally in Os-CO bonding. Now that we have set the stage, we will discuss the results of our molecular orbital (MO) calculations in an attempt to resolve the questions: to what extent the t2g-like pairs are involved in cluster bonding and, which, if any, of the 3c/2e models is closest to the MO results. Figure 1 shows the MO diagram for both Os3(C0)9S2 and Os3C0)9Se2. The relative energies of the 5 3p and Se 4p are shown on the left, while in the high lying orbitals of the Os(C0)3 fragment are shown on the right. Hoffmann and coworkers have shown how carbonyl fragments such as Os(C0)3 retain much of their octahedral parentage.9 Thus, the orbitals in Figure 1 are labelled "t2g" for the three Sd orbitals stabilized by the carbonyls 1t* (7t acceptor orbital), "eg" for the two Sd orbitals destabilized by the carbonyls lone pair (cr donor orbital), and "at" for the unused 6s/6p hybrid. Together the "eg" and "at'' orbitals form the three unused hybrids on the Os(C0)3 fragment while the "t2g" orbitals are mainly spectator orbitals according to the isolobal analogy with BH and SEPC. There are eight electrons to distribute among these orbitals and the Os(C0)3 fragment has six in the "t2g" and two in the "eg"· Thus, according to the SEPC formalism Os(C0)3 has two electrons and three orbitals to contribute to cluster bonding. The 5/Se p orbitals contribute four electrons and three orbitals to cluster bonding. As Figure 1 shows there is strong mixing of the 5 orbitals with the "t2g", "eg", and "at" orbitals. This mixing will only be important, i.e. produce a net bond, if it results in an occupied bonding orbital and an unoccupied antibonding orbital. At low energy we would expect to find six 5-0s bonding orbitals. However, Figure 1 only shows five low-lying 5-0s orbitals; the sixth, 6a', is found in the midst of the "t2g" orbitals. It is easily identified because of its large energy shift when 5 is replaced by Se. This orbital is plotted in Figure 2, where one can see that it has strong X-X antibonding character which is responsible for its high energy.

X

(a)

(b)

Os(COI3

Figure 1. Molecular orbital diagrams for (a) Os3(C0)9S2 and (b) Os3(C0)9Se2. The energy values were obtained from Fenske-Hall calculations. 7a" is the HOMO. (Reprinted with permission from reference 6. Copyright 1988 American Chemical Society.)

268

_.-·

....· ..··

.....·····

···... ···... ··.._ ··. ··..

·····-·-· ...

Figure 2. Orbital plot of 6a' in the mirror plane. The lowest contour values are 2.44 x w-4 e au-3, and each succeeding contour differs from the previous one by a factor of 2.0. (Reprinted with permission from reference 6. Copyright 1988 American Chemical Society.)

Figure 3. Orbital plot of 7a" in the metal plane of Os3(CO)gSe2. The lowest contour values are 2.44 x w-4 e au-3, and each succeeding contour differs from the previous one by a factor of 2.0. (Reprinted with permission from reference 6. Copyright 1988 American Chemical Society.)

The highest occupied molecular orbital (HOMO) is the 7a". A plot of this orbital, Figure 3, shows it to be a 3c/2e bond between the apical Os and the basal Os's.6 The remaining nine molecular orbitals are derived from the "t2g" pairs. The MO calculations indicate that the dominant bonding is that shown in 4. However, the Os-Os bonds are shorter than one would expect for a bond order of 1/2. A closer examination of the overlap population reveals net bonding between the "a1" and "t2g" orbitals of the Os(C0)3 fragments. This bonding occurs mainly in the 9a' orbital, the second highest occupied orbital, and in the 1a' orbital, the

269

lowest energy orbital shown in Figure 1. Such bonding in the ai system could help account for the nearly typical Os-Os bond lengths that are observed. Interestingly, the high-lying 6a', which is primarily involved with the apical Os, does not provide as much Os-X bonding as any of the other five Os-X bonding orbitals. Thus, the Os-X bonds to the apical Os may be somewhat weaker. This high-lying X-Os orbital also plays an important role in the chemistry of these clusters as it seems to be involved in directing the attack of electrophiles such as PtL2 to the X-Os edge.B For these clusters the reported photoelectron spectra are consistent with our orbital description, but with many closely spaced levels a detailed assignment was not possible.6 In conclusion, the principle bonding features are well represented by the SEPC model with the major form of possible resonance structures being 4. However, the EAN structures, 2 and 3, and the alternative SEPC structures, 5 and 6, appear also to contribute in some smaller way.

In this section we will concentrate on the bonding of carbonyl and nitrosyl derivatives of Group 9 (Co, Rh, Ir). In the solid state, both <11 5 -CsMes)2Co2(~-C0>2 and (1l 5-CsHs>2Co2(~­ N0)2 appear to have planar M2(E0)2 cores as shown in 7

7

In solution, the Co2(N0)2 derivative apparently has a puckered geometry. In our description of the bonding we will concentrate on the planar form. Some previous theoretical calculations suggested the possibility of a triplet ground state,11 but the singlet nature of the ground state is consistent with the PE spectrum7 and was recently confirmed by magnetic measurements.12 If one enforces the 18-electron rule the Co2(C0)2 dimer would have a double Co = Co bond while the Co2(N0)2 dimer would have a single Co - Co bond. On the other hand if we view this dimer from the isolobal analogy: CO and NO are 2 and 3 electron donors, respectively (isolobal to BH and BH-, respectively), and the CoCp fragment supplies 2 electrons (isolobal to BH). The isolobal principle is limited here because B4H4 (isolobal to Cp2C02(C0)2) would be tetrahedral and B4H42- (isolobal to Cp2Cp2(N0)2) would be planar, whereas Cp2C02(C0)2 is planar and Cp2C02(N0)2 may have a folded structure. However, the isolobal electron counts do provide us with the number of electrons available for the principal bonds which confer stability to these molecules. Thus, the Cp2Co2(C0)2 dimer has four pairs available for a cluster bonding and Cp2Co2CN0>2 has five pairs available. In constrast, the 18-electron rule assigns six pairs to "cluster" bonds in Cp2Co2(C0)2,: four pairs for C - Co bridge bonds and two pairs for Co = Co bonds; and five pairs to duster bonds in CP2Co2(N0)2: four pairs for N- Co bonds and one pair for the Co -Co bond. For Cp2Co2CC0)2 , the two approaches differ widely in the number of pairs that they assign to cluster bonding. Even for Cp2C02(N0)2,. where the number of pairs is identical, the placement of the pairs is very different. Planar B4H42- or C4H42+ would have two electrons in the 1t system and would

270

be aromatic, but the 18-clectron rule would suggest a Co -Co cr bond in addition to the four inplane Co - N bonds, i.e. all five pairs in the molecular plane. The molecular orbital (MO) energy level diagram, which is for the Rh dimers7, is shown in Figure 4. The left side shows the upper energy levels of the CpRh fragment; the lowest energy orbitals shown, let, are the degenerate p donor orbitals of the Cp- ring. Above those, the e2 and lat are the remains of the "t2g" like 4d orbitals. At even higher energy are the 2e1 and 2at which are destabilized by the donation from the Cp- ring. The 2et are mainly 4dxz and 4dyz, while the 2at is primarily an "sp" hybrid on Rh. These CpRh fragment orbitals interact weakly with each other and strongly with the orbitals of the bridging ligands which are shown on the right of Figure 2. Since NO is a poorer cr donor and better 1t acceptor than CO, both its 5cr donor orbital and its 27t acceptor orbital lie lower in energy than the corresponding orbitals of CO. The center has two columns of levels, one for the CO dimcr, the other for the NO dimer. At low energy are the out-of-phase and in-phase combinations of the 5cr donor orbitals, lau and lag respectively. The lau is stabilized by the metal fragments 2e1 (dyz) while the lag is stabilized by the 2at orbitals. The next 4 orbitals are the result of the interaction of the Cp-Rh "1t" bonding orbitals. These are mainly Cp in character and contribute little to the net "cluster" bonding. The next 6 orbitals are the bonding and antibonding combinations of the "t2g" orbitals. The next two orbitals, 3bg and 4bu, are the two linear combinations of the 2e1 which are

....m.. co

CpRh

EO

Figure 4. Molecular orbital diagram for CpzRh2(EOl2 (EO = CO, NO). The energy values were obtained from Fenske-Hall calculations. The dashed lines indicate small but significant contributions from the fragment orbitals to the molecular orbitals. (Reprinted with permission from reference 7. Copyright 1988 American Chemical Society.)

271

strongly stabilized by interactions with the 21t orbitals. These two orbitals are occupied in both dimers. The Sag orbital, which is the antibonding combination of the dxz's, is occupied only in the nitrosyl dimer. The MO calculations suggest that the primary "cluster" bonding in Cp2Rh2(C0)2 involves the lau, lag, 3bg and 4bu. These orbitals contain the four pairs identified by the isolobal analogy as being involved in cluster bonding. One might expect to find all four pairs involved in Co-C a bonds, but the 4bu is an out-of-plane orbital. Figure S displays orbital drawings for these four orbitals and the Sag which is occupied in the NO dimer. The lag is the totally bonding combination of the CO a donor and the "sp" hybrid on the metal. The lau and 3bg are antibonding and bonding combinations of the metal dyz orbitals. Figure S shows them as linear combinations which emphasize that the net bonding is equivalent two 3c/2e C-Rh-C bonds. The 4bu is a totally bonding "1t" orbital similar to the lowest energy 1t orbital of C4H4. This orbital contributes to the planarity of the carbonyl dimer. Occupation of the Sag breaks any direct metal-metal bonding in the 4bu and would reduce the barrier to "folding" the dimer. Thus, Cp2Co2(N0)2 appears planar in the solid but folded in solution. For the NO dimer, one could form linear combinations of the 4bu and Sag, the new orbitals would be 3c/2e N-Rh-N "1t" bonds similar to their a counterparts, lau ± 3bg. Although the MO calculations suggest that the primary "cluster" bonding is confined to these orbitals, they also indicate some important secondary bonding effects. The MO diagram, Figure 4, shows the interaction of the la1 orbitals on each dimer to form the 3ag, M-M bonding, and the 3bu, M-M a antibonding. The isolobal analogy suggests that the la1 orbitals are "spectator" orbitals not involved in cluster bonding. This result would be the situation if the 3bu antibond cancelled the 3ag bond, but it doesn't. The 2a1 fragment orbital stabilizes the 3bu converting it, at least partly, into a non-bonding orbital. Thus, there is net a M-M bonding in this "t2g" mainfold. A second result, also unexpected from the isolobal analogy, is that the CO and NO 1t• orbitals contribute strongly to the mainly metal 2bu. This contribution is clearly evident in substantial stabilization of the 2bu which occurs when the CO's are replaced by NO's. This orbital corresponds to the "missing" fourth in-plane M-EO bonding orbital. The four M-Eb bonds can be made from the lau, lag, 2bu, 3bg· The 3ag contributes a M-M a bond and the 4bu contributes a M-M 1t bond. Now we are back to 6 electron pairs involved in cluster bonding as suggested by the 18-electron rule. In the NO dimer the bonding 4bu and anti bonding Sag cancel 5aQ

~

~

a 4b,

~

0

cP

~~

~~

cP

Iau+ 3b;

lao

®X)

~

Ia,- 3bo

~ ~

(:)®

Figure S. Valence description of interfragment bonding. The view is of the yz plane (molecular plane). Only orbitals of the metals and proximal atoms of the bridging ligands are shown. (Reprinted with permission from reference 7. Copyright 1988 American Chemical Society.)

272

each other forming two lone-pairs on the metal fragments. Here, the remaining 5 pairs form 4 M-NO bonds and one M-M bond. CONCLUSIONS We have come "full circle" on the issue of the isolobal analogy and skeletal electron pair counting vs. the 18-electron rule. One can generalizes our observation by stating that careful application of the isolobal principle and skeletal electron pair counting will lead to the enumeration of the principle bonding in the cluster, but the pairs which are still missing from those required by the 18-electron rule may provide important secondary bonding interactions. Finally, our unpublished ab initio calculations on these dimers shows that even the detailed analysis given above is an oversimplification of the "true" electronic structure. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

N.V. Sidgwick, "The Electronic Theory of Valency", Cornell Univ. Press, Ithaca, N.Y. (1927). K. Wade, Adv. Inorg. Radiochem., 18:1 (1976); D.M.P. Minges, Acct. Chern. Res., 17:311 (1984). J.W. Lauher, J. Am. Chern. Soc., 100:5305 (1978); J.W. Lauher, J. Am. Chern. Soc., 101:2604 (1979). M. Elian and R. Hoffmann, Inorg. Chern., 14:365 (1975). R. Hoffmann, Angew. Chern. Int. Ed. Engl., 21:711 (1982). G.L. Griewe and M.B. Hall, Inorg. Chern., 27:2250 (1988). G.L. Griewe and M.B. Hall, Organomet., 7:1923 (1988). W.N. Lipscomb, "Boron Hydrides," W.A. Benjamin, New York (1963); W.N. Lipscomb, Science, 196:1047 (1977). M. Elian, M.M.-L. Chen, D.M.P. Mingos, and R. Hoffmann, J. Am. Chern. Soc., 98:7240 (1976). R.D. Adams, T.S.A. Hor, and I.T. Horvath, Inorg. Chern., 23:4733 (1984). J. Demuynck, P. Mougenot, and M. Benard, J. Am. Chern. Soc., 109:2265 (1987). D.J. Berg and R.A. Anderson, J. Am. Chern. Soc., 110:4849 (1988).

273

EXPERIMENTAL MEASURES OF METAL-METAL SIGMA, PI, AND DELTA BONDING FROM PHOTOELECTRON SPECTROSCOPY Dennis L. Lichtenberger and Roy L. Johnston Laboratory for Electron Spectroscopy and Surface Analysis Department of Chemistry University of Arizona Tucson, AZ 85721 INTRODUCTION Since the discovery, in 1964 by Cotton and co-workers, that the anion [Re 2Cl 8 ] 2- contains a quadruple metal-metal bond, 1 the field of multiple metal-metal bonds has grown dramatically. 2 From the very beginning, CI

1/C

I

CI

I

/Re

c

CI

c

2-

17 I

(I

Re

CI

interest has focused on the nature of the bonding between the metals. In a seminal paper, Cotton proposed that the quadruple bond is composed of one u, two w and one 6-component arising from overlap of the d orbitals on the metal atoms, as shown in Figure 1. 10 Molecular Orbital (MO) calculations of varying degrees of sophistication have subsequently shown that these symmetry interactions are important to the metal-metal bonding in a

-M

=£:~ N ·~:

.

a

~

Mz

M IE M

Figure 1. Spectrum of d-based MO's for quadruple bonded metal dimers Metal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J.P. fackler, Jr. Plenum Press, New York, 1990

275

number of different metal dimer complexes. 3 However, there has been considerable disparity among these calculations with regard to the calculated stability of these levels, their contributions to the total metalmetal bond strength, and mixing with other states. These quadruple bonded metal dimers lie at the foundation of studies of bonding and reactivity in metal systems. Our understanding of bond breaking and bond making chemical transformations is rooted in our understanding of the simplest examples of these bonds. For example, models of covalent a bonding trace to the hydrogen molecule and models of p~ bonding trace to the nitrogen molecule. The bonding in more complex molecules can not be understood any more completely than these simplest examples. Likewise, metal-metal dimers provide the most basic examples of 6 bonding. Although bare metal diatomic molecules might be considered the simplest examples, most of the information on metal-metal interactions has been obtained from the study of ligated species. Examples are the tetracarboxylates of chromium, molybdenum, and tungsten (type I), and the phosphine-halides of molybdenum, tungsten, and rhenium (type II).

