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
546' . 6--d c20 90-34855
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 . . . . . . 1

F. A. Cotton

The Multiple Metal-Metal Bond: Twenty-Five Years of

Synthetic Serendipity and Structural Discovery 7
Richard A. Walton

Theoretical Studies of Dinuclear Compounds with

Multiple Metal-Metal Bonds . . . . . . 19
Bruce E. Bursten and William F. Schneider

Uses of Metal Clusters in Homogeneous and Heterogeneous

Catalysis . . . . . . . . . . . . . . . . . . . . 41
Donald J. Darensbourg

Reactivity of Dinuclear and Tetranuclear Clusters of

Molybdenum and Tungsten . . . . . . . . . . 55
Malcolm H. Chisholm

Clusters and Their Implications for Catalysis . . . . . . . . . . . 75

Richard D. Adams

Metal Clusters in the Solid State . . . . . . . . . . . . . . . . . 91

Robert E. McCarley

CLUSTERS IN CATALYSIS

Nature of Bimetallic Clusters . . . . . . . . . 103

J. H. Sinfelt

Thermochemical Aspects of Organotransition Metal

Chemistry. Insights Provided by Metal-Ligand
Bond Enthalpies . . . . . . . . . . . . . . . 113
Michel R. Gagne, Steven P. Nolan, Afif M. Seyam,
David Stern and Tobin J. Marks

Metal Clusters and Supported Metal Catalysts 127

B. C. Gates

Boldface denotes symposium speaker

vii
 Mechanistic Features of Carbonyl Cluster Rearrangement 141
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 . . . . . . . . . . . . . 161
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 . . . . . . . . . . . . . . . 171
Richard J. Lagow

CLUSTERS IN MATERIALS

Organometallic Chemical Vapor Deposition of GaAs

and Related Semiconductors Using Novel
Organometallic Precursors . . . . . , . . . 195
Alan H. Cowley

Surface Chemistry of Mixed-Metal Systems 205

D. V. Goodman

Organometallic Chemical Vapor Deposition

of Aluminum Nitride and Aluminum Metal 215
David C. Boyd, RichardT. Haasch, Kwok-Lun Ho,
Jen-Vei Hwang, Roland K. Schulze, John F. Evans,
Wayne L. Gladfelter and Klavs F. Jensen

Solid State Carbon-13 NMR of Metal Carbonyls 231

Brian E. Hanson

Surface Chemistry of Metal and Semiconductor Clusters . . . . . . . 249

R. E. Smalley

BONDING AND SPECTROSCOPY IN CLUSTERS

The Electronic Structure of Metal Dimers and

Metal Clusters: The Eighteen-Electron
Rule vs. Skeletal Electron-Pair Counting 265
Michael B. Hall

Experimental Measures of Metal-Metal Sigma, Pi,

and Delta Bonding from Photoelectron Spectroscopy 275
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 299
Alan L. Balch

ABSTRACTS

The Preparation and Characterization of New

Heteropolyoxofluorometalate Anion, [Few 17 o56 F6H5 ] 8 - 311
Sadiq H. Wasfi

viii
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 312
Yun Chi, Sue-Lein Wang and Shie-Ming Peng

Thermal Constants and Structure of Tin Clusters 314

Richard W. Schmude, Jr., Karl A. Gingerich
and Joseph E. Kingcade, Jr.

Reactivity and Isomerization of Mo 2 (ALLYL) 4 315

Reed J. Blau, Ron-Jer Tsay and Su-Inn Ho

Surface Coordination/Organometallic Chemistry of

Monometal and Bimetallic Electrocatalysts 316
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 318
R. T. Carlin and R. A. Osteryoung

The Maximum Strength of the Chemical Bond between

Two Metal Atoms . . . . . . . . . . . 319
K. A. Gingerich

Reactions of (~ 7 -c 7 H 7 )M(~ 5 -c 5 H 5 ), M- Ti or Zr,

with Carboxylic and Dithiocarboxylic Acids 320
S. A. Duraj, M. T. Adras, R. A. Martuch and S. SriHari

Model Hydrodesulfurization Systems: Reactions of

Sulfur Containing Molecules on Ni(llO) 321
D. R. Huntley

Electrochemical Studie~ of Triangular Niobium Cluster,

Nb30 2 (so4 ) 6 .3H2o -, in Sulfuric Acid . . . . . 322
V. Sayers, T. Batten, M. May and V. Katovic

Reactivity of Dithioethers toward [Re 2H8 ] 2 - . 323

J. Gregory Jennings and Gregory L. Powell

Theoretical Investigations of the Metal-Metal

Interactions within the Trinuclear
Au 2Pt(CH 2 (S)PH 2 ) 4 Complex . • • . . . . 324
Andrew L. Sargent and Michael B. Hall

Structural and Theoretical Studies on Heteronuclear

Transition-Metal Clusters Containing the
Alkylidyne Ligand . . . . . . . 325
P. Sherwood, M. B. Hall. J. C. Jeffery and F. G. A. Stone

Electronic Structure and Nature of Bonding in Transition

Metal Dimers • • . . . . . . . . . . . . . . . . . 326
Irene Shim

Theoretical Calculations on the Interaction of Bridging

Carbonyls with Transition Metal Dimers . . . . . 327
Charles Q. Simpson II and Michael B. Hall

ix
 Bimetallic Hydroformylation Catalysis . . 328
Scott A. Laneman and George G. Stanley
252 cF-Plasma Desorption Mass Spectra of Very
Large Clusters . . . . . 329
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 330
N. M. J. Brodie, Lezhan Chen and A. J. Poe

Mixed Pd-Au and Pt-Au Cluster Compounds . 332

Louis H. Pignolet

Stability of Small Biclusters of Transition Metals

with Semi-Conductors . . . . . . . . 333
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 . . . . . . . . . . . . . 334
Preston J. MacDougall and Michael B. Hall

The Effect of Carbonyl Ligands on Osmium and Ruthenium

Metal-Metal Bonds . . . . . . . . . . 335
Ann E. Miller and William A. Goddard III

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. I then wondered wheth-
er it would make sense to seek out more. However, it was not until I
had acquired the knowledge and equipment necessary for X-ray crystal-
lography that I could actually embark on such a program. It also
happened that chance favored a prepared mind. 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
1
ta me.

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
3-
while attempting to explore further the chemistry of the Re 3Cl 12 ion.
2-
We 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
 3 4 5
1844. The earliest reports ' of compounds with Mo-Mo quadruple bonds
appear to be those of several molybdenum(!!) carboxylates (whose struc-
tures 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 con-
3-
taining 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
2- 2- 8
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. This is sum-
marized in Fig. 2. 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

I
120
110
100
90 ~
I
0
80 ti
70 ~
m
60 ~
50 LL
0
40 ffi N = e xp LQI57(Y-1952)]

~~.~ 1965 1970

YEAR
1975 1980 1985

Fig. 1. The approximate number of publications on all types of

M-M multiple bonds each year, 1964-1988.

200

Mo
100 w
80 Rh
Re
60
40 Ru
pt
Os
20

10
Tc
5

To 65 "66-'70 71-75 76-80 "81-85 86-90 (est)

Fig. 2. 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 be-
cause 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<c structures. 10
L 0 f course, t h ere are even greater problems
for the "simple" diatomic Cr 2 itself, and for neighboring ones such as
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, ~.

512.
6. H. Blondel, Ann. Chim. Phys., 1905, !L 110.
7. J. D. Eakins, D. G. Humphreys and C. E. Mellish, J. Chern. Soc.,
1963, 6012.
B. V. G. Tronev and S. M. Bondin, Dokl. Akad. Nauk SSSR, 1952, 86, 87.

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 characteriza-
tion of the [Re3X12l3- (!) and [Re2Xs] 2 - (!) anions, and the recognition
that these species (and closely related ones) contain Re-Re bonds of

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 metal-
metal 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 7

Edited by J. P. Fackler, Jr.
Plenum Press. New York, 1990
 o=o M=o M 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

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 some-

thing 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-dithia-
hexane (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 unsymmet-
rical 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 metal-
metal 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.
Re~ Re). This point will be considered again a little later. Suffice it
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 com-
plexes ([Re2XsJ 2 -, Re2(02CCH3)4X2 (X= Cl or Br), [Mo2Clg] 4 - and Mo2(02-
CCH3)4). These were the first measurements of the v(M-M) modes of metal-
metal multiple bonds. Although my contribution to this work was completed
by early 1969, the results of this study were not published until mid-
1971.12 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

3 4
QRe Qc1 or Br Q)py

10
 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 config-
uration 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 0

0 0
-5
-
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 five-
membered metallacycle ring can best be rationalized in terms of contri-
butions 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 tem-

plate 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 2,5-
dithiahexane 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 com-
pounds 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 dispro-

portionation 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 <z). 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. 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.
2. 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.
3. F. A. Cotton and M. H. Chisholm, Chemical & Engineering News, 60:40
(1982).
4. There are many excellent reviews on metal cluster chemistry. See,
for example, G. Schmid, Structure and Bonding (Berlin), 62:51
(1985).
5. 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.
6. 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).
7. M. L. Listemann and R. R. Schrock, Organometallics, 4:74 (1985).

16
 8. F. A. Cotton and R. A. Walton, Inorg. Chern., 5:1802 (1966).
9. F. A. Cotton, C. Oldham and R. A. Walton, Inorg. Chern., 6:214 (1967).
10. M. J. Bennett, F. A. Cotton and R. A. Walton, J. Am. Chern. Soc.,
88:3866 (1966).
11. M. J. Bennett, F. A. Cotton and R. A. Walton, Proc. Roy. Soc.,
303:175 (1968).
12. W. K. Bratton, F. A. Cotton, M. Debeau and R. A. Walton, J. Coord.
Chern., 1:121 (1971).
13. R. Colton, R. Levitus and G. Wilkinson, J. Chern. Soc., 4121 (1960).
14. D. G. Tisley and R. A. Walton, Inorg. Chern., 12:373 (1973).
15. H. D. Glicksman and R. A. Walton, Inorg. Chim. Acta, 19:91 (1976).
16. See for example, V. I. Spitsyn, A. F. Kuzina, A. A. Oblova and S. V.
Kryuchkov, Russ. Chern. Revs., 54:373 (1985).
17. R. A. Wheeler and R. Hoffmann, J. Am. Chern. Soc., 108:6605 (1986).
18. See ref. 1, p 275.
19. F. A. Cotton and J. T. Mague, Inorg. Chern., 3:1094 (1964).
20. S.M.V. Esjornson, P. E. Fanwick and R. A. Walton, unpublished results
(1988).
21. J. R. Ebner and R. A. Walton, Inorg. Chern., 14:1987 (1975).
22. P. Brant, D. J. Salmon and R. A. Walton, J. Am. Chern. Soc., 100:4424
(1978).
23. F. A. Cotton, K. R. Dunbar, L. R. Falvello, M. Tomas and R. A.
Walton, J. Am. Chern. Soc., 105:4950 (1983).
24. F. A. Cotton, B. A. Frenz, J. R. Ebner and R. A. Walton, Inorg.
Chern., 15:1630 (1976).
25. T. R. Ryan and R. E. McCarley, Inorg. Chern., 21:2072 (1982).
26. R. A. Walton, A.C.S. Symp. Ser., 155:207 (1981) and references cited
therein.
27. R. A. Walton, Polyhedron, in press (1989).
28. A. C. Price and R. A. Walton, Polyhedron, 6:729 (1987) and references
cited therein.
29. F. A. Cotton, K. R. Dunbar, A. C. Price, W. Schwotzer and R. A.
Walton, J. Am. Chern. Soc., 108:4843 (1986).
30. F. A. Cotton, K. R. Dunbar, L. R. Falvello and R. A. Walton,
Inorg. Chern., 24:4180 (1985).
31. 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).
32. P. E. Fanwick, A. C. Price and R. A. Walton, Inorg. Chern., 27:2601
(1988).
33. T. J. Barder, F. A. Cotton, L. R. Falvello and R. A. Walton,
Inorg. Chern., 24:1258 (1985).
34. D. Esjornson, P. E. Fanwick and R. A. Walton, Inorg. Chern., 27:3066
(1988).
35. D. Esjornson, D. R. Derringer, P. E. Fanwick and R. A. Walton,
Inorg. Chern., submitted for publication.
36. M. Bakir, D. Esjornson, P. E. Fanwick, K. S. Jones and R. A. Walton,
unpublished results (1989).
37. A. R. Chakravarty, F. A. Cotton, A. R. Cutler and R. A. Walton,
Inorg. Chern., 25:3619 (1986).
38. See ref. 1, p 49.
39. M. H. Chisholm, J. C. Huffman and W. G. VanDer Sluys, J. Am. Chern.
Soc. ,109:2514 (1987).
40. 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 Cl iz-
1 ,,....,_Cl 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 19

Ediced by J.P. Fackler, Jr.
Plenum Press, New York, 1990
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

(a) (b) (c) (d)

E

a9(z2)

Xx M-L<J

D
D D
M-X<J
Xx

D M-X<J

D
M
x•... M .••• x X,,,,M,,,,L
x-' ""x L,. ""x
Figure 1. 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).

20
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 octachloro-

dirhenate(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 a-
only 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

a 19(a)

....,X I ,...,X
X X
IM·~M···
x/1 x/1
X X
Figure 2. Qualitative molecular orbital diagram of the interaction of
two MX4 fragments to generate the four components of an M-M
quadruple bond.

22
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 non-
bonding, 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 semi-
staggered 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 metal-
ligand 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 P p

I~ \~
L X X L

M:::':__M··'' ' ' l. .~x

M--M

x~lL L~lX J\x;l

p~_,l
2 3 4

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 metal-

metal 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
Cl PMe3

M~~Mo.•

MeJP
~ICI CI;1PMe3
5 6

24
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

-4.0
5b2(11)
·3.0
5tJ:! (yz)
5b1 (11)
~-I'•F'T'"T'~(8)

7a 1 (a) -5.0

-4.0 5b1 (xz)

7a 1 (z9
-6.0

-5.0

-7.0

-6.0

·7.0 7a 1 (a)
4a1 (a)

-8.0
HO PH
1,..-PH•I.J>H 9_.,,PH,IPH}:.I
Mo...i-Mo Mo...i-Mo

HO OH'IPH, H,J>'Cl
H,P"I :l Cl.,1PH,
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 (.i-
Pr0)4MoMo(dmpe)z (8; dmpe- 1,2-bis(dimethylphosphino)ethane).z4

n :: :-
I *,p I . ,p I
RO PPh3 OR
,.oR
......~
...... ~

' M"a-3-Mi,.
.....:..; 4
Re--Re
Cl~~ Cl~~PPh3 '.~pOR
. .,.IOR
RO '
PJ
7 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
H~
""'illil ~...
M6"-·-M~. .
(eV) ,I ~1
Hy> Hy> OH OH
-1.0
10e(7t")

-2.0

-3.0

-4.0
4a, (z2j

-5.0

-6.0

-7.0 7a1 (a) . ~oH

.
7a1 (a)
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
L
I
L

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 metal-
based 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
 (a) (b) (c)
E

0
M-Lo
M-Lo

D
M"'+ L'"'"·M-L
L_.... L~7M......_L
Figure 5. 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 ) metal-
based 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 L L L L L
I
M I \"'
M-'-l
,~
!
M
I\L
L L
....
'i.
I
L
I
L

Figure 6. Qualitative molecular orbital correlation diagram showing the

interaction of two planar Mla fragments to generate a metal-
metal 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
Mo--Mo
3 ''
OH OH HO
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

(aV)
-4

-5

-6

Mo-MoCJ Mo-MoCJ
-7

-8

D
D D
01t
-9

-10

-11
C-HCJ
Mo-NCJ
D
Mo-OCJ

10 11 12
Figure 7. 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
d,..2-y2,
1rinteractions 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.