Type II There are several advantages to studying the ligated metal dimers over the bare metal species. Although the presence of ligands reduces the symmetry of the dimers from D~, the local (virtual) symmetry is high enough to enable the assignment of a, w and 6 labels to the components of the multiple bond. For theoretical calculations, the larger number of atoms is compensated by the ligand-induced destabilization of a number of the virtual levels so that there is less configuration interaction. 30 • 4 These ligated metal dimer complexes are stable and can, therefore, be studied with a variety of experimental techniques, including high resolution spectroscopy in the gas phase. 5 In addition, ligated metal dimers offer the opportunity to study the effects of chemical perturbations on the metal-metal interaction. The present understanding of bonding in multiple bonded metal dimers has involved major contributions from preparative, theoretical and physical (spectroscopic and magnetic) investigations. 2 One of the most informative experimental measures of metal-metal a, w and 6 bonding has come from photoelectron spectroscopy (PES). 6

PHOTOELECTRON SPECTROSCOPY: BACKGROUND The photoelectron experiment affords us valuable direct evidence as to the energies and bonding of the various electronic states of the molecular ion. These states are generated by the ejection of electrons from the ground electronic state of the neutral molecule by absorption of a

276

photon of energy hu: M0 (initial) + hu

--""'

(1)

The ionization or binding energy Er is merely the difference in energy between the excited electronic state and the initial state of the molecule: hu

(2)

where EK(e-) is the experimentally measured kinetic energy of the ejected electron. The relationship between the ionization energy and the theoretical (one-electron) molecular orbital energies is given by Koopmans' theorem, 7 which states that the ionization energy is the negative of the orbital energy (e 1 ): (3)

provided that there is negligible relaxation (energy lowering due to elec· tron redistribution) of the molecular ion state. Relaxation effects may be large in metal complexes, 8 so care must be taken when assigning ioniza· tion features in a PE spectrum with the aid of an MO calculation, especi· ally when the ionizations are closely spaced. The most significant information from the photoelectron experiment is often independent of Koopmans' theorem. For example, analysis of the ionization bandshapes and vibrational fine structure enables the bonding nature of the orbitals to be determined. 6 • 9 A broad vibrational progression is a direct measure of the bonding contribution of the electron associated with that molecular state. The ionization energy is also a well-defined thermodynamic quantity. Molecular ionization energies are directly related to bond energies through thermodynamic cycles. 10 Thus electronic structure and bonding information can be obtained from photo· electron spectroscopy without Koopmans' theorem, and the results are true independent tests of the results of theoretical methods. PHOTOELECTRON SPECTROSCOPY OF QUADRUPLE BONDED METAL DIMERS Figure 2 shows the gas phase ultraviolet PE spectra obtained for the quadruple bonded acetate bridged dimers M2 (0 2 CCH 3 ) 4 of chromium, molybdenum and tungsten. The ionizations above 10 eV are largely associated with the acetate ligands and change very little between the spectra of the different metal complexes. The ionizations below 10 eV are associated with the metal valence d levels and show striking differences from Cr 2 to Mo 2 to W2 . The first ionization of the chromium dimer is extremely broad. A close-up examination of this band (Figure 3) shows that it is comprised of three peaks in an approximate 1:2:1 intensity pattern. Some vibrational structure is evident on the top of the band, but the three broad peaks overlap too much to obtain much specific information. The low valence region of the molybdenum complex has two clearly identified peaks. The first has been attributed to ionization from the 5 bonding (2b 28 ) MO of the dimer, 11 as predicted by MO calculations on the quadruple bonded systems (see Figure 1). 10 • 3 The second is assigned to the~ ionization state. One problem in this system has been locating the ionization associated with the a state. The spectrum of the tungsten complex shows three separated ionizations in the low valence region. The first two have again been assigned to the 5 and~ ionizations, and the third has been assigned to the a ionization (vide infra). The sharpness of the third ionization and its close proximity to the ~ ionization are unusual for ionization of a a bonding electron.

277

Ionization Energy (eV)

9

13

17

Figure 2.

He(I) PE spectra of M2C02CCH3) complexes.

Ionization Energy (eV)

9

8

'

'

'

'

'

'

'

' \

\

...... _:::~j-:·.:·:····················-.-·· ... '.\"' ~~ ~~

..... ~,·

Figure 3.

278

'......... . ........... . .,"

...

First ionization band of Cr2C02CCH3)4.

Of the many interesting questions which have arisen out of the PES study of quadruple bonded metal dimers, we will focus on understanding how electron occupation of the u, w, and 6 bonds influences the metal-metal bond length and strength. Knowledge of these influences is crucial if we are to advance our understanding of the fundamentals of bonding in these systems. As a first observation, the w ionization in each case shows a broad vibrational progression indicating substantial bond distance changes with w ionization. Thus the w bonds contribute significantly to the overall bond strength. The nature of the 6 and u bonds is much less apparent. The following sections will focus on the nature of the 6 and u bonds in multiple bonded metal-metal complexes. THE 6 IONIZATION Figure 4 shows a close up of the 2Bz8 (6) ionization band of Mo 2 (0 2 CCH 3 ) 4 . We have observed an extensive vibrational pro~ression in this broad band. 11c The frequency of this vibration in the B28 ion state was found to be lower (360 cm- 1 ) than that of the symmetric metal-metal stretch (406 cm- 1 ) in the neutral molecule 1A18 state. 14 • A Franck-Condon normal mode analysis of the 2 B28 ionization band structure was performed, 9 • 12 based on the assumption that the fine structure is due to the pure symmetric 13 (a18 ) Mo-Mo stretch. This assumption follows from spectroscopic studies on molybdenum acetate and related dimers in both the ground (6 2 ) and excited (6 1 ) configurations. 14 MO calculations on these systems also indicate that the 6 orbital has more than 80% metal character. 3 The Franck-Condon analysis leads to a calculated bond length increase of 0.18 A upon ionization of the 6 M0. 11 c The potential energy curves for the 1A18 and 2 B28 states are shown schematically in Figure 5. The large increase in equilibrium bond length accompanying the electronic transition results in the vertical transition to a high (v' - 7) excited vibrational state of the cation. The "Mo-Mo" vibrational frequency and calculated bond length for the cation state are given in Table I, together with the experimentally determined bond length of the neutral molecule in the ground ( 1A18 ) state. 15 For comparison, the Mo-Mo bond length resulting from exciting an electron from the 6 orbital to the 6* orbital (thereby generating the 1A2u

2 B28

ionization energy 7. 5

1

(eV) 6.5

279

Table I.

Compilation of stretching frequencies and bond lengths for [Mo 2(0 2CCH 3 ) 4 ]n+ (n - 0, 1).

Electronic State

Electronic Configuration

Formal Bond Order

1A1s (ground)

0271'462

4.0

406 8

2.093,b 2. 079°

1A2u ( 6-+6*)

o 2 '11' 4 66*

3.0

390d

2.19 (calc.) •

2Bzs (6 loss)

0271'46

3.5

360f

2.26 (calc.) f

VHo-~

(em- )

~-Ho (A)

Table adapted from reference llc. The calculated Mo-Mo bond lengths are obtained by adding the calculated increments to the experimentally determined (gas phase) bond length from reference 15b. a Reference 14a. b Reference 15a. c Reference 15b. d Reference 16. • References 14a and 17. f Reference llc.

state of the neutral molecule 16 ) is also shown. This transition changes the bond order from 4 to 3 in comparison to the change from 4 to 3.5 with ionization. Despite the greater change in formal bond order upon 6-+6* excitation, the bond distance lengthening is less (0 .11 A) 148 •17 and the vibrational frequency remains higher (390 cm- 1) 16 than for the 6 ionization. This has been explained in terms of the increased effective oxidation states of the metals with ionization, which contracts the metal d orbitals and reduces the bonding in the remaining occupied metal-metal orbitals. llc Table II lists some experimentally determined bond lengths for a number of multiple bonded molybdenum dimers with formal bond orders of 4.0, 3.5 and 3.0, so that the structural effect of removing one or two electrons from the 6 bonding MO can be compared.

75

~

73

~

71

z

6.9

~ w

0

2 ~ STATE

~

z

N

Q

NEUTRAL I""'T'"'1

-0.2 0 0 2

Mo-Mo DISPLACEMENT (A)

Figure 5. 280

Franck-Condon analysis of the 2B28 ionization.

Table II.

Mo-Mo bond lengths in a number of dimers; the effect of 6 ionization.

Dimer Mo 2 (0 2CCH 3 ) 4 [Mo2(S04)d4Mo 2 (form) 4a [Mo 2 (S0 4)d J[Mo 2(form) 4]+

Electronic Configuration

Formal Bond Order

a21r4 62

4.0

2.093(1), 2.079(3) 2.111(1) 2.085(4)

a27f46

3.5

2.167(1) 2.122(3)

18b. 20 19

a27f4

3.0

2.223(2)

21

a

[Mo2(HP04)4]2- b

• form- (~-tolyl)NCHN(~-tolyl). molecules. 21

19

~o-Mo (A)

Ref. 15 18 19

b This complex possesses axial water

From the Table, it is apparent that the bond lengthening upon removal of successive electrons from the 6 orbital (e.g. +0.056 A on going from [Mo 2(S0 4) 4] 4- to the trianion 18 · 20 and +0.057 A on going from the trianion to the closely related [Mo 2 (HP0 4) 4] 2- ion) 21 is much smaller than that calculated from the PE spectrum of molybdenum acetate (0.18 A). 11 c It seems, therefore, that the Franck-Condon analysis has overestimated the magnitude of bond lengthening accompanying the removal of an electron from the 6 MO. In the following sections we shall discuss possible reasons for these inconsistencies between spectroscopic and crystallographic measures of the lengthening of the Mo-Mo bond upon ionization. 1. The Hot Band Argument: The possibility that the presence of hot bands (i.e. transitions from excited vibrational levels of the ground electronic state) 22 are complicating the vibrational fine structure observed for molybdenum acetate was discussed in the original presentation of the analysis. 11 c It was concluded that the presence of hot bands does not affect the spacing between the apparent adiabatic (v-0 ~ v'-0) and vertical (assigned as 0~7) ionizations. This assertion was later questioned by Miskowski and Brinza. In a subsequent analysis they included contributions from as high as the fourth vibrationally excited level (v-4) and eventually concluded that a better upper limit on the bond lengthening upon 6-ionization would be 0.11 A, 23 i.e. approximately the same as that calculated for the 6 ~ 6* transition. 148 · 17 However, a simple Boltzmann distribution calculation indicates that less than 1% of the molecules have v ~ 4 at the temperature at which the PES data was collected (~ 200"C 473 K). Figure 6 shows the relationship between a vibrational profile at 0 K and the effects of hot bands (v - 400 cm- 1 ) at 473 K. The figure reveals that the addition of hot bands has little effect on the determination of the number of vibrational quanta between the adiabatic and vertical ionizations, even at 473 K. The Miskowski analysis also assumes that the vibrational frequencies are the same in the neutral molecule and the positive ion, and this assumption will over-estimate the hot band effect. It is also not certain that the adiabatic ionization is observed, and if it occurs at lower IP than assumed then the apparent bond distance change is even greater. In any event, even with over-estimation of the effects of hot bands, the analysis still results in bond distance changes that are greater than those found in the structural studies. 2. Effective Mass Correction: If, instead of treating the vibrational progression as arising solely from motion of the molybdenum atoms, we rather consider the motion of two rigid Mo0 4 units, the increase in effective mass leads to a reduction in the calculated bond length increase 281

0 K

Figure 6.

473 K

Comparison of ionization band profiles simulated at 0 K (shaded bars) and 473 K (solid bars).

to 0.13 A for the 2 B28 ion state. 11c In a similar analysis, Crosby et al. obtained a smaller value (0 . 08 A) for the bond lengthening in the excited 1A2u state (o-+o*) of the neutral molecule . 24 Thus, the inclusion of hot bands and changing the vibrating units does not reduce the calculated bond length sufficiently to be consistent with the data that we have for the oxidized multiply bonded

the effective mass of increase in metal-metal solid state structural dimers (Table II).

3 Inclusion of Metal-Ligand Vibrations : In addition to the factors mentioned above, it is possible that the vibrational progression observed in the o ionization contains symmetric M-0 stretch and Mo-Mo-0 de f ormation components. This would make the actual Mo-Mo displacement a fraction of that calculated, since the normal coordinate mode is a weighted combination of the activated internal coordinate modes. Thus, the a 18 internal coordinate modes of v(Mo-Mo), v(Mo-0) and o(Mo0 4 ) may combine to give a "missing mode effect" (MIME). 25 The symmetric Mo-O stretch and Mo -Mo-0 deformation are believed to lie ±300 cm- 1 from the symmetric Mo-Mo stretch, 144 • 8 • 26 but mixing may be induced under the perturbation of the electronic transition. Support for the contribution of metal-oxygen component(s) in t h e fi ne structure of the o -band comes f rom solid sta te s t ructural studies . Table III lists the M-M and M-X (where X is the ~-donor ligand, such as oxygen or chlorine) bond lengths for a number of multiple bonded dimers which are related by the removal of one or both electrons from either t he 6 bonding or 6* antibonding orbital (in the case of the technetium and rheni um complexes). Conside ring first the series of molybdenum compounds, going from o2 to 61 results in a Mo-Mo bond increase of 0 . 056 A, but the decrease in Mo-O bond length is larger (0 . 074 A). 18b Going from o 1 t o o 0 , the changes i n bond length are approximately equal in magnitude (an increase of 0.056 A in Mo -Mo and a decrease of 0.052 A in Mo-O) . 18b · 21 For the technetium chloride anions the removal of the 6* electron actual ly leads to an increase in the Te-Te bond (by anything from 0 . 016 A to 0.046 A). 27 - 29 Finally, for the rhenium phosphine chlorides, removal of the first (metal-metal antibonding) 6* electron results in a Re-Re bond shortening of 0 . 023 A, while t he Re - Cl bond shrinks by 0.057 A. 30 Remova l

282

Table III.

Comparison of M-M and M-X bond length changes on oxidation of a number of multiply bonded dimers.

Dimer

Electron Configuration

Formal M-M Bond Order

~-M (A)

~-X (A)

[Moz(S04)414[Moz(S04)413[Mo 2 (HP0 4) 412-

(]27f452 (]27f45 (]27f4

4.0 3.5 3.0

2.111(3) 2.167(1) 2.223(2)

2.139(4)" 2.065(3) 2.013(9)

lBb lBb 21

[ Tc 2Cl 8] 3 -

CJZ1f4525*

3.5

2.364(2)

[ Tc 2Cl 8 12-

CJZ'If40Z

4.0

2.105(1), 2.117(2) 2 .15l(l)b 2.133(3) 0

2.320(4) 2.338(9)

27 28 29

Re 2Cl 4( PMe 2Ph) 4 [Re 2Cl 4(PMe 2Ph) 41+ [ Re 2Cl 4( PMe 2Ph) 4 12+

(]27f4525*2 (]27f4525* CJZ'If402

3.0 3.5 4.0

2.241(1) 2.218(1) 2.215(2)

2.387(l)d 2.334(4) 2.291(7)

Ref.

30 30 30

M-X bond lengths are averaged. The e.s.d.'s given are averaged over all the M-X bonds also. b Major isomer (69.1%). c Minor isomer (30.9%). d X = Cl for all theRe complexes. a

of the second 5* electron has an even smaller effect on the Re-Re bond (it increases by only 0.003 A, but the high e.s.d. for the dication makes this number unreliable), with the Re-Cl bond shrinking by 0.039 A. The structural details reveal that the removal of a 5 (or 5*) electron generally results in a shortening of the M-X bond which is equal to or greater than the change in the M-M bond length. While part of the change in the M-0 bonds can undoubtedly be attributed to a reduction in the size of the metal ion upon oxidation (i.e. a tightening of the valence orbitals), the major contribution probably comes from a reduction in the M(d~)-X(p~) repulsion. This repulsion occurs due to the antibonding M-X interaction which is present in both the 5 and the 5* MO's when X is a 1r-donor ligand. 3 M-X

~

Interactions

The importance of the M-X(1r) interaction can be seen by studying dimers of the type M2X4(PR 3) 4 (M- Mo, W, Re; X- Cl, Br, I). ForM- Mo and W (quadruple bonded 52 systems) and Re (electron rich triply bonded 5 2 5* 2 systems) the geometry is always local Dzd (local symmetry includes only those atoms directly bonded to the metals). 31

P R3

X

I /X I /p R3 M===M 7 R3P

ll

P R3

I

X

283

This eclipsed conformation is expected for the Mo and W molecules, which possess a 6 bond, but for the Re complexes the occupation of 6* should favor the staggered conformation, as observed for triple bonded [Os 2X8 ] 2 " (X- Cl, Br), which has D4d symmetry. 32 The observed conformation of the rhenium phosphine halides is favored because eclipsing the phosphines with halides keeps the phosphines on the two ends of the molecule as far apart as possible and thereby minimizes steric repulsions between them. 33 PE spectra have been recorded for a number of these M2X4 (PR 3 ) 4 systems. 34 Figure 7 shows a comparison of the gas phase ultraviolet PE spectra of M2 Cl 4 (PMe 3 ) 4 , with M- Mo and Re. 340 The most noticeable feature in this comparison is the appearance of a new band at low binding energy (i.e. low ionization energy) in the rhenium case. This is due to ionization from the 6* MO, which is empty for molybdenum (and tungsten) 34 a,b and doubly occupied for rhenium. 34 h.c Figure 8 shows the effect of varying the halide (X) on the PE spectrum of the molybdenum species. The effect of replacing Cl by Br is to shift the w and 6 bands to higher binding energy, while the M-P(a) ionization moves to lower binding energy (see Table IV).

Ionization Energy leV) 16.0

12D

14.0

10.0

8.0

6.0

Figure 7.

Table IV.

Dimer

The effect of changing halide ligands on the ionization energies of the metal-based MO's of Mo 2X4 (PMe 3 ) 4 • Peak Assignment 6 w

M-P(a) 6 w

M-P(a)

284

Ionization Energy (eV) 6.44 7. 71 8.40 6.53 7.73 8.33

Table V.

The effect of changing the halide ligands on the ionization energies of the 6 and 6* MO's of Re 2X4 (PMe 3 ) 4 .