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)

-1

-2

-3

-4

-5

Mo

Figure 8. 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).

32
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 Hartree-
Fock 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 bond-
centered 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: 1x( •2_ 3ll, f y( y2- 3x2]

Figure 9. Schematic of the possible diatomic interactions between f

orbitals on adjacent atoms. (From ref. 48).

34
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 H1C CH1 CH1 CH1 CH1
I . u--u
I l" ,~
I
ru
H1C H,'t
~
u,c
....~
H,'t
I
en,
u
I
CH,

Figure 10. 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, ~. 330-333.

(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, ~. 334-
336.

36
(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 w-
donating 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, ~. 271-176.

(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, 99-
104.

(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, 510-
512.

(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)

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, 7736-
7738. (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. (b) Kok, R.
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
(42) Goodgame, M. M.; Goddard, W. A. G., III J. Phys. Chern. 1981, 85, 215-
217.

(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 1979, 18, 1507-1509.

(53) VanDer Sluys, W. G.; Burns, C. J.; Huffman, J. C.; Sattelberger, A.

P. J. Am. Chern. Soc. 1988, 110, 5924-5925.

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 Homogeneous
(Heterogeneous Catalysis) Catalysis

Scheme 1

Reactivity Patterns Structural Models

on Metal Surface
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 41

Edited by J.P . Fackler, Jr.
Plenum Press, New York, 1990
 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 mix-
tures 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 190-
2600C 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

1
(Scheme 3) involving separately the tetranuclear and trinuclear ruthenium clusters,

H2 Ru3(C0)12 " \ H -

CO C0 2

Schfme3

42
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 organo-

metallic 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).

,0
[MtH + C02 [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 ligand-
assisted 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

HCOOMe j
+ 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 con-
verting 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.

COz + Hz + MeOH HCOOMe + HzO (4)

MeOH + CO - HCOOMe (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 carbony-
latlon 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 PH2 !Co turnovers8

!Co2

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 250 250 100 4.1
HR11a(C0f1 250 106

8 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 equa-
tion 7. However, unlike the mononuclear group 6 metal carbonyl hydrides which read-
ily 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 cat-
alytic 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 rear-
rangement required for the formation of A. This is seen upon modelling the decarboxy-
lation 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
o, . .
/c, /c\ (8)
I ~
Ru--Ru
~
Ru--Ru
A B

This operation places the formate hydrogen atom much closer to the metal center adja-
cent 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 (9)

46
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
,c,
(1) 0 H :
I ,.; ::
!
HCr(CO); + C02 ~ -Cr-
II""' I (5) :
........ -- --.--------------------------------------------- ------ ~

l
Scheme7

(2) (3) (4)

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 methoxy-
carbonyl 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 tempera-
ture. 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, alu-
mina, 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 clus-
ters 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 pro-
posed to be [Ru(C0)2X2ln. where X is a surface oxygen atom of alumina and n is un-
known.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 180°C

RuClj3Hp 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
I
I
I
B
I
10 I
I
::> I

..,
a:
I
I
I

~ 50 I
-.. I
I

~ I
I

~
I
40 I
w I

>
~
u 20
~

200 2~ 300 3110 400

TEMPERATURE I"Cl
Figure 4. Catalytic activity of catalyst precursor prepared by VP
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

Figure 5. Catalytic activity of catalyst precursor prepared by high

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

24
a::" A
"'
o-
w 20
(D
a::
0
(fJ
0 16
<{

0
u B
(fJ 12
w
_j
0
:::;:
':'
Q
><

10 15 20 25
PRESSURE (mm Hgl

Figure 6. CO chemisorption results for high pressure synthesized catalyst: a)He-

activated 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 in-
trazeolite 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 ruthe-
nium 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

1.) E. L. Muetterties, Bull. Soc. Chim. Bel~ .. 84, 959 (1975); .!b!d... 85, 451 (1976).

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

52
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
(1983). (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.
: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
oxidative addition and reductive elimination sequences. Principles
pertaining to the design of a catalytic system are discussed in light
of the above and specific attention is given to the hydrogenation and
isomerization of alkenes by W4 (H) 2 (OPri) 14 . Recent studies of the
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
further reaction with aldehyde or ketone. A pinacol coupling is ruled
out. Recent studies of the activation of carbvn monoxide by W2 (0R) 6
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
compounds for studies of reactivity. 1. The metal atoms provide a
template for the assembly of substrates . 1 2. The metal-metal bonds
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
centers. 2 3. The presence of two or more metal atoms, especially in
cases where the metal atoms are different, raises the possibility of
multisite activation of substrates. 3 4. The binding of organic
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
"ensemble effect" a reaction having a requirement for a specific
number of metal atoms. 5

Metal-Metal Bonds and Clusters in Chemistry and Catalysts 55

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
 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 square-
planar 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)

(b) (1)

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 1,2-
diorganoderivatives 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, 2-

dimetallacyclohex-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.

22"C
(i)
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
hydride derived from the R'OH hydrogen atom. Support for this general
scheme came from the alcoholysis reaction shown in eq. 7 where the
Mo~Mo alkoxide was isolated and subsequently structurally
characterized. Presum~bly in reaction 7 the bulky alkene, Me 2C-CH 2
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
(L - py, PMe 3 ). 38 It also reacts with ethylene to give Mo 2Et(OPri) 5
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
reactivity with M - W but not with M - Mo. Our first encounter with
this type of facile oxidative addition to the (W=W) 6+ center was during
studies of alcoholysis reactions of W2 (NMe 2 ) 6 • Addition of bulky
alcohols such as ButOH gave W2 (0But) 641 but the less sterically
demanding alcohols with more acidic hydroxyl protons MeOH and EtOH gave
tetranuclear compounds W4 (OR) 16 , where R - Me and Et, contain-
ing tungsten in oxidation state +4. 42 No intermediates were detected
in alcoholyses reactions employing W2 (NMe 2 ) 6 and ROH where R - Et or
Me. The reactions proceed very rapidly at room temperature. Studies
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 of

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
(9)

When reaction 9 was carried out at 0°C the initially formed

products were W2 (0Pri) 6 (HNMe 2 ) 2 and the carbido cluster
W4 (C)(NMe)(OPr 1 ) 12 . The latter compound is formed inca 5-10% yield
and is apparently a product of a degradation reaction of a coordinated
NMe 2 ligand. 44 The mechanism leading to W4 (C)(NMe)(OPri) 12 is not
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 1,2-

dimethoxyethane (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 re-
evaluation 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
equivalent 183 W nuclei : 1 J 1 aaw_ 1 H "' 100 Hz . 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 toluene-
d8. 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 dihydrido-
tetratungsten complex is a catalyst for olefin isomerization, eq . 13,
and, in the presence of both olefins and dihydrogen, olefin

W4 (H) 2 (OPri) 14
a-olefins ------------------------~..
~ internal olefins (13)
hexane/benzene, 22oc

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
(15)
hexane ...

Walton and his coworkers have studied extensively the cleavage of

M~+ units in their reactions with isocyanides, e.g. eq. 16. 51

22·c
MeOH, PF 6 -
... 2Mo(CNR)~+ (16)

Schrock and coworkers were the first to observe a metathesis of

-----1··
M=M and C=C bonds in the cleavage reaction shown in eq. 17. 52 • 53

o·c
hexane 2(But0) 3W=CR
( 17)
R- Me, Et, Pr 1

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 and
Aldehydes

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
(19)
hexane

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 W-
C(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 Cross-

Coupling of Ketones/Aldehydes by W2 (0CH 2 But) 6py2 .a

Aldehyde/Ketone Olefin

51%

21%

o-CHO 3 34%

0< 4 44%

66%
5

b
6 36%

~b
~H H
7 18%

a 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 points. The isolated yields given in Table 1 are not
optimized. From following the course of various reactions by NMR
spectroscopy employing 13 C labelled ketonic carbon atoms the idealized
yield should 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 cyclo-
propenones.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. Bridging Carbonyl Adducts of Molybdenum and

Tungsten Alkoxides.

Compound 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 305

66
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

0 o-
1
~ c
/"-..
M=M
+
M- '
/- M

22•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
support the qualitative description presented above. The CO ~*
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 + c..o

0
g
/""-
M=M

(22)

(ii) -+

The previously described reduction of ketones to oxo-alkylidenes

by W (OR) compounds led us to think that the reaction between
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./
0 .·
r l '?l f' i·.
o::::::._~0 ·w ,./ o
j~'\??'w
wtz:J-w·/

366.5 365.5 364 5 349 348 347 346

ppm ppm

Fig . 2. 13 C 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.

68
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
metal atoms of the butterfly. For a certain W4 (JJ- 4 -C) moiety lacking
any element of symmetry there are two large, W-wingtip-C, and two
small, W-backbone-C, values of 1 Jtasw_13c· For a W4 (C)(O)(OR) 12
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

other compounds formed in reaction 24. The implication was, however,
that it took two W=W centers to cleave carbon monoxide. So why not
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
the temperature range -78"C to 22"C. Care must be taken to limit the
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
•
(25)
hexane

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
compounds. We have as yet not been able to obtain a suitable single
crystal for an X-ray diffraction study. Typically, the crystals showed
diffraction to only small angles as was found for W4 (OCH 2R) 12
compounds. This is probably a result of hexagonal close-packed
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
the type characterized by Shriver? 74 To address this problem Charles
Hammond 75 employed further 13 C NMR spectroscopic studies using a 60:40
tsctGo: tsctso mixture. Using high field and high resolution 13 C NMR
spectroscopy there is a readily measurable isotope chemical shift
effect. The 13 C resonances for the mixture of 13 C16 0: 13 C18 0 employed
in these studies are shown in Figure 3. The chemical shift separation
is 0.050 ppm. The compounds W2 (0But) 6 (JJ.-CO), W4 (JJ.-C0) 2 (DCH 2But) 12
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~

184.550 184.500
PPH

Fig. 3. 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.

70
 0
II
/c""
W=W

291.0 290.5 290.0 313.40 313.30 313.20 366.0 365.5 365.0

PPM PPM PPM

Fig. 4. 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
MHz. (a) W2 (0-t-Bu) 6 (JL-CO), (b) the central resonances (no
satellites due to 1J1a 3w_1 3c are shown) of; [W2 (JL-CO)(OCH 2 -t-
Bu)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. It is no accident that in nature the Fe 4 S4 cluster in ferredoxin
serves as an electron transfer reservoir.
3. In the limiting case this can be taken to be a dinuclear early-
late 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).
4. J. T. Yates and M. R. Albert, in: "The Surface Scientists Guide
to Organometallic Chemistry," ACS Publishers, Washington,
D.C. (1987).
5. 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).
6. Y. Doi, K. Koshizuka, and T. Keii, Inorg. Chem. 21:2732 (1982).
7. G. SUss-Fink, J. Organomet. Chem. 193:C20 (1980).
8. G. SUss-Fink and J. Reiner, J. Mol. Catal. 16:231 (1982).
9. J. A. Smiejla, J. E. Gozum, and W. L. Gladfelter, Organometallics
5:2154 (1986).
10. G. SAs-Fink, Angew. Chem. Intl. Ed. Engl. 21:73 (1982).
11. 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).
12. J. L. Vidal and R. C. Schoening, Inorg. Chem. 21:438 (1982).
13. G. C. Demitras and E. L. Muetterties, J. Am. Chem. Soc. 99:2796
(1977).
14. H.-K. Wang, H. W. Choi, and E. L. Muetterties, Inorg. Chem.
20:2661 (1981).
15. J. C. Bricker, C. C. Nagel, and S. G. Shore, J. Am. Chem. Soc.
104:1444 (1982).
16. J. C. Bricker, C. C. Nagel, A. A. Bhattacharayya, and S. G. Shore,
J. Am. Chem. Soc. 107:377 (1985).
17. M. W. Payne, D. L. Leussing, and S. G. Shore, J. Am. Chem. Soc.
109:617 (1987).
18. P. C. Ford, Ace. Chem. Res. 14:31 (1981).
19. R. D. Adams, H. Kim, and S. Wang, J. Am. Chem. Soc. 107:6107
(1985).
20. R. D. Adams, J. E. Babin, and M. Tasi, Inorg. Chem. 25:514 (1986).
21. See also Chapters by R. D. Adams and W. L. Gladfelter in this
volume.
22. 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).
23. For a novel low temperature route to MoW alloys see A. K.
Cheetham, Nature 288:469 (1980).
24. A. B. Brignole and F. A. Cotton, Inorg. Syn. 13:81 (1972).
25. M. H. Chisholm, D. A. Haitko, and C. A. Murillo, Inorg. Syn. 21:51
(1982).
26. M. H. Chisholm, D. A. Haitko, and J. C. Huffman, J. Am. Chem. Soc.
103:4046 (1981).
27. 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).

72
28. M. H. Chisholm, M. J. Hampden-Smith, J. C. Huffman, and K. G.
Moodley, J. Am. Chem. Soc. 110:4070 (1988).
29. M. H. Chisholm, M. J. Hampden-Smith, K. A. Stahl, J. C. Huffman,
J. D. Martin, and K. G. Moodley, Polyhedron 7:1991 (1988).
30. M. H. Chisholm and M. W. Extine, J. Am. Chem. Soc. 99:782 (1977);
idem, ibid, 99:792 (1977).
31. M. J. Chetcuti, M. H. Chisholm, K. Folting, D. A. Haitko, and J.
C. Huffman, J. Am. Chem. Soc. 104:2138 (1982).
32. J. K. Kochi in "Organometallic Mechanisms and Catalysis,"
Academic Press, Chpt. 11 and references cited therein (1978).
33. M. J. Chetcuti, M. H. Chisholm, K. Folting, J. C. Huffman, and J.
Janos, J. Am. Chem. Soc. 104:4684 (1982).
34. 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).
35. M. H. Chisholm, H. T. Chiu, and J. C. Huffman, Polyhedron 3:475
(1984).
36. M. H. Chisholm, D. L. Clark, J. C. Huffman, and W. G. Van Der
Sluys, J. Am. Chem. Soc. 109:6817 (1987).
37. M. H. Chisholm, and R. J. Tatz, Organometallics 5:1590 (1986).
38. M. H. Chisholm, K. Folting, J. C. Huffman, and R. J. Tatz, J. Am.
Chem. Soc. 106:1153 (1984).
39. 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).
40. M. H. Chisholm, C. C. Kirkpatrick, and J. C. Huffman, Inorg. Chem.
20:871 (1981).
41. 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).
42. M. H. Chisholm, J. C. Huffman, C. C. Kirkpatrick, and J. Leonelli,
J. Am. Chem. Soc. 103:6093 (1981).
43. M. H. Chisholm, J. C. Huffman, and C. A. Smith, J. Am. Chem. Soc.
108:222 (1986).
44. M. H. Chisholm, D. L. Clark, J. C. Huffman, and C. A. Smith,
Organometallics 6:1280 (1987).
45. K. J. Ahmed, M. H. Chisholm, K. Folting, and J. C. Huffman, J. Am.
Chem. Soc. 108:989 (1986).
46. M. H. Chisholm, D. L. Clark, K. Folting, J. C. Huffman, and M. J.
Hampden-Smith, J. Am. Chem. Soc. 109:7750 (1987).
47. M. H. Chisholm, D. L. Clark, and M. J. Hampden-Smith, J. Am. Chem.
Soc. 110:574 (1989).
48. 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).
49. (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).
50. M. H. Chisholm, F. A. Cotton, M. W. Extine, and R. L. Kelly, J.
Am. Chem. Soc. 101:7645 (1979).
51. R. A. Walton, ACS Symp. Ser. 155:207 (1981) and references
therein.
52. R. R. Schrock, M. L. Listemann, and L. G. Sturgeoff, J. Am. Chem.
Soc. 104:4291 (1982).
53. R. R. Schrock and M. L. Listemann, Organometallics 4:74 (1985).
54. M. H. Chisholm, D. M. Hoffman, and J. C. Huffman, J. Am. Chem.
Soc. 106:6794 (1984).
55. 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. M. A. Harmer, T. R. Halbert, W. -H. Pan, C. L. Coyle, S. A. Cohen,
and E. I. Stiefel, Polyhedron, 5:341 (1986).
57. T. P. Blatchford, M. H. Chisholm, K. Folting, and J. C. Huffman,
J. Chern. Soc., Chern. Commun. 1295 (1984).
58. 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).
59. F. A. Cotton, R. A. Walton, D. DeMarco, and L. R. Falvello, J. Am.
Chern. Soc. 104:7375 (1982).
60. M. H. Chisholm and J. A. Klang, J. Am. Chern. Soc. llO:xxx (1989).
61. R. R. Schrock, J. Am. Chern. Soc. 98:5399 (1976).
62. T. C. Bryan and J. M. Meyer, J. Am. Chern. Soc. 109:7213 (1987).
63. M. H. Chisholm, D. M. Hoffman, and J. C. Huffman, Inorg. Chern.
22:2903 (1983).
64. M. H. Chisholm, J. C. Huffman, J. Leonelli, and I. P. Rothwell, J.
Am. Chern. Soc. 104:7030 (1982).
65. R. C. Benson, W. H. Flygare, M. Oda, and R. Breslow, J. Am. Chern.
Soc. 95:2772 (1973).
66. E. M. Kober and D. L. Lichtenberger, J. Am. Chem. Soc. 107:7199
(1985).
67. 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).
68. D. F. Shriver and C. P. Horwitz, Adv. Organometal. Chem. 23:219
(1984).
69. M. H. Chisholm, Polyhedron 2:681 (1983).
70. P. J. Blower, M. H. Chisholm, D. L. Clark, and B. W. Eichhorn,
Organometallics 5:2125 (1986).
71. J. S. Bradley, G. B. Ansell, and E. W. Hill, J. Am. Chem. Soc.
101:7417 (1979).
72. M. H. Chisholm, K.· Felting, C. E. Hammond, J. C. Huffman, and M.
J. Hampden-Smith, J. Am. Chem. Soc. 110:3314 (1988).
73. M. H. Chisholm, C. E. Hammond, and M. J. Hampden-Smith, J. Am.
Chem. Soc., submitted.
74. 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.
75. For a review for 13 C18 0 isotope shifts see: P. E. Hensen, Ann.
Rep. NHR Spectrosc. 15:105 (1983).