Dimer

Peak Assignment

Ionization Energy (eV)

Re 2 Gl 4 (PMe 3 ) 4

6* 6

5.66 6.51

Re 2 Br 4 (PMe 3 ) 4

6* 6

5.72 6.56

Re 2 I 4 (PMe 3 ) 4

6* 6

5.80 6.56

One would expect from a charge potential point of view that substituting Gl by the less electronegative Br should lead to a reduced positive charge on the metal and therefore those MO's with an appreciable amount of metal character should be less tightly bound, as is observed for the M-P(u) ionization. The fact that the ~ and 6 ionizations shift instead to higher ionization energies indicates that the ~ donation effect is dominating the charge potential or electronegativity effect. Thus, the poorer ~ donor ability of Br, compared to Gl results in less X(p~)~M(d~) destabilization of the metal-metal ~ and 6 bonding orbitals and consequently higher ionization energies. 340 Figure 9 shows the effect of replacing Gl by Br and I on the PE spectrum of the Re 2 Gl 4 (PMe 3 ) 4 molecule. From the Figure (see also Table V), we can see that both the 6 and 6* bands move to higher binding energy on going from Cl to Br to I. 340 This is the same trend as observed for the molybdenum dimers and again illustrates the importance of ~ interactions between the halides and the metal atoms. The ~ ionization is complicated by splitting due to spin-orbit coupling, 34 but there is a small shift to lower binding energy, as observe.d for the M-P(u) band, so the charge effect appears to be winning out over the ~ donation effect for the ~ ionization of the rhenium dimers.

9.3

Figure 8.

8.3

7.3

6.3

Low binding energy PE spectra of Mo 2 X4 (PMe 3 ) 4

7.1

Figure 9.

6.6

5.6

The 6 and 6* ionizations of Re 2X4 (PMe 3 ) 4 285

·o~d·

11(e)

Figure 10 .

Symmetry interactions of 6 and 11 orbitals with halogen p11 orbitals.

For both molybdenum and rhenium, the change in the ionization energies of the 6 (and 6*) bands are greater than for the 11 band. This can be understood in terms of the 6 and 6* MO's possessing more ligand character 3 and, therefore being more susceptible to a change in the 11-donor ability of the halide. As shown in Figure 10, the 6 and 6* MO's have contributions from all four 11-donor (X) ligands, while each component of the 11 set has contributions from only two X ligands. The extent of interaction of the metal levels with the 11-symmetry orbitals of bridging carboxylates remains to be demonstrated . Evidence for this interaction comes from X-ray photoelectron spectroscopy (XPS) 35 studies on M2 (0 2CCF 3 ) 4 (M- Mo, W) . 36 Figure 11 shows the carbon ls region of these dimers . The shoulder on the low ionization side of the methyl band (the most intense feature in both spectra) is attributed to a satellite peak from the carboxylate carbon primary ionization which arises from a shake-up process . This shake-up is primarily associated with a metal-to-ligand charge transfer (MLCT) process involving electron transfer from the valence 6-bonding orbital to a virtual 11* level of the carboxy late. The same process has been observed in some other Mo 2 (0 2 CR) 4 dimers . 37 The MLCT is greater for tungsten than molybdenum, as evidenced by the greater intensity of the satellite peak on the low ionization energy side of the methyl band and also the appearance of another shake-up peak on the high ionization energy side of the methyl band. 36 Summarizing the data relating to the vibrational fine structure observed for the 6 ionization in multiply bonded systems . The hot band argument may explain some of the apparent lengthening but is probably less important than contributions from metal-ligand vibrations . The solid state structural data reveals that the magnitudes of the M-X bond length changes are equal to or greater than those of the M-M bond. The ultraviolet PES data on the M2X4 (PR 3 ) 4 dimers show that 11-donation from the halide ligands is important in determining the binding energy of the 6 MO . Finally, the XPS data indicates that the 6 band of the tetracarboxylate dimers is involved in metal-to-ligand charge transfer into 11* orbitals of the bridging ligands. 286

lonizalion Energy (eV)

302

Figure 11.

2 8

294

290

X-ray PES of the carbon ls core region of Mo 2 (0 2 CCF 3 ) 4 and W2 (0 2 CCF 3 ) 4

THE o IONIZATION Turning once more to the PE spectra of the quadruple bonded acetate dimers of Cr, Mo and W (Figure 2), we see that for chromium the metal 3d ionizations form a single broad band, while for molybdenum there are two distinct bands, which were assigned as 6 and w (in order of increasing binding energy) by Green et a1 .. 11• The question which arises, therefore, is "where is the M-M o band?" In the tungsten spectrum, three low binding energy bands can be distinguished. Could the sharp band with the highest binding energy (which would be consistent with the MO calculati·ons) 3 be the o band? 36 In a previous PES study of W2 (0 2CCF 3 ) 4 , a similar sharp

ionization (with full width at half maximum (fwhm) of 0.3 eV) was assigned to the o ionization. 38 This sharp ionization feature at higher binding

energy than the w band is present in all W2 (0 2 CR) 4 complexes so far studied. 36 The sharpness of this band is, however, at odds with traditional thoughts as to the strong bonding nature of the o orbital, since this would be expected to be accompanied by the excitation of metal-metal stretch vibrations and hence, to give rise to a broad ionization.

Since three peaks are observed for tungsten and only two for molybdenum, it was thought that the second peak in the molybdenum spectrum may in fact contain two overlapping bands. 36 To test this hypothesis, the PE spectrum of MoW(0 2 CC(CH 3 ) 3 ) 4 was taken and compared with those of the Mo 2 and W2 analogs. 36 The results are shown in Figure 12. The Figure clearly shows that on going from the W2 dimer to MoW the sharp ionization band starts to merge with the w band and that for the Mo 2 dimer only one band is observed. By extrapolation, we can conclude that the distinct o and w ionizations observed in W2 are completely coalesced in Mo 2 • In an attempt to establish whether or not the second band in the molybdenum spectra is a superposition of two ionization features, PES data was collected for a number of Mo 2 (02CR) 4 species. By varying the R group 287

Ionization Energy (eV)

8

Figure 12.

7

6

5

He(I) PE spectra of M2 (0 2 CGMe 3 ) 4

it was hoped that the a and ~ bands would separate due to their presumed different response to inductive effects. Unfortunately, no such separation was ever observed. 36 One piece of supporting evidence for the assertion that the a ionization lies under the ~ ( 2 Eu) band in the molybdenum case is the fine structure in the " ionization band.

The " ionization band shows vibra-

tional fine structure on the high binding edge of the band but not on the low binding edge as shown in Figure 13. 36 There is also a slight bump on this low binding energy edge which suggests the presence of the a band.

Ionization Energy (eV)

94

....Ul c:

;:)

0

u

288

9.0

8.6

8.2

Triple-bonded metal-alkoxides. At this point, in order to understand the position of the a ionization in molybdenum acetate and its analogs, it is informative to consider the PES of a closely related family of dimers. The M2X6 dimers of molybdenum and tungsten (X - OR, NR2 , R) adopt the staggered (D 3 d) conformation shown in III. These molecules possess a triple bond (a 2w4 ). The PE spectrum of Mo 2 (0CH 2CMe 3 ) 6 enabled Green et al.

III to assign both the a and the w ionizations. 39 Kober and Lichtenberger subsequently compared the PE spectra of four M2 (0R) 6 dimers (M - Mo, R CHMe 2 , CMe 3 , CH 2CMe 3 ; M- W, R- CMe 3 ). The spectra of the molybdenum and tungsten complexes with R- CMe 3 are reproduced in Figure 14. 40 In agreement with the work of Green et al. , 39 they found that two peaks can be distinguished in the metal region (5.5-8.5 eV). The relative areas of the peaks are approximately 2:1 (lower binding energy band first, see Table VI), which is consistent with their assignment as wand a respectively. Theoretical calculations of various levels of sophistication agree with the assignment of the first two bands as w and a. 3 • 39 • 41

Mo=Mo

WaW

80

Figure 14.

7.0

6.0

PE spectra of M2 (0CMe 3 ) 6 .

The data in Table VI show that the a ionization ( 2A8 ) is much narrower (fwhm = 0.4 eV) than thew ionization ( 2 Eu) (fwhm 0.7 eV). Another notable feature is that the triple bonded tungsten compound has a considerably larger a-w separation (1.52 eV) than the corresponding molybdenum compound (1.01 eV). In addition, the a-w separation in the tungsten alkoxide dimer is more than twice that measured for W2 (0 2 CCMe 3 ) 4 (0.75 eV). 36 The fact that the a-w separation is greater for tungsten dimers than their molybdenum analogues, combined with the much smaller a-w separation in the quadruple bonded tungsten dimers, clearly makes the coalescence of the a and w bands for the quadruple bonded molybdenum compounds a reasonable proposition.

=

289

Table VI.

Peak positions, widths and relative areas of the metal ionizations of some M2 (0R) 6 dimers.

Dimer Mo 2 (OCHMe 2 ) 6 Mo 2 ( OCH 2 CMe 3 ) 6 Mo 2 (0CMe 3 ) 6 W2 (0CMe 3 ) 6

Peak Position Band 1 (7r) 7.33 7.46 6.79 6.27

[0.69] [0.69] [0.74] [0.82]

[fwhm] (eV) Band 2 (a) 7.99 8.08 7.80 7.79

[0.35] [0.36] [0.43] [0.40]

Area(Band 1)/Area(Band 2) 2.4 2.1 2.0 2.0

Thin film surface studies of metal-metal bonds. In theory, gas phase PES of molecules of the type Mo 2 (0 2 CR) 4L2 , 2 with axially bound 2-electron donor ligands (e.g. H2 0, PR' 3 ) would be of great interest, since the donor orbitals will overlap with the metal a-bonding MO and may destabilize it sufficiently for it to be distinguishable from the 1r band. Unfortunately, it is found that these molecules lose their axial ligands before subliming. In the solid state, however, it is known that the quadruple bonded acetate (etc.) dimers associate, with long intermolecular M--0 contacts at the vacant axial positions (vide infra), which make these systems worth studying. It has been found for Cr 2 (0 2 CCH 3 ) 4 that there is a significant difference in Cr-Cr bond length between the gas phase (1.966(14) A) 42 and the solid state (2.288(2)A). 43 The lengthening of the Cr-Cr bond in the solid state is due to quite strong axial interaction between the chromium atoms and the oxygens on neighboring dimers, which are only 2.33 A away, as illustrated below. 43 This may be compared with 2.00 A for a common

covalent Cr-0 bond. In the case of Mo 2 (0 2CCH 3 ) 4 the Mo-Mo bond lengthens only from 2.079 A (gas phase) 15b to 2.093 A (solid state) 15 ". The difference is smaller for molybdenum than chromium, which is consistent with the greater sensitivity of the Cr-Cr bond length to the presence and nature of axially coordinated ligands. 28 • 44 The observed perturbation of metal-metal bonding in the solid state suggested the interesting possibility of looking at the solid state PE spectrum of the molybdenum dimers to determine whether the intermolecular Mo--O interactions are sufficiently large to destabilize the a MO, so that its ionization band no longer overlaps the 1r band. To study intermolecular influences on the M-M bonding in quadruple bonded dimers, thin (~ 100 A thick) films of the M2 (0 2 CCH 3 ) 4 dimers (M- Cr, Mo, W) were deposited in ultrahigh vacuum on gold foil. 45 Even at 100 A, these films are thick enough to regard as bulk material. 290

Ul

12

8

C1 11'

Figure 15.

6

Thin film and gas phase PE spectra of M2 (0 2 CCH 3 ) 4 •

The ultraviolet (He I) PE spectra of the acetate dimer films are presented in Figure 15, together with the gas phase PE spectra obtained for the same compounds. 4 ~ For the chromium dimer, the film shows a shift of the broad metal band (which presumably contains the a, ~ and 6 ionizations) to lower binding energy relative to the gas phase spectrum (keeping the high binding energy acetate bands constant) . This is due to the previously mentioned lengthening of the Cr-Cr bond in the solid state, which arises from the intermolecular Cr--0 interaction and reduces a, ~ and 6 bonding overlap. For tungsten, there is a significant shift of the presumed a band to lower binding energy, relative to the ~ and 6 ionizations, so that it coalesces with the~ band. This large shift in the a band is again due to the intermolecular M--0 interaction discussed above . A close-up of the metal ionization region (6-10 eV) of the molybdenum acetate spectrum is shown in Figure 16, together with the three asymmetric Gaussian peaks which were used to simulate the spectrum. 4 ~ There is a pronounced shoulder at the low binding edge of the 11' band for the film, whereas there was only the slightest hint of a shoulder in the gas phase (see Figure 13) . 36 This is again consistent with a shift of the a band to lower binding energy relative to the ~ and 6 bands and is once more due to intermolecular axial Mo--O antibonding overlap. 291

8

10

Figure 16.

Triple bonded

M 21Q 2~4 R' 2

7

e

Close-up of thin film PE spectrum of Mo 2 (0 2CCH 3 ) 4 .

complexes,

Although it has not proven possible to take the gas phase PES of any axially ligated M2 (02 CR) 4L2 molecules, where L is a 2-electron-donating ligand, the spectra of the closely related molecules Mo 2 (0 2CR) 4 (np) 2 (np neopentyl; R- H, CH 3 , CF3 ) have been obtained. 46 Dimers of the type M2 (0 2 CR) 4R' 2 (M- Mo, W) are M(III) (d3 ) complexes . They are related to the M(II) (d4 ) M2 (0 2CR) 4 dimers by the formal oxidative addition of R' 2 . 46

2R

Theoretical calculations on a model tungsten complex46 indicate that the singly occupied frontier orbitals of the R' radicals interact with the filled u(d) and empty u"(d) orbitals of the tetracarboxylate dimer , · generating two occupied u~ MO's which are mostly localized on R' and two empty u~· MO's which have mostly metal character. Thus the oxidative addition effectively involves the transfer of two electrons from the metal u orbital to the R' ligands, thereby generating a triply bonded species with the unusual electron configuration ~ 4 5 2 . Actually, the electron transfer is far from complete, since both of the u~ orbitals have more than 20% metal character (based on the calculation forM- W). 46 An interesting feature of this oxidative addition is that the metal-metal distance is essentially unchanged from that of the parent quadruple bonded complex. The PE spectra of W2 (02 CCH3 ) 4 (np) 2 and W2 (0 2CCH 3 ) 4 are compared in Figure 17. 46 In agreement with the qualitative MO arguments presented above and the SCF-Xa-SW calculations, the metal u ionization disappears on going to the axially ligated compound. A shoulder at high binding energy is assigned to one of the u~ ionizations, while the second is merged with the ~band. The o band moves to higher binding energy, reflecting an increase in the positive charge of the tungsten atoms which have been formally oxidized. In the analogous molybdenum system, as shown in Figure 18, the broad ("u+~") band of the acetate dimer becomes narrower upon oxidative addition, as the u component is removed, and new features (uM0 c) appear on either side of it. 292

CJ+"II'

\

V

~

11'·ac ,

'

\_A

9.0

8.0

7.0

7.0

Ionization Enern, eV

Ionization Enern, eV

Figure 17. PE spectra of W2 (0 2 CCH 3 ) 4 (np) 2 (top) and W2 (0 2 CCH 3 ) 4 (bottom).

8.0

9.0

Figure 18. PE spectra of Mo 2 (0 2CCH 3 ) 4 (top) and Mo 2 (0 2CCH 3 ) 4 (np) 2 (bottom).

THE VALENCE u NON-BOND? The assignment of the u ionization is now supported by a number of experimental observables. However, the basic problem still remains that the characteristics of this ionization are at odds with the supposed strong metal-metal interaction in these quadruple bonded metal-metal dimers. First is the problem of the close energy proximity of the u ionization to the 1r ionization. In the Mo 2 (02 CCH 3 ) 4 case, the u ionization is most likely at lower ionization energy than the 1r ionization, which is not consistent with stronger u bonds than 1r bonds. The second problem is the narrow bandwidth of the u ionization, which is characteristic of ionization of non-bonding electrons. The non-bonding nature of this ionization is supported by the oxidative addition studies cited above, where the bonding of R' groups in the axial positions results in a 1r 4&2 configuration but does not significantly change the metal-metal distance. Some insight into the factors contributing to the low IP of the u ionization is provided by comparison of the ionizations of the quadruple bonded carboxylates with the triple bond alkoxide complexes. The u ionization is destabilized relative to the 1r ionization in the quadruple bonded M2 (0~CR) 4 complexes in comparison to the triple bonded M2 (0R) 6 complexes. 3 b, 36 • 41 Fenske-Hall calculations indicate that destabilizing influence of the ligands on the u MO is smaller for the triple bonded alkoxide complexes, because the alkoxide ligands are bent away from the xy plane (perpendicular to the M-M bond) towards the node in the atomic d.z

~

I

0

0

~ orbital, thereby reducing the overlap. 41 In addition, there are four oxygen atoms on each metal interacting with the d.z in the case of the carboxylates, and only three in the case of alkoxides. The calculations also indicate that the M-M 1r bonding MO's are destabilized more in M2 (0R) 6 by M-0 u and 11'-type interactions. 293

Table VII.