74
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 H
'c
l
R R
c/
' /
u/ 'u

,(~\-/) M
M---M
DI

A B

Metal-Metal Bonds and Clusters in Chemistry and CatalysiS 75

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
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
\ "
'

' /\"t' /
I
c--c ' Os--Os
I

-l~~:}Y\-
llO"C
-~~~'l\' (I)

,,;'
........
...... 'Y . . . .
-CO R
,c,l /c'
,' W R

H Cp
Cp

' / \ "

Os 6(C0) 11 + PhC- CPh

hv

-2 co
''?f~'fJ
Os-
.' . . I, '
/ 'I __
'-- c,
- -Os
( 2)
_os~ v i s , Ph
I c
I
Ph

,...cu,tol
o--c
\
'os-1
~~W~p\
/1\. · llO"C (3)
"" I ". / '\ .,.... -2 co
-Os---os
I\ I'

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 [p-
C(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/ R
Os 1
,\ /I"-. l_.c~H
Os Os~ \\ R
/1 /I ~c~
H
us•c j-co
\/ l
-os-c
,\/
Os
k~ 1\..-a
os- '
/I !\" H
R=H;
-CO
1
\
\co
R:alkyl

o._
...._I_.
//\"'-~
,\os:----..... /
___ o._
/'\,H,/"
c-c
a"' 'a
2 3

Figure 1. 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 (p-
PPh2)(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 I
)t::_
P,

'I
(4)
f /~I, so•c I/
/Ru:..-
:R~<. iu~ -c,n, '\ / Ru~
I 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. 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.

78
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
Me Me
'le--N\/ H \/
'_.-N
' ,_
/o'~s/"- --"1\,. \----~05-
l~n~// -os-
(5)
H~
/,,Os
/i'
7
6

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.

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 3-
sc6H4)(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
activation on the phenyl ring of the benzenethiolato
ligand at the decarbonylated osmium atom route b.
The carbene ligand in 9 was not transformed in the formation of
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 ,
13. 19 The replacement of fluorine atoms for the hydrogen atoms
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)(p-

H)2' 14. Reprinted with permission from J. Am.
Chern. Soc., 1987, 109, 1414. Copyright (1987)
American Chemical Society.

83
process. 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
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 metal-
metal 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 [p-
C(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-...........___ /

H\v !'
Os--C
/H?,oi
12s•c

/\ '\-R
I
R

10 c 14
Scheme III

84
 C • H Activation in NEt3

\ HI NEt 2
' / II
Os-e-e
'1/,\/
os........_ /H 'u
,..., I -.. . . . . II
Os
/I'

Et 2N
' e-e\//
,
H

I_.._.. /,.....u'
I" __
{)s-
',.....os~
os
u-- /\'

17 16

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(p-
H)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(diethyl-
amino)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(p-
H)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) Adams, R. D.; Horvath, I. T. Prog. Inorg. Chern. 1985, 33,

127.
2) Muetterties, E. L.; Krause, M. J. Angew. Chern. Int. Ed.
Engl. 1983, 22, 135.
3) Muetterties, E. L. Catal. Rev. Sci. Engl. 1981, 23, 69.
4) Muetterties, E. L. Pure Appl. Chern. 1978, 50, 941.
5) Laine, R. M. J. Mol. Catal. 1982, 14, 137.

87
6) Ugo, R.; Psara, P. J. Mol. Catal. 1983, 20, 53.
7) Ford, P. c. Accts. Chern. Res. 1981, 14, 31.
8) 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.
14) a) Bergman, R. G-i Seidler, P. F •; Wenzel, T. T. J. Am.
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) Shvo, Y.; Laine, R. D. J. Chern. Soc., Chern. Commun.
1980, 753.
b) 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. It is readily seen that M-M bonding should be
favored in a compound~ when the radius of X is small, the X/M ratio n

Metal-Metal Bonds and Clusters in Chemzstry and Catalysts 91

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
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 ~Mo 4 o 6 which
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 ~G"<O

for Ta and W, and ~G">O for Nb and Mo. 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) 14.5

MM'Mo ob M - Zn, Fe 14.5-14.75

4 7
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 edge-
shared 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

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 8-
coordinate (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. For
example resistivity measurements on InMo 4o621 and Sn0 . 9Mo 4 o6 23 show
typical metallic behavior, while ZnMo 8o10 25 shows the resistivity-
temperature dependence of a semiconductor with Egap - O.OSeV over the
range T < 200K. Above 200K the latter compound shows conductivity typical
of a metal. 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
formulation (Ins7+) (In 6B+>< Mo 18 o28
7->< Mo 22 oa-
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. From the Pauling bond order
sum for all Mo-Mo bonds in the cluster unit, ~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. Also the possibility of
forming other compounds ~Mo 8 o 14 , with M 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. ~· Soc. Lond., A308, 141 (1982).
3. A. Simon, Ang:ew. Chern. Int. ,M. ~. ZQ, 1 (1981).
4. J.D. Corbett, Ace. Chern. Res. 14, 239 (1981).
5. C. Perrin, S. Ihmaine and M. Sergent, New~. Chern. !l, 321 (1988).
6. A. Perrin and M. Sergent, New~. Chern. 12, 337 (1988).
7. A. Simon, Ang:ew. Chern. Int. Ed. Engl. 1J.., 159 (1988).
B. A. Perrin, C. Perrin and M. Sergent, ~. Less-Common~. 137, 241
(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
(1979).
20. C.C. Torardi and R.E. McCarley, ~. Less-Common Met. 116, 169
(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, J. Solid State Chern. 11. 407
(1988).
29. S.J. Hibble, A.K. Cheetham, A.R.L. Bogle, H.R. Wakerley and D.E.
Cox, J. Am. Chern. Soc. 110, 3295 (1988).
30. P. Gougeon, C.D. Carlson and R.E. McCarley, to be published.

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 103

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
 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

2~. and E is the kinetic energy of the photoelectron. In the treatment
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
Ru-cu
2o-.------.,

0.2

"">< 0.0
"'
0 10
"" X
-0.2 :;
C<:
0
LJ..
Vl
-0.4 z
<[
C<:
1--
2o-.-------.
LJ..
0
0.2 w
0
=>
1-
~ z
..... 0.0
l:)
<[ 10
>< :;

""
-0.2 v
0
2 4 b 8 10
K, _A-1
12 14 1b 0 2
.
R, A
4 b

Fig. 2. 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

-0.50

-
>l
.....
':- 0.4
"' Cu-Ru
0.2

0.0

-0.2

-0 .44---.--,.......,......,..--.-"'T"'...,.....T""""""'~~--.-...,.....-j
2 4 6 8 10 12 14 16
K, A-1
Fig. 3. 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. Sinfelt, J.H. Chern. and Eng. News 1972 (July 3 issue), 50, 18.
2. Sinfelt, J.H. J. Catalysis 1973, 29, 308.
3. Sinfelt, J.H., Ace. Chern. Res. 1977, 10, 15.
4. Sinfelt, J.H. Bimetallic Catalysts: Discoveries. Concepts. and
Applications; Yiley: New York, 1983.
5. Sinfelt, J.H. Ace. Chern. Res. 1987, 20, 134.
6. Sinfelt, J.H. U.S. Patent 3.953.368; 1976.
7. Cotton, F.A. Inorg. Chern. 1965, ~. 334.
8. Spenadel, L.; Boudart, M. J. Phys. Chern. 1960, 64, 204.
9. Sinfelt, J.H.; Yates, D.J.C. J. Catalysis 1967, ~. 82.
10. Yates, D.J.C.; Sinfelt, J.H. J. Catalysis 1967, ~. 348.
11. Yilson, G.R.; Hall, Y.K. J. Catalysis 1970, 17, 190.
12. Sinfelt, J.H. Annual Review of Materials Science 1972,z. 641.
13. Hildebrand, J.H.; Scott, R.L. The Solubility of Nonelectrolytes, 3rd
ed.; Reinhold: New York, 1950; p.335.
14. Schwab, G.M. Disc. Faraday Soc. 1950, ~. 166.
15. Dowden, D.A.; Reynolds, P. Disc. Faraday Soc. 1950, ~. 184.
16. Couper, A.; Eley, D.D. Disc. Faraday Soc. 1950, ~. 172.
17. Best, R.J.; Russell, W.W. J. Amer. Chern. Soc. 1954, 76, 838.
18. Hall, ~.K.; Emmett, P.H. J. Phys. Chern. 1959, 63, 1102.
19. Hall, W.K.; Emmett, P.H. J. Pbys. Chern. 1958, 62, 816.
20. Mott, N.F.; Jones, H. Theory of the Properties of Metals and Alloys;
Oxford University Press: Oxford, 1936.
21. Sinfelt, J.H.; Barnett, A.E.; Dembinski, G.W. U.S. Patent 3,442.973;
1969.

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/calori-
metric 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 (3)

Metal-Metal Bonds and Clusters in Chemistry and Catalysis 113

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
 AHrx - D(LnM-R) + D(H-X) - D(LnM-X) - D(R-H) (4)

LnM-R + X2 - - > LnM-X + RX + AHrx (5)

AHrx - D(LnM-R) + D(X2) - D(LnM-X) - D(R-X) (6)

advantages of titration calorimetry include the ability to monitor reac-

tion stoichiometries in situ as an additional check on the analysis chem-
istry 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 desir-
able 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 oxida-
tion state.6,8-10 While this transferability approximation seems reason-
able, 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-R + x 2 - - > LnM-X + RX .6Hrx(5) (7)

LnM-X ---> LnM + 1/2 x2 .6Hrx(8) (8)

X• ---> 112 x 2 1/2 D(X-X) (9)

R-X ---> R• + X• D(R-X) (10)

LnM-R ---> LnM + R• D(LnM-R) (11)

,
for (Me3SiC5H4 )3U-Rll and Cp2Ln-R12,13 complexes (Cp '15-Me5C5; Ln- Sm,
Eu, Yb).

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 corre-
sponding 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 transfer-
ability.

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

• Cp',SmX
0 x,smx

Cl Br
Fig. 1. 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
ep 20

0
Sm Eu Yb

Fig. 2. 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 115
'til Br
•
"
.!< 105 aS"Pr
H
xI 95 • I
;:::
cp NMe 2
• •
85
R

75
40 50 60 70 80 90 100

D(Sm-X), kcaJ/mol
Fig.3, 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).

116
 100

80
0
8
'-
a; 60
C)
.:.:
><I 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.

115 Cl•

0 105
8
'-
a;
0
95
."'

><I 85
...
N
q 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.

0 25
E
:::::..
0
_,.u 20
.-::.
:I:
u
.
I 15
3Cl
10
.......
:I:
I
::IE
c 5

Sm Th u Zr,Hf Mo,W' Mo' Mn Ir Ir' Pt

Fig. 6. 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).

118
 (12)

represents a considerable improvement over the original Pauling formula-

tion and should be applicable. Here the x's are Pauling electronega-
tivities. 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

-e-0.219(XM-XH)2) (13)

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 electro-
negativity 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.
15 Cp' 1 ZrX, Cp 1loloX,
,...::. Cp',H!X, Cp,WX,
~
:I:
u
I
:::; loln(CO),X
0 10
Cp'lrPX,

-::::-
1
~ 5
0

Sm Th U Zr,Hf Mo,W Mo' Mn Ir Ir' Pt

Fig. 7. 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
 12
0
E
~
c 8
u
.Yo

,......
:I:
.. 4
' '
(J
I

= 2~~--- -
:::::!l 0
a XcH,
-4 From Matcha Formalism
I'.
I
a:::::!l -8
L,,-~-r,-r,-ro~~-r~-r,-r,-r
1.20 1.45 1.70 1.95 2.20
xt.A
Fig. 8. Correlation between D(M-H) - D(M-CH3) and XM calculated using
eq.(l3) and metal Pauling electronegativities.

0 20
E
.........
0 16
u
~

..-....., 12
I
(J
I 8
:::::!:
.........
Q
4 From Matcha Formalism

I 0
:::::!:
......... 1.20 1.45 1.70 1.95 2.20
Q

xt.A
Fig. 9. 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.

120
 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 consider-
able 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 else-
where,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 environ-

ment, 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
~
u
.
H

100
xI CH(TMS) 2
* NMe
ES
Q
90
*
2

1r-C3Hs

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 least-
squares fits to the data points.

121
 Table 2. Product and Turnover Frequency Data for Organolanthanide-
Catalyzed Hydroaminations/Cyclizations.

Entry Substrate Pwduct N1(hr- 1) Catalyst

1,1
1 H 2N~
z. '6 1 13(25°C)
140(60°C)
1a
la

1,1
2 H2N~ :!.
'+ .8. 125(25°C)
75(80°C)
<1(80°C)
la
lb
lc
H

--(:}
NH2
3 ~ ! .2. 84(25°C) la

4 H2N~
s. l) lJl 5(60°C) la

5 ():)--- 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 embodi-
ment of this strategy, we have recently disclosed the facile and regie-
specific organolanthanide-catalyzed hydroamination/cyclization of
N-unprotected amino olefins,40 heretofore difficult homogeneous trans-
formations 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 note-
worthy. 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 excep-
tion 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 ggm-
dimethyl 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 coordin-
ation 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

-{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. T. J. Marks, ed., " Metal-Ligand Bonding Energetics in

Organotransition Metal Compounds", Polyhedron Symposium-in-Print,
(1988). 1.
2. 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.
3. J. A. Connor, Top. Curr. Chern. 71:110 (1977).
4. J. Halpern, Ace. Chern. Res. 15:238 (1982).
5. J. U. Mondal and D. M. Blake, Coord. Chern. Rev. 47:204 (1983).
6. J. W. Bruno, T. J. Marks, and L. R. Morss, J. Am. Chern. Soc.
105:6824 (1983).
7. L. E. Schock and T. J. Marks, manuscript in preparation.
8. D. C. Sonnenberger, L. R. Morss, and T. J. Marks, Organometallics
4:352 (1985).
9. J. W. Bruno, H. A. Stecher, L. R. Morss, D. C. Sonnenberger, and T.
J. Marks, J. Am. Chern. Soc. 108:7275 (1986).
10. L. E. Schock and T. J. Marks, J. Am. Chern. Soc. 110:7701 (1988).
11. L. E. Schock, A. M. Seyam, M. Sabat, and T. J. Marks, in ref. 1, pp
1517-1530.
12. S. P. Nolan, D. Stern, and T. J. Marks, J. Am. Chern. Soc., in press.
13. S. P. Nolan, D. Stern, and T. J. Marks, Abstracts, 196th ACS National
Meeting, Los Angeles, CA, Sept. 25-30, 1988, !NOR 378.
14. J. E. Huheey, "Inorganic Chemistry", 2nd ed., Harper and Row, New
York (1978), pp. 824-850.
15. 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.
16. 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.
17. A. R. Dias, M. S. Salema, and J. A. Martinho-Simoes, J. Organomet.
Chern. 222:69 (1981).
18. 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.
19. 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.
20. J. U. Mondal and D. M. Blake, Coord. Chern. Rev. 47:205 (1982).
21. P. 0. Stoutland, R. G. Bergman, S. P. Nolan, and C. D. Hoff, ref. 1,
pp. 1429-1440.
22. R. L. Brainard and G. M. Whitesides, Organometallics 4:1550 (1985).
23. J. B. Schilling, W. A. Goddard, III, and J. L. Beauchamp, J. Am.
Chern. Soc. 109:5573 (1987).
24. J. W. Bruno, M. R. Duttera, C. M. Fendrick, G. M. Smith, and T. J.
Marks, Inorg. Chim. Acta 94:271 (1984).
25. T. Ziegler, V. Tschinke, L. Versluis, and E. J. Baerends, in ref. 1,
pp. 1625-1657.
26. L. Pauling, "The Nature of the Chemical Bond," 3rd ed., Cornell
University Press, Ithaca, New York, (1960), Chapter 3.

124
27. J. E. Huheey, "Inorganic Chemistry," 3rd ed., Harper and Row, New
York (1983) pp. 144-160.
28. B. E. Douglas, D. H. McDaniel, and J. J. Alexander, "Concepts and
Models of Inorganic Chemistry," 2nd ed., Wiley, New York (1983)
Chapt. 24.
29. J. Mullay, Struct. Bond (Berlin) 66:1 (1987), and references therein.
30. 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.
31. 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.
32. H. E. Bryndza, L. K. Fong, R. A. Paciello, W. Tam, and J. E. Bercaw,
J. Am. Chern. Soc. 109:1444 (1987).
33. H. Mauermann and T. J. Marks, Organometallics 4:200 (1985).
34. G. Jeske, H. Lauke, H. Mauermann, P. N. Swepston, H. Schumann, and T.
J. Marks, J. Am. Chern. Soc. 107:8091 (1985).
35. G. Jeske, L. E. Schock, H. Mauermann, P. N. Swepston, H. Schumann,
and T. J. Marks, J. Am. Chern. Soc. 107:8103 (1985).
36. G. Jeske, H. Lauke, H. Mauermann, H. Schumann, and T. J. Marks, ~
Am. Chern. Soc. 107:8111 (1985).
37. Cp2LaNMe2 catalyzes ethylene polymerization (D. Hedden and T. J.
Marks, unpublished results).
38. 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).
39. For the addition of NH3 to H2C=CH2, fiG 0 ~ -4 kcal/mol.
40. M. R. Gagn~ and T. J. Marks, J. Am. Chern. Soc., in press.
41. Y. Tamaru, M. Hojo, H. Higashima, and Z. Yoshida, J. Am. Chern. Soc.
110:3994 (1988), and references therein.
42. 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.
43. A. L. Casalnuovo, J. C. Calabrese, and D. Milstein, J. Am. Chern. Soc.
110:6738 (1988).
44. 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 hetero-
geneous catalysis, a metal cluster is a small, discrete group
of metal atoms (referred to here as a metal aggregate) dis-
persed 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 dif-
ficult 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 struc-
tures 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, investi-
gations 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 127

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
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 exam-
ple, 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 organo-
metallic 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 ship-

in-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

Precurs or Prepare d by
Adsorpt ion of [Os 3 (COh2 l co 2-
[0s5C(C0 )14] /MgO
on MgO 400 275°C, 1 atm
4 h

C, 1 atm
He 150°C 4 h
275°C, 1 atrn 2 h
2 h
A ~
[H 3 Os 4 (C0) 12 ]-/MgO

Precurs or Prepare d CO or
from CO + H2
H2 0sC16+ Mg0400 {0Mg} 2 0s (CO) x CO + H2 , 275°C, 11 atrn
200-250° C
1 atrn x = 2 and 3 12 h

275°C, 1 atm [H 3 0s 4 (C0) 12 ]- /MgO

COCO+ or
H2 l 4-5 h

CO + H 2 (1:3 molar ratio)

275°C, 10 atm
[Os 10 C(C0)24 ] 2 -/MgO
> 2 days
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.
MgO . l 0 , l l , l 2 of the
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
~
® Re 10n

MgO 0 surface OH

0
(100l
face 0 atom

0 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 MgO-
supported 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

experiments with an extraordinarily robust cluster,
[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
 Table 1. Performance of supported osmium catalysts for CO hydrogenation at low conversions.

Predominant Avg. Os Product

form of Catalyst Catalyst particle distribution, mol%
catalyst precursor support size,A T, °C P,atm 104xTOFa C1 C2 C3 C4 Cs Ref.

Os/y-Al203b H20sCl6 y-Al203 70C 275 10 70 74 9 8 5 2 34

[H30S4(C0)12l-{Mg O} [H20s (CO) 4] MgO - 275 10 1 97 2 1 - - 10

? [H20s (CO) 4] basic Y <12d 300 19 >0.04e 31 35 19 14 2 23

zeolite

a Turnover frequency, s-1

b Os aggregates on y-Al203
c Determined by transmission electron microscopy
d Approximate diameter of zeolite supercage
e Only lower limit determined.

w
(11
w
en

Table 2. Selectivities and stabilities of zeolite-supporte d CO hydrogenation catalysts.

Product Approximate
Catalyst Catalyst distribution, mol% catalyst
precursor support T, °C P,atm C1 C2 C3 C4 C5 lifetime, h Ref.

Fe atomsa NaY 254 1 19 2 9 47 23 -1 20

Ru(NH3)6Cl3 LaY 252 14 34 5 6 20 34 -1 19
[H20S (CO) 4] NaY 300 19 47 24 12 15 2 >480 23

a Fe in toluene made from metal vapor.

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 impregna-
tion 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 be-
tween 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 Mononuclear Re Trinuclear Re Re metal

reaction complex on MgO, ensemble on MgO aggregates
prepared from prepared from on MgO
[HRe (CO) 5 ] [H 3 Re 3 (C0) 12 ]

Propylene
hydrogenation 3.5 x 1o-2 2. 1 x 1o-2 -1o-1
at 80°C

Cyclopropane
hydrogenolysis
to give methane 4. 7 X lQ-4 -3 X lQ-2
+ ethane at 200°C

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 deter-
minations of these structures by techniques including EXAFS
spectroscopy are now allowing determination of metal-support
bonds in supported catalysts (placing the issues of metal-
support 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. B. C. Gates, L. Guczi, and H. Kn6zinger, eds. "Metal

Clusters in Catalysis," Elsevier, Amsterdam (1986).
2. B. F. G. Johnson, ed., "Transition Metal Clusters," Wiley,
Chichester (1980) .
3. M. Moskovits, ed., "Metal Clusters," Wiley, New York
(1986).
4. B. C. Gates and H. H. Lamb, J. Mol. Catal., in press.
5. 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.
6. M. Ichikawa, Polyhedron, 2 (1988) 2351.
7. H. H. Lamb, B. C. Gates, and H. Kn6zinger, Angew. Chern.
Int. Ed. Engl., £I (1988) 1127.
8. J. M. Basset, A. Theolier, D. Commereuc, and Y. Chauvin,
J. Organomet. Chern.,~ (1985) 147.
9. G. Bergeret, P. Gallezot, P. Gelin, Y. Ben Taarit, F.
Lefebvre, C. Naccache, and R. D. Shannon, J. Catal.,
l..Q.1 (1987) 279.
10. H. H. Lamb and B. C. Gates, J. Am. Chern. Soc., 1na (1986)
821.
11. H. H. Lamb, T. R. Krause, and B. C. Gates, J. Chern. Soc.
Chern. Commun. (1986) 821.
12. H. H. Lamb, A. S. Fung, P. A. Tooley, J. Puga, T. R.
Krause, M. J. Kelley, and B. C. Gates, to be published.
13. L. D'Ornelas, A. Theolier, A. Choplin, and J.-M. Basset,
Inorg. Chern., 22 (1988) 1261.
14. T. R. Krause, Ph.D. Thesis, University of Delaware, 1987.
15. c. Dossi, R. Psaro, and R. Ugo, J. Organomet. Chern., ~
(1988) 259.
16. P. Dufour, L. Huang, A. Choplin, R. Sanchez-Delgado, A.
Theolier, and J.-M. Basset, J Organomet. Chern., ~
(1988) 243.
17. J. L. Vidal and R. C. Schoening, Inorg. Chern., Zl (1982)
438.
18. P. A. Jacobs in B. C. Gates, L. Guczi, and H. Kn6zinger,
eds., "Metal Clusters in Catalysis," Elsevier,
Amsterdam (1986) p. 357.
19. H. H. Nijs, P. A. Jacobs, and J. V. Uytterhoeven, J. Chern.
Soc. Chern. Commun. (1979) 180; 1095.
20. L. F. Nazar, G. A. Ozin, F. Hugues, J. Godber, and D.
Rancourt, J. Mol. Catal., Z1 (1983) 313.

139
21. 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.
29. M. Primet, J Chern Soc. Faraday Trans.I, 1.1 (1978) 2570.
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.

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 141

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
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
b c
0~
d

Figure 1: The Platonic Solids

e

a
b
c
d
e
f
The convex de~ahedra:
Tetrahedron (4-hedron)
Triangular dipyramid (6-hedron)
Octahedron (8-hedron)
Pentagonal dipyramid (1()-hedron)
12-lledron
14-lledron
G/' 0;_\
f

g 16-hedron
h Icosahedron (20-hedron) 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 (Bl (Figure 7) leads to the sequence:

rat _. {9-11 ~ 110-21 ~ rn-3} ~ £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

-
{8-2} {9-3}
{6} { 7-1} (3)
(1)
(2)

Figure 6: Multi-edge Cleavage for 6-coordination

145
 ® {8}
___..
r:i1lJ ~
{9-1 }
{1 0-2 }

\
{ 13-5}
..- {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) <10-1}

Note: {6-2} square

{9} tricapped trigonal prism

{9-1} = bicapped trigonal prism

{9-2} nonocapped trigonal prism

<9-3} = trigonal prism

Figure 10: General Scheme for Pseudorotation

150
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
 (a) (b)

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
suggesting the existence of alternative· isomeric forms. These and other
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
governing nuclear motion. The existence of these readily excited states
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:
(81 -) (9-ll -) (10-21 the polyhedra become increasingly more open and
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
CO's corresponding to the two polyhedral caps in l9-l). In contrast, 2b,
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

have been found to be highly fluxional. There is very little energy
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
b

a 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. 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).
2. B. F. G. Johnson, to be published.
3. J. F. Blount and L. F. Dahl, Inorg Chern , 1373 (1964)
4. C. H. Wei and L. F. Dahl, J 8mer Chern Soc , 91:1351 (1969).
5. B. F. G. Johnson, J Chern Soc Chern Commun , 211 (1976); R.E.
Benfield and B. F. G. Johnson, J Chern Soc , Dalton Trans , 1554
(1978) .
6. 0. A. Gansow, A. R. Burke, and W. D. Vernon, J Affier Chern Soc.,
94:2550 (1972).
7. F. A. Cotton and J. M. Troup, J Affier Chern Soc , 96:4155 (1974) and
references therein.
8. F. A. Cotton and J. M. Troup, "Symposium on Metal Carbonyl
Chemistry", Ettal (1974).
9. B. F. G. Johnson, J Chern Soc . Chern Commun , 703 (1976).
10. F. H. Carre, F. A. Cotton, and B. A. Frenz, Inorg Chern, 15:380
(1976) .
11. C. E. Anson, R. E. Benfield, A. W. Bott, B. F. G. Johnson, D. Braga,
and E. A. Marseglia, unpublished observations.
12. 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).
13. 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)

160
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 chem-

istry concerned with the partial oxidation of organic molecules. There are
two grounds for this. 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 to-

ward performing selective organic oxidations homogeneously using soluble
oxides in an effort to mimic and understand the corresponding heterogene-
ous 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 ad-
dressed 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 161

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
 Table I. Major Products of the Usage of Heterogeneous Catalysisa

Global
Reactant(s) Product production,
k I r

Crude Oil Hydrocarbon Fuels 1 10 12

X

so 2 , o 2 Sulfuric Acid 1.4x 10 11

N2, H2 Ammonia 9 X 1011

NH 3 , o 2 Nitric Acid 2.5 X 1010

CO, H2 Methanol 1. 5 X 1010

Ethylene Oxide 1 X 1010
C2H4, 02
Unsaturated Vege- Hydrogenated Vege- 8 X 10 9
table Oils, H2 table Oils

C2H4 Polyethylene 6 X 10 9
CH 30H, 02 Formaldehyde 5 X 10 9
c3H6 , NH 3 , 0 2 Acrylonitrile 3 X 10 9
E._- Xylene, 02 Phthalic Anhydride 2 X 10 9
_g_-Butane, 02 Maleic Anhydride 4 X 10 8

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 consid-
ered, not in terms of achieving partial oxidation but in terms of synthe-
sizing soluble polyoxovanadates that might have features normally associ-
ated only with solid vanadium oxide surfaces.

METHANOL DEHYDROGENATION

Formally speaking, the oxidation of methanol to formaldehyde is a de-

hydrogenation 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 metha-
nol, methoxide oxidation at 150°-250°C, and catalyst reoxidation at still
higher temperatures. A detailed mechanism for the first two steps was pro-
posed in 1977 by Edwards, Nicolaidis, Cutlip, and Bennett. 3 This mecha-
nism, shown in Scheme II, proposes dissociative adsorption of methanol

162
across a molybdenum-oxygen double bond followed by oxidation involving par-
ticipation 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 mechan-
ical computations. These calculations support addition of methanol across
an Mo=O bond followed by an oxidation step that involves an adjacent molyb-
denum center. Unfortunately, there is no experimental precedent for the
formation of stable surface methoxides of the type observed in methanol ox-
idation 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 ob-
viates 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 0 HO OCH3 0

I
Mo--0--Mo
I + \1 Mo--0--Mo
II

HO OCH 3 0 HO OCH OH

\1 II
Mo--0--Mo
\1 2

Mo--0--Mo--
I
Scheme II

163
 OCH 3 HO
. . . _ _...o........._ / "'-../ '\./
Mo
Mo Mo + ~ Mo +
#~ #~ #~ #~
0 0 0 0 0 0 0 0

OCH 3 0~
"'-../ "'-..--~cH 2
Mo ~ -•------'~~ Mo
0
#~
0 0
#'\.OH
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. The geom-
etry of the ethoxide complex 6 is shown in a.