Metal-metal valence d and outer core s and p overlap integrals.

(a) Mo 1 W'z

0.116 0.138

0.117 0.134

0.111 0.116

C,.. > <6 >

<sis>

---------------·---·--- ---

0.155 0.181

0.046 0.051

0.028 0.028

0.005 0.004

These ligand interactions are not able to account for the narrow bandwidth of the a ionization. One important factor that may account for the narrow bandwidth, as well as the low IP, arises from the close contact between the metal atoms in the triple and quadruple bonded complexes. This close contact leads to substantial overlap between the valence nd 2 z orbital on one atom and the outer core npz (and ns) orbital of its neighbor. Jn,p,J~b.~0.4l The overlap integralsu between the valence d orbitals and outer core orbitals are listed in Table VII.~ 1 From the Table, it is apparent that the overlap between dzz and the outer core Pz is roughly of the same magnitude as the d.z-d 2 z overlap, with the d 2 z-s overlap being somewhat smaller. Also the d.z-d 2 2 (a) overlap is smaller than the d,..-d,.. (11") overlap. Ziegler's LCAO-HFS calculations agree that a-ionization should lead to no change or a slight decrease (S 0.01 A) in the M-M bond length. 3n,p The small a overlap coupled with the repulsive interaction with the outer core Pz results in the a MO's energy being close to that of the 'If's. These LCAO-HFS calculations indicate that with a-ionization the valence(d.z)-outer core(pz) repulsion is reduced as well as the formal bond order (d.z-d.z attraction). These interactions, which have opposite influences on the M-M bond length, cancel out approximately in these calculations. Ziegler's calculations also indicate a small increase in the bond length (= 0.04 eV) upon 6 ionization and a much larger increase (0.15 A) on 11" ionization. 3P A slightly different explanation for the role of the outer core np. orbitals is offered by the results of ab initio calculations 48 and FenskeHall calculations.~ 1 These calculations indicate that the symmetric combination of the outer core np. orbitals mixes with the valence da orbital as shown in Figure 19. The result is the same as for the first valence ionization of N1 , where the mixing of the pa combination with the 2sa combination leads to a valence non-bonding a orbital that ionizes at lower energy than the P'~~" bonding orbital. Interestingly, the binding energy of the N2 2sa combination is almost the same as the Mo 4pz level at 38 eV. We are currently carrying out further investigations into the nature of these bonding interactions by a variety of spectroscopic techniques. Photoelectron spectroscopy will continue to play an important role in revealing the electronic structure and bonding in metal chemistry. This report has shown that, in addition to providing unique experimental information, the photoelectron observations also often provide special challenges to our fundamental models of chemical bonding.

ACKNOWLEDGMENTS D.L.L. acknowledges support by the U.S. Department of Energy (Division of Chemical Sciences, Office of Basic Energy Sciences, Office of Energy Research, DE-AC02-80ER10746), the National Science Foundation (CHE8519560), and the Materials Characterization Program, Department of 294

7

8

a

eV 9

~1 Figure 19.

4p-'

I

I

I

I

I

I

I

7r

\

\

\

\

\

\

\

\

4p

Valence d ionization energies of Mo 2 (0 2 CCH 3 ) 4 and interaction with the outer core 4p levels.

Chemistry, University of Arizona. R.L.J. acknowledges the Science and Engineering Research Council (U.K.) for the award of a N.A.T.O. postdoctoral fellowship. D.L.L. would also like to acknowledge his co-workers C. H. Blevins, II, G. D. Hinch, J. G. Kristofski and E. M. Kober, and especially the fruitful collaborations and discussions with F. A. Cotton, M. H. Chisholm, H. B. Gray, A. P. Sattelberger, R. A. Walton and their coworkers. Thanks are also due to many members of the D.L.L. research group and to D. K. Myers for typing the manuscript.

REFERENCES 1.

2.

3.

(a) Cotton, F.A.; Curtis, N.F.; Harris, C.B.; Johnson, B.F.G.; Lippard, S.J.; Mague, J.T.; Robinson, W.R.; Wood, J.S. Science 1964, 145, 1305. (b) Cotton, F.A.; Harris, C.B. Inorg. Chern. 1965, 4, 330. (c) Cotton, F.A. Inorg. Chern. 1965, 4, 334. (a) Cotton, F.A.; Walton, R.A. "Multiple Bonds Between Metal Atoms" Wiley: New York, 1982, and references therein. (b) Cotton, F.A. Chern. Soc. Rev. 1975, 4, 27. (c) Cotton, F.A. Chern. Soc. Rev. 1983, 12, 35. (d) Cotton, F.A.; Walton, R.A. Struct. Bond. 1985, 62, 1. (a) Norman, J.G., Jr.; Kolari, H.J. J. Chern. Soc., Chern. Commun. 1974, 303. 295

4. 5.

6.

7. 8.

9. 10.

11.

12.

296

(b) Norman, J.G., Jr.; Kolari, H.J. J. Am. Chem. Soc. 1975, 97, 33. (c) Norman, J.G.; Kolari, H.J. J. Chem. Soc., Chem. Commun. 1975, 649. (d) Norman, J.G., Jr.; Kolari, H.J.; Gray, H.B.; Trogler, W.C.; Inorg. Chem. 1977, 16, 987. (e) Cotton, F.A.; Kalbacher, B.J. Inorg. Chem. 1977, 16, 2368. (f) Block, T.F.; Fenske, R.F.; Lichtenberger, D.L.; Cotton, F.A. J. Coord. Chem. 1978, 8, 109. (g) Hillier, I.H.; Garner, C.D.; Mitcheson, G.R.; Guest, M.F. J. Chem. Soc., Chem. Commun. 1978, 204. (h) Guest, M.F.; Garner, C.D.; Hillier, I.H.; Walton, I.B. J. Chem. Soc., Faraday Trans. 2 1978, 11, 2092. (i) Benard, M. J. Am. Chem. Soc. 1978, 100, 2354. (j) Bursten, B.E.; Cotton, F.A.; Hall, M.B. J. Am. Chem. Soc. 1980, 102, 6348. (k) Atha, P.M.; Hillier, I.H.; Guest, M.F. Chem. Phys. Lett. 1980, 75' 84. (1) Goodgame, M.M.; Goddard, W.A. J. Phys. Chem. 1981, 85, 215. (m) Arriata-Perez, R.; Case, D.A. Inorg. Chem. 1984, 23, 3271. (n) Ziegler, T. J. Am. Chem. Soc. 1985, 107, 4453. (o) Hall, M.B. Polyhedron 1987, 6, 679. (p) Ziegler, T.; Tschinke, V.; Becke, A. Polyhedron 1987, 6, 685. (q) Bursten, B.E.; Clark, D.L. Polyhedron 1987, 6, 695. (r) Davy, R.D.; Hall, M.B. J. Am. Chem. Soc. 1989, 111, 1268. (a) Atha, P.M.; Hillier, I.H. Hol. Phys. 1982, 45, 285. (b) Cotton, F.A.; Shim, I. J. Phys. Chem. 1985, 89, 952. (a) Cr 2 : Michalopolous, D.L.; Geusic, M.E.; Hansen, S.G.; Powers, D. E.; Smalley, R.E. J. Phys. Chem. 1982, 86, 3914. (b) Atha, P. M.,; Hillier, I. H. Hol. Phys. 1982, 46, 437. (c) Mo 2 : Efremov, Y.M.; Samoilova, A.N.; Kozhukhovsky, V.B.; Gurvich, L.V. J. Hol. Spectrosc. 1978, 73, 430. (a) Turner, D.W.; Baker, C.; Baker, A.D.; Brundle, C.R. "Molecular Photoelectron Spectroscopy" Wiley: London, 1970. (b) Rabelais, J.W. "Principles of Ultraviolet Photoelectron Spectroscopy" Wiley: New York, 1977. (c) Ghosh, P.K. "Introduction to Photoelectron Spectroscopy" Wiley: New York, 1983. (d) Lichtenberger, D.L.; Kellogg, G.E. Ace. Chem. Res. 1987, 20, 379. Koopmans, T. Physica (Amsterdam) 1934, 1, 104. Bohm, M.G. J. Chem. Phys. 1983, 78, 7044. Hubbard, J.L.; Lichtenberger, D.L J. Am. Chem. Soc. 1982, 104, 2132. (a) Lichtenberger, D.L.; Kellogg, G. E. in "Gas Phase Inorganic Chemistry" (Ed. Russell, D.H.), Plenum: New York, 1989, Chapter 8, pp. 245-277. (b) Lichtenberger, D.L.; Darsey, G.P.; Kellogg, G.E.; Sanner, R.D.; Young, V.G., Jr.; Clark, J.R. J. Am. Chem. Soc. in press. (a) Green, J.C.; Hayes, A.J. Chem. Phys. Lett. 1975, 31, 306. (b) Coleman, A. W.; Green, J. G.; Hayes, A. J.; Seddon, E. A.; Lloyd, D. R.; Niwa, Y. J. Chem. Soc., Dalton Trans., 1979, 1057. (c) Lichtenberger, D.L.; Blevins, C.H., II J. Am. Chem. Soc. 1984, 106, 1636. (a) Herzberg, G. "Molecular Spectra and Molecular Structure" Van Nostrand Reinhold: New York, 1950; Vol.l, pp. 101-103. (b) Eland, J.H.D.; Danby, C.J. Int. J. Hass. Spectrom. Ion Phys. 1968, 1, 111. (c) Yersin, H.; Otto, H.; Zink, J.I.; Gliemann, G. J. Am. Chem. Soc. 1980, 102, 951. (d) Hipps, K.W.; Merrell, G.A.; Crosby, G.A. J. Phys. Chem. 1976, 80, 2232. (e) Mazur, U.; Hipps, K.W. J. Phys. Chem. 1980, 84, 194. (f) Fordyce, W.A.; Brummer, J.G.; Crosby, G.A. J. Am. Chem. Soc. 1981, 103, 7061.

(g) Tutt, L.; Tanner, D.; Heller, E.J.; Zink, J.l. Inorg. Chem. 1982, 21, 3859. 13. Totally symmetric vibrational modes normally dominate the vibrational progressions seen in PE spectra: (a) Calabro, D.C.; Hubbard, J.L.; Blevins, C.H., II; Campbell, A.C.; Lichtenberger, D.L. J. Am. Chem. Soc. 1981, 103, 6839. (b) Stevens, A. E.; Feigerle, C.S.; Lineberger, W.C. J. Am. Chem. Soc. 1982, 104, 5026. (c) Lee, S.-Y. J. Chem. Educ. 1985, 62, 561. 14. (a) Bratton, W.K.; Cotton, F.A.; Debeau, M.; Walton, R.A. J. Coord. Chem. 1971, 1, 121. (b) Cowman, C.D.; Gray, H.B. J. Am. Chem. Soc. 1973, 95, 8177. (c) Mortola, A.P.; Moskowitz, J.W.; Rosch, N.; Cowman, C.D.; Gray, H.B. Chem. Phys. Lett. 1975, 32, 283. (d) Cowman, C.D.; Trogler, W.C.; Gray, H.B. Isr. J. Chem. 1977, 15, 308. (e) Trogler, W.C.; Solomon, E.l.; Trajberg, I.B.; Ballhausen, C.J.; Gray, H.B. Inorg. Chem. 1977, 16, 828. (f) Fanwick, P.E.; Martin, D.S.; Cotton, F.A.; Webb, T.R. Inorg. Chem. 1977, 16, 2103. (g) Hutchinson, B.; Morgan, J.; Cooper, C.B.; Mathey, Y.; Schriver, D.F. Inorg. Chem. 1979, 18, 2048. 15. (a) Solid State: Cotton, F.A.; Mester, Z.C; Webb, T.R. Acta Cryst. 1974, B30, 2768. (b) Gas Phase: Kelley, M.H.; Fink, M. J. Chem. Phys. 1982, 76, 1407. 16. Manning, M.C.; Trogler, W.C. Inorg. Chem. 1982, 21, 2797. 17. Martin, D.S.; Newman, R.A.; Fanwick, P.E. Inorg. Chem. 1979, 18, 2511. 18. (a) Angell, C.L.; Cotton, F.A.; Frenz, B.A.; Webb, T.R. J. Chem. Soc., Chem. Commun. 1973, 399. (b) Cotton, F.A.; Frenz, B.A.; Pederson, E.; Webb, T.R. Inorg. Chem. 1975, 14, 391. 19. Cotton, F.A.; Feng, X.; Matusz, M. Inorg. Chem. 1989, 28, 594. 20. Cotton, F.A.; Frenz, B.A.; Webb, T.R. J. Am. Chem. Soc. 1973, 95, 4431. 21. Bino, A.; Cotton, F.A. Inorg. Chem. 1979, 18, 3562. 22. Lichtenberger, D.L.; Copenhaver, A.S., J. Chem. Phys., in press. 23. Miskowski, V.M.; Brinza, D.E. J. Am. Chem. Soc. 1986, 108, 8296. 24. Fordyce, W.A.; Brummer, J.G.; Crosby, G.A. J. Phys. Chem. 1981, 103, 7061. 25. (a) Heller, E.J. J. Chem. Phys. 1975, 62, 1544. (b) Heller, E.J. J. Chem. Phys. 1978, 68, 3891. (c) Kulander. K.C.; Heller, E.J. J. Chem. Phys. 1978, 69, 2439. (d) Heller, E.J. Ace. Chem. Res. 1981, 14, 368. (e) Tutt, L.; Tanner, D.; Heller, E.J.; Zink, J.I. Inorg. Chem. 1982, 21, 3859. 26. (a) Ketteringham, A.P.; Oldham, C. J. Chem. Soc., Dalton Trans. 1973, 1067. (b) Ketteringham, A.P.; Oldham, C.; Peacock, C.J. J. Chem. Soc., Dalton Trans. 1976, 1640. 27. Counter ion- Y3 +: Cotton, F.A.; Davison, A.; Day, V.W.; Friedrich, M.F.; Orvig, C.; Swanson, R. Inorg. Chem. 1982, 21, 1211. 28. (a) Counter ion- [NH 4 ]+: Cotton, F.A.; Shive, L.S. Inorg. Chem., 1975, 14, 2032. (b) Counter ion- K+: Cotton, F.A.; Pedersen, E. Inorg. Chem., 1975, 14, 383. 29. Counter ion- [(n-C 4H9 ) 4N]+: Cotton, F.A.; Daniels, L.; Davison, A.; Orvig, C. Inorg. Chem. 1981, 20, 3051. 30. Cotton, F.A.; Dunbar, K.R.; Falvello, L.R.; Tomas, M.; Walton, R.A. J. Am. Chem. Soc. 1983, 105, 4950. 31. Cotton, F.A.; Fanwick, P.E.; Fitch, J.W.; Glicksman, H.D.; Walton, R.A. J. Am. Chem. Soc. 1979, 101, 1752. 32. (a) Fanwick, P.E.; King, M.K.; Tetrick, S.M.; Walton, R.A. J. Am. Chem. Soc. 1985, 107, 5009. 297

33. 34.

35.

36. 37.

38.

39.

40. 41. 42. 43. 44.

45.

46.

47. 48.