This complex, when heated to at least 180°C in acetonitrile or nitroben-

zene 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

3- 7 2
electron-donating than the P3o9 ligand. Formation of a coordinated 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.

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 util-
izes 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 hydro-
gens since the presence of allylic hydrogens leads to a great loss in se-
lectivity.

Several homogeneous catalytic systems have been developed that are

capable of o2 epoxidation of olefins. 9 Mechanistic details of these pro-
cesses are obscure, however, due in part to an inability to isolate reac-
tion intermediates. The iridium olefin complex d undergoes selective o2
oxidation via intermediates that may be related to those involved in heter-
ogeneous and homogeneous epoxidatione Specifically, complex d reacts with

2-

o2at ambient temperature according to equation (3) to form a single pro-

duct, 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-

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

Observation of this intermediate draws a clear analogy between the irid-

ium phosphate system under discussion and the heterogeneous catalytic sys-
tem of Scheme IV: if the cyclic intermediate f were to convert to an epox-
ide g instead of the allylic complex e, a homogeneous catalytic cycle

2-

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
~ 0
+ 3 H2o (4)

+ 7/2 02
~ 0
+ 4 H2o (5)

Little is known about catalyst-substrate interactions in these sys-

tems 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 interac-
tion 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 recently-
16
reported soluble inclusion complex [CH 3CN C: v 12 o32 4- )], h:

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

4-
v12032 cage of corner- and edge-sharing VOS square pyramids. Efforts
are currently being made to determine whether less polar guest molecules
4-
can bind to the v 12 o32 cage and to investigate the ability of the
vl2032
4- cage to undergo redox
reaction chemistry.

168
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,
2327 (1986).
3. J. Edwards, J. Nicolaidis, M. B. Cutlip, and C. 0. Bennett, J. Catal.
50, 24 (1977).
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-
vanced Science Institute Series, Series C; Mathematical and Physical
Sciences, Vol. 231: Surface Organometallic Chemistry: Molecular Ap-
proaches to Surface Catalysis; J. M. Basset, B. C. Gates, J. P. Candy,
A. Chaplin, M. Leconte, F. Quignard, and C. Santini, eds.; Kluwer:
Dordecht, 1988; p. 173.
7. 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) S-
VP05: 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 pre-
pared the compound dilithiomethane, 5 but his report had suffered a credibil-
ity problem when others could not reproduce it and was not widely accepted. 1

Metal-Metal Bonds and Clusters in Chemistry and Catalysts 171

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
 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
desired. 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
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 n

(CH 3Li)n (22) 1-4

n
(CH 3Li)nLi (22) + 7 1-4
n
(CH 3 ) Li 1 (22) - 7 2-4
n n- n

Table 2. Major Ions from the Flash Vaporization Mass Spectrum of CH 2Li 2

Positive Ion m/e n

(cH 2Li 2 )n (28) 1-6

n
(CH2)nLi2n-1 (28) - 7 2-6
n
(CH 2Li 2 )nLi (28) + 7 2-5
n
(CH 2Li 2 )nH (28) + 1 1-3
n
(CLi2)nH2n-1 (28) - 1 1-4
n
(C2Li2)n (38) 1-3
n

Table 3. Major Ions from the Flash Vaporization Mass Spectrum of HCLi 3

Positive Ion m/e n

(CH3Li)n (22) 1-4

n
(CH3Li)nLi (22)
n
+7 1-4
(CH2Li 2 )n (28) 1-4
n
(CH 2Li 2 )nLi (28)
n
+7 1-4
(CHLi 3 )n (34) 1-4
n
(CHLi 3 )nLi (34)
n
+7 1-4
(CLi 4 )n (40) 1-4
n
(CLi 4 )nLi (40)
n
+7 1-4
(C2Li2)n (38) 1-4
n
(CH 3 ) Li 1 (22) - 7 2-6
n n- n
(CH2)nLi2n-1 (28) - 7 1-6
n
CnHnLi3n-1 (34) - 7 1-5
n
CnLi4n-1 (40) - 7 1-4
n

173
 Table 4. Major Ions from the Flash Vaporization Mass Spectrum of CLi 4

Positive Ion m/e n

(CLi 4 )n (40) 1-4

n
(CLi2)nLi2n-1 (40) - 7 1-4
n
n +
(CLi 4 )nLi (40) 7 1-4
(C2Li2)n (38)n 1-3
(C 2Li)nLin_ 1 (38)n - 7 1-3

Table 5. 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 31
c2Li 2 38
c2Li 3 45
c2Li 4 52
c3Li 2 50
c3Li 3 57
c3Li 4 64
c3Li 5 71
(C3Li4)2 128
(C3Li4)3 192

spectrum is that the sample was not highly pure and contained other
species such as dilithiomethane). Note the dilithiomethane gives species
as high as n = 6. 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
Table 6. Positive Ions in the Flash-Heated High-Resolution Mass Spectrum
of Halide-Free Methyllithium

Rel
Calcdb Measd Error, Rel
a Mass ppm Intensityc
Ion Mass

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

~e =methyl. bBased on C 12.000 ... , H = 1.0078246, and Li = 7.016001

amu. cintensities varied ± 2-3% from spectrum to spectrum and were
determined by comparison of sample line densities to standard Ar 2+/Ar+·
spectra whose line densities were known relative to integrated ion
current.

Table 7. Comparison of the Positive Ion Mass Spectrum of Methyllithium

with Those of Ethyllithium and tert-Butyllithiuma

CH b
3

not found R5 Li 6 + (24)

not found R4Li 5+ (1.3)
R3 Li 4+ (8) R3Li 4+ (47.5) R3Li 4+ (43)
R2 Li 3+ (4) R2Li 3+ (15) R2 Li 3+ (4.5)

RLi 2+ (100) RLi 2+ (100) RLi 2+ (100)

aRelative intensities in parentheses. 70-eV electrons used, except where

noted. bThis work. cReference 8. Intensities are for saturated
ethyllithium vapor. 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
Polyllthium Power Feed