298

(b) Agaskar, P.A.; Cotton, F.A.; Dunbar, K.R.; Falvello, L.R.; Tetrick, S.M.; Walton, R.A. J. Am. Chem. Soc. 1986, 108, 4850. (a) Ebner, J.R.; Walton, R.A. Inorg. Chem. 1975, 14, 1987. (b) Cotton, F.A.; Frenz, B.A.; Ebner, J.R.; Walton, R.A. J. Am. Chem. Soc. 1976, 15, 1630. (a) Cotton, F.A.; Hubbard, J.L.; Lichtenberger, D.L.; Shim, I. J. Am. Chem. Soc. 1982, 104, 679. (b) Root, D.R.; Blevins, C.H., II; Lichtenberger, D.L.; Sattelberger, A.P.; Walton, R.A. J. Am. Chem. Soc. 1986, 108, 953. (c) Hinch, G.D., Ph.D. Dissertation, University of Arizona, 1989. (a) Siegbahn, K.; Nordling, C.; Johansson, G.; Hedman, J.; Heden, P.F.; Hamrin, K.; Gelius, U.; Bergmark, T.; Werme, L.O.; Manne, R.; Baer, Y. "ESCA Applied to Free Molecules" North-Holland: Amsterdam, 1969. (b) Briggs, D. "Handbook of X-ray and Ultraviolet Photoelectron Spectroscopy" Heyden: London, 1977. (c) Jolly, W.L. Ace. Chem. Res. 1983, 16, 370. Blevins, C.H. II, Ph.D. Dissertation, University of Arizona, 1984. (a) Atha, P.M.; Ford, P.C.; Garner, C.D.; MacDowell, A.A.; Hillier, I.H.; Guest, M.F.; Saunders, V.R. Chem. Phys. Lett. 1981, 84, 172. (b) Atha, P.M.; Berry, M.; Garner, C.D.; Hillier, I.H.; MacDowell, A.A. J. Chem. Soc., Chem. Commun. 1981, 1027. (c) Atha, P.M.; Hillier, I.H.; MacDowell, A.A.; Guest, M.F. J. Chem. Phys. 1982, 77, 195. (d) Atha, P.M.; Campbell, J.C.; Garner, C.D.; Hillier, I.H.; MacDowell, A.A. J. Chem. Soc., Dalton Trans. 1983, 1085. (a) Bancroft, G.M.; Pellach, E.; Sattelberger, A.P.; McLaughlin, K.W. J. Chem. Soc., Chem. Commun. 1982, 752. (b) Sattelberger, A.P. in "Inorganic Chemistry: Towards the 21st Century"; Chisholm, M.H., Ed.; American Chemical Society: Washington DC, 1983, p. 291. (a) Cotton, F.A.; Stanley, G.G.; Kalbacher, B.J.; Green, J.C.; Seddon, E.; Chisholm, M.H. Proc. Natl. Acad. Sci. U.S.A. 1977' 74' 3109. (b) Bursten, B.E.; Cotton, F.A.; Green, J.C.; Seddon, E.A.; Stanley, G.G. J. Am. Chem. Soc. 1980, 102, 4579. Kober, E.M.; Lichtenberger, D.L. J. Am. Chem. Soc. 1985, 107, 7199. Kober, E.M.; Lichtenberger, D.L. unpublished work. Ketkar, S.N.; Fink, M. J. Am. Chem. Soc. 1985, 107, 338. Cotton, F.A.; DeBoer, B.G.; LaPrade, M.D.; Pipal, J .R.; Ucko, D.A. Acta Cryst. B 1971, 27, 1644. (a) Kristofzski, J.G., Ph.D. Dissertation, University of Arizona, 1988. (b) Cotton, F.A.; Extine, M.W.; Rice, G.W. Inorg. Chem. 1978, 17, 176. (c) Cotton, F.A.; Ilsley, W.H.; Kaim, W. J. Am. Chem. Soc. 1980, 102, 3464. (d) Wiest, R.; Benard, M. Chem. Phys. Lett. 1983, 98, 102. (a) For experimental details regarding film deposition, thickness calibration and spectral measurement see Reference 44a. (b) Lichtenberger, D.L.; Kristofski, J.G. J. Am. Chem. Soc. 1987, 109. 3458. (a) Chisholm, M.H.; Clark, D.L.; Huffman, J.C.; VanDer Sluys, W.G.; Kober, E.M.; Lichtenberger, D.L.; Bursten, B.E. J. Am. Chem. Soc. 1987, 109, 6796. (b) Braydich, M. D.; Bursten, B. E.; Chisholm, M. H.; Clark, D. L. J. Am. Chem. Soc. 1985, 107, 4459. Hall, M.B.; Fenske, R.F. Inorg. Chem. 1972, 11, 768. (a) Bernholz, J.; Holzwarth, N. A. W. Phys. Rev. Lett., 1983, SO, 1451. (b) Sundholm, D.; Pyykk6, P.; Laaksonen, L. Finn. Chem. Lett. 1985, 51.

FORMATION,

STRUCTURE AND

LUMINESCENT PROPERTIES

OF METAL-META!.

BONDED COMPOUNDS OF TBE LATE TRANSITION METAL AND POST TRANSITION METAL

IONS

Alan L.

Balch

Department of Chemistry University of California Davis, CA 95616 ABSTRACT

Metal-metal bond forming reactions involving interactions of planar dB metal ions (Ir(I), Rh(I), Pt(II)) with a variety of low coordinate transition metal and post transition metal ions are described. Many of the examples involve metal complexation by the metallomacrocycles M2(C0)2Cl2{~-(Ph2PCH2)2ASPh}2 (where M is Rh or Ir). INTRODUCTION

Metallomacrocyclesl, macrocycles containing metal ions as an intrinsic part of their cyclic structure, offer an unusual environment for the formation of metal-metal bonds with conventional and nonconventional bonding. In this article, the formation of metal-metal bonds within the cavities of the metallomacrocycles Ir2(C0)2Cl2(~-dpma)2, 1,2 and Rh2(C0)2Cl2(~-dpma)2, 2, (dpma is bis(diphenylphosphinomethyl)phenylarsine) are considered. These complexes are also compared to related trinuclear complexes obtained by other means.

1

M

= Ir

2 M = Rh

The metallomacrocycles 1 and 2 are readily prepared in high yield and isomeric purity by the reaction of dpma with either Ir(C0)2Cl(ptoluidine) or Rh2(~-Cl)2(C0)4.2 This reaction makes use of the fact that, almost without exception, metal ions preferentially coordinate to phosphorus rather than arsenic. The structure of the rhodium metallomacrocycle 2 is shown in Figure 1. It possesses two nonMetal-Metal Bonds and Clusters in Chemistry and Catalysis Edited by J. P. Fackler, Jr. Plenum Press, New York, 1990

299

Figure 1

300

Two views, the lower stereoscopic, of the metallomacrocycle Rh2(C0)2Cl2(~-dpma)2, 2, as determined from X-ray diffraction.

interacting planar rhodium ions that are separated by 5.428(1) A and two uncoordinated arsenic atoms whose lone pairs are directed away from the center of the open cavity.3 Presumably the iridium analog is similar. A variety of metal ions, including Rh(I), Ir(I), Pd(II), Cu(I), Ag(I), Au(I), Au(III), Hg(II), In(I), Tl(I), Sn(II), Pb(II), Sb(III), and Bi(III), have been found to coordinate to the binding sites of 1 and 2. These sites include both the arsenic and iridium atoms within the macrocycle and also the chloride and carbon monoxide ligands that are bound to rhodium or iridium. Interestingly, at least nine different types of structures have been identified in the metal ion complexes formed from 1 and 2. This article will concern the case where metalmetal bonding occurs and will discuss certain of the features of the electronic spectra of these compounds. METAL-METAL

BOND

FORMATION

THROUGH OXIDATIVE-ADDITION

Metal halogen bonds can add to low valent transition metal centers to give new metal-metal bonds. For example, mercury(!!) chloride reacts with Ir(CO)Cl(PPh3l2 to form six-coordinate Ir(HgCl) (CO)Cl2(PPh3)2 with an Ir-Hg bond length of 2.570(1) A.4 Addition of mercury(!!) chloride to 1 results in the uptake of two moles of HgC12.5 The 31p NMR of the product reveals that it is unsymmetrical with 51 = 18.1 ppm, &2 = 16.6Hz, J(P, Hg) = 31.9 Hz. For the latter resonance, the satellites are consistent with oxidative addition of one mercury (II) chloride to one iridium ion within 1. The structure of the product, [Ir2(HgC1)2(C0)2Cl3(~-dpma)2]Cl, as determined by X-ray crystallography, is shown as A in Figure 2, where three different examples of structures formed by oxidative-addition reactions are shown. As indicated by the NMR work, one mercury (II) chloride molecule has undergone oxidative addition to iridium wnile the second has formed the Hgcl+ unit that is suspended between the two arsenic atoms, a chloride bridge to iridium and the other iridium center. The Ir(2)-Hg(l) distance (2.788(3) A) is somewhat longer than the Ir(l)-Hg(2) distance (2.573(3) A). Both appear to involve significant metal-metal bonding. Treatment of 2 with bis(benzonitrile)palladium (II) chloride yields the brown adduct [Rh2Pd(C0)2Cl3(~-dpma)2]+ whose structure is shown in part B of Figure 2.6 Again the cation is unsymmetrical with the palladium disposed so that it is bonded to Rh(2) with a normal Rh-Pd single bond (2.699(1) A) but bridged through chloride so that the Pd··Rh (1) distance is 0. 467 A longer. One rhodium is six-coordinate while the other retains the four-coordinate geometry found in the parent, 2. NMR studies of [Rh2Pd(C0)2Cl3(~-dpma)2]+ in dichloromethane solution reveal that it is fluxional. Two sharp doublets are seen in the 3lp NMR at -25° C, but at 45° C these coalesce into a single doublet. This implies that the two end environments of the cation become equivalent in the fluxional process. Consequently the Pd-Rh bonding is mobile and can be formed and broken so as to render the two rhodium ions equivalent as shown in Eq. 1.

p..........--...As

J

rrI/If

+

/"-A·---r I (\:

r

+

CI-P.d~lh-CI I ··--.. Rh" I ' 'c I /1 I cf

C'J

C l - ' \ . . - d - Cl"

0

Rh

~As-../

p

P_/As............_,.p

(1)

Cl

301

+ A

B

Figure 2

302

The structures of three cations with metal-metal bonds formed through oxidative addition of M-Cl bonds to a second metal. A. {Ir2(HgC1)2(C0)2(~-dpma)2} Ir(1)-Hg(2) 2.573(3) A, Ir(2)-Hg(1) 2.788(3) A. B. {Rh2Pd(C0)2Cl3(~-dpma)2}+ Pd-Rh(2), 2.699(1); Pd···Rh(1) 3.199(1) A. c. {Ir2AU(C0)2C14(~-dpma)2}+ Au-Ir(1) 2.812(2); Au-Ir(2), 2.806(2) A.

The two preceding examples show single-center two-fragment oxidative additions. Other examples of this sort of reactivity are known7 and it has been established that bifunctional phosphine ligands can facilitate such reactions, particularly for Pd-Cl additions to other transition metals.B The constructions of 1 and 2 are such, however, that oxidative addition of a single substrate to both metal ions should be possible and indeed the reactions above were examined with a view toward obtaining such an addition. This goal has been realized in the next example. The reaction of AuCl4- with 1 proceeds in a three-fragment two-centered fashion (Eq. 2).9

~ n+ n+

M

M +

\.____.)

X·Y·X

(2)

It yields the symmetrical cation shown in part C of Figure 2. At the center, a planar gold is bound to two arsenic atoms and to two iridium ions. ·Each iridium has undergone addition of an AuCl unit and is sixcoordinate. The two Ir-Au bond lengths are nearly equivalent and are in the range expected for single bonds between these metals. COMPLEXATION OF

POST

TRANSITION METAL

IONS

Although coordination of transition metal ions via the arsenic atoms of 1 and 2 virtually assured success in the formation of heterotrinuclear species involving these ions, the ability of 1 to bind post transition metal ions, especially Tl(I) and Pb(II), was particularly unexpected.10 Complex 1 is capable of binding non-transition metal ions in a chemical selective fashion. It recognizes and binds ions with an s2 electronic configuration (Tl(I), In(I), Sn(II), Pb(II), Sb(III) and Bi(III)) to give highly colored, frequently luminescent products. Ions of similar size from Groups 1 and 2 (e.g. Rb(I), Ba(II)) show no interaction with 1. Figure 3 shows some qualitative data regarding the interaction of 1 with antimony (III) fluoride. The spectrum is typical for a complex containing a Vaska type IrP 2 (CO)Cl unit. In 1these two units do not interact electronically to any significant extent. Trace A, the dotted .line, shows the spectrum of 1 in dichloromethane. Trace B, the solid line, shows the effect of shaking the sample with solid antimon5 (III) fluoride. A new, intense absorption at 540 nm (with E = l x 10 ) is readily apparent. The effect of the addition of indium (I) chloride on 1 is similar. A new absorption band is readily apparent at 610. Characterization of the species responsible for these intense colors (pink for antimony, green for indium) is in progress. However, analogous species in the related ions have been thoroughly studied.

303

•u t:

!

•0 J:J c 400

500

600

700

Wavelength (nm)

Figure 3

Absorption spectra of dichloromethane solutions of Ir2(C0)2Cl2(~-dpma)2, 1: A alone, B after shaking with solid antimony (III) fluoride,

The structures of two adducts of 1 with post transition metal ions are shown in Figure 4. The tin (II) adduct, [Ir2(SnCl) (C0)2Cl2(~­ dpma)2)+,11 is at the top, while the thallium complex, [Ir2Tl(C0)2Cl2(~­ dpma)2) [N03],10 is shown at the bottom. Both structures have idealized C2v symmetry with the snc1+ or Tl+ ions suspended at the center of the metallomacrocycle and connected to it exclusively through bonds to the iridium ions. In both cases the two Ir-M distances are similar. For the tin adduct they are about 0.2 A shorter than found for the thallium adduct (Ir-Sn 2.742(1) A, 2.741(1) A vs. Ir-Tl 2.958(1), 2.979(1) Al. In contrast the As···M separations (Sn···As, 3.06, 3.17 A; Tl···As, 3.295(3), 3.308(3) A) are much longer and longer than expected for As-Tl or As-Sn bonding. A molecular orbital diagram based on symmetry considerations for the Ir-Tl-Ir core of [Ir2Tl(C0)2Cl2(~-dpmal21+ is presented in Figure 5.10 This diagram focuses on a simple picture, the interaction of the filled iridium dz2 and thallium s orbitals and the empty pz on both iridium and thallium. These orbitals are aligned roughly along the Ir-Tl-Ir bond axis. In C2v symmetry mixing of levels of like symmetry will result in stabilization of the filled a1, b2 and a1 orbitals through mixing with their empty counterparts. This will impart stability to the Ir-Tl-Ir chain despite the fact that in the absence of mixing, the formal Ir-Tl bond order is zero. Nevertheless, these complexes are quite stable and the Ir-Tl interaction is substantial. For example, 1 is capable of extracting thallium from an aqueous TlN03 solution into a dichloromethane solution with the production of the orange thallium adduct. However, binding of snc1+ to 1 is even stronger and a transmetallation reaction can be effected. Shaking a dichloromethane solution of [Ir2Tl(C0)2Cl2(~­ dpma)2)N03 with solid tin(II) chloride results in a rapid color change of the solution from orange of the Tl+ adduct to the deep blue-green of the Sncl+ adduct. The thallium can also be removed from [Ir2Tl(C0)2Cl2(~­ dpma)2)+. Treatment of a dichloromethane solution of the orange thallium adduct with dibenzo-18-crown-6 results in the formation of a yellow solution of free 1 with the crown ether binding the Tl+.

304

+

+

Figure 4

The structures of top [Ir2(SnCl) (C0)2Cl2(~-dpma)21+ and bottom, [Ir2Tl(C0)2Cl2(~-dpmal2l [N03] as determined by x-ray crystallography. In the lower structure we believe that the nitrate ion is ion paired with the cation. The Tl···O distances are long, 2.883, 2.798 A.

305

Tl

2 lr

Figure 5

A qualitative molecular orbital diagram for the bent Ir-TlIr unit in [Ir2Tl(C0)2Cl2(~-dpma)2]+ which utilizes the outof-plane dz2 and pz orbitals on iridium and the s and pz orbitals on thallium for bonding. The z axis on each iridium is coincident with Ir-Tl bonds.

85 picoseconds

n•'

:

~

I\

j \

Absorption

:

:

Emission

\ : \

: . .::. .

:

:i I

--· 300

Figure 6

306

500

~ ~

:

1.4 microseconds

\:

,#'·.,

/

\\

i \

\, '~

nm

700

\

\

./ \\'..

..,.··

....__ ........•

/

...... __ 900

The electronic absorption (solid line) of [IrzTl(CO)zClz(~­ dpma)z] [N03] in dichloromethane solution at 25° c. The dotted line shows the uncorrected emission spectrum obtained with excitation at 510 nm. The lifetimes of the two emissions are indicated above for each.

The absorption and emission spectra of the thallium adduct in dichloromethane solution at 25° C are presented in Figure 6. An excitation spectrum obtained for either emission parallels closely the absorption spectrum presented here. The lifetimes observed for the two emissions are also recorded in Figure 6. Based on the small Stokes shift and the short lifetime, the emission at 580 nm is assigned to fluorescence while the longer-lived emission at 814 nm results from phosphorescence. The lifetime of the phosphorescence is within a factor of three of that of the well-studied Pt2(~-P205H2l44- 10 and the opportunity to observe binuclear reactions involving the excited state of the thallium adduct exists. RELATED

CHAINS

WITH VARYING ELECTRONIC CONFIGURATIONS

A number of trinuclear complexes with similarly strong electronic absorption and emission properties have been prepared and structurally characterized in this laboratory. These are shown in Figure 7 along with prominent absorption and emission maxima. The rhodium trimer13 shown at the top left is one of several examples of short chains involving d8 metal ions. A bondin~ picture for the metal-metal interactions in such chains has focused on 4,15 the interaction among the out of plane d 2 2 and Pz orbitals (with the z being the Rh···Rh···Rh axis) shown in Figure 8. This of course is very similar to the picture developed in Figure 6 for the Tl+adduct of 1. The essential point is that the symmetry properties of the d 2 2 or s orbitals that are involved in metal-metal bonds make these orbitals interchangeable. Thus this bonding model for d8d8d8 chains can be expected to be useful after modification for a variety of chains of metal ions so long as each is able to contribute a filled orbital (dz2 or s) and an empty orbital (Pzl. The Tl+ and snc1+ adducts are examples of d8s2d8 chains. In both the dadada and the d8s2d8 cases, the intense absorption in the electronic spectra of the complexes is assigned to a fully allowed transition from the filled level involving either the dz2_ dz2-dz2 or the dz2-s-dz2 orbitals to the empty orbital made out of contributions from the empty Pz orbitals. The Au+ adduct of 1, which has been characterized cystallographicallyl4 and is shown at the lower left of Figure 7 is an example of yet another type of chain, this one a d 8dloda sequence. Since Au+ is a d10 ion it contains only filled d orbitals and one of these can function to provide the necessary contribution to the bonding scheme. Again the complex cation is intensely colored and strongly luminescent.