High
Vacuum I
Environment Low Temperature
I Matrix for

~~~rature _)J #
Various
Spectroscopic
Investigations

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
chemistry around the world. 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
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
bridging lithium s1tes, and that those olefins with 1,1-dilithio
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
Pople, and co-workers of the existence of fluctional cis-tetrahedra l,
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 • Very
2
little difference in energy (-.8 Kcal/mol- 1 ) between planar and
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
z,10 and cis-C H Li , 8. 10 However structure l will be
3 4 2 extraordinarily
difficult to observe and certainly would be a challenging candidate for
our matrix collaboration since lithium halide elimination would occur at
temperatures on the other side of -80 •c.

178
 H
H-'

lr
Ll

\/'
lr H lr lr

H--e
I~ e--H e \I li

l~\
e
H lr
lr
1\
F H1
H
I, 2- Orlrthraethane Tetr•lrthromethane Orfluarcrdrlrthrornethane I, 1-0rlrthrocyclopropane

5 I 7
•
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
..."'··... Lr

··.. I/
li
' \ I
I
I \
\
\
H/c
/ \ e c
Lr---e e---li
\ / \ /
\ I \ I
'u' \
li
I

I, I - Oilrthroethylene Perlrthroethrlene Dilithroacetvlene

' 10 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:

1---------1-----
Li CLis

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

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 lowest eigenvalue

2 D.,.,h -s180
3 0 3h - 28 120
4 Td - 38 109.47
5 2
03h -l/ 2 ( 8 180 + 28 120) [ 1 + {1 - 8 ( 8 180 8 120 - 38 90)/(S180
+ 2S )2}1/2]
120
6 oh -(S180 + 48 90)
8 oh -( 38 70.53 + 38 109.47 + 8 180)

aEach eigenvalue is expressed in terms of overlap integrals S between

8
orbitals separated by angle 8 (as measured from the cage center). For Li
2s STOs at 2.0 l 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
 Table 9. SCF Total Energies and Carbon Populations for Some Lithiated Carbon Species CLi , as Calculated with
3-21G and Dunning-Hay (DH) Split Valence Basis Sets,f n

Mulliken natural population

C-Li bond calcd total energy (au) population on C on C
Species Symmetry length clt) 3-21G DHa 3-21G DHa DHa

1.929 -67.13154 -67.520562 6.81 7.09 9.13

CLi 4 Td
2- 1.929 -67.396262 7.05 8.97
CLi4 Td
CLi 5 2.087 -74.59934 -75.001260 6.81 7.18 9.08
0 3h a
2.056
c
CLi 5- 2.087 -75.000853 7.14 9.19
D3h a
2,056
c
CLi + 2.087 -74.45440b -74.885704 (6.02)c 7.26 9.19
5 D3h a
2.056
c
2.058 -82.00344 -82,483723 6.93 7.28 9.44
CLi 6 oh
CLi 2+ 2.094 -81.63028d -82.104959 (6,05)e 7.53 9.34
6 oh

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

assessed as being much more negative in the natural population analysis than in the Mulliken population analysis.
(XI
 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
CLi 5 , CLi 6 , and CLi 8 .
our laboratory when we were studying the vapor species associated with

Table 10. Pentacoordinate Carbocations CLi 5 _nHn+

Calculated Measured
Ion Mass Mass Source of Ion

CLi5 47.08000 47.08019 (CLi 4 )n

CHLi 4+ 41.07183 41.07196 (CHLi 3 )n
CH 2Li 3+ 35.06365 35.06380 (CH2Li 2 )n, (CH 3Li) 4
CH 3Li 2+ 29.05548 29.05559 (CH 2Li 2 )n, (CH 3Li) 4
CH4Li+ 23.04730 [sought in CH4 +
Li reaction]

182
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 0
c 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
 .....

4r\'r-er...
0

o
00
~
0
eLi
c
0
4
,...
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) A
in CH 3Li 31 Both the lithium and carbon atoms were found to be in
ordered sites.

~axis

Qc
u

Figure 4 . CD 2 coordination in CD 2Li 2 •

184
 Table 11. Structural Parameters CD 2Li 2 .

Space Group: 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): wR = 0.098, R = 0.077, c 2 = 1. 70. a

p p

Atom X y z U(isotropic)x10

C(l) 0.0 0.25 0.4267(8) 0.00(3)

D(1) -0.034(3) 0.323(2) 0.336(1) 10.6(9)
D(2) -0.019(6) 0.044(2) 0.406(1) 6.1(6)
Li(1) 0.189(2) 0.461(3) 0.028(1) 0. 00(7)

aStructure Refinement with GSAS by Bob Von Dreele of Los Alamos.

Selected Bond Distances and Angles.

Atoms Distance (A) Atoms Angle

C(l)-D(l) 1.09(1) D(l)-C(l)-D(l) 100.2(5)

C(1)-D(2) 1.18(1) Li(1)-C(1)-Li(1) 153.6(7)
C(1)-Li(1) 2.63(1)
2.17(1)
2.73(1)
2.53(2)
Li( 1 )-D(l) 2.36(2)
2.15(1)
1. 72(2)
2. 28( 1)
2.03(2)
Li(l)-D(2) 2.16(1)
2.38(2)
1.99(1)
2.06(2)
Li(l) -Li( 1) 2.26(2)
2.86(1)
2.78(1)
3.09(1)

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-
lithium approach [2.17(2) !]
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

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)

;H· ..
~C-H···M -c-·····M
I \
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
·
· d us~ng
were o b ta~ne 1H- 13 c cross po 1 ar~zat~on
· · mag~c · ·
· ang 1e sp~nn~ng
(CPMAS) techniques in collaboration with C.S. Yannoni. 39 The weak signal
13 7
observed in the proton-decoupled C spectrum of CH 2 ( Li) 2 at 150 •c
shown in Figure Sa is due to broadening from dipolar coupling with
the 7Li nucleus. This type of broadening was eliminated through use of
.
d ecoup 1 ~ng tee hn~ques,
. 40 t h roug h mag~c . . 40 an d t hroug h
. ang 1e sp~nn~ng

(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)

13
Figure 6. Proton-decoupled solid-phase 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
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
1 6 13 6
TMS. The H and Li decoupled
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
increasing lithium substitution. 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
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
<:
::J
0 u;
~ (b) Diithiomethone
2500 1000

1500
600

291 287 283 279 58 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. Binding Energies, Chemical Shifts, and Calculated Charges.

C 1s C 1s Calcd Li 1s Calcd
Compound BE, eV shift, eV qc BE, eV qLi

C3H6 285.0a 0.0 -0.03

CH 3Li 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)

aUsed as reference. b
Average of monomer, dimer, and trimer structures.
Standard deviation in parentheses.

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 H Mo3S•, /Lo

2
'
H/ 'L•
c
/ 150"!;
96%
.. c
H/ 'Lo
• M"4S•

Me3So H Mo3Si Lo
'-./ 170"C ..
""-/
c + IMo3 Sil:! CHz
2
/c'-.. 95%
Me3Si/ "LI
Mo3So Lo

'/
ICH313 c H (CH3 13C u
2 c '
H/ '-u
/ 180"C
88% > c
H/ ' L i
• (CH 314C

(CH 3 12 C Li (CH 3lzC

I'/
ICH3lzC H
2 I"/ c 180"C
90% .,. /c""- • I '-cHz
(CH.Y~/ '-.Li (CHyzC Ll ICH3lzC/

.~"
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
lithium by Schleyer and Pople are beginning to create for such dilithio
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
R
::.:.....u~
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. J.A. Gurak, J.W. Chinn, Jr. and R.J. Lagow, ~Am. Chern. Soc.
104:2637 (1982).
2. L.A. Shimp, J.A. Morrison, J.A. Gurak, J.W. Chinn, Jr., and R.J.
Lagow, ~Am. Chern. Soc. 103:5951 (1981).
3. (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).
4. (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). - - -- --
5. K. Ziegler, K. Nagel, and M. Patheiger, ~ Anorg. Allgern. Chern.
282:345 (1955).
6. F.J. Landro, J.A. Gurak, J.W. Chinn, Jr., and R.J. Lagow,~
Organomet. Chern. 249:1 (1983).
7. J.W. Chinn, Jr. and R.J. Lagow, Organometallics 3:75 (1984).
8. (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).
9. M.Y. Darensbourg, B.Y. Kimura~E. Hartwell, and T.L. Brown, ~Am.
Chern. Soc. 92:1236 (1970).
10. ~Collins, J.D. Dill, E.D. Jemmis, Y. Apeloig, P.v.R. Schleyer,
and J.A. Pople, ~Am. Chern. Soc. 98:5419 (1976).
11. E.D. Jemmis, P.v.R. Schleyer, and J.A. Pople, ~ Organomet. Chern.
154:327 (1978).
12. Y. Apeloig, P.v.R. Schleyer, J.S. Brinkley, J.A. Pople, and W.A.
Jorgensen, Tetrahedron Lett. 43:3923 (1976).
13. E.D. Jemmis, D. Poppinger, P.v.R. Schleyer, and J.A. Pople, J. Am.
Chern. Soc. 99:5796 (1977). --

193
14. 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).

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 195

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
 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

r
was provided by the metathetical reaction of lithium phosphide or arsenide with the
appropriate metal trichloride which afforded compounds of 1:3 stoichiometry, 1.4

t-Bu2

/M~
t-Bu 2E E-t-Bu 2

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 --~

197
 '
t-Bu t-Bu
Me Me

R
' M'\./ R Sb /
'/ '\.I
R-E E-R t-Bu2sb....._ / "\. /'
Me-.M
I I
M-Me
_;~Ja Ga,
Cl " / Sb-t-Bu2
Me" " ' / 'Me Sb
E
/' t-Bu
/ 't-Bu
R R

1:1 1:2

M=Ga,ln;
E=P,As,Sb

1:4 1:3

E-t-Bu2

~
t-Bu2E/ ~E-t-Bu2
Figure 1. Ill-Y Compounds of Various Stoichiometries

198
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 t-Bu

CI., 'Sb,
/ ••• Sb-t-Bu2
II,, / " t.''
In In

t-Bu2Sb / " Sb/~ Cl

t-Bu
/"- t-Bu

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

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 t-Bu
"-/
As
Me ·.,, / ' ..• Me
'Ga/ 'Ga''
Me,;::?' "'-. / ~Me
As

t-Bu
/"'-. t-Bu

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

1. H. M. Manasevit, Alwl. Pbys. Lett. 12:156 (1974).

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,
New J. Chern. 12:553 (1988).
7. 0. T. Beachley, J.P. Kopasz, H. Zhang, W. E. Hunter, and J. L. Atwood, L.
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 205

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
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
A.u 1 ML Cu/Au
<l

100 200 300 400 100

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
<I

TEMPERA TURE-K

Figure 2. TPD results corresponding to CO adsorbed to

saturation levels on the clean Ru(OOOl) sur-
face, 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
enhancement of -40. Above this coverage, the rate falls to an
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
activation barrier of approximately 5 kcaljmol [31]. In the
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
cu-free surface, is suppressed. Further, Cu reduces the
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 [39-
40] 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

8.0

•.
!
c

-.•
0

u
8.0

.
Gl
~ 4.0
c-;
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 dehydro-
genation to benzene. PT ; 101 Torr. H2 /
cyclohexane ; 100. T ; 650 K. (Reproduced
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 • .:I IlL NI/W ( 1 101
:1! 0 .I IlL NI/W (1101 n
A 1.0 IlL NI/W 11101
0 .4 IlL NI/W ( 1001
• . 7 IlL NIIW ( 1001
+ .!I IlL NI/W ( 1001

1.4 1.6 1.8 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 ;
4/1. (Reproduced with permission from Ref. 41.
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 0.1 ML Nl
o 0.4 ML Nl
• 1.2 ML Nl
10- 4
1.5 1.7 1.9 2.1

Figure 5. Arrhenius plots of the rates of ethane hydro-

genolysis 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).

Table 1. Comparison of Strained-Metal Overlayer Systems

Adsorbate Substrate Atom Density Pseudomorphicj Change

Mismatch/ML Epitaxial in co
Layers Desorption
T

Cu Ru ( 0001) 6% 1/2 50K

cu W(llO) 20 1/1 80
Ni W(llO) 21 1/1 -50
Ni W(100) 42 1/1 -50
Ni Ru (0001) 15 1/1 50
Pd W(llO) 10 1/1 -200
Pd W(100) 35 2/2 -170
Pd Ta (110) 18 1/1 -230
Fe W(llO) 9 1/2 -50
Fe W(100) 35 2/2 -60

212
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
(1973).
2. J. H. Sinfelt, "Bimetallic Catalysts: Discoveries, Concepts,
and Applications", John Wiley & Sons, NY, NY, 1983.
3. G. M. Schwab, Disc. Faraday Soc. 8, 166 (1950).
4. J. T. Yates, Jr., c. H. F. Peden, and D. W. Goodman, J.
catal., 94, 576 (1985).
5. D. w. Goodman, J. T. Yates, Jr., and c. H. F. Peden, Surf.
sci., 164, 417 (1985).
6. D. 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 215

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
 1. Qualitative and quantitative analysis of ill products.
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 A He-Ne
laser interferometry assembly 19. 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 Rcil!:l!H Tr;:mg (°C) ~AI ~N ~c P'r;Q

[Et2AIN3]3 quartz 500 43 45 10 2
[Et2AIN3l3 metal 480 44 45 10 1
[Me2AIN3l3 quartz 500 43 44 11 2
[Et2Al(NH2)l3 quartz 550 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 P-hydrogen elimination mechanism
operative in cleaving the Al-Et bond. 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: Figure 2 shows

the dependence of the growth rate on substrate temperature. 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 1-l:!C
Et 2 AIN 3 (g) _____. Et 2 AJN 3 (s) _____. 'AIN (s)
I
Et

Scheme 1.

800 700 600 500

10
l.81J. m/hr
Ea=5.23kcal/mol 1.4
1.0
9
Ill
Ea=26.4kcal/mol 0.6
...
I
~

.d
......
-<
~

"' 8
Azide!H 2 50 seem
H 2 Makeup 7.5 seem 0.2
Precursor Temp 50°C
Reactor Pressure 3 torr
7+---~--~--T---~--~--T---T-~T-~
0.9 1.1 1.3 1.5
1000/T(K-1)

Figure 2

220
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, e. g.
simple dissociation of the donor-acceptor bond. 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 fi-
hydrogcn elimination of the NMe2 ligand. 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

H.. I
~'''Al--H (1)
H.;' I
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
readily cleaved). In fact, from the early studies describing these compounds, it
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
appearance of the film were dependent on the specific reaction conditions. At
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. The behavior usually observed is best described with a
specific example. 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
180 °C, the deposition rate was 0.9 ~m/min. In depositions where the reactor
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
for thinner films. Figure 4b, however, shows the striking effect caused by
pretreating the surface with TiCl4 and growing the films at 100 °C. The films
show nearly complete preference for growing with the (111) face parallel to
the surface. While Figure 4b was taken from a film grown on Si(lOO), similar
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
the final film. Those films grown on Si(lOO) wafers or glass slides that were
pretreated with TiCI4 exhibited a much higher number of crystallites in this
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 30.0
r::
.!
.5 25.0
.>1.
ilD..
20.0
,g
.>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 adhesion of
the AI films to polyimide was always good.

224
 5.00
a

4.00
,.....
.,
"0

.,
c:
0
:I 3.00
0
.r::.
~

.,"u
'!l
c:
:I
2.00

0
u

1.00

0.00
Ll \_ ~ k
30.0 50.0 70.0 90.0
two theto

5.00,-------,-------------------------~----------------~

4.00

.,
,.....
"0
c:
.,0 3.00
:I
0
.r::.
~
u
.,"
'
!l
c:
:I
2.00

0
u

, .00

) ~
0.004-------~--~~--r--------.--~~~~~----~n-----~ h
30.0 50.0 70.0 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

3.50
.,
..._
"C
3.00
.,c
I:

"0 2.50
2,
u
.," .L.OO
.......
!l
""u0 1.50

, .00

0.50
...._....._
II.
0.00
30.0 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) (4)

3 AIH(s) ~ 3/2 H2 + 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. Although there is an emerging understanding of the surface
growth 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. Current work in our laboratories also

228
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.
1982, 25(12), 62.
15. Green, M. L.; Levy, R. A.; Nuzzo, R. G.; Coleman, E. Thin Solid Films 1984, 114,
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.; Smith, J. D.; Wharmby, D. H. W. J. Chern. Soc. 1969, (A), 1738.
26. Cowley, A. H.; Benac, B. L.; Ekerdt, J. G.; Jones, R. A.; Kidd, K. B.; Lee, J. Y.;
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).
32. Schmidt, D. L.; Hellmann, R. U. S. Patent No. 3 462 288, (1969).
33. Baum, T. H.; Larson, C. E.; Jackson, R. L. Abstracts of Papers, Fall Meeting,
Boston, MA; Materials Research Society: Pittsburgh, PA, 1988; 84.12.
34. Powder Diffraction File, International Center for Diffraction Data,
Swarthmore, PA, Card #4-0787.
35. Heitsch, C. W. Nature, 1962, 195, 995.
36. Nechiporenko, G. N.; Petukhova, L. B.; Rozenberg, A. S. Izv. Akad. Nauk SSSR,
Ser. Khimicheskaya 1975, 1697.
37. Paul, J.; Hoffmann, F. M. Surf. Sci. 1988,194, 419

230
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 231

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
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, <cr>, 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
all possible orientations of spins occur in a powdered sample. (Details
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=<CJ> + Ga (2)

m""~~
<2\
54.7°
~ sample

0
Fig. 1. Experimental geometry for the removal of chemical shift
anisotropy through magic angle spinning. B0 represents
the applied magnetic field. Rapid spinning at the angle
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
in solution. This has been demonstrated for several metal carbonyls
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
for many metal carbonyls. [26] Figure 2 illustrates the situation when a
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 0 zoo 400 400 zoo 0

Fig.2 (a) The breadline NMR for RhG(CO)l6· [22] (b) Magic angle
spinning NMR for Rh6(C0)16. [26] The centerbands in the MAS
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
(downfield) to the right of 0 ppm on the x axis.

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
complexes. [29-31] Although many mechanisms have been proposed to
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
mechanism has been proposed. [29] The basis for this comes, in part,
from the recognition that the structure of many complexes of the type
HML 4 can be described as a hydride capped tetrahedron. [29] The Berry
psuedorotation and the hydride tunnelling mechanisms are shown
schematically in Figure 3.

Iron pentacarbonyl. Iron pentacarbonyl has been studied by

breadline NMR at temperatures from 4.2 to 213 K. [12] Evaluation of
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
psuedorotation better described the experimental data. [12] The exchange
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 40 20 0 KHz 215.0 208.1 ppm

Fig 4. (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 axial-

equatorial 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~l- The 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"

228 220 ppm

Fig. 5. Magic angle spinning NMR spectra of [HFe(C0)4]-.[35]

Fig. 6. 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]

Fe 2 cco) 9 , [24] and co 2 (COl8 [14] have each been investigated by 13c MAS
NMR. Of these the cobalt carbonyl dimer is unique in showing dynamic
behavior in the solid state. [14] The structure of dicobaltoctacarbonyl
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
the spectrum. [14] Two signals are clearly seen, at 234 and 182 ppm,
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
terminal structures proposed to exist in solution. [40,41] The problem
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)

Fig. 7. (a) Representation of the molecular structure in