Notice that this cation is not

that shown in C of Figure 2, since it lacks the two axial chloride ligands. However, the Au+ adduct, [Ir2Au(C0)2Cl2(~-dpmal2l+, may be converted into Ir2Au(C0)2Cl4(~-dpma)2+ by oxidation with molecular chlorine or carbon tetrachloride. A particularly novel example which we believe is related to this phosphine-bridged trinuclear compound is shown in the lower right side of Figure 7. This is Tl2Pt(CN)4,15 an example of an s 2d 8s 2 chain which consists of a planar Pt(CNl42- unit capped by two Tl+ions. In the crystalline solid the complex is located at a center of symmetry and the Tl-Pt distances are both 3.140(1) A. This is somewhat longer than the Tl-Ir distances (2.958(1), 2.979(1) Al found in the Tl+ adduct of 1. Aside from the Tl-Pt bond, the thallium is bare: there are no other groups bound to thallium. This material crystallizes in anhydrous form upon mixing TlN03 with K2[Pt(CN)4l in aqueous solution. The formation of Tl-Pt bonds here is particularly striking because Pt(CN)42- units normally crystallize in stacks or columns with Pt···Pt separations that range from 3.09 ~o 3.75 A. Twenty crystallographically characterized examples of such stacks are known. Here, however, thallium breaks up

307

d8d8d8

absorption

647 (30,000)

absorption

516 (33,000)

emission

815

emission

580,814

P_.-......._As.........-.....f •

r---TI--r'Co c'........._

/........-c'

cf=/

"'-.....--A-__. p s2d&s2

Figure 7

absorption

508 (32,000)

absorption 370

emission

606

emission

444

Related trinuclear complexes with d8d8d8, d8dl0d8, d8s2d8, and s2d8s2 configurations.

~--· -::r·

-~

-a2u -•2u

~·h .

[ it~

J: -~

*~

*~ ~M~

;M~ '; M '!

;M":. :,M-:.

o:.M: Figure 8

308

Qualitative molecular orbital diagram showing the interactions of the filled dz 2 and empty Pz orbitals for monomeric ML 4 de complexes and the result of stacking to form dimeric M2L8 and trimeric M3 Ls in D4h symmetry.

that tendency to stack. Tl2Pt(CN)4 is stronqly luminescent in the solid state. The blue emission results in a single sharp band centered at 444 nm. Again the bonding can be described by considering interaction of the filled thallium s and platinum dz2 orbitals and the empty pz orbitals on each. ACKNOWLEDGEMENTS

Dr. Phil E. Reedy, Jr., Dr. Alan Fossett, Vince Catalano, Steve Reimer, Ella Fung, Professor Jeff Nagle, Professor Mauro Ghedini, and Francesco Neve have been involved with the synthetic and spectroscopic work. Mark Chatfield and Professor R. Rosenfeld helped with the lifetime studies. Dr. Marilyn Olmstead and Doug Oram were responsible for the xray structures. The National Science Foundation (CHE-8519557) has been generous with financial support.

REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

A.L. Balch. Pure and Applied Chemistry, 60: 555 (1988). A.L. Balch, L.A. Fossett, M.M. Olmstead, D.E. Oram, P.E. Reedy, Jr. J. Am. Chern. Soc , 107: 5272 (1985). A.L. Balch, M. Ghedini, D.E. Oram, P.E. Reedy, Jr. Inorg. Chern , 26: 1223 (1987). P.D. Brotherton, C.L. Raston, A.H. White, S.B. Wild. J. Chern. Soc. Dalton Trans , 1799 (1976). P.E. Reedy, Ph.D. Thesis, University of California, Davis, 1987. D.A. Bailey, A.L. Balch, L.A. Fossett, M.M. Olmstead, P.E. Reedy, Jr. Inorg. Chern., 26: 2413 (1987). A.L. Balch. Ann. New York Acad. Sci., 313: 651 (1978). A.L. Balch in Homogeneous Catalysis with Metal Phosphine Complexes. L.H. Pignolet, ed. Plenum Press, New York, P.l67. A.L. Balch, D.E. Oram, P.E. Reedy, Jr. Inorg. Chern., 26: 1836 (1987). A.L. Balch, J.K. Nagle, M.M. Olmstead, P.E. Reedy, Jr. JAm· Chern, .s.o.c........ 109: 4123, (1987) . A.L. Balch, M.M. Olmstead, D.E. Oram, P.E. Reedy, Jr., S.H. Reimer. J. Am· Chern. Soc., in press. D.M. Roundhill, H.B. Gray, C.-M. Che. Accounts Chern. ReSearch, 22: 55 (1989). Balch, A.L., L.A. Fossett, J.K. Nagle, M.M. Olmstead. J. Am· Chern. ~. 110: 6732 (1988). K.R. Mann, N.S. Lewis, R.M. Williams, H.B. Gray, J.G .. Gordon, II. Inorg. Chern., 17: 828 (1978). A.L. Balch, L.A. Fossett, M.M. Olmstead, P.E. Reedy, Jr. Organometallics,_5: 1929 (1986). A.L. Balch, J.K. Nagle, D.E. Oram, P.E. Reedy, Jr. J. Am. Chern. ~. 110: 454 (1988). J.K. Nagle, A.L. Balch, M.M. Olmstead. J. Am· Chern. Soc., 110: 319 (1988).

309

THE PREPARATION AND CHARACTERIZATION OF NEW HETEROPOLYOXOFLUOROMETALATE ANION, [FeW17 o56 F6H5 ] 8 Sadiq H. Wasfi Department of Chemistry, Delaware State College Dover, Delaware 19901 The ammonium salt of the anion [Few17 o56 F6H5 ] 8 - has been isolated in the crystalline form. Chemical analysis, FABMS, visible and FTIR spectroscopy indicate that the anion has the Dawson structure with one of the w+ 6 in the original 2:18 replaced by Fe+ 3 ion. The anion appears to be isomorphous with that reported by Baker et al in the Thir~ Chemical Congress of North America, June 5-10, 1988, [Znw17 o56 H4 NaF 6 ] -. Single crystal structure study is now underway.

Figure 1. Structure of the anion. The shaded area represents the replacement of a tungsten by a co+2 or a Fe+3 ion.

311

FACILE EXCHANGE OF TERMINAL, DOUBLY BRIDGING, AND QUADRUPLY BRIDGING CARBONYL LIGANDS IN SOLUTION:

CRYSTAL STRUCTURE AND SOLUTION DYNAMICS

Yun Chia, Sue-Lein Wanga and Shie-Ming Pengb aDepartment of Chemistry, National Tsing Hua University Hsinchu 30043, Taiwan bDepartment of Chemistry, National Taiwan University Taipei 10764, Taiwan In the series of the tetranuclear mixed-metal clusters LWM3 (C0) 12H, L = c5H5 , c5Me 5 and M = Os, Ru, the cluster cores adapt different geometries from tetrahedron (isomer a), distorted tetrahedron (isomer b) to butterfly arrangement (isomers c and d) in the solid state (see scheme 1). For the last two butterfly isomers, a CO ligand adapted a w-bonded ~ 4 -~ 2 bridging mode (as a four electron donor) to stabilize the cluster. In solution stfte, these isomers undergo rapid interconversion on the time scale of H NMR spectroscopy, suggesting that the w-bonded CO ligands is in equilibrium with the regular, terminal or doubly bridging CO ligands. Hydrogenation of LWM3 (C0) 12 H in refluxing toluene solution produces LWM 3 (C0) 11H3 which adapted a tetrahedral cluster core structure. The fluxionality of these hydride complexes in solution will be discussed.

312

L(COh

/1

w-H I

(CO)~~ /!M(C0) 3 0

M = Os,L=C;H5; M

= Os, L = C5Me5;

M = Ru, L = C5 Hs;

la (0.75)

lb (0.25)

2a (0.32)

2b (0.68)

3a (0.17)

3b (0.28)

4b (0.16)

M = Ru, L = C5Me5;

M = Ru, L = C 5H 5;

M

=Ru, L = C5Me5;

C-M-C (CO)z 0

3c (0.55) 4c (0.22)

4d (0.62)

Scheme 1. Note: The numbers in parentheses are their relative ratio in solution state.

313

THERMAL CONSTANTS AND STRUCTURE OF TIN CLUSTERS

Richard W. Schmude, Jr.,a Karl A. Gingerich,a and Joseph E. Kingcade, Jr.b aDepartment of Chemistry, Texas A&M University, College Station, Texas 77843 bDivision of Natural Science, Blinn College, Brenham, Texas 77833 The Knudsen effusion method coupled with high temperature mass spectroscopy was used in evaluating thermal constants for the gaseous (n=4-7) molecules. The enthalpies for the following reaction was determined: SDn(g) = nSn(g) for n = 4-7. The resulting atomization energy, heats of formation and entropy are for the linear molecules: 4H,, 0

4Hr,o

4S,,29B

KJ/mol

KJ/mol

J/K·mol

Sn4

714

491

288

Snn

314

Sns

908

598

358

Sn6

1126

681

432

Sn7

1387

721

518

S~

REACTIVITY AND ISOMERIZATION OF Mo 2 (ALLYL) 4

Reed J. Blau, Ron-Jer Tsay, and Su-Inn Ho Chemistry Department, University of Texas at Arlington, Arlington, Texas 76019

Mo 2 (~ 3 -~ 2 -allyl) 2 (~ 3 -allyl) 2 , A, is an organometallic molybdenum dimer with a quadruple Mo-Mo bond. ButOH reacts with A to form propene concurrently with an NMR-detectable intermediate containing two types of t-butoxide ligands. At 0 °C, t~e intermediate is readily transformed into an isolable species, Mo 4 (~ -~ 2 -allyl) 4 (~ 2 -0But) 4 , with a proposed tetrametallacyclobutadiyne structure. The reaction of A with acetylacetone yields a pinkish produ§t with concom~tant evolution of propene having the structure, M~ 2 (~ -~ 2 -allyl) 2 (~ -acac) 2 . Acetylacetonate~ have replaced the terminal ~ -allyls structurally in A leaving the Mo 2 (~ -~ 2 allyl)2 core intact. Carbon monoxide promotes the heretofore unobserved isomerization of green A to a violet isomer with configurational changes in the bonding of the bridging allyl ligands to the dimetal center. In addition, c~rbon monoxide induces allyl dimerization upon reaction with A yielding (~ -l,5-hexadiene)Mo(C0) 4 and eventually Mo(C0) 6 and free 1,5hexadiene as products.

315

SURFACE COORDINATION/ORGANOMETALLIC CHEMISTRY OF MONOMETAL AND BIMETALLIC ELECTROCATALYSTS Ginger M. Berry, Michael E. Bothwell, Beatriz G. Bravo, George J. Cali, John E. Harris, Thomas Mebrahtu, Susan L. Michelhaugh, Jose F. Rodriguez and Manuel P. Soriaga* Department of Chemistry, Texas A&M University, College Station, TX 77843 The interaction of selected organic and inorganic functional groups, which are reversibly electroactive and strongly surface-active, with monometal (Rh, Pd, Ir, Pt, Au) and mixed-metal (Au-Pt, Ag-Pt) electrocatalysts has been studied to help establish the interfacial organometallic/coordination chemistry of these metals in aqueous solutions. Experimental measurements were based upon thin-layer electrochemical and ultra-high vacuum surface spectroscopic methods; the latter included low-energy electron diffraction, Auger electron spectroscopy, X-ray photoelectron spectroscopy and thermal desorption mass spectrometry. The results to date indicate the following trends: (i) Electrode surface phenomena can be modeled in terms of monometal and cluster coordination/organometallic chemistry. (ii) Chemisorption is analogous to oxidative addition; desorption is similar to reductive elimination. (iii) Chemisorption of an electroactive center favors its oxidized state over the reduced form. (iv) Chemisorption of electroinactive anionic reagents forms polyprotic surface acids. (v) Substrate-mediated interactions between pendant electroactive centers may arise if the redox group itself is surface-active. (vi) Substratemediated adsorbate-adsorbate interactions can be viewed similarly to mixed-valence complexes. (vii) The electrocatalytic reactivity of an adsorbed material is dependent upon its initial mode of binding. (viii) The strong dependence of the electrochemical properties of chemisorbed redox centers on the electrode material makes it suitable for the study of mixed-metal interfaces. REFERENCES

1. 2. 3. 4. 5.

316

T. Mebrahtu, J. F. Rodriguez, B. G. Bravo and M. P. Soriaga. J. Electroanal. Chem. 219 (1987) 327. B. G. Bravo, T. Mebrahtu, J. F. Rodriguez and M. P. Soriaga. J. Electroanal. Chem. 220 (1987) 281. J. F. Rodriguez, B. G. Bravo, T. Mebrahtu and M. P. Soriaga. Inorg. Chem. 26 (1987) 2760. B. G. Bravo, T. Mebrahtu and M. P. Soriaga. Langmuir. 3 (1987) 595. J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J. Electroanal. Chem. 233 (1987) 283.

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25.

J. F. Rodriguez and M. P. Soriaga. J. Electrochem. Soc. 135 (1988) 616. B. G. Bravo, J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J. Phys. Chem. 91 (1987) 5660. B. G. Bravo, S. L. Michelhaugh and M. P. Soriaga. J. Electroanal. Chem. 241 (1988) 199. J. F. Rodriguez, J. E. Harris, M. E. Bothwell, T. Mebrahtu and M. P. Soriaga. Inorg. Chim. Acta. 148 (1988) 123. M. P. Soriaga. J. Electroanal. Chem. 240 (1988) 309. J. F. Rodriguez, M. E. Bothwell, J. E. Harris and M. P. Soriaga. J. Phys. Chem. 92 (1988) 2702. T. Mebrahtu, G. M. Berry and M. P. Soriaga. J. Electroanal. Chem. 239 (1988) 375. M. P. Soriaga. In M. P. Soriaga, Editor, "Electrochemical Surface Science: Molecular Phenomena at Electrode Surfaces." ACS Books: Washington, DC. Chapter 1 (1988). T. Mebrahtu, G. M. Berry, B. G. Bravo, S. L. Michelhaugh and M. P. Soriaga. Langmuir. 4 (1988) 1147. B. G. Bravo, S. L. Michelhaugh, T. Mebrahtu and M. P. Soriaga. Electrochim. Acta. 33 (1988) 1507. Mebrahtu, G. M. Berry and M. P. Soriaga. J. Electroanal. Chem. 247 (1988) 241. M. P. Soriaga, G. M. Berry, M. E. Bothwell, B. G. Bravo, G. J. Cali, J. E. Harris, T. Mebrahtu, S. L. Michelhaugh and J. F. Rodriguez. In M. P. Soriaga, Editor, "Electrochemical Surface Science: Molecular Phenomena at Electrode Surfaces." ACS Books: Washington, DC. Chapter 36 (1988). M. E. Bothwell, J. F. Rodriguez and M. P. Soriaga. J. Electroanal. Chem. 252 (1988) 453. M. E. Bothwell and M. P. Soriaga. J. Electroanal. Chem. 260 (1989) 193. J. E. Harris, M. E. Bothwell, J. F. Rodriguez and M. P. Soriaga. J. Phys. Chem. 93 (1989) 2610. J. E. Harris and M. P. Soriaga. Electrochimica Acta. In press (1989). G. M. Berry, B. G. Bravo, M. E. Bothwell, G. J. Cali, J. E. Harris, T. Mebrahtu, S. L. Michelhaugh, J. F. Rodriguez and M. P. Soriaga. Langmuir. 5 (1989) 707. J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J. Electroanal. Chem. In press (1989). B. G. Bravo, S. L. Michelhaugh and M. P. Soriaga. Langmuir. In press (1989). G. J. Cali, M. E. Bothwell and M. P. Soriaga. J. Electroanal. Chem. In press (1989).

317

AMBIENT-TEMPERATURE CHLOROALUMINATE MOLTEN SALTS:

SOLVENTS FOR CHLORO

COMPLEX ELECTROCHEMISTRY AND FOR REDUCTIVE CONDENSATION SYNTHESES R. T. Carlin and R. A. Osteryoung Department of Chemistry, State University of New York at Buffalo, Buffalo, New York

14214

The ambient-temperature molten salt A1Cl 3 :1-ethyl-3-methylimidazolium chloride (ImCl) provides a low-temperature medium for the study of chloro complexes when the AlC1 3 : ImCl molar r!tio is < ~ (basic melt). The electrochemical interconversions of Mo 2c1 9 -, Mo 2c1 8 -, and Mo 2cl 8H3 - in a basic melt demonstrate the ability of this solvent to maintain the integrity of chloro complexes in solution. When the AlC1 3 :ImCl molar ratio is> 1 (acidic melt), the molten salt contains the strongly Lewis-acidic species Al 2c1 7 - which functions as an effective chloride acceptor. Recently, it has been demonstrated by J. S. Wilkes and co-workers that Na metal reacts with an acidic melt producing a metallic precipitate, believed to be an Al/Na alloy, and soluble NaCl. This alloy, when combined with the chloride ~ccepting properties of the acidic melt, can be employed as a reagent to induce reductive condensation reactions providing low-valent metal clusters from high-valent metal chlorides. Preliminary investigations on the reduction of NbC1 5 and TaC1 5 in an acidic melt at room temg~rature indicate formation of the hexanuclear chloro clusters, M6c1 12 .

318

THE MAXIMUM STRENGTH OF THE CHEMICAL BOND BETWEEN TWO METAL ATOMS

K. A. Gingerich Department of Chemistry, Texas A&M University, College Station, Texas

77843-3255

Over the past 20 years we have measured, at Texas A&M University, a large number of bond energies between two ligand free transition metal atoms, using high temperature equilibrium mass spectrometry. The results of these measurements, together with predictions by empirical models of bonding, permit a definite estimate of the possible maximum strength of the_~hymical bond between two ligand-free metal atoms of G40 ± 40 kJ moL . The possible maximum strength of the chemical bonds between pairs of atoms in dinuclear transition metal complexes will also be discussed.

REFERENCES 1.

K. A. Gingerich, Faraday Disc. Chem. Soc., 14 (1980) 109.

319

REACTIONS OF (~ 7 -c 7 H 7 )M(~ 5 -c 5 H 5 ), M- Ti OR Zr, WITH CARBOXYLIC AND DITHIOCARBOXYLIC ACIDS S. A. Duraj, M. T. Adras, R. A. Martuch, and S. SriHari Department of Chemistry, Cleveland State University, Cleveland, OH 44115 New reactions involving titanium and zirconium sandwich compounds will be presented. Thus, sandwich ~-complexes of the type (~ 7 -c 7 H 7 )M(~ 5 c5H5) where M = Ti or Zr, readily react with carboxylic or dithiocarboxylic acids to give compounds of the general formula: (~ 5 c5H5)M(Y2CR)3, M = Ti, Y = S, R = CH 3 or M- Zr, Y = 0, R = C5H6 . X-ray crystallography of these corpounds will be displayed together with other spectroscopic data such as H and 13 c NMR, IR, etc.

320

MODEL HYDRODESULFURIZATION SYSTEMS:

REACTIONS OF SULFUR CONTAINING

MOLECULES ON Ni(llO) D. R. Huntley Oak Ridge National Laboratory, Oak Ridge, Tennessee

37831-6201

Reactions of sulfur containing molecules on the Ni(llO) surface have been studied under ultrahigh vacuum conditions using such surface analysis tools as high resolution electron energy loss spectroscopy (HREELS), x-ray photoelectron spectroscopy (XPS), temperature programmed reactions (TPR) and Auger electron spectroscopy (AES). Mechanistic information on the desulfurization reaction of methanethiol was obtained. The reaction products were gaseous methane and hydrogen, and surface sulfur and carbon. Methylthiolate fragments were identified as intermediates by vibrational spectroscopy (HREELS) and XPS. Recent results on the decomposition of H2 s and on the interaction of H2 with sulfided Ni(llO) will also be presented.

321

ELECTROCHEMICAL STUDIES OF TRIANGULAR NIOBIUM CLUSTER, Nb 302(S04)6·3H20 5 -, IN SULFURIC ACID. W. Sayers, T. Batten, M. May and V. Katovic Wright State University, Department of Chemistry, Dayton, Ohio

45435

The electrochemical properties of Nb 3o2 (so 4 ) 6 ·3H 2o5 - cluster anion in 9 M sulfuric acid were investigated using DC polarography, cyclic voltammetry, spectroelectrochem istry, and constant potential electrolysis. It was found that Nb 3o2 (so4 ) 6 ·3H 2o5 - in 9 M H2so 4 displays a reduction wave at E = -1.30 V vs. Hg/Hg 2 so 4 electrode, and a large irreversible oxidation wave at Epa = -0.30 V. Cyclic voltammetry indicates that the reduction wave involves a reversible one-electron process. Controlled potential electroG~sis at E- -1.4 V produces a green Nb(3.33) anion, Nb 3o2 (so 4 ) 6 ·3H 2o which is stable on the time scale of bulk electrol6~is. Electrochemical reoxidation of Nb(3.33) anion Nb 3o2 (so 4 ) 6 ·3H 2o atE= -0.75 V consumes 1 mole/ mol Nb 3 cluster and regenerates the red Nb(3.66) Nb 3o2 (so 4 ) 6 ·3H 2o5 - anion. Further oxidation at -0.2 V involves an irreversible four electron process to form Nb(V) species.

322

REACTIVITY OF DITHIOETHERS TOWARD [Re 2X8 ] 2 -

J. Gregory Jennings and Gregory L. Powell Department of Chemistry, Abilene Christian University, Abilene, TX 79699 The reactivities of 3,6-dithiaoctane (dto) and 2,5-dithiahexane (dth) toward quadruply bonded octahalodirhenate(III) ions have been investigated und~: a variety of conditions. At or below so•c, the reaction of [Re 2cl 8 ] with dto in ethanol yields a novel quadruply bonded anion, [Re 2cl 7 (dto)]-. At higher temperatures, this reaction produces the paramagnetic triply bonded compound Re 2cl 5 (dto) 2 . Similarly, (Re 2cl 7 (qth)]- and Re 2cl 5 (dth) 2 have been isolated from reactions between [Re 2 cl 8 ] -and dth in ethanol. The complexes [Re 2cl 7 (LL)]-, in which LL - dto or dth, have been shown to be intermediates in the synthesis of Re 2cl 5 (LL) 2 . No such intermediates have been detected in the reactions of dto and dth with quadruply bond~d [Re 2Br 8 ] 2 - to produce Re 2Br 5 (LL) 2 . No reaction occurs between [Re 2x8 ] - and dto in dichloromethane, acetone, or acetonitrile. The X-ray crystal structures of (n-Bu4 N)(Re 2cl 7 (dto)]-, (n-Bu4 N)[Re 2cl 7 (dth)]-, Re 2cl 5 (dto) 2 , and Re 2Br 5 (dto) 2 will be presented.

323

THEORETICAL INVESTIGATIONS OF THE METAL-METAL INTERACTIONS WITHIN THE

Andrew L. Sargent and Michael B. Hall Department of Chemistry, Texas A&M University, College Station, TX

77843

Unparameterized Fenske-Hall molecular orbital calculations were performed on the title complex and its chlorine-oxidized analogue to help determine why the Au-Pt bondlengths decrease from 3.03 A to 2.66 A upon oxidation. The results indicate that the oxidation withdraws electron density from a metal-metal antibonding molecular orbital while populating a previously unoccupied metal-metal bonding orbital.

324

STRUCTURAL AND THEORETICAL STUDIES ON HETERONUCLEAR TRANSITION-METAL CLUSTERS CONTAINING THE ALKYLIDYNE LIGAND P. Sherwooda, M. B. Halla, J. C. Jefferyb, and F. G. A. Stoneb aDepartment of Chemistry, Texas A&M University, College Station, Texas

77843

bDepartment of Inorganic Chemistry, University of Bristol, Cantock's Close, Bristol BS8 lTS, U.K. The synthetic studies of F. G. A. Stone and co-workers have led to the characterization of a large number of alkylidyne-bridged di- and trinuclear complexes. The results of structural studies on some of these species will be presented, together with an analysis of these structural results and some of the spectroscopic data, based on extended Huckel, Fenske-Hall and ab initio calculations. Of particular interest are the varia£~ons in the geometrical C n.m.r. chemical shift of the parameters of the cluster cores, and the The r~kylidyne carbon atom, when the cluster electron count is changed. C n.m.r. data is also found to be dependent on the organic substituent at the alkylidyne carbon, and some of this data will be discussed in the light of the theoretical results.

325

ELECTRONIC STRUCTURE AND NATURE OF BONDING IN TRANSITION METAL DIMERS

Irene Shim Chemistry Department B, The Technical University of Denmark, DTH 301, DK-2800 Lyngby, Denmark All electron ab initio Hartree-Fock and configuration interaction methods have been applied to elucidate the electronic structure and the nature of bonding of the transition metal dimers cu 2 , Ni 2 , Co 2 , Fe 2 , Ag 2 , Pd 2 , Rh 2 , Nb 2 . Towards the end of the transition metal series the d electrons tend to localize around the nuclei, and they therefore participate only slightly in the bond formation. However, due to the various exchange couplings the d electrons give rise to large numbers of extremely close-lying potential energy curves. Towards the middle of the transition metal series the d electron participation in bond formation appears as larger populations in the bonding natural orbital relative to the antibonding natural orbitals. The chemical bonds of all the molecules considered are very complicated, and even when the d electrons participate significantly in the bond formation the molecular orbital picture is inadequate for describing the bonds.

326

THEORETICAL CALCULATIONS ON THE INTERACTION OF BRIDGING CARBONYLS WITH TRANSITION METAL DIMERS Charles Q. Simpson II and Michael B. Hall Department of Chemistry, Texas A&M University, College Station, Texas

77843

Molecular orbital calculations at the ab initio Hartree-FockRoothaan level of theory were performed to determine what electronic properties cause a bridging carbonyl to adopt a particular bridging mode, i.e. linear semibridging, or symmetrical bridging. A set of rules is then developed to predict when a bridging carbonyl will assume a certain bridging mode.

327

BIMETALLIC HYDROFORMYLATION CATALYSIS

Scott A. Laneman and George G. Stanley Department of Chemistry, Louisiana State University Baton Rouge, Louisiana 70803-1804 Bimetallic transition metal complexes based on a new binucleating tetratertiary-phosphine ligand system (Et 2PcH 2cH 2 )(Ph)PCH 2P(Ph)(CH 2cH 2PEt 2 ), eLTTP, have been prepared and characterized. The Rh(I) bimetallic complexes meso- and racemicRh2Cl2(C0)2(eLTTP) have been synthesized and characterized and the racemic structure characterized. The hydroformy~ation activity of the norbornadiene derivative, Rh 2 (norb) 2 (cO)z(eLTTP) +(norb- norbornadiene), with respect to simple 1-alkenes and functionalized alkenes such as vinyl acetate are reported. The Rh 2 (eLTTP) catalyst system is unusual for several reasons: 1) excess phosphine ligand is not required to maintain catalyst stability; 2) unusually high normal/branched product aldehyde ratios are seen in the hydroformylation of 1-alkenes (18:1 normal to branched, at 120 psi and 80"C); 3) the hydroformylation catalysis is markedly faster than that seen for monometallic model system Rh(norb)(depe)+ (depe- Et 2PCH 2CH 2PEt 2 ) and 4) the vinyl acetate hydroformylation reaction does not suffer from the extensive product, reactant and catalyst decomposition reactions typically seen for Rh(I)/PR 3 catalyzed systems. The enhanced rate for the bimetallic verses monometallic system is believed to be due to bimetallic cooperativity in the form of intramolecular hydride transfer and elimination of the product aldehyde.

328

252 cF-PLASMA DESORPTION MASS SPECTRA OF VERY LARGE CLUSTERS

J. P. Fackler, Jr., C. J. McNeal,

and R. E. P. Winpenny

Department of Chemistry, Texas A&M University, College Station, Texas 77843 The 252 cf plasma desorption mass spectra of gold clusters of molecular weight ~ 17000 a.m.u are reported and explained in terms of structures involving vertex sharing icosahedra. The spectra seen depend on the surface from which the ions are desorbed. From a gold surface peaks are seen due to clusters of stoichiometry Au 25 (PPh 3 ) 12 cl 6 ; Au46 (PPh 3 ) 1 2Cl6 and Au 67 (PPh 3 ) 14 cl 8 . On a nitrocellulose surface peaks are seen up to m/z 56000. A condensation of clusters appears to be occurring and a mechanism by which this condensation takes place is proposed.

329

SYSTEMATIC KINETIC STUDIES

pF

ASSOCIATIVE AND DISSOCIATIVE REACTIONS OF

SUBSTITUTED METAL CARBONYL CLUSTERS:

THE INTIMATE MECHANISMS

N. M. J. Brodie, Lezhan Chen, and A. J. Poe Department of Chemistry, University of Toronto, Toronto, Ontario MSS lAl, Canada Systematic kinetic studies 1 of associative reactions 2 of metal carbonyl clusters can provide reactivity profiles that are as usefully characteristic of the clusters as their crystal structures. A given cluster can be quantitatively characterized by its sensitivity to the electronic and steric nature of a series of P-donor nucleophiles, 2 by ~ts standard or "intrinsic" reactivity (I.R.) towards nucleophilic attack, and by the distribution between substitution an~ fragmentation products formed as a consequence of nucleophilic attack. Thus the electronic and steric profiles for associative reactions of Ru 3 (C0) 11 L show that bondmaking in the transition states decreases along the series L = P(OEt) 3 > CO > P-n-Bu 3 and this correlates with increasing intrinsic reactivities of the clusters. This must result from a balance between opposing steric and electronic effects but the exact way in which these operate is not yet clear. The tendency for fragmentation to result from associative attack by P-n-Bu 3 on Ru 3 (C0) 11 L increases along the series L- CO << P(OEt) 3 < PPh 3 < P-n-Bu 3 which suggest that steric effects are dominant but that electronic effects can be significant (PPh 3 < P-n-Bu 3 ). The importance of Ru-P bond lengths in contributing to the steric effects is shown by the virtually linear increase of d(Ru-P) by ca. 0.2 A as the cone angle of the P-donor increases from 100-170". 4 Reactions of Rh 4 (C0) 10 (PCy 3 ) 2 and Rh 4 (CO)g{P(OCH 2 ) 3CEt)} 3 show that associative reactions of the former involve less bond making, and that its intrinsic reactivity i~ ca. 100 times greater. It can be estimated that the I.R. of Rh4 (C0) 12 is greater still by another 2 orders of ma1nitude. This is ca. 10 7 times greater than that for Ir 4 (co) 12 and ca. 10 · 5 times greater than that for Ru 3 (C0) 12 . The importance of M-M bond strengths and the nuclearity of the clusters is therefore evident. A similar analysis of data for dissociative reactions of 12 _nLn (n - 1 and 2) shows that log 6k 1 increases linearly with 5 ( CO)(the net electron donor ability of L) and with the cone angle of DHta in the literature for L, so that the data fit well to eq (1). reactions of Co 4 (C0) 11 L7 and Ir 4 (C0) 12 _n~ (L - 1-3) also seem to fit Ry~(C0)

(1)

330

eq (1) but data over a wider range of 6( 13 co) and 9 are required before this can be fully substantiated. It is evident that data for both associative and dissociative reactions can be analyzed quantitatively in these ways and that the electronic and steric profiles can provide empirical data for characterizing the clusters and also provide a means of probing the intimate details of the reaction mechanisms. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8.

A. J. Poe, Chapter 4 in "Metal Clusters", Ed. M. Moskovits, Wiley, New York (1986). A. J. Poe, Int. J. Chem. Kin., 20:467 (1988). N. M. J. Brodie and A. J. Poe, Inorg. Chem., 27:3156 (1988). Bruce et al., J. Organomet. Chem., 347:157 (1988); M. N.J. Brodie, L. Chen, A. J. Poe, and J. F. Sawyer. Acta Cryst., Part C. cryst. Str. Comm., 1989, C45, 0000. J. R. Kennedy, F. Basolo, and W. C. Trogler, Inorg. Chim. Acta, 146:75 (1988). G. Bodner, M. May, and L. McKinney, Inorg. Chem., 19:1951 (1980). D. J. Darensbourg, D. J. Zalewski, and T. Delord, Organometallics, 3:1210 (1984). D. C. Sonnenberger and J. D. Atwood, J. Am. Chem. Soc., 104:2113 (1982) and Organometallics, 1:694 (1982). D. J. Darensbourg and B. J. Baldwin-Zuschke, J. Am. Chem. Soc., 104:3906 (1982).