crystalline Co2 (CO) 8. [39] (b) The MAS NMR spectra at 40 and
-40 °C for Co2CC0)8· [14] The low temperature spectrum was
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
 240 220 200
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.

225.2
!calc I
(225.8)
·~:::~J== 224.1 (224.1)

215.6 ----::.::::::=···
••••

----··_::_.-_.._··-~:::::::::=··---- 213.2
210.4 _ _ _ _ _ _ 1~----······-------~----
(213 0)
209.6 (210.4)
203.9
201.3
199.2
__ ..............
==~~=-~~:l-:.::::~::::t====
.
202.2
201.4
(202.1)
(201.5)

-so •c 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. 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.

244
to apparently anomalous chemical shifts for the carbonyl carbons. [15]
Recently a room temperature spectrum for Co4(COl12 at 68 MHz has been
obtained and shown to be consistent with the crystal structure. [51] The
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
24 °C. The line shape changes over this temperature range are
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. Cotton, F. A.; Danti, A. ; Waugh, J. S.; Fessenden, R. W.; J. Chern

Phys. ; 1958, 29, 1427.
2. (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.
3. Meakin, P.; Jesson, J. P.; J. Am. Chern. Soc.; 1973, 95, 1344.
4. Berry, R. S.; J. Chern. Phys., 1960, 32, 933.
5. (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.
6. (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.
7. Dorn, H. C., Hanson, B. E., Motell, E., Inorg. Chim. Acta., 1981,
54, L71.
B. Duncan, T. M.; Yates, J. T., Jr.; Vaughan, R. W.; J. Chern. Phys.,
1980, 73, 975.
9. Cotton, F. A.; Hunter, D. L.; White, A. J.; Inorg. Chern.; 1975, 14,
703; and references therein.
10. Kendrick, R. D.; Maresch, G. G.; Yannoni, C. S.; submitted J. Mag
Res.
11. Spiess, H. W., Mahnke, Ber. Bunsen Gesell.; 1972, 76, 990.
12. Spiess, H. w., Groesca, R.; Haeberlen, U.; Chern. Phys.; 1974, 6,
226.
13. Sheline, R. K., Mahnke, H., Angew. Chern. Int. Ed. Engl.; 1975, 14,
314.

246
14. Hanson, B. E., Sullivan, M. J., Davis, R. J., J. Am. Chern. Soc.,
1984, 106, 251.
15. Hanson, B. E., Lisic, E. C., Inorg. Chern., 1986, 27, 716.
16. Hanson, B. E., Lisic, E. C., Petty, J. T., Iannaconne, G.; Inorg.
Chern., 1986, 25, 4082.
17. Lisic, E. C., Ph.D. Dissertation, VPI&SU, 1986.
18. (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
19. Campbell, A. J.; Cottrell, C. E.; Fyfe, C. A.; Jeffrey, K. R.;
Inorg. Chern., 1976, 15, 2690.
20. Wagner, G. w., Hanson, B. E., Inorg. Chern.; 1987, 26, 2019.
21. Benn, R.; Grundey, H.; Nolte, R.; Erker, G.; Organornetallics, 1988,
7, 777.
22. Gleeson, J. W., Vaughan, R. W., J. Chern. Phys, 1983, 78, 5384.
23. Yannoni, C. S., Ace. Chern. Res., 1982, 15, 208
24. Dorn, H. C., Hanson, B. E., Motell, E., J. Organornet. Chern, 1982,
224, 181.
25. Airne, S.; Botta, M.; Gobetto, R.; Osella, D.; Inorg. Chirn. Acta,
1988, 146, 151.
26. Walter, T. H., Reven, L., Oldfield, E., J. Phys. Chern., in press.
27. Hasselbring, L., Lamb, H., Dybowski, C.; Gates, B.C.; Rheingold,A.;
Inorg.Chirn. Acta, 1987, 127, 149;
28. Herzfeld, J., Berger, A. E., J. Chern. Phys., 1980, 73, 6021.
29. Meakin, P.; Muetterties, E. L.; Jesson, J. P.; J. Am. Chern. Soc.;
1972, 94,
30 Meakin, P.; Jesson, J. P.; Tebbe, F. N.; Muetterties, E. L.; J. Am.
Chern. Soc.; 1971, 93, 1797.
31. Shapley, J. R.; Osborn, J. A.; Ace. Chern. Res., 1972, 6,
32. Wilkinson, J. R.; Todd, L. J.; J. Organornet. Chern.; 1974,-77, 1
33. Hanson, B. E.; submitted J. Am. Chern. Soc.
34. Whitmire, K. H.; personal communication.
35. Whitmire, K. H.; Hanson, B. E.; submitted J. Am. Chern. Soc.
36 Smith, M. B.; Bau, R.; J. Am. Chern. Soc.; 1973, 95
37. Rothwell, W. P.; Waugh, J. S.; J. Chern. Phys.; 1981, 74, 2721.
38. G. Wagner, personal communication.
39. (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.
40. Hanson, B. E.; chapter 2 in "Advances in Dynamic Stereochemistry,"
M. Gielen, ed. Freund, London, 1985.
41. Johnson, B. F. G.; J. C. S. Chern. Commun.; 1976, 211.
42 Dixon, W. T.; J. Chern. Phys.; 1982, 77, 1800.
43. Keller, H. J.; Wawersik, H.; z Naturforsch. B., 1965, 20, 938.
44. Connor, J. A.; Topics in Current Chemistry; 1977, 71, 71.
45. Johnson, B. F. G.; J. c. s. Chern. Commun., 1976, 703.
46. Wei, C. H.; Dahl; L. F.; J. Am. Chern. Soc.; 1969, 91, 1351.
47. Cotton, F. A.; Troup, J.; J. Am. Chern. Soc., 1974, 96, 4155.
48. Hanson, B. E.; submitted Inorg. Chern.
49. Hanson, B. E.; Lisic; E. C.; Petty, J. T.; Iannaconne, G. A.;
Inorg. Chern.; 1986, 25, 4062
50. 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.
51. Airne, S.; Botta, M.; Gobetto, R.; Hanson, B. E.; Inorg. Chern.; in
press.
52. Wei, C. H.; Dahl, L.; J. Am. Chern. Soc., 1966, 88, 1821.
53. Wei, C. H.; Inorg. Chern., 1969, 8, 2384.
54. 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 level of technology involved, or the increasing level of
intricacy and sophistication this technical evolution has 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 249

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
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 in any physical or
chemical process one might wish to study. 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 low-

lying 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 held together
predominantly by interaction of the 4s atomic orbitals. 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. While this has been a fine formula for
successful science in small molecules, it is clearly useless for a
molecule with 10 10 electronic states. Some other approach will be

250
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 ion cores bound together by a sea of electrons
(metallic). 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. So a natural early topic of concentrated
research has been carbon, silicon, and gallium arsenide. This short
paper summarizes a few of the recent developments along these lines,
emphasizing one of the most promising new techniques: ion 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

apparatus recently developed at Rice. Extensive discussion of the design
details of this machine have appeared elsewhere 4 ' 7 , along with several
early examples of its application to the study of dissociative
chemisorption of H2 on transition metal clusters 8 . Here, however, the
supersonic cluster beam was arranged to produce clusters of silicon and
germanium.

Fig. l. Schematic of the supersonic cluster beam FT-ICR

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 pulsed supersonic nozzle. Using an arrangement
originally developed in our group for the study of 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. In such a
nozzle the hot atoms 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 several hundred atoms per cluster. Although the
techniques were originally developed the for
spectral and
10
photofragmentation study of the positive cluster ions , the negatively
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
center of the magnet. Fortunately, as described in an earlier paper
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. Drift of the cluster ions out of the trap by motion
along the 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, a considerable energy spread is present in the
cluster ion packet as 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 neon as the
thermalizing gas held at 1 x 10- 5 torr in 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, with
excellent signal to noise. 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
function of cluster size 13 •14 . Figure 3 shows a section of this
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.

1.0 1.4 1.6 1.8 2.0 2.2

KILOHERTZ

Fig. 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
frequency ICR peaks are due to 13 12
c c59+ and 13 c2 12 c 5s+
respectively.

255
 47

50 49
45
!)4 53 52 51 46
44

....... ~
ollro.,j ~
1500
~ ~ lw.ll ~
1400
~
_I
1lOO
~--
MASS (AMU)

Figure 3. 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 that each

cluster mass has its own cyclotron frequency, 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

,,111111 IIll, ,IIlii IIIlLI

880 860 840 820 800 780
MASS (AMU)

880 860 840 820 800 780

MASS (AMU)

Fig.4. FT-ICR of positive germanium clusters in 11-12 atom size

range. The boxed in region of the top panel received
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 problem, of course, is that the desired single mass cluster is
exposed to the RF as well and one has to worry that off-resonant
excitation of this cluster will leave its cyclotron motion highly
excited.

A beautiful solution to this problem of off-resonant excitation

loss of the desired mass is the "Stored Waveform Inverse Fourier
Transform" (SWIFT) technique first suggested by Alan Marshal and his
coworkers at Ohio State 15 . Here one calculates the discrete inverse
Fourier transform of the RF power spectrum necessary eject the unwanted
masses, stores this waveform in a large memory, and then uses it to
generate an analog time-dependent excitation voltage that is sent to the
excitation plates of the ICR cell. The result is an extremely
complicated RF waveform which is guaranteed to minimize off-resonant
excitation of the desired cluster. 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 of most 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. Note that some silicon
clusters such as 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
 51 48

52 49 45

47 46 44

1~00 1400 1300

.J....
' J . .I. j
.......-~o<lo..o.
_{ I
1$00 1400 1300
MASS (AMU)

Fig. 5. Surface chemistry probe of silicon clusters in the 44-54

atom size range. The top panel shows the (SWIFTed)
contents of the IGR cell before reaction. The bottom
panel shows the result on this bare silicon cluster
distributi on of a 5 second exposure to l x 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. Note that even though most
clusters are quite reactive, there are a few special ones that are
relatively inert, namely those at 11,25,39, 45, and 64 atoms. We are
currently in the process of measuring the absolute reaction rates for
these surface chemisorption reactions.

Such dramatic dependence of reactivity upon cluster size has been

one of the hallmarks of the new cluster studies. It is a particularly
striking effect in dissociative chemisorption reactions of species like

100

\ I
/1
80
J

60
...,.,.,
"0

~~
c..,
0

"" 40

20

0 10 20 30 40 so 60 70
# Si atoms

Fig. 6. Measured fractional depletion of bare silicon clusters due

to chemisorption of NH3 at 2 x 10- 7 torr during a 5 s
reaction exposure (roughly 35 NH3 collisions for each
cluster ion).

260
H2 and N2 on small transition metal clusters. But it is quite surprising
to find it here. First, sharp variations in reactivity are being seen
here for very 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 reaction chemistry. Yet that clearly cannot be
happening. 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 surface so that not a single active
chemisorption site remains. 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
conceivable that quite detailed and predictive calculations will be
available over the next few years for silicon clusters throughout the 2-
100 atom size range. As yet there is no compelling explanation for the
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 . A strikingly beautiful model with full tetrahedral
symmetry has be suggested by Kaxiras 24 as a candidate for the Si45+
cluster. 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

beginning to come out of the new generation of cluster beam studies. As
mentioned a number of times above, transition metal surface chemistry has
been another area of great activity. For example, in cluster ICR
experiments similar to those discussed above for silicon, we have
discovered that there is a dramatic variation of reactivity for
dissociative chemisorption of H2 as a function of cluster size 4 • 8 . This
reactivity sensitivity to size appears at least to first order to be
roughly independent of net charge on the cluster, but highly responsive
to changes in cluster geometry. The 19-th cluster of niobium is
particularly interesting. It appears to have roughly equal likelihood of
being found in either a very reactive form that chemisorbs H2
successfully on nearly every collision, and another form which is less
reactive by over 5 orders 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 packing arrangements
available to metal clusters of 19 atoms 25 . 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 (semiconductor clusters), the US Army Research Office
(gallium arsenide clusters), the 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 265

Edited by J.P. Fackler, Jr.
Plenum Press, New York, 1990
 0 /0
'c l c
";/
o,~/~,//o
o-c-r
c
x-,s- c-oc
I
0 I
0

One can think of 1 as an electron precise duster which satisfies the EAN rule at all atoms (Se-
at 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.

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.

4 5

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.

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

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 in-
plane 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 ~~
~~
Iau+ 3b;
cP Ia,- 3bo

lao ®X)
~ (:)®

~
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. N.V. Sidgwick, "The Electronic Theory of Valency", Cornell Univ. Press, Ithaca, N.Y.
(1927).
2. K. Wade, Adv. Inorg. Radiochem., 18:1 (1976); D.M.P. Minges, Acct. Chern. Res., 17:311
(1984).
3. J.W. Lauher, J. Am. Chern. Soc., 100:5305 (1978); J.W. Lauher, J. Am. Chern. Soc.,
101:2604 (1979).
4. M. Elian and R. Hoffmann, Inorg. Chern., 14:365 (1975).
5. R. Hoffmann, Angew. Chern. Int. Ed. Engl., 21:711 (1982).
6. G.L. Griewe and M.B. Hall, Inorg. Chern., 27:2250 (1988).
7. G.L. Griewe and M.B. Hall, Organomet., 7:1923 (1988).
8. W.N. Lipscomb, "Boron Hydrides," W.A. Benjamin, New York (1963); W.N. Lipscomb,
Science, 196:1047 (1977).
9. M. Elian, M.M.-L. Chen, D.M.P. Mingos, and R. Hoffmann, J. Am. Chern. Soc., 98:7240
(1976).
10. R.D. Adams, T.S.A. Hor, and I.T. Horvath, Inorg. Chern., 23:4733 (1984).
11. J. Demuynck, P. Mougenot, and M. Benard, J. Am. Chern. Soc., 109:2265 (1987).
12. 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 CI 2-

1/C I 17 I

I (I
/Re Re
c c
CI 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) calcula-
tions 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 275

Edited by J.P. fackler, Jr.
Plenum Press, New York, 1990
number of different metal dimer complexes. 3 However, there has been
considerable disparity among these calculations with regard to the calcu-
lated stability of these levels, their contributions to the total metal-
metal 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 under-
standing 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. Like-
wise, 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 tetra-
carboxylates 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 multi-
ple 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 physi-
cal (spectroscopic and magnetic) investigations. 2 One of the most inform-
ative 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 mole-
cular 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 theoret-
ical (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 progres-
sion 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 differ-
ent 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)

17 13 9

Figure 2. He(I) PE spectra of M2C02CCH3) complexes.

Ionization Energy (eV)

9 8

'
'
'
'
' \

' \

'
'
...... _:::~j-:·.:·:····················-.-·· ... '.\"'
~~ ~~
..... ~,· '......... . ........... . .," ...

Figure 3. First ionization band of Cr2C02CCH3)4.

278
 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 over-
all 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 per-
formed, 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 spec-
troscopic 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 charac-
ter. 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
2 B28cation 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

ionization energy (eV)

7. 5 1 6.5

279
 Table I. Compilation of stretching frequencies and bond lengths for
[Mo 2(0 2CCH 3 ) 4 ]n+ (n - 0, 1).

Electronic Electronic Formal VHo-~ ~-Ho

State Configuration Bond Order (em- ) (A)

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

Table adapted from reference llc. The calculated Mo-Mo bond lengths are
obtained by adding the calculated increments to the experimentally deter-
mined (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 ioniza-
tion. This has been explained in terms of the increased effective oxida-
tion 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
~
w
z 6.9
0 2 ~ STATE
~
z
N

NEUTRAL
I""'T'"'1
-0.2 0 0 2

Mo-Mo DISPLACEMENT (A)

Figure 5. Franck-Condon analysis of the 2B28 ionization.

280
Table II. Mo-Mo bond lengths in a number of dimers; the effect of 6
ionization.

Dimer Electronic Formal ~o-Mo (A) Ref.

Configuration Bond Order

Mo 2 (0 2CCH 3 ) 4 a21r4 62 4.0 2.093(1), 2.079(3) 15

[Mo2(S04)d4- 2.111(1) 18
Mo 2 (form) 4a 2.085(4) 19

[Mo 2 (S0 4)d J- a27f46 3.5 2.167(1) 18b. 20

[Mo 2(form) 4]+ a 2.122(3) 19

[Mo2(HP04)4]2- b a27f4 3.0 2.223(2) 21

• form- (~-tolyl)NCHN(~-tolyl). 19 b This complex possesses axial water

molecules. 21

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 ques-
tioned 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 vibration-

al 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 473 K

Figure 6. 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 effective mass of
the vibrating units does not reduce the calculated increase in metal-metal
bond length sufficiently to be consistent with the solid state structural
data that we have for the oxidized multiply bonded 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 combina-
tion 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 Formal M-M ~-M (A) ~-X (A) Ref.

Configuration Bond Order

[Moz(S04)414- (]27f452 4.0 2.111(3) 2.139(4)" lBb

[Moz(S04)413- (]27f45 3.5 2.167(1) 2.065(3) lBb
[Mo 2 (HP0 4) 412- (]27f4 3.0 2.223(2) 2.013(9) 21

[ Tc 2Cl 8] 3 - CJZ1f4525* 3.5 2.105(1), 2.364(2) 27

2.117(2) 28
[ Tc 2Cl 8 12- CJZ'If40Z 4.0 2 .15l(l)b 2.320(4) 29
2.133(3) 0 2.338(9)

Re 2Cl 4( PMe 2Ph) 4 (]27f4525*2 3.0 2.241(1) 2.387(l)d 30

[Re 2Cl 4(PMe 2Ph) 41+ (]27f4525* 3.5 2.218(1) 2.334(4) 30
[ Re 2Cl 4( PMe 2Ph) 4 12+ CJZ'If402 4.0 2.215(2) 2.291(7) 30

a 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.

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*) elec-
tron 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

ll
M===M
R3P
7
I
P R3 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 phos-
phines 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 ioniza-
tion 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 14.0 12D 10.0 8.0 6.0