331

MIXED Pd-Au AND Pt-Au CLUSTER COMPOUNDS

Louis H. Pignolet Department of Chemistry, University of Minnesota, Minneapolis, MN 55455 A large number of bimetallic transition metal-gold cluster compounds have been synthesized (see for example New. J. Chem •. 1988, 12, 505). These clusters are stabilized by phosphine ligands and range in size from two to ten metal atoms. The details of the syntheses and structures of some of these compounds will be presented. Recently, we have been successful in ~~thesizing the firs~ Pd-Au clusters.· For example, [PdAu 6 (PPh 3 ) 7 J and [PdAu8 (PPh 3 ) 8 ] + were made in high yield and characterized by NMR and FABMS. The latter cluster and its CO adduct have also been characterized by x-ray crystallography. Analogous Pt-Au clusters have been made and will also be presented. Some of these compounds show interesting reactivity and in several cases catalytic behavior has been observed.

332

STABILITY OF SMALL BICLUSTERS OF TRANSITION METALS WITH SEMI-CONDUCTORS

J. E. Kingcade, Jr., I. Shim, and K. A. Gingerich Department of Chemistry, Texas A&M University, College Station, Texas

77843

The characterization of the stability, structure and nature of bonding of small biclusters between £r~nsition metals and semiconductors is of considerable current interest. High temperature equilibrium mass spectrometry has been used to obtain atomization energies of small clusters of transition metals, such as scandium, yttrium, nickel, palladium, copper and gold with silicon and germanium. The experimentally determined bond energies will be reviewed and critically discussed in terms of empirical models and theoretical ab initio calculations. For selected diatomic molecules, such as PdGe, PdSi and NiSi all electron ab initio Hartree-Fock (HF) and configuration interaction (CI) computations have been performed to elucidate the electronic structure and the nature of bonding. This work has been supported by the National Science Foundation, the Robert A. Welch Foundation, and by NATO Grant No. RG 85/0448. REFERENCES 1. 2. 3.

M. D. Morse, Chem. Rev. 86 (1986) 1049. K. A. Gingerich, Faraday Disc. Chem. Soc., 14 (1980) 109. K. A. Gingerich, I. Shim, S. K. Gupta and J. E. Kingcade, Jr., Surface Sci., 156 (1985) 495.

333

THE TOPOLOGY OF THE TOTAL CHARGE DENSITY IN BINUCLEAR TRANSITION-METAL COMPLEXES THAT FORMALLY CONTAIN METAL-METAL BONDS Preston J. MacDougall and Michael B. Hall Department of Chemistry, Texas A&M University, College Station, Texas

77843

The quantum theory of atoms in molecules and the topological theory of molecular structure define atoms and bonds, respectively, in terms of the electronic charge distribution, p(r) [1]. This identification enables X-ray crystallographers and theoretical chemists to discuss the properties of atoms and bonds using a familiar and common language. The results from the first such analysis of transition-metal complexes will be presented. The binuclear systems Mn2 (C0) 10 , Fe 2 (C0) 9 and co 2 (C0) 8 have been studied to determine whether or not their corresponding molecular structures contain metal-metal bonds. At a level of theory that is generally found to be sufficient for predicting chemical structures (i.e. the network of topologically defined bond paths that link the atoms as opposed to the precise geometrical structure of the molecule), only Mn 2 (C0) 10 possesses a metal-metal bond. REFERENCES 1.

334

R. F. W. Bader Ace. Chem. Res., 18, 9 (1985).

THE EFFECT OF CARBONYL LIGANDS ON OSMIUM AND RUTHENIUM METAL-METAL BONDS

Ann E. Miller and William A. Goddard III Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, California 91125 Metal carbonyl clusters have been studied intently to determine their bonding, reactivity, and catalytic activity. While determining the properties of the electronic structure which influence the reactivity of H2os 3 (co) 10 we have studied the effect that carbonyl ligands have on metal-metal bonds. The results of this study suggest that the nature of the metal-metal bond is dependent on the angle of the carbonyl ligand, and consequently, that the geometry of the carbonyl ligands around the cluster is dictated by the stabilization of the metal-metal interactions. These results have led to the study of ~ -2 and ~ -3 carbonyl ligands. The system involves a carbonyl ligand bound to a Ru 3 cluster modeling three Ru atoms on the Ru(OOl) surface. The goal for this work is to ascertain the exact mode of bonding of the GO to the cluster, and to determine the effects of the bridging ligand on the metal-metal bonds. The insights into metal-metal and metal-ligand interactions gained from this work will help in the design of new clusters and catalysts.

335

CONTRIBUTORS

Adams, R.D. Adras, M. T. Balch, A.L. Batten, T. Benefield, R.E. Berry, G.M. Blau, R.J. Bothwell, M.E. Batt, A. Boyd, D.C. Braga, D. Bravo, B.G. Brodie, N.M.J. Bursten, B.E. Cali, G.J. Carlin, R.T. Chen, L. Chi, Y. Chisholm, M.H. Cotton, F.A. Cowley, A.H. Darensbourg, D.J. Day, V.W. Duraj, S.A. Evans, J .F. Fackler, Jr., J.P. Gagne, M.R. Gates, B.C. Gingerich, K.A. Gingerich, K.A. Gingerich, K.A. Gladfelter, W.L. Goddard Ill, W.A. Goodman, D.W. Haasch, R.T. Hall, M.B. Hall, M.B. Hall, M.B. Hall, M.B. Hall, M.B. Hanson, B.E. Harris, J.E. Ho, K.-L. Ho, s.-1. Huntley, D.R. Hwang, J. -W.

University of South Carolina Cleveland State University University of California-Davis Wright State University University of Kent Texas A&M University University of Texas-Arlington Texas A&M University University of Cambridge University of Minnesota University of Bologna Texas A&M University University of Toronto Ohio State University Texas A&M University State University of New York-Buffalo University of Toronto National Tsing Hua University Indiana University Texas A&M University University of Texas-Austin Texas A&M University Crystalytics Company Cleveland State University University of Minnesota Texas A&M University Northwestern University University of Delaware Texas A&M University Texas A&M University Texas A&M University University of Minnesota California Institute of Technology Texas A&M University University of Minnesota Texas A&M University Texas A&M University Texas A&M University Texas A&M University Texas A&M University Virginia Polytechnic Institute Texas A&M University University of Minnesota University of Texas-Arlington Oak Ridge National Laboratory University of Minnesota

1

320 299 322 141 316 315 316

141

215 141 316 330 19 316 318 330 312 55

1

195

41

161 320 215 329

ll3 127 314 319 333 215 335 205 215 265 324 325 327 334 231 316 215 315 321 215 337

Jeffery, J.C. Jennings, J.G. Jensen, K.F. Johnson, B.F.G. Johnston, R.L. Katovic, V. Kingcade, Jr., J.E. Kingcade, Jr., J.E. Klemperer, W.G. Lagow, R.J. Laneman, S.A. Lictenberger, D.L. Lockledge, S.P. MacDougall, P.J. Main, D.J. Marks, T.J. Marseglia, E.A. Martuch, R.A. May, M. McCarley, R.E. McNeal, C.J. Mebrahtu, T. Michelhaugh, S.L. Miller, A. E. Nolan, S.P. Osteryoung, R.A. Peng, S.-M. Pignolet, L.H. Poe, A.J. Powell, G.L. Rodger, A. Rodriguez, J.F. Rosenberg, F.S. Sargent, A.L. Sayers, W. Schmude, Jr., R.W. Schulze, R.K. Seyam, A.M. Sherwood, P. Shim, I. Shim, I. Simpson, II, C.Q. Sinfelt, J.H. Smalley, R.E. Soriaga, M.P. SriHari, S. Stanley, G.G. Stern, D. Stone, F.G.A. Tsay, R.-J. Walton, R.A. Wang, R. -c. Wang, S.-L. Wasfi, S.H. Winpenny, R.E.P. Yaghi, O.M.

338

Texas A&M University Abilene Christian University University of Minnesota University of Cambridge University of Arizona Wright State University Texas A&M University Texas A&M University University of Illinois University of Texas-Austin Louisiana State University University of Arizona University of Illinois Texas A&M University University of Illinois Northwestern University Cavendish Laboratory Cleveland State University Wright State University Iowa Sate University Texas A&M University Texas A&M University Texas A&M University California Institute of Technology Northwestern University State University of New York-Buffalo National Tsing Hua University University of Minnesota University of Toronto Abilene Christian University University of Cambridge Texas A&M University University of Illinois Texas A&M University Wright State University Blinn College University of Minnesota Northwestern University Texas A&M University The Technical University of Denmark The Technical University of Denmark Texas A&M University Exxon Rice University Texas A&M University Cleveland State University Louisiana State University Northwestern University University of Bristol University of Texas-Arlington Purdue University University of Illinois National Tsing Hua University Delaware State College Texas A&M University University of Illinois

325 323 215 141 275 322 314 333 161

171

328 275 161 334 161 113 141 320 322 91 329 316 316 335 113 318 312 332 330 323 141 316 161 324 322 314 215 113 325 326 333 327 103 249 316 320 328 113 325 315 7

161 312 311 329 161

INDEX

actinide diatomic molecules, 35 alkylidyne ligand, 325 aluminum nitride, 215 aluminum nitride, 218 ambient-temperature chloroaluminate molten salts, 318 anticuboctahedral, 151 Bailar Twist, 150 Berry psuedorotation, 234 bimetallic catalyst, 205, 206 bimetallic clusters, 103 bond energies, 319 bridging carbonyls with transition metal dimers, 327 bulk nickel, 250

c60+,

255 carbide, tungsten, 68 carbon cluster, 254 carbon dioxide, 43 carbon monoxide activation, 66 carbonyl ligands on osmium and ruthenium metalmetal bonds, 335 carbonylation, 44 charge density in binuclear transition-metal complexes, 334 chemical shift anisotropy, 232, 233 chemical vapor deposition, 195 chromium(!!) acetate, 1 CLi 5+, 186 cluster carbonyls, 141 cluster chains, 93 cluster condensation, 60 Co 2Rh 2 (C0) 12 , 246 Co 2 (C0) 2 , 270 Co 2 (C0) 8 , 157 Co 2 (N0) 2 , 270 Co 3Rh(C0) 12 , 246 Co 4 (C0) 12 , 157 Co 4 (C0) 12 , 246 copper-ruthenium clusters, 52

Co-Co bond enthalpy, 239 Cp 2Rh 2 (E0) 2 (EO- CO, NO), 271 cr 2 , 33 Cr 2 (o 2ccH 3 ) 4 , 278 Cs 3 [Re 3cl 12 J, 1 cubeoctahedral intermediate, 158 CufRu system, 209 cyclopropane, 1,1-dilithio, 171 decaosmium cluster, 131 dehydrogenation, 162 delta, 6 ~ 6* transition, 281 delta bonding, 275 diaminomethanes, 78 dimetalatetetrahedrane, 62 dirhenium(II) complexes, 11 dithiocarboxylic acids, 320 dodecahedron, 149 DRIFfS cell, 49 eighteen-electron rule, 265 electron delocalization, 100 electronegativity, 119 electronic microcircuits, 216 electronic structure in transition metal dimers, 326 ensemble, 105 ensemble effect, 56 epoxidation, 165 ethane hydrogenolysis, 213 EXAFS, 106, 133 Fe 2 (C0) 9 , 151 Fe 3 (C0) 12 , 239 Fischer-Tropsch reactions, 76 fluoren-9-one, 64 fluxionality, 148 gallium-arsenic dimers, 200 gallium-phosphorus, 200 germanium clusters, 257 H3Ru4 (C0) 12 , 43 heteronuclear transition-metal clusters, 325 heteropolyoxofluorometalate anion, 311 339

hexalithiobenzene, 171 hexalithioethane, 171 hydride tunnelling, 235 hydrides of Os, Ru, 312 hydroaminations, 122 hydrodesulfurization, 321 hydroformylation catalysis, 328 hydrogenation, 62 indium phosphide, 195 indium phosphide, epitaxial, 203 interpenetrating icosahedra, 262 Ir, [Ir 2 Tl(C0) 2 c1 2 (~·dpma) 2 ] [N0 3 ], 305 Ir, [Ir 2 (SnCl)(C~) 2 cl 2 (~·dpma) 2 ] , 305 Ir, {Ir 2Au(C0) 2Cl 4 (~·dpma) 2 }+, 302 Ir, {Ir 2 (HgC1) 2 (C0) 2 (~·dpma) 2 }, 302 Ir(HgCl)(CO)Cl 2 (PPh 3 ) 2 , 301 Ir-Cu, 110 isocyanides, 62 lanthanide-alkyl bonds, 121 late transition metal ions, 299 Li, (CLi 2H2 ) 4 , 171 Li, (CLi 4 ) 3 , 171 lithium arsenide, 197 lithium phosphide, 197 M2 (o 2ccH 3 ) 4 of chromium, molybdenum and tungsten, 277 magic angle spinning, 233 McMurray reagent, 63 metal carbonyl clusters, 149, 231 metal carbonyls, dinuclear, 238 metallic aluminum films, 216 metallomacrocycles, 299 metal-ligand bond enthalpies, 113 metal-metal bonded arrays, 91 metal-metal multiple bonds, 1' 2' 7' 19 Metal-Metal triple bond, 9 methanation catalyst, 208 methyl formate, 45 mixed-metal (Au-Pt, Ag-Pt) electrocatalysts, 316 Mo, (H0) 4MoMo(PH 3 ) 4 , 26 MO calculations, 272 Mo 2 , 32 Mo 2 (ALLYL) 4 , 315 Mo 2 (o 2 ccF 3 ) 4 , 287 Mo 2 (o 2 CR) 4 , 4 Mo 2 (0CH 2 CMe 3 ) 6 , 289 Mo 2 (PH 3 ) 4 cl 4 , 24 Mo 6 o 12 ~ 97 Mo 8o14 -, 99 molecular beam epitaxy, 195 molybdenum, 55 340

molybdenum ternary oxides, 92 molybdenum(!!) carboxylates, 2 molybdic anhydride, 164 molybednum sulfide, 98 MoW(0 2CC(CH 3 ) 3 ) 4 , 287 Mo-Mo bond lengths, 281 multicenter transformations, 75 Ni 10 , 250 nickel on tungsten, 210 niobium, 19th cluster of, 262 niobium ternary oxides, 92 nitriles, 62 Ni-Au, 105 Ni-Cu, 104 Ni-Cu, 105 octachlorodirehate(III) dianionorganometallicjcovrdination chemi~try, 316 Os 10 c(C0) 24 ·, 129 Os 3 (C0) 9 x 2 ~ 265 Os 5 C(C0) 14 -, 129 Os 6 6(C0) 18 , 76 osmium hydride cluster, 137, 312 osmium sulfur-containing cl~ster, 87 Os-Cu, 110 oxametallacyclobutane, 167 oxidations, 161 oxidative addition, dinuclear, 59 oxides, 161 oxo-alkylidenes, 67 Pd-Ag, 105 Pd-Au, 104 Pd-Au and Pt-Au cluster compounds, 332 pentacoordinate complexes, 234 photoelectron spectroscopy, 275 phthalic anhydride, 167 plasma desorption mass spectra, 329 polycrystalline Al films, 223 polylithium, 171 post transition metal ions, 299 pseudorotation4 141, 144, 148 Pt 2 (~-P 2 o 5 H 2 ) 4 ·, 307 Quaduruple bond, 8 Re 2Br 82- , 2 Re 2c1 4 (PR 3 ) 4 , 28 Re 3c1 9 (dppm) 3 , 11 reactivity of dithioethers toward [Re 2X ] 2 -, 323 reductive e~imination, dinuclear, 59 Re-Cu, 110 Rh, [Rh 2 Pd(C0) 2 c1 3 (~·dpma) 2 ]+, 301, 302 Rh 2 (o 2CR) 2 , 4 Rh4 (C0) 12 , 246 Rh-Cu, 109 Ru 2 (o 2CR) 4 , 4

Ru 3 (C0) 12 , 50 Ru 6C(C0) 16 2 -, 129 RuC1 3 , 48 ruthenium catalyst, 43 Ru-Cu, 106 Ru-Cu, 109 semiconductor clusters, 249 semiconductors, 95 Si +, 261 sii~con clusters, 255 skeletal electron-pair counting, 265 s~ (n- 4-7), 314 soiid state carbon-13, 231 solid state clusters, 91 spin lattice relaxation times, 232 stability of small biclusters, 333 stored waveform inverse Fourier transform, 258 structure sensitive reactions, 105 supersonic cluster beam ICR, 252 systematic kinetic studies of metal carbonyl clusters, 330

Tc 2c1 83- , 2 technetium clusters, 12 tetralithioethylene, 171 tetralithioethylene, 179 tetralithiomethane, 171 tetraruthenium cluster, 77 thin film PE spectrum of Mo 2 (o 2ccH 3 ) 4 , 292 Tl 2Pt(CN) 4 , 307 triangular niobium cluster, 322 triirondodecacarbonyl, 240 trilithiomethane, 171 trinuclear Au2Pt(CH 2 (S)PH 2 ) 4 , 324 triosmium clusters, 77 tungsten, 55

311

Xa-SW calculations, 23, 24, 26 ynamine (aminoacetylene), 85 zeolite cages, 129 zeolites, 50

341