Figure 7.

Table IV. The effect of changing halide ligands on the ionization

energies of the metal-based MO's of Mo 2X4 (PMe 3 ) 4 •

Dimer Peak Assignment Ionization Energy (eV)

6 6.44
w 7. 71
M-P(a) 8.40

6 6.53
w 7.73
M-P(a) 8.33

284
 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* 5.66
6 6.51

Re 2 Br 4 (PMe 3 ) 4 6* 5.72
6 6.56

Re 2 I 4 (PMe 3 ) 4 6* 5.80
6 6.56

One would expect from a charge potential point of view that sub-
stituting 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 conse-
quently 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 8.3 7.3 6.3 7.1 6.6 5.6

Figure 8. Low binding energy PE Figure 9. The 6 and 6* ionizations

spectra of Mo 2 X4 (PMe 3 ) 4 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 ener-
gies 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 contribu-
tions 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 ultra-
violet 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 2 8 294 290

Figure 11. 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 tradi-
tional 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 molyb-
denum, 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 7 6 5

Figure 12. 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 separa-
tion was ever observed. 36

One piece of supporting evidence for the assertion that the a ioniza-
tion 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 9.0 8.6 8.2

....Ul
c:
;:)
0
u

288
 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 agree-
ment 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 7.0 6.0
Figure 14. 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 ioniza-
tions of some M2 (0R) 6 dimers.

Dimer Peak Position [fwhm] (eV) Area(Band 1)/Area(Band 2)

Band 1 (7r) Band 2 (a)

Mo 2 (OCHMe 2 ) 6 7.33 [0.69] 7.99 [0.35] 2.4

Mo 2 ( OCH 2 CMe 3 ) 6 7.46 [0.69] 8.08 [0.36] 2.1
Mo 2 (0CMe 3 ) 6 6.79 [0.74] 7.80 [0.43] 2.0
W2 (0CMe 3 ) 6 6.27 [0.82] 7.79 [0.40] 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 dif-
ference 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 inter-
molecular 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' 6

Figure 15. 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 ioniza-
tions) 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 ioniza-
tions, 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
 10 8 7 e

Figure 16. Close-up of thin film PE

spectrum of Mo 2 (0 2CCH 3 ) 4 .

Triple bonded M 21Q 2~4 R' 2 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 9.0 8.0 7.0

Ionization Enern, eV Ionization Enern, eV

Figure 17. PE spectra of Figure 18. PE spectra of

W2 (0 2 CCH 3 ) 4 (np) 2 (top) and Mo 2 (0 2CCH 3 ) 4 (top) and
W2 (0 2 CCH 3 ) 4 (bottom). 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.

<d.zl d.z>(a) <d.zlp.> <d.zl s> <d.,.. I d.,..> C,.. > <d,.y I d,.y> <6 > <p.lp.> <sis>
---------------·---·--- ---
Mo 1 0.116 0.117 0.111 0.155 0.046 0.028 0.005
W'z 0.138 0.134 0.116 0.181 0.051 0.028 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 Fenske-
Hall 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 (Divi-

sion 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
I 7r \
9 I \
I \
I \

~1
I \
\
I \
I \
4p-' 4p

Figure 19. 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. postdoc-
toral 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 co-
workers. 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. (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.
2. (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.
3. (a) Norman, J.G., Jr.; Kolari, H.J. J. Chern. Soc., Chern. Commun.
1974, 303.

295
 (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.
4. (a) Atha, P.M.; Hillier, I.H. Hol. Phys. 1982, 45, 285.
(b) Cotton, F.A.; Shim, I. J. Phys. Chem. 1985, 89, 952.
5. (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.
6. (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 Spectro-
scopy" 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.
7. Koopmans, T. Physica (Amsterdam) 1934, 1, 104.
8. Bohm, M.G. J. Chem. Phys. 1983, 78, 7044.
9. Hubbard, J.L.; Lichtenberger, D.L J. Am. Chem. Soc. 1982, 104, 2132.
10. (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.
11. (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.
12. (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.

296
 (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
 (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.
33. (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.
34. (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.
35. (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.
36. Blevins, C.H. II, Ph.D. Dissertation, University of Arizona, 1984.
37. (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.
38. (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.
39. (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.
40. Kober, E.M.; Lichtenberger, D.L. J. Am. Chem. Soc. 1985, 107, 7199.
41. Kober, E.M.; Lichtenberger, D.L. unpublished work.
42. Ketkar, S.N.; Fink, M. J. Am. Chem. Soc. 1985, 107, 338.
43. Cotton, F.A.; DeBoer, B.G.; LaPrade, M.D.; Pipal, J .R.; Ucko, D.A.
Acta Cryst. B 1971, 27, 1644.
44. (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.
45. (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.
46. (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.
47. Hall, M.B.; Fenske, R.F. Inorg. Chem. 1972, 11, 768.
48. (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.

298
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 non-
conventional 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(p-
toluidine) 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 non-

Metal-Metal Bonds and Clusters in Chemistry and Catalysis 299

Edited by J. P. Fackler, Jr.
Plenum Press, New York, 1990
 Figure 1 Two views, the lower stereoscopic, of the metallomacrocycle
Rh2(C0)2Cl2(~-dpma)2, 2, as determined from X-ray
diffraction.

300
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 metal-
metal 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
rrI/If +
r/"-A·---r
I (\: +

C'J
I ··--..J CI-P.d~lh-CI
Rh" I
Rh
' (1)

I I
C l - ' \ . . - d - Cl"
'c
~As-../
0 cf
/1 p P_/As............_,.p
Cl

301
 +

Figure 2 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.

302
 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 six-
coordinate. 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
 2 lr Tl

Figure 5 A qualitative molecular orbital diagram for the bent Ir-Tl-

Ir unit in [Ir2Tl(C0)2Cl2(~-dpma)2]+ which utilizes the out-
of-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\
Absorption j \ Emission

:
: \
: .
:

.::. .
\
~
~
: : 1.4 microseconds
:i \: ,#'·.,
/ \
I \\ / \

i \ ./ \\'..
..,.··
--· \,
'~ ....__ ........• ...... __

300 500 700 900

nm

Figure 6 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.

306
 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

absorption 508 (32,000) absorption 370

emission 606 emission 444

Figure 7 Related trinuclear complexes with d8d8d8, d8dl0d8, d8s2d8,

and s2d8s2 configurations.

-~

-::r·
~--· -a2u

J:
-~

[
-•2u

~·h . it~ *~
*~
~M~ ;M~
;M":.
'; M '!
:,M-:.

o:.M:
Figure 8 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.

308
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 x-
ray structures. The National Science Foundation (CHE-8519557) has been
generous with financial support.

REFERENCES
1. A.L. Balch. Pure and Applied Chemistry, 60: 555 (1988).
2. A.L. Balch, L.A. Fossett, M.M. Olmstead, D.E. Oram, P.E. Reedy, Jr.
J. Am. Chern. Soc , 107: 5272 (1985).
3. A.L. Balch, M. Ghedini, D.E. Oram, P.E. Reedy, Jr. Inorg. Chern ,
26: 1223 (1987).
4. P.D. Brotherton, C.L. Raston, A.H. White, S.B. Wild. J. Chern. Soc.
Dalton Trans , 1799 (1976).
5. P.E. Reedy, Ph.D. Thesis, University of California, Davis, 1987.
6. D.A. Bailey, A.L. Balch, L.A. Fossett, M.M. Olmstead, P.E. Reedy,
Jr. Inorg. Chern., 26: 2413 (1987).
7. A.L. Balch. Ann. New York Acad. Sci., 313: 651 (1978).
8. A.L. Balch in Homogeneous Catalysis with Metal Phosphine Complexes.
L.H. Pignolet, ed. Plenum Press, New York, P.l67.
9. A.L. Balch, D.E. Oram, P.E. Reedy, Jr. Inorg. Chern., 26: 1836
(1987).
10. A.L. Balch, J.K. Nagle, M.M. Olmstead, P.E. Reedy, Jr. JAm· Chern,
.s.o.c........ 109: 4123, (1987) .
11. A.L. Balch, M.M. Olmstead, D.E. Oram, P.E. Reedy, Jr., S.H. Reimer.
J. Am· Chern. Soc., in press.
12. D.M. Roundhill, H.B. Gray, C.-M. Che. Accounts Chern. ReSearch, 22:
55 (1989).
13. Balch, A.L., L.A. Fossett, J.K. Nagle, M.M. Olmstead. J. Am· Chern.
~. 110: 6732 (1988).
14. K.R. Mann, N.S. Lewis, R.M. Williams, H.B. Gray, J.G .. Gordon, II.
Inorg. Chern., 17: 828 (1978).
15. A.L. Balch, L.A. Fossett, M.M. Olmstead, P.E. Reedy, Jr.
Organometallics,_5: 1929 (1986).
16. A.L. Balch, J.K. Nagle, D.E. Oram, P.E. Reedy, Jr. J. Am. Chern.
~. 110: 454 (1988).
17. 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 geome-
tries 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 lig-
ands 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 dis-
cussed.

312
 L(COh

/1 w-H
(CO)~~ /!M(C0) 3
C-M-C
I

0 (CO)z 0

M = Os,L=C;H5; la (0.75) lb (0.25)

M = Os, L = C5Me5; 2a (0.32) 2b (0.68)
M = Ru, L = C5 Hs; 3a (0.17) 3b (0.28)
M = Ru, L = C5Me5; 4b (0.16)

M = Ru, L = C 5H 5; 3c (0.55)
M =Ru, L = C5Me5; 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 S~
(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:

Snn 4H,, 0 4Hr,o 4S,,29B

KJ/mol KJ/mol J/K·mol

Sn4 714 491 288

Sns 908 598 358
Sn6 1126 681 432
Sn7 1387 721 518

314
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 acetylace-
tone 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 1,5-
hexadiene 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) Substrate-
mediated 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. T. Mebrahtu, J. F. Rodriguez, B. G. Bravo and M. P. Soriaga. J.

Electroanal. Chem. 219 (1987) 327.
2. B. G. Bravo, T. Mebrahtu, J. F. Rodriguez and M. P. Soriaga. J.
Electroanal. Chem. 220 (1987) 281.
3. J. F. Rodriguez, B. G. Bravo, T. Mebrahtu and M. P. Soriaga. Inorg.
Chem. 26 (1987) 2760.
4. B. G. Bravo, T. Mebrahtu and M. P. Soriaga. Langmuir. 3 (1987) 595.
5. J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J. Electroanal.
Chem. 233 (1987) 283.

316
 6. J. F. Rodriguez and M. P. Soriaga. J. Electrochem. Soc. 135 (1988)
616.
7. B. G. Bravo, J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J.
Phys. Chem. 91 (1987) 5660.
8. B. G. Bravo, S. L. Michelhaugh and M. P. Soriaga. J. Electroanal.
Chem. 241 (1988) 199.
9. J. F. Rodriguez, J. E. Harris, M. E. Bothwell, T. Mebrahtu and M. P.
Soriaga. Inorg. Chim. Acta. 148 (1988) 123.
10. M. P. Soriaga. J. Electroanal. Chem. 240 (1988) 309.
11. J. F. Rodriguez, M. E. Bothwell, J. E. Harris and M. P. Soriaga. J.
Phys. Chem. 92 (1988) 2702.
12. T. Mebrahtu, G. M. Berry and M. P. Soriaga. J. Electroanal. Chem.
239 (1988) 375.
13. M. P. Soriaga. In M. P. Soriaga, Editor, "Electrochemical Surface
Science: Molecular Phenomena at Electrode Surfaces." ACS Books:
Washington, DC. Chapter 1 (1988).
14. T. Mebrahtu, G. M. Berry, B. G. Bravo, S. L. Michelhaugh and M. P.
Soriaga. Langmuir. 4 (1988) 1147.
15. B. G. Bravo, S. L. Michelhaugh, T. Mebrahtu and M. P. Soriaga.
Electrochim. Acta. 33 (1988) 1507.
16. Mebrahtu, G. M. Berry and M. P. Soriaga. J. Electroanal. Chem. 247
(1988) 241.
17. 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).
18. M. E. Bothwell, J. F. Rodriguez and M. P. Soriaga. J. Electroanal.
Chem. 252 (1988) 453.
19. M. E. Bothwell and M. P. Soriaga. J. Electroanal. Chem. 260 (1989)
193.
20. J. E. Harris, M. E. Bothwell, J. F. Rodriguez and M. P. Soriaga. J.
Phys. Chem. 93 (1989) 2610.
21. J. E. Harris and M. P. Soriaga. Electrochimica Acta. In press
(1989).
22. 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.
23. J. F. Rodriguez, T. Mebrahtu and M. P. Soriaga. J. Electroanal.
Chem. In press (1989).
24. B. G. Bravo, S. L. Michelhaugh and M. P. Soriaga. Langmuir. In
press (1989).
25. 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-methyl-

imidazolium 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 condensa-
tion 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 interme-
diates 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 reac-
tion 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 tri-
nuclear 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

parameters of the cluster cores, and the C n.m.r. chemical shift of the
r~kylidyne carbon atom, when the cluster electron count is changed. The
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-Fock-

Roothaan 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 racemic-
Rh2Cl2(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 hydro-
formylation reaction does not suffer from the extensive product, reactant
and catalyst decomposition reactions typically seen for Rh(I)/PR 3 cata-
lyzed 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 bond-
making 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 elec-
tronic 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

Ry~(C0)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
L, so that the data fit well to eq (1). DHta in the literature for
reactions of Co 4 (C0) 11 L7 and Ir 4 (C0) 12 _n~ (L - 1-3) also seem to fit

(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. A. J. Poe, Chapter 4 in "Metal Clusters", Ed. M. Moskovits, Wiley,

New York (1986).
2. A. J. Poe, Int. J. Chem. Kin., 20:467 (1988).
3. N. M. J. Brodie and A. J. Poe, Inorg. Chem., 27:3156 (1988).
4. 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.
5. J. R. Kennedy, F. Basolo, and W. C. Trogler, Inorg. Chim. Acta,
146:75 (1988).
6. G. Bodner, M. May, and L. McKinney, Inorg. Chem., 19:1951 (1980).
7. D. J. Darensbourg, D. J. Zalewski, and T. Delord, Organometallics,
3:1210 (1984).
8. 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. M. D. Morse, Chem. Rev. 86 (1986) 1049.

2. K. A. Gingerich, Faraday Disc. Chem. Soc., 14 (1980) 109.
3. 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 molec-
ular 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. R. F. W. Bader Ace. Chem. Res., 18, 9 (1985).

334
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 lig-

ands. 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. University of South Carolina 1

Adras, M. T. Cleveland State University 320
Balch, A.L. University of California-Davis 299
Batten, T. Wright State University 322
Benefield, R.E. University of Kent 141
Berry, G.M. Texas A&M University 316
Blau, R.J. University of Texas-Arlington 315
Bothwell, M.E. Texas A&M University 316
Batt, A. University of Cambridge 141
Boyd, D.C. University of Minnesota 215
Braga, D. University of Bologna 141
Bravo, B.G. Texas A&M University 316
Brodie, N.M.J. University of Toronto 330
Bursten, B.E. Ohio State University 19
Cali, G.J. Texas A&M University 316
Carlin, R.T. State University of New York-Buffalo 318
Chen, L. University of Toronto 330
Chi, Y. National Tsing Hua University 312
Chisholm, M.H. Indiana University 55
Cotton, F.A. Texas A&M University 1
Cowley, A.H. University of Texas-Austin 195
Darensbourg, D.J. Texas A&M University 41
Day, V.W. Crystalytics Company 161
Duraj, S.A. Cleveland State University 320
Evans, J .F. University of Minnesota 215
Fackler, Jr., J.P. Texas A&M University 329
Gagne, M.R. Northwestern University ll3
Gates, B.C. University of Delaware 127
Gingerich, K.A. Texas A&M University 314
Gingerich, K.A. Texas A&M University 319
Gingerich, K.A. Texas A&M University 333
Gladfelter, W.L. University of Minnesota 215
Goddard Ill, W.A. California Institute of Technology 335
Goodman, D.W. Texas A&M University 205
Haasch, R.T. University of Minnesota 215
Hall, M.B. Texas A&M University 265
Hall, M.B. Texas A&M University 324
Hall, M.B. Texas A&M University 325
Hall, M.B. Texas A&M University 327
Hall, M.B. Texas A&M University 334
Hanson, B.E. Virginia Polytechnic Institute 231
Harris, J.E. Texas A&M University 316
Ho, K.-L. University of Minnesota 215
Ho, s.-1. University of Texas-Arlington 315
Huntley, D.R. Oak Ridge National Laboratory 321
Hwang, J. -W. University of Minnesota 215

337
Jeffery, J.C. Texas A&M University 325
Jennings, J.G. Abilene Christian University 323
Jensen, K.F. University of Minnesota 215
Johnson, B.F.G. University of Cambridge 141
Johnston, R.L. University of Arizona 275
Katovic, V. Wright State University 322
Kingcade, Jr., J.E. Texas A&M University 314
Kingcade, Jr., J.E. Texas A&M University 333
Klemperer, W.G. University of Illinois 161
Lagow, R.J. University of Texas-Austin 171
Laneman, S.A. Louisiana State University 328
Lictenberger, D.L. University of Arizona 275
Lockledge, S.P. University of Illinois 161
MacDougall, P.J. Texas A&M University 334
Main, D.J. University of Illinois 161
Marks, T.J. Northwestern University 113
Marseglia, E.A. Cavendish Laboratory 141
Martuch, R.A. Cleveland State University 320
May, M. Wright State University 322
McCarley, R.E. Iowa Sate University 91
McNeal, C.J. Texas A&M University 329
Mebrahtu, T. Texas A&M University 316
Michelhaugh, S.L. Texas A&M University 316
Miller, A. E. California Institute of Technology 335
Nolan, S.P. Northwestern University 113
Osteryoung, R.A. State University of New York-Buffalo 318
Peng, S.-M. National Tsing Hua University 312
Pignolet, L.H. University of Minnesota 332
Poe, A.J. University of Toronto 330
Powell, G.L. Abilene Christian University 323
Rodger, A. University of Cambridge 141
Rodriguez, J.F. Texas A&M University 316
Rosenberg, F.S. University of Illinois 161
Sargent, A.L. Texas A&M University 324
Sayers, W. Wright State University 322
Schmude, Jr., R.W. Blinn College 314
Schulze, R.K. University of Minnesota 215
Seyam, A.M. Northwestern University 113
Sherwood, P. Texas A&M University 325
Shim, I. The Technical University of Denmark 326
Shim, I. The Technical University of Denmark 333
Simpson, II, C.Q. Texas A&M University 327
Sinfelt, J.H. Exxon 103
Smalley, R.E. Rice University 249
Soriaga, M.P. Texas A&M University 316
SriHari, S. Cleveland State University 320
Stanley, G.G. Louisiana State University 328
Stern, D. Northwestern University 113
Stone, F.G.A. University of Bristol 325
Tsay, R.-J. University of Texas-Arlington 315
Walton, R.A. Purdue University 7
Wang, R. -c. University of Illinois 161
Wang, S.-L. National Tsing Hua University 312
Wasfi, S.H. Delaware State College 311
Winpenny, R.E.P. Texas A&M University 329
Yaghi, O.M. University of Illinois 161

338
INDEX
actinide diatomic molecules, 35
alkylidyne ligand, 325
aluminum nitride, 215
aluminum nitride, 218
ambient-temperature chloro-
aluminate 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 metal-
metal bonds, 335
carbonylation, 44
charge density in binuclear
transition-metal com-
plexes, 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) dianion-
organometallicjcovrdination
